JP3582655B2 - Component mounting sequence optimizing method, device and component mounting machine - Google Patents

Description

  The present invention relates to a method and the like for determining an optimal order when electronic components are mounted on a substrate such as a printed wiring board by a component mounter, and in particular, a component including a work head that sucks a plurality of components and mounts the components on the substrate. The present invention relates to optimization of a component mounting order for a mounting machine.

In a component mounter that mounts electronic components on a substrate such as a printed wiring board, the mounting order of target components is optimized in order to achieve a shorter tact (mounting time). Specifically, as one of the optimizations of the mounting order, for example, it is necessary to optimize the arrangement order of each component cassette in the component cassette group provided in the component mounting machine.
As a conventional technique for that purpose, for example, there is a component mounting order optimization method disclosed in Patent Document 1. In this method, (1) the component cassette group is divided into groups according to the mounting speed applied to the component, and the sum of the mounting points when two component cassettes in the same group are combined is equal. (2) The cassette groups are arranged in the order of the mounting speed by appropriately combining those having a large number of mounting points and those having a small number of mounting points on the same substrate. Determines the arrangement order of the cassettes by arranging them in pairs, and (3) thereafter, performs optimization processing using only the mounting order of the components as a parameter.

As a result, complicated optimization based on the two parameters of the cassette arrangement order and the component mounting order is avoided, and optimization in a short time with a single parameter is realized.
JP 05-104364 A

However, such a conventional optimization method is based on the premise that the working head sucks only one component from the component cassette and mounts it on the board, and thus a plurality (for example, 10) of components is used. There is a problem that it cannot be applied to a component mounter provided with a high-performance work head (multi-mounting head) that sucks and mounts on a substrate.
In particular, with the recent rapid increase in demand for electronic devices such as mobile phones and notebook computers, a component mounter equipped with a highly productive multi-mounting head that sucks and mounts a plurality of components onto a substrate has been developed. Therefore, a new method of optimizing a component mounting order corresponding to such a high-performance component mounter is desired.
Therefore, the present invention is optimized by a component mounting order optimization method corresponding to a high-productivity component mounting machine, that is, a component mounting order optimization method, an apparatus and a method thereof that enable higher productivity. It is an object of the present invention to provide a component mounter or the like that mounts components in the following order.

  More specifically, for example, it is an object of the present invention to provide a method of optimizing a component mounting order corresponding to a component mounter having a multi-mounting head that sucks a plurality of components and mounts them on a substrate.

In order to achieve the above object, the present invention provides a method of picking up a component group with a mounting head capable of picking up to n (≧ 2) components from a row of component cassettes storing the components, and using an XY robot. A component mounter that moves the mounting head and mounts the component group on a board, and picks up and moves a group of components by the mounting head, and then mounts the component group on the board. When a group is set as a task, the mounting head completes mounting all the components belonging to one task on the board, and among the unmounted tasks, the task that moves the minimum distance to pick up the component. Is implemented next.
In other words, the component mounting order optimizing method according to the present invention employs a sequence of component cassettes storing component tapes each having a group of components of the same type as one component tape, and picking up components from the sequence and mounting them on a board. A method of optimizing the mounting order of components by executing a computer, targeting a component mounter having a mounting head that performs a series of operations of picking up, moving, and mounting components by the mounting head. In a task group consisting of a plurality of tasks in which a task is a component mounted by one series of operations in the above, the position of the mounting head immediately after all the components belonging to a certain task have been mounted on the board, If the position of the mounting head when picking up a component belonging to a task is set as the final mounting point and the position of the mounting head Any one of the first tasks belonging to the loop is selected, the final mounting point of the first task is specified, and among all the tasks except the first task, the task pickup position is the final mounting of the first task. Identifying the second task closest to the point, identifying the final implementation point of the second task, identifying the third task, and so on, until the first task is identified again. Iteratively, when the first task is specified again, the shortest cyclic partial path specifying step of specifying a set of the tasks as the shortest cyclic partial path; and the shortest cyclic partial path specifying step until all the tasks belonging to the task group are specified. Includes an iterative step of repeating the traveling partial path specifying step, and an arranging step of arranging the shortest traveling partial paths specified by the repetition in the repeating step And wherein the door.

Here, the component mounting order optimizing method further has a shortest distance that the mounting head has to move to start the next shortest cyclic partial path after completing the mounting of all components belonging to a certain shortest cyclic partial path. Thus, the method may include a step of determining the first task in each of the shortest traveling partial paths and the order of the shortest traveling partial paths, and changing the order of the tasks so that the determined tasks are arranged.
In order to achieve the above object, a component mounting order optimizing method according to the present invention is based on a sequence of component cassettes containing component tapes each having a group of components of the same type as one component tape. A method for optimizing the order of mounting components by executing a computer on a component mounter having a mounting head that sucks components and mounts the components on a substrate. A task group generating step of generating a task group including a sequence of a plurality of tasks each having a component mounted by one series of operations in the series of mounting operations, and a task for each of the generated task groups; Order the tasks in the task group so that the time required to mount all the components that make up the group is minimized. And a task replacement step for optimizing the mounting order of components corresponding to the obtained task order, and the position of the mounting head immediately after all the components belonging to a certain task have been mounted on the board. Is set as the final mounting point of the task, and the position of the mounting head when picking up a component belonging to a certain task is set as the picking position of the task. One task is selected, the final mounting point of the first task is specified, and among all the tasks except the first task, the task at which the task adsorption position is closest to the final mounting point of the first task is selected. Identifying two tasks, and so on, identifying the final implementation point of the second task, and identifying the third task, until the first task is identified again. When the first task is specified again, a shortest-path partial path specifying step of specifying a group of the tasks as the shortest-path partial path; and the shortest-path partial path until all the tasks belonging to the task group are specified. The method includes a repetition step of repeating the specifying step, and an arrangement step of arranging the shortest cyclic partial paths specified by the repetition in the repetition step.

  Furthermore, the present invention can be realized as an optimizing device having the steps of the above-described optimization method as functional means, or as a component mounter that mounts components in a mounting order optimized by the steps of the above-described optimization method. The present invention can also be realized as a program that causes a computer to execute the steps of the above-described optimization method, and a computer-readable recording medium that stores such a program.

According to the present invention, the moving distance when the mounting head finishes mounting components in one task and returns to mount the next task is reduced, so that the total mounting time in the entire task group is reduced.
That is, from the arrangement of the component cassettes storing the components, the component group is sucked by the mounting head capable of sucking a maximum of n (≧ 2) components, and the mounting head is moved by the XY robot and mounted on the board. When the mounting head finishes mounting all the components belonging to one task on the board, the mounting head next mounts the task that moves the minimum distance to pick up the component. The optimization of the component mounting order corresponding to a component mounter having high productivity such as a component mounter having a head is realized.

  Embodiments of the present invention will be described in detail with reference to the drawings and the drawings. The meanings of the main technical terms used here are described in the text and in “5. Explanation of Terms”.

1 Component Mounting System 1.1 Configuration of Component Mounting Machine 1.2 Restrictions in Component Mounting Machine 1.2.1 Multi-Mounting Head 1.2.2 Component Recognition Camera 1.2.3 Component Supply Unit 1.2.4 Components Cassette 1.2.5 Other restrictions 1.3 Optimization device 1.3.1 Hardware configuration of optimization device 1.3.2 Software configuration of optimization device 2 Operation of optimization device (Overview)
2.1 Creation of parts group 2.2 Line balance processing 2.3 Optimization for small parts 2.4 Task group generation method 2.5 Pruning method 2.6 Random selection method ("greedy method")
2.7 Intersection elimination method 2.8 Return optimization method 2.9 Optimization for general-purpose parts 3 Operation of optimization device (detailed version)
3.1 "Mowing method"
3.1.1 Overview of "task group generation method" 3.1.2 Issues of "task group generation method" 3.1.3 "Pruning method"
3.1.4 Optimization of small parts using the "reaping method" 3.1.5 Related individual processing 3.2 "Intersection elimination method"
3.2.1 Overview of the Greedy Method 3.2.2 Issues of the Greedy Method 3.2.3 Crossing Resolution Method
3.2.4 Related individual processing 3.3 "Return optimization method"
3.3.1 Examination of component mounting operation 3.3.2 Necessity of optimization of “return” process 33.3 “Return optimization method”
3.3.4 Related individual processing 3.4 Array fixing processing 3.4.1 Overview 3.4.2 Related individual processing 3.5 Support for LL size board 3.5.1. Overview 3.5.2 Replacement of component tapes on Z axis 3.5.3 Change of suction method 3.5.4 Related individual processing 3.6 Support for XL size board 3.6.1 Overview 3.6.2 Related individual processing 3.7 Load balance processing 3.7.1 Overview 3.7.2 Level of balance adjustment method 3.7.3 Related individual processing 3.8 Line balance processing 3.8.1 Overview 3.8.2 Balance adjustment Method level 3.8.3 Related individual processing 3.9. Details of individual processing by the optimizer 3.9.1 "Pruning method"
3.9.2 Cassette division by parallelogram 3.9.3 Cassette division by rectangle 3.9.4 Core processing method with given number of cassettes 3.9.5 Task generation processing of small parts 3.3.9. 6 "Intersection Resolution Method"
3.9.7 "Return optimization method"
3.9.8 Overall flow (start from histogram)
3.9.9 Arrangement relationship between fixed parts and “mountain” in cassette block 3.9.10 Fixing array: Judgment of availability of fixed destination 3.9.11 Fixing array of double cassette 3.9.12 LL restriction : Change of adsorption method (1)
3.9.13 LL restriction: Change of adsorption method (2)
3.9.14 LL constraint: Replacement of component tape on Z axis (1)
3.9.15 LL constraint: Replacement of component tape on Z axis (2)
3.9.16 Support for XL size substrates (XL restrictions)
3.9.17 Load level balance adjustment processing ("mountain" unit)
3.9.18 Load level balance adjustment processing (part tape)
3.9.19 Processing to move mountain from front sub-equipment to rear sub-equipment 3.9.20 Processing to move component tape from front sub-equipment to rear sub-equipment 3.9.21 Processing from front sub-equipment to rear sub-equipment Processing to move the mounting point 3.9.22 Swap processing in line balance processing 3.9.23 "Mowing method" for double cassette
3.9.24 Algorithm for nozzle replacement 3.10 Screen display example 3.10.1 Main screen 3.10.2 Open screen 3.10.3 Optimization detailed information screen 3.10.4 Cassette number setting screen 3. 10.5 Parts division number setting screen 3.10.6 Nozzle number setting screen 3.10.7 Nozzle station selection screen 3.10.8 Option setting screen 3.10.9 Z axis information screen 3.10.10 Nozzle station Information screen 4 Operation of the optimization device (application)
4.1 Optimization of small parts 4.1.1 Optimization of Z array without dividing parts 4.1.2 Optimization in distribution processing to left and right blocks 4.1.3 Estimation of the number of double cassettes used 4.1 .4 Fixed pair of double cassettes 4.1.5 Optimization algorithm considering NG head 4.2 Simultaneous optimization of multiple NC data 4.3 Optimization of general-purpose parts (introduction of rule base)
4.3.1 Preemption method 4.3.2 Task division 4.3.3 Task fusion 4.3.4 Task replacement 4.4 Optimization considering nozzle restrictions 4.4.1 Change nozzle arrangement on nozzle station 4.4.2 Optimization of small parts when less than 10 nozzles are used 5 Explanation of terms The explanation of each item shown in the above table of contents is as follows.

1 Component Mounting System FIG. 1 is an external view showing the entire configuration of a component mounting system 10 according to the present invention. The component mounting system 10 includes a plurality of component mounters 100 and 200 constituting a production line for mounting electronic components while sending the circuit board 20 from upstream to downstream, and various databases upon starting production. And an optimizing device 300 for optimizing a mounting order of necessary electronic components based on the obtained NC data, downloading the obtained NC data to the component mounters 100 and 200, and setting and controlling the NC data.

  The component mounter 100 includes two sub-equipments (a front sub-equipment 110 and a rear sub-equipment 120) that perform component mounting simultaneously and independently, or cooperatively (or alternately) with each other. Each sub-equipment 110 (120) is an orthogonal robot type mounting stage, and has two component supply units 115a and 115b each having an array of a maximum of 48 component cassettes 114 for storing component tapes, and a maximum of 10 components from the component cassettes 114. A multi-mounting head 112 (10-nozzle head) having ten suction nozzles (hereinafter, also simply referred to as "nozzles") capable of picking up and mounting individual components on the substrate 20, and moving the multi-mounting head 112 The XY robot 113 includes a component recognition camera 116 for two-dimensionally or three-dimensionally inspecting a suction state of the component sucked by the multi-mount head 112, a tray supply unit 117 for supplying tray components, and the like.

Note that a “component tape” is actually a plurality of components of the same component type arranged on a tape (carrier tape), and is supplied in a state wound on a reel (supply reel) or the like. It is mainly used to supply relatively small-sized components called chip components to a component mounter. However, in the optimization process, the “component tape” is data that specifies a set of components belonging to the same component type (a plurality of these components arranged on a virtual tape). In some cases, a component group (one component tape) belonging to one component type is divided into a plurality of component tapes by a process called “part division”.
Parts supplied by the component tape are called taping parts.
The component mounter 100 is, specifically, a mounter having both functions of a component mounter called a high-speed mounter and a component mounter called a multifunctional mounter. A high-speed placement machine is a facility that features high productivity, mainly for mounting electronic components of □ 10 mm or less at a speed of about 0.1 second per point. Equipment for mounting electronic components, odd-shaped components such as switches and connectors, and IC components such as QFP and BGA.

  In other words, the component mounter 100 is designed so that almost all types of electronic components (components to be mounted, from a chip resistor of 0.6 mm × 0.3 mm to a connector of 200 mm) can be mounted. By arranging the required number of component mounters 100, a production line can be configured.

1.1 Configuration of Component Mounter FIG. 2 is a plan view showing a main configuration of the component mounter 100 to be subjected to component mounting order optimization according to the present invention.
The shuttle conveyor 118 is a moving table (collection conveyor) for placing components taken out from the tray supply unit 117 and transporting the components to a predetermined position where the components can be suctioned by the multi-mounting head 112. The nozzle station 119 is a table on which replacement nozzles for supporting various types of component types are placed.

The two component supply units 115a and 115b constituting each sub-equipment 110 (or 120) are disposed on the left and right sides of the component recognition camera 116, respectively. Therefore, the multi-mounting head 112 that has picked up the component in the component supply unit 115a or 115b moves to the mounting point of the board 20 after passing through the component recognition camera 116, and performs an operation of sequentially mounting all the picked up components. repeat.
Here, one operation (suction, movement, mounting) in a series of operations of suction, movement, and mounting of components by the multi-mounting head 112, or a component group mounted by such one operation) Is called a “task”. For example, according to the 10-nozzle head 112, the maximum number of components mounted by one task is 10. Note that the “sucking” here includes all sucking operations from the start of sucking the component to the movement of the head. For example, ten sucking operations (up and down operations of the multi-mount head 112) are performed by one sucking operation. This includes not only the case of picking up a component but also the case of picking up 10 components by a plurality of suction operations.

FIG. 3 is a schematic diagram showing a positional relationship between the multi-mounting head 112 and the component cassette 114. The multi-mounting head 112 is a work head called a “gang pick-up system”, and can mount up to ten suction nozzles 112a to 112b. Can be adsorbed at the same time (with one vertical movement).
The component cassette 114 called "single cassette" is loaded with only one component tape, and the component cassette 114 called "double cassette" is loaded with two component tapes (provided that the feed pitch (2 mm or 4 mm) is the same). (Limited to component tape). The position of the component cassette 114 (or component tape) in the component supply units 115a and 115b is referred to as “value on the Z axis” or “position on the Z axis”, and the leftmost end of the component supply unit 115a is “1”. Is used. Therefore, determining the mounting order for taping components is equivalent to determining the arrangement (position on the Z axis) of component types (or component tapes or component cassettes 114 containing the component tapes).

FIG. 4A shows a specific configuration example of the component supply units 115a, b and 215a, b of the sub-equipment 110, 120, respectively, and FIG. 4B shows the number of mounted various component cassettes 114 in the configuration. And a table showing positions on the Z axis.
As shown in FIG. 4A, each of the component supply units 115a, 115b, 215a, and 215b can mount a maximum of 48 component tapes (the positions thereof are Z1 to Z48, Z49 to Z49, respectively). Z96, Z97-Z144, Z145-Z192). Specifically, as shown in FIG. 4B, by using a double cassette containing two component tapes each having a tape width of 8 mm, a maximum of 48 types can be provided in each component supply unit (A block to D block). Parts can be mounted. As the component (component cassette) has a larger tape width, the number of cassettes that can be mounted on one block decreases.

The component supply units 115a and 215a (blocks A and C) on the left side of each sub-equipment are referred to as "left blocks", and the component supply units 115b and 215b on the right side of each sub-equipment (blocks B and D). Is also called a “right block”.
FIGS. 5A and 5B are a diagram and a table showing an example of the position (Z-axis) of the component supply unit to which the ten nozzle heads can suck. Note that H1 to 10 in the drawing indicate (positions) of the nozzles mounted on the 10-nozzle head.
Here, since the interval between the nozzles of the 10-nozzle head corresponds to the width (21.5 mm) of one double cassette, the Z number of the component sucked by one vertical movement is every other (only odd numbers). Or even only). Further, due to the movement restriction in the Z-axis direction of the 10-nozzle head, as shown in FIG. 5B, a nozzle that cannot be sucked to a component (Z-axis) forming one end of each component supply unit. (“-” In the figure) exists.

Next, a detailed structure of the component cassette 114 will be described with reference to FIGS.
A plurality of chip-type electronic components 423a to 423d as shown in FIGS. 6A, 6B, 6C, and 6D are continuously formed at predetermined intervals on a carrier tape 424 shown in FIG. It is stored in the concave portion 424a, the upper surface thereof is covered with a cover tape 425, packaged, and supplied to the user in a taping form (component tape) in which a predetermined amount is wound around a supply reel 426.
Such a taping electronic component 423d is used by being mounted on a component cassette 114 as shown in FIG. 8, and a supply reel 426 is rotatably mounted on a reel side plate 428 coupled to a main body frame 427 in FIG. Installed. The carrier tape 424 drawn from the supply reel 426 is guided by a feed roller 429, and is provided in the electronic component automatic mounting device (not shown) in which the electronic component supply device is mounted, in conjunction with the operation thereof. The feed lever 430 of the electronic component supply device is moved in the direction of arrow Y1 in the figure by a feed lever (also not shown), and the ratchet 432 is rotated by a fixed angle via a link 431 attached to the feed lever 430. Then, the feed roller 429 interlocked with the ratchet 432 is moved by a constant pitch (for example, a feed pitch of 2 mm or 4 mm).

Further, the carrier tape 424 is peeled off the cover tape 425 at the cover tape peeling portion 433 in front of the feed roller 429 (on the supply reel 426 side), and the peeled cover tape 425 is wound around the cover tape take-up reel 434. The carrier tape 424 from which the cover tape 425 has been peeled off is conveyed to the electronic component take-out unit 435, and from the electronic component take-out unit 435 that opens in conjunction with the ratchet 432 at the same time that the feed roller 429 carries the carrier tape 424. The chip-type electronic component 423d stored in the storage recess 424a is suctioned and taken out by a vacuum suction head (not shown). Thereafter, the feed lever 430 is released from the pressing force by the feed lever, and returns to the original position in the Y2 direction by the urging force of the tension spring 436, that is, returns to the original position.
When the above series of operations is repeated, the used carrier tape 424 is discharged to the outside of the electronic component supply device, cut into small pieces by a cutter (not shown) linked with the operation of the electronic component automatic supply device, and discarded. It is configured to be.

In the case where the component cassette 114 is of a double cassette type that stores two carrier tapes 424, the component cassette 114 can supply the two stored carrier tapes 424 only at the same feed pitch. I do.
The operational characteristics of the component mounter 100 are summarized as follows.
(1) Nozzle Replacement When the nozzle required for the next mounting operation is not in the multi-mounting head 112, the multi-mounting head 112 moves to the nozzle station 119 and performs nozzle replacement. The types of nozzles include, for example, types S, M, and L, depending on the size of the component that can be sucked.
(2) Component Suction The multi-mounting head 112 moves to the component supply units 115a and 115b and sucks electronic components. When ten components cannot be simultaneously picked up at a time, a maximum of ten components can be picked up by performing suction up and down operations a plurality of times while moving the suction position.
(3) Recognition Scan The multi-mounting head 112 moves at a constant speed on the component recognition camera 116, captures images of all electronic components sucked by the multi-mounting head 112, and accurately detects the position where the components are picked up.
(4) Component mounting Electronic components are sequentially mounted on the board 20.

  By repeating the above operations (1) to (4), all the electronic components are mounted on the board 20. The operations (2) to (4) are basic operations in mounting components by the component mounter 100, and correspond to “tasks”. In other words, a maximum of ten electronic components can be mounted on the board by one task.

1.2 Constraints in Component Mounter The purpose of optimizing the component mounting order is to maximize the number of boards produced by the component mounter 100 per unit time. Therefore, a preferable optimization method (optimization algorithm) is to select ten electronic components that can be efficiently mounted on the board, as can be seen from the above-described functional and operational characteristics of the component mounter 100. The algorithm is such that they are simultaneously sucked from the component supply unit and are sequentially mounted in the shortest path. The component mounting order determined by such an optimization algorithm can ideally improve productivity by about 10 times compared to the case of component mounting using only one nozzle.

However, any component mounting machine has a limiting factor in determining the component mounting order in terms of mechanism, cost, operation, and the like. Therefore, in practice, the optimization of the component mounting order is to maximize the number of boards produced per unit time as much as possible while observing various restrictions.
Hereinafter, main restrictions in the component mounter 100 will be listed. Note that the details of the constraints are also described in the sections describing individual optimization algorithms.

1.2.1 Multi-Mounting Head The multi-mounting head 112 has ten mounting heads that perform suction and mounting operations independently arranged in a line, and a maximum of ten suction nozzles can be attached and detached. With a series of the suction nozzles, a maximum of 10 parts can be simultaneously suctioned by one suction up and down operation.

It should be noted that when referring to the individual working heads (working heads for picking up one component) constituting the multi-mounting head, they are simply referred to as “mounting heads (or“ heads ”)”.
Due to the structure in which the ten mounting heads constituting the multi-mounting head 112 are arranged in a straight line, there are restrictions on the movable range of the multi-mounting head 112 during component suction and component mounting. Specifically, as shown in FIG. 5B, when the electronic components are sucked at both ends of the component supply unit (near the left end of the left block 115a and near the right end of the right block 115b), the mounting head that can be accessed is not. Limited.
Also, when the electronic component is mounted on the substrate, the movable range of the multi-mounting head 112 is limited. This is a constraint that occurs when components are mounted on a board called “LL size board” or “XL size board” which is larger in size in the vertical or horizontal direction than usual.

1.2.2 Component Recognition Camera The component mounter 100 includes, as the component recognition camera 116, a 2D camera that captures a two-dimensional image and a 3D camera that can also detect height information. The 2D camera includes a 2DS camera and a 2DL camera depending on the size of a viewable field of view. The 2DS camera has a small field of view but is capable of high-speed imaging, and the 2DS camera is characterized by a large field of view up to 60 × 220 mm. The 3D camera is used to three-dimensionally inspect whether all the leads of the IC component are bent.

  The recognition scan speed at the time of imaging an electronic component differs depending on the camera. If a part using a 2DS camera and a part using a 3D camera are in the same task, the recognition scan needs to be performed twice at each speed.

1.2.3 Component Supply Unit There are two types of electronic component packages: a method called taping for storing electronic components in a tape shape, and a method called tray for storing electronic components in a plate that is partitioned according to the size of the components. There is.
The supply of the components by taping is performed by the component supply units 115a and 115b, and the supply by the tray is performed by the tray supply unit 117.
The taping of electronic components is standardized, and there is a taping standard from 8 mm width to 72 mm depending on the size of the component. By setting such a tape-shaped component (component tape) in a component cassette (tape feeder unit) corresponding to the tape width, it is possible to take out the electronic component continuously in a stable state.

The component supply unit for setting the component cassette is designed so that component tapes having a width of up to 12 mm can be mounted at a 21.5 mm pitch without gaps. When the tape width is 16 mm or more, the tapes are set with a necessary gap depending on the tape width. In order to simultaneously pick up a plurality of electronic components (by one vertical movement), the pitches of the mounting head and the component cassette need only match. For parts with a tape width of up to 12 mm, simultaneous suction of 10 points is possible.
Each of the two component supply units (the left block 115a and the right block 115b) constituting the component supply unit can mount up to 48 component tapes having a width of up to 12 mm.

1.2.4 Component Cassettes Component cassettes include a single cassette that stores only one component tape and a double cassette that can store up to two component tapes. Two component tapes housed in a double cassette are limited to component tapes having the same feed pitch (2 mm or 4 mm).

1.2.5 Other Restrictions The restrictions on the component mounter 100 include not only the constraints resulting from the structure of the component mounter 100 as described above, but also the following due to circumstances at the production site where the component mounter 100 is used. There are also operational restrictions.
(1) Fixing the arrangement For example, in order to reduce the work of replacing component tapes manually, for a specific component tape (or a component cassette containing the same), the position in the component supply unit to be set (on the Z axis) Position) may be fixed.
(2) Restrictions on resources The number of component tapes that can be prepared for the same component type, the number of component cassettes storing component tapes, the number of double cassettes, the number of suction nozzles (the number of each type), and the like are limited to a certain number. May be done.
1.3 Optimizing device The optimizing device 300 is provided when a production target (a board and components to be mounted thereon) and a production tool (a component mounting machine with a limited resource, a sub-equipment) are provided. In addition, the apparatus determines a component mounting order for manufacturing a board in a time as short as possible (increase the number of boards that can be manufactured per unit time).

Specifically, in order to minimize the mounting time per board, a component cassette containing any component tape is arranged at which position (Z axis) of which component mounter (sub equipment) and The multi-mounting head of the mounting machine (sub-equipment) simultaneously picks up as many components as possible from the component cassette in any order and mounts the picked-up multiple components at any position (mounting point) on the board in any order This is a device that determines on the computer whether to do so (searches for an optimal solution).
At this time, it is required to strictly observe the above-mentioned restrictions of the target component mounter (sub equipment).

1.3.1 Hardware Configuration of Optimizing Device The optimizing device 300 is realized by executing the optimizing program according to the present invention by a general-purpose computer system such as a personal computer, and is connected to the actual component mounter 100. If not, it also functions as a stand-alone simulator (a tool for optimizing the order of component mounting).

FIG. 9 is a block diagram illustrating a hardware configuration of the optimization device 300 illustrated in FIG. The optimization apparatus 300 is capable of performing a line tact (part of the tact for each sub-equipment constituting a line, the largest tact in the component mounting of a target board) under various restrictions based on the specifications of each equipment constituting the production line. In order to minimize the tact time, the parts to be mounted in each sub-equipment and the mounting order of the parts in each sub-equipment are determined for all the parts provided from the component mounting CAD apparatus and the like, and the optimum NC is determined. This is a computer device that generates data, and includes an arithmetic control unit 301, a display unit 302, an input unit 303, a memory unit 304, an optimization program storage unit 305, a communication I / F (interface) unit 306, a database unit 307, and the like. You.
The “tact” is the total time required to mount the target component.

The arithmetic control unit 301 is a CPU, a numerical processor, or the like, loads and executes a necessary program from the optimization program storage unit 305 to the memory unit 304 according to an instruction from a user, and executes each component according to the execution result. 302 to 307 are controlled.
The display unit 302 is a CRT, an LCD, or the like, and the input unit 303 is a keyboard, a mouse, or the like. These are used by the optimization device 300 and an operator to interact under the control of the arithmetic control unit 301. Used for A specific user interface is as described in a screen display example described later.
The communication I / F unit 306 is a LAN adapter or the like, and is used for communication between the optimization device 300 and the component mounters 100 and 200.
The memory unit 304 is a RAM or the like that provides a work area for the arithmetic control unit 301. The optimization program storage unit 305 is a hard disk or the like that stores various optimization programs that realize the functions of the optimization device 300.

The database unit 307 is a hard disk or the like that stores input data (mounting point data 307a, component library 307b, and mounting device information 307c) used for the optimization processing by the optimization device 300, mounting point data generated by the optimization, and the like. It is.
10 to 12 show examples of the mounting point data 307a, the component library 307b, and the mounting device information 307c, respectively.
The mounting point data 307a is a collection of information indicating mounting points of all components to be mounted. As shown in FIG. 10, one mounting point pi includes a component type ci, an X coordinate xi, a Y coordinate yi, and control data φi. Here, “component type” corresponds to the component name in the component library 307b shown in FIG. 11, and “X coordinate” and “Y coordinate” are coordinates of the mounting point (coordinates indicating a specific position on the board). The “control data” is restriction information (a type of a suction nozzle that can be used, a maximum moving speed of the multi-mounting head 112, and the like) on mounting of the component. The NC data to be finally obtained is a sequence of mounting points that minimizes the line tact.

The component library 307b is a library in which information unique to each of all component types that can be handled by the component mounters 100 and 200 is collected. As shown in FIG. 11, the component size, tact ( The information includes a tact unique to a component type under certain conditions) and other constraint information (a type of a suction nozzle that can be used, a recognition method by the component recognition camera 116, a maximum speed ratio of the multi-mount head 112, and the like). In this drawing, the external appearance of the component of each component type is also shown for reference.
The mounting device information 307c is information indicating a device configuration for each sub-equipment constituting the production line, the above-described restrictions, and the like. As shown in FIG. The information includes nozzle information on the type of suction nozzles that can be mounted on the head, cassette information on the maximum number of component cassettes 114, tray information on the number of trays stored in the tray supply unit 117, and the like.

  These pieces of information are data called as follows. That is, equipment option data (for each sub-equipment), resource data (the number of cassettes and nozzles available for each equipment), nozzle station arrangement data (for each sub-equipment with a nozzle station), initial nozzle pattern data (for each sub-equipment) , Z-axis arrangement data (for each sub-equipment). Regarding resources, the number of nozzles of each type such as SX, SA, S, etc. is assumed to be 10 or more.

1.3.2 Software Configuration of Optimization Device One of the features of the optimization program stored in the optimization program storage unit 305 is that electronic components are roughly classified into “small components” and “general-purpose components”. The point is that different optimization algorithms were applied.
The number of electronic components mounted on a substrate is, for example, about 1000 when the number is large, and 90% of the components are chip components having a component size of □ 3.3 mm or less (hereinafter, such small-size components are Called "small parts.") Small components are components such as resistors and capacitors, and the component size can be limited to several patterns. The taping is a part that is all 8 mm wide and can simultaneously suction 10 points. The conditions to be satisfied by the small parts are, for example, as follows.
・ Part size is □ 3.3mm or less.
-The component height is 4.0 mm or less.
-The component recognition camera is 2DS.
-The component tape width is 8 mm.

On the other hand, the remaining 10% of components are odd-shaped components such as connectors and ICs (hereinafter, large-size components that do not satisfy the condition of small components are referred to as “general-purpose components”). Depending on the parts, there are many parameters to be considered during optimization because they are supplied in trays or require special nozzles.
Therefore, for small parts, the goal is an algorithm that can generate a maximum of 10 simultaneous suction tasks and can execute optimization processing at high speed. On the other hand, for general-purpose components, optimization is performed using a flexible algorithm that derives the optimal mounting order while changing the state (one of the possible mounting orders) using the mounting time in task units as an evaluation function. The goal is to raise the level.
FIG. 13 is a functional block diagram of the optimization program stored in the optimization program storage unit 305 shown in FIG. The optimization program is roughly divided into a component group generation unit 314, a line balance optimization unit 315, and a state optimization unit 316. Although not shown, the optimization program also includes a GUI (graphical user interface) function for interacting with the user.

The component group generation unit 314 classifies all the mounted components specified by the mounting point data 307a stored in the database unit 307 into, for example, nine component groups in terms of component thickness. Specifically, a component table indicating the number of components for each component type is created by referring to all component types indicated by the mounting point data 307a, and all component types are referenced by referring to component sizes in the component library 307b. Each component type is associated with one of a plurality of component groups. Then, it notifies the line balance optimizing unit 315 of the classification result (such as the type and number of parts belonging to each part group).
The line balance optimizing unit 315, based on the information of the component groups notified from the component group generation unit 314, observes that components are mounted in order from a component group having a smaller component thickness, and minimizes a line tact. Optimize line balance (level tact for each sub-equipment). For this purpose, three function modules (a first LBM unit 315a, a second LBM unit 315b, and a third LBM unit 315c) that operate in cooperation with the state optimization unit 316 are provided.

The reason why the component group having a small component thickness is preferentially mounted is to facilitate the movement of the multi-mounting head 112 when mounting the component on the board and to improve the quality of mounting.
The first LBM unit 315a roughly distributes the plurality of component groups notified from the component group generation unit 314 in units of task groups so that the tact time in each sub-equipment is substantially equal. That is, the line balance is optimized by coarse adjustment. Here, the “task group” refers to a group of tasks, and corresponds to a range of a component group in which the mounting order of components can be changed for optimization.
The second LBM unit 315b minimizes the line tact by moving the task group for each sub-equipment roughly allocated by the first LBM unit 315a between the sub-equipment. That is, the line balance is optimized by fine adjustment.

The third LBM unit 315c optimizes the line balance for the state (task group distribution) optimized by the second LBM unit 315b in the same procedure as that of the second LBM unit 315b for each component type (component tape). I do.
For each of the plurality of component groups generated by the component group generating unit 314, the state optimizing unit 316 determines a task group constituting each component group, and determines an optimal state (Z of each component tape) for each determined task group. The values on the axis and the mounting order of the components (mounting points) on each component tape are determined, and optimization is performed on small components (for example, components belonging to five of nine component groups). A small component optimizing unit 316a, a general component optimizing unit 316b for optimizing general components (for example, components belonging to the remaining four of the nine component groups), and a small component optimizing unit 316a and an optimization engine unit 316c that executes calculation processing common to optimization in the general-purpose component optimization unit 316b. The “state” refers to an individual mounting order that can be taken by a target component or component type (component tape).

Note that the small component optimization unit 316a determines a task group and optimizes a state using an algorithm that is simple and suitable for high-speed processing, while the general-purpose component optimization unit 316b uses a sophisticated and intelligent algorithm. To optimize the state. This is because it is generally known that the total number of small components mounted on a substrate such as a mobile phone is extremely large (for example, a ratio of 9: 1) as compared with general-purpose components, as described above. This is because, by performing optimization using a corresponding algorithm, a more optimal solution can be obtained in a shorter time as a whole.
The optimization engine unit 316c performs optimization calculation based on a heuristic but deterministic algorithm (hill-climbing method) based on parameters given from the small component optimization unit 316a and the general component optimization unit 316b, and a stochastic method. However, it performs an optimization calculation based on an algorithm (multi-canonical method) that searches for an optimal solution globally.

FIG. 14 is a schematic flowchart when the operation control unit 301 executes the optimization program stored in the optimization program storage unit 305 shown in FIG. That is, this drawing is a typical flow of processing by each functional block shown in FIG. 13 and corresponds to a flowchart corresponding to main processing by the optimization device 300.
Here, basically, the processing is sequentially performed from the upper step (the processing in the rectangular frame) downward. The nesting-displayed portion indicates that the parent step is realized by nested child steps (or repetitions thereof).
As shown in the figure, the entire optimization process S310 includes the following six large steps S311 to S316.
(1) Reading of mounting point data (S311)
First, all mounting point data 307a is read from the database unit 307 into the memory unit 304 or the like. Related data (component library 307b, mounting device information 307c) is also read as needed.
(2) Creating a parts list (S312)
Each mounting point data 307a is linked with information on a component to be mounted (component library 307b), so that if all the mounting point data 307a is read, a component list describing what components are to be mounted and how many points are to be mounted. Can be created.
(3) Generation of a part group (S313)
Next, a component group is generated from the component list. The “parts group” is a grouping of a parts list according to the size of parts, and is roughly classified into small parts and general-purpose parts. The small parts are further subdivided into the following three parts groups according to the size, for example.
G1: 0.6 mm × 0.3 mm size component G2: 1.0 mm × 0.5 mm size component G3: 1.6 mm × 0.8 mm size or larger component (4) Initial distribution to front and rear sub-equipment (S314)
The standard mounting time is determined for each electronic component, and the component type (component tape) is assigned to the front and rear sub-equipment 110 and 120 so that the accumulated value of the standard mounting time for the component allocated to each facility becomes substantially the same. Distribute. After allocating the components to the front and rear sub-equipment 110 and 120, the component tape is further allocated to one of the left and right blocks in units of a component group or the like.

(5) Line balance processing (S315)
The optimization process of the small component and the optimization process of the general-purpose component are sequentially executed (S320, S321). Then, the component tapes are arranged in the component supply units 115a and 115b in consideration of the arrangement and fixing (S322). Subsequently, the mounting time is calculated for each of the front and rear sub-equipment 110 and 120. If the front and rear sub-equipment has a poor balance, the part is moved between the front and rear sub-equipment 110 and 120 (S323), and the small part and the general-purpose part are again returned. Perform optimization processing. Further, optimization is performed in consideration of the mounting point (the mounting position of the component on the board), that is, optimization by an intersection elimination method (S324) described later and optimization by a return optimization method (S325).

Note that the flowchart of FIG. 14 shows a processing procedure when a representative one selected from among a plurality of methods (“cutting method”) is adopted for optimization of small parts (S320). ing.
(6) Output of optimization result (S316)
When all the above processes are completed, the following data is output.
-Mounting order and task configuration of electronic components-Layout of component supply units 115a and 115b (arrangement of component tapes)
Resource use status of feeder, nozzle, etc. Predicted mounting time for each of the front and rear sub-equipment 110, 120 The correspondence between these steps and the respective functional blocks shown in FIG. 13 is as follows. In other words, steps S311 to S313 are processing mainly by the component group generation unit 314, and step S314 is processing mainly by the first LBM unit 315a and the second LBM unit 315b of the line balance optimization unit 315. The processing is mainly performed by the third LBM unit 315c and the state optimizing unit 316 of the line balance optimizing unit 315, and the step S316 is mainly performed by the line balance optimizing unit 315 and a user interface unit (not shown).

The details of these steps are as described in “Operation of the optimization device (overview)”, “Operation of the optimization device (detail)”, and “Operation of the optimization device (application)” to be described later. is there.
The “HC method” in the figure means a hill-climbing method and is an algorithm that is heuristic but deterministically seeks an optimal solution. The “MC method” means a multi-canonical method and is stochastic. An algorithm that searches for the optimal solution globally.
More specifically, the optimization of the component mounting order is a process of searching for a mounting order that satisfies certain conditions (such as the constraints described above) and has the shortest mounting time from among a finite number of possible mounting orders. Mathematically, it corresponds to processing for finding a solution (optimum solution) under certain conditions in an optimization problem.
The "hill climbing method (HC method)" is one of the solutions called a local search method. First, one solution that satisfies a condition is selected, and thereafter, the solution is transformed according to a certain order (here, This is a method in which the result (in this case, the mounting time) is improved while the condition is satisfied, and the process is repeated if the result is improved.

The “multicanonical method (MC method)” is one of the so-called global search methods. First, one solution that satisfies the condition is selected, and various unbiased deformations are added to the solution. , The probability that the result is improved (entropy is reduced) while satisfying the conditions is evaluated for each type of deformation, and the process of improving the result with the highest probability from among those deformations is repeated. This is a method that ends when the result does not improve even after adding.
Note that these “hill climbing method” and “multi-canonical method” both attempt to greedily modify the immediately preceding solution, and if the results are improved and satisfy certain conditions, the solution is It is common in adoption and belongs to one of the approaches called "greedy method".
Further, the optimization device 300 is a device for optimizing the mounting order of components by information processing on a computer based on a dedicated program, and therefore, in this specification, the optimization device is a product (component, task, task To "move" a group, a component cassette, a component tape, etc. means "to rewrite data (such as data specifying a component mounting order) held in a storage device such as a memory or a hard disk".

2 Operation of Optimizer (Overview)
Next, a basic operation of the optimization device 300 in the component mounting system 10 configured as described above will be described.

2.1 Creation of Component Group The component group generation unit 314 converts all the mounted components specified by the mounting point data 307a stored in the database unit 307 from the point of component thickness as shown in FIG. Nine parts groups G [1] to G [9]. This processing corresponds to step S313 in FIG.

  Specifically, by referring to all the component types indicated by the mounting point data 307a, a component table indicating the number of components for each same component type as shown in FIG. 15B is created, and a component library 307b is created. By referring to the component sizes in, all the component types are associated with any of the nine component groups G [1] to G [9]. Then, it notifies the line balance optimizing unit 315 of the classification result (such as the type and number of parts belonging to each part group).

2.2 Line Balance Processing FIG. 16 is a diagram illustrating a state in which the first LBM unit 315a of the line balance optimization unit 315 allocates task groups to sub-equipment. This processing corresponds to step S314a in FIG.
The first LBM unit 315a arranges all the task groups in a line so that the part group with the thinner part comes first, and the tact of each sub-equipment is sequentially represented by the following equation from the top in the arrangement. The task groups are sorted in order from the upstream sub-equipment so that the task group is closer to
θ = (total tact on all parts groups) / total number of sub equipment N
The “total tact for all component groups” is specified by referring to the mounting point data 307a and the component library 307b, and the “total number N of sub-equipment” is specified by referring to the mounting device information 307c. Is done.
FIG. 17 is a diagram illustrating a state of optimizing the line balance (moving the task group) by the second LBM unit 315b, and a graph 405a illustrates a tact distribution (a task group distribution state to each sub-equipment) before the optimization. And a graph 405b shows how the task group is moved by the optimization, and a graph 405c shows the tact distribution after the optimization. This processing corresponds to Step S314b in FIG.
Here, in the tact distribution shown in this figure, the vertical axis indicates the size of the tact, and the horizontal axis indicates the arrangement of all the sub-equipment (here, six) constituting the production line (from upstream to downstream). ), And the task group is shown as a block “TGn-m” whose height is the tact. n indicates the part group numbers 1 to 9 to which the task group belongs, and m is a number for distinguishing the task groups belonging to the same part group.

In each of the sub-equipment, those belonging to the component group having a small component thickness are mounted first on the divided task groups. However, the order of a plurality of task groups belonging to the same component group is not restricted. For example, the sub-equipment [3] may be mounted in the order of TG3-3 → TG3-1 → TG3-2.
FIG. 18 is a flowchart showing a procedure for optimizing the line balance by the second LBM unit 315b shown in FIG.
The second LBM unit 315b firstly sets the sub-equipment [Smax] with the maximum tact for each sub-equipment in the initial state (distribution of task groups) shown in the graph 405a of FIG. 17 generated by the first LBM unit 315a. The smallest sub-equipment [Smin] is specified (S500). For example, Smax = 5 and Smin = 2 are specified.

Then, the tact of the sub-equipment [Smax] is stored as the line tact LT (S501). For example, LT is stored as sub-equipment [5].
Next, for the sub-equipment [i] from the sub-equipment [Smin] to the sub-equipment [Smax-1], a movable task group is moved between two adjacent sub-equipments in order (S502 to S507). .
That is, one task group is temporarily moved from the sub-equipment [i + 1] to the sub-equipment [i] (S503), and it is confirmed whether the tact of the sub-equipment [i] is still smaller than the line tact LT. (S504).
As a result, the task group is actually moved only when it is confirmed that the task group is small (S505). That is, the tact of the sub-equipment [i] and the sub-equipment [i + 1] is updated. For example, the task group TG3-1 is moved from the sub-equipment [3] to the sub-equipment [2]. It should be noted that a task group that is a candidate to be moved is preferentially selected from those belonging to a component group having a small component thickness.

When such movement of the task group is repeated in order from the sub-equipment [Smin] to the sub-equipment [Smax-1] (S502 to S506), finally, it is determined whether the tact of the sub-equipment [Smax] has decreased. That is, it is determined whether one or more task groups have been moved from the sub-equipment [Smax] to the sub-equipment [Smax-1] (S507).
As a result, if it has decreased, it is determined that there is still room for optimization, and the same optimization (S500 to S507) is repeated again. If not, further optimization is performed. Is determined to be difficult, and the process ends (S507).
When there are a plurality of task groups that can be moved, there is a degree of freedom in selecting a movement target, so that various combinations of task groups to be moved are tried within the allowable calculation time.

In this way, by sequentially trying to move the task group between the sub-equipment with the minimum tact and the sub-equipment with the maximum tact, reduction of the maximum tact (line tact), that is, optimization of the line balance is realized. .
When the above optimization is completed, the third LBM unit 315c next sets the second LBM unit in units of component types (component tapes) with respect to the state (task group distribution) optimized by the second LBM unit 315b. The line balance is optimized in the same procedure as in 315b.
In other words, the second LBM unit 315b moves between adjacent sub-equipments in units of task groups (S503, S505), but the third LBM unit 315c replaces the task groups with the component types constituting each task group. (Component tapes) are moved in units of sub-equipment. Therefore, the step of increase / decrease in tact between the two sub-equipments is smaller than that in the case of the second LBM unit 315b, and more detailed optimization is performed. Thereby, the line tact LT can be further reduced.

2.3 Optimization for Small Parts FIG. 19 is a flowchart showing a schematic procedure of optimizing the order of mounting small parts by the small part optimizing unit 316a of the state optimizing unit 316, and includes two large steps.

First, the small component optimization unit 316a generates a suction pattern for all mounted components (S520). This is equivalent to determining the arrangement in units of component types (component tapes), that is, determining the arrangement (Z-axis) of the component cassettes 114.
Here, the “suction pattern” is a two-dimensional diagram as shown in FIG. 20, in which the vertical axis is the order of suction of components by the multi-mounting head 112 and the horizontal axis is the arrangement of the component cassettes 114 (component tapes). (Z-axis) shows one or more sets of components that are simultaneously adsorbed by the multi-mount head 112. Each component (mounting point) to be suctioned is indicated by a unit rectangle (square or rectangular).
In FIG. 20, for convenience of explanation, a suction pattern for a four-nozzle head is shown, and a unit in which a maximum of four unit rectangles are connected horizontally is mounted (suction / movement / mounting). (That is, a task), and a set of connected tasks surrounded by a circle corresponds to a task group. Therefore, this figure shows a total of three independent task groups.

The generation of such a suction pattern corresponds to determining the relative arrangement of the component tapes so that the multi-mounting head can simultaneously suction as many components as possible. In other words, all the component tapes are determined. Is divided into a plurality of independent array groups (task groups).
Next, as shown in FIG. 19, the small component optimizing unit 316a reduces the total tact of each task group (component tape group having a fixed arrangement) determined in step S520. The mounting order of the components constituting each component tape is determined (S521). This is because the distance from the immediately preceding mounting point in the same task differs depending on which mounting point, even if the component is taken out (adsorbed) from the same component cassette 114, This corresponds to shortening the moving distance (mounting time) of 112.

2.4 Task Group Generation Method The first specific algorithm for generation of the suction pattern in FIG. 19 (S520) is the “task group generation method”.

  This method is a method of repeatedly generating a task group including a sequence of a number of component types (component tapes) within a certain range (less than twice the number of suction nozzles). It consists of large steps (first and second steps). FIG. 21 is a diagram for explaining the first and second steps. The component histogram 406a is a component histogram in which target components are arranged (sorted) in the order of component tapes having the largest number of components. , Diagram 406b is the suction pattern generated by the first and second steps.

[First step]
In this step, the first half processing for generating one task group, that is, the component histograms are arranged in the right direction (Z-axis direction) in the order of the component tapes with the largest number of components. In particular,
(i) A component tape having the largest number of components (one component tape) among component tapes that have not been arranged is placed on the Z axis.
(ii) A component tape of the second number of components (two-component tape) is placed on the right side.
(iii) A third component tape (three component tape) is placed on the right of the two component tape.
(iv) Hereinafter, this is repeated up to the number L of suction nozzles of the multi-mounting head 112 (here, “4”).
As a result, four component tapes 400 are taken out from the component histogram 406a and arranged at the location 400 shown in the diagram 406b.

[Second step]
In this step, the component histograms are arranged in the left direction such that the number of simultaneous suctions of the task whose number of simultaneous suctions is less than L becomes L with respect to the diagram generated in the first half processing. In particular,
(i) The number of parts of the L-component tape is subtracted from the number of parts of the one-component tape.
(ii) A component tape (L + 1 component tape) having a component number less than or equal to the obtained component number difference and closest to the component number difference is placed on the left of one component tape.
(iii) Subtract the number of components of the (L-1) component tape from the two-component tape.
(iv) A component tape having a component number that is equal to or less than the component number difference and that is closest to the component number is placed on the left of the (L + 1) component tape.
(v) Hereinafter, this is repeated (L-1) times.
As a result, the two component tapes 401a and 401b in the component histogram 406a are taken out and arranged at the location 401 shown in the diagram 406b. Thus, one suction pattern including the component tape 400 and the component tape 401 is completed. As a result, the relative Z-axis is determined for the task group including the six types of component tapes.

The generation of the task group by the above first and second steps is repeated until the target component tape runs out.
Here, if there are no unplaced component tapes satisfying the condition of the second step, the following three steps (third to fifth steps) are performed instead of the first and second steps. ). FIG. 22 is a diagram for explaining the third to fifth steps. A component histogram 415a shows a portion of the entire component histogram that is not arranged (portion surrounded by a solid line), and a diagram 415b shows The suction patterns generated by the third to fifth steps are shown.

[Third step]
In this step, the unplaced component histogram is shaped to generate a partial histogram. In particular,
(i) Find the minimum value of the number of component tapes that have not been placed yet.

(ii) Subtract (minimum value-1) from the number of components of each component tape that has not been placed yet.
As a result of such a subtraction process, the number of components in the unplaced component histogram becomes a component histogram 415a surrounded by a thick solid line. Hereinafter, the following fourth and fifth steps are performed using the number of components in the component histogram 415a. Proceed.

[4th step]
This step corresponds to the first step described above. In particular,
(i) A component tape having the largest number of components (one component tape) among component tapes that have not been arranged is placed on the Z axis.
(ii) A component tape of the second number of components (two-component tape) is placed on the right side.
(iii) A third component tape (three component tape) is placed on the right of the two component tape.
(iv) Hereinafter, this is repeated up to the number L of suction nozzles of the multi-mounting head 112 (here, “3”).
As a result, the three component tapes 410 are extracted from the component histogram 415a and are arranged at the location 410 shown in the diagram 415b.

[Fifth step]
This step corresponds to the above-described second step. In particular,
(i) Subtract the value of (number of components of L component tape -1) from the number of components of one component tape.
(ii) A component tape (L + 1 component tape) having a component number equal to or less than the component number difference and closest to the component number difference is placed on the left of the one component tape.
(iii) Subtract the number of components of the L component tape from the (L + 1) component tape.
(iv) A component tape that is less than or equal to the component number difference and closest to the component number difference is placed on the left of the (L + 1) component tape.
(v) Hereinafter, this is repeated L times.
As a result, the three component tapes 411 in the component histogram 415a are taken out and arranged at the location 411 shown in the diagram 415b. Thus, one suction pattern including the component tape 410 and the component tape 411 is completed. As a result, even with respect to the component tapes left in the first and second steps, that is, component tapes having a small difference in the number of components, a task group including a group of tasks that can be simultaneously picked up is generated. Is determined relative to the Z axis.

2.5 Pruning Method The second specific algorithm for generating the suction pattern (S520) in FIG. 19 is the "pruning method". This processing corresponds to steps S320a to S320d in FIG.
This method is based on arranging a component histogram arranged in the order of component tapes having a large number of components on the Z-axis as it is, and performing the above-described suction only for a portion where the maximum number (L) of components cannot be simultaneously suctioned. This is a method to apply a pattern generation method, and includes the following two large steps (first and second steps).

[First step]
In this step, the task of extracting a task consisting of L parts from the parts histogram is repeated (removed).
FIG. 23 and FIG. 24 are diagrams for explaining the first step in the trimming method. FIG. 23 is a component histogram 450 in which all components to be mounted are arranged in the order of the component tape having the largest number of components. FIG. 24 shows a case where components are picked up from the component histogram 450 of FIG. 23 in units of L (here, 10) components (tasks when the maximum number of components are picked up at the same time). FIG. FIGS. 23 and 24 correspond to steps S320a and S320b in FIG. 14, respectively.

  In the trimming, the L component list (one of 〇, △, and × is selected) so that the component tapes with a small number of components are eliminated first, that is, components are eliminated from the rightmost component tape in the component histogram. (A list of 10 rectangles). This is repeated until it cannot be removed in units of L parts arrangement.

[Second step]
In this step, a diagram according to the above-described task group generation method is generated for the component histogram including the remaining components after the above-described pruning.
FIG. 25 and FIG. 26 are diagrams for explaining the second step in the trimming method. FIG. 25 is a diagram in which the parts left after the trimming in the first step are reconstructed in descending order of the number of parts. FIG. 26 is a diagram illustrating a state in which a diagram according to the above-described task group generation method is generated for the reconstructed component histogram 451. FIGS. 25 and 26 correspond to step S320d in FIG.

Note that the width (the number of component tapes) of the reconstructed component histogram 451 is always (L-1) or less according to the processing content of the first step.
In the second step, specifically, the following processing is performed.
(i) For the components remaining after the pruning, a component histogram 451 shown in FIG. 25 is generated, and the total number of components (here, 100) is calculated.
(ii) Divide the calculated total number of parts by L (here, 10), and aim to create a suction pattern using the obtained value (here, 10) as the number of tasks.
(iii) For this purpose, as shown in FIG. 26, for a component tape having a number of components larger than the obtained number of tasks (10), a portion of the excess 451a (or a portion of the excess portion divided). ) Is cut out and complemented and placed on the left side of the component histogram 451.

FIG. 27 shows the suction pattern 452 for the component tape for which the Z-axis has been determined by the trimming method in the first and second steps. As shown in this figure, all the components are constituted only by the task in which the maximum number (10) of components are simultaneously picked up, and can be efficiently mounted with the maximum simultaneous suction rate.
FIG. 28 is a component histogram 453 (reconstructed without changing the Z axis) corresponding to the suction pattern 452 shown in FIG.
As can be seen from the histogram 453, according to the trimming method, the tendency that component tapes having a large number of components are arranged at the left position is maintained. This means that the reaping method requires that the moving path of the multi-mounting head 112 (for the right block 115b, after picking up a part, the moving path always passes in front of the two-dimensional camera placed on the left end of the right block 115b. ) Is considered (the total moving distance is reduced, that is, the total tact is reduced).

  The left block 115a may be subjected to processing symmetric in the Z-axis direction in the processing described above. In other words, after arranging the component tapes in ascending order of the number of components, the task may be cut down in the same procedure to generate a diagram.

2.6 Random selection method ("greedy method")
A first specific algorithm for optimizing the mounting order (S521) in FIG. 19 is a random selection method. This processing corresponds to Step S320e in FIG.
This method is a greedy method that repeats the process of replacing two randomly selected mounting points if the total tact becomes smaller in one task group.
FIG. 29 is a flowchart showing the procedure for optimizing the order of mounting components by the random selection method, and FIG. 30 shows how two mounting points are interchanged by the random selection method.

First, the small component optimization unit 316a calculates the total tact in the initial state (S530). Note that the state here is a state in which the mounting order for all components (mounting points) constituting one task group is determined in a fixed pattern. Therefore, the total tact for one state is uniquely determined from the information 307a to 307c stored in the database unit 307.
Next, two are randomly selected from all the mounting points (S531), and the total tact (temporary tact) when the order of the selected two mounting points is changed is calculated (S532). FIG. 30 shows an example of a state in which the mounting points B2 and B4 are exchanged.
Then, it is determined whether the calculated temporary tact is smaller than the tact in the state immediately before (S533).
As a result, if smaller, the two mounting points are exchanged (S534). In other words, the current state and the total takt are updated and stored as those obtained when the mounting points are exchanged. Then, it is determined whether or not the end condition at that time (the tact in that state is smaller than the target tact specified in advance by the operator or a certain processing time has been reached) (S535). , The process ends.

  On the other hand, if the tact does not decrease even after the two mounting points are replaced (No in S533), and if the termination condition is not satisfied (No in S535), the same processing is performed again until the termination condition is satisfied. It repeats (S531-S533-S535). In this way, according to the random execution method, the tact for each task group is reduced according to the execution time spent, and the component mounting order is optimized.

2.7 Intersection Resolution Method A second specific algorithm for optimizing the mounting order (S521) in FIG. 19 is the intersection resolution method. This processing corresponds to step S324 in FIG.
This method eliminates the intersection of polygonal lines (paths) obtained by connecting the mounting points for each task with a straight line, instead of randomly selecting the two mounting points to be replaced. This is a method of selecting and replacing the mounting points that satisfy the criteria of performing.

FIG. 31 is a diagram showing how the placement order of components is optimized by the intersection elimination method for the three tasks 455a to 455c including five mounting points. The diagram 457 shows the state before the intersection of the polygonal lines is eliminated. (A distribution of polygonal lines for each task), and a diagram 458 shows a mounting order after intersections of polygonal lines are resolved. Mounting points of the same component type (component tape) are indicated by circles of the same pattern.
First, the small component optimizing unit 316a specifies all the intersections in the initial state by referring to the mounting point data 307a and the like of the database unit 307. However, the intersection here is the intersection of a line connecting two mounting points that belong to the same task and is continuously attached, and a similar line that belongs to another task. The component types (component tapes) of the components used for the points are limited to those having the same line segments.

Next, for all the specified intersections, the connection of the line segments is sequentially changed so as to eliminate the intersections. Note that before and after the cancellation, the component types of the components located at both ends of each line segment are not changed. Therefore, the connection change of this line segment is uniquely determined, and the arrangement of the component types constituting each task is determined by the connection change. Does not change.
Such an intersection eliminating method eliminates useless movement of the multi-mounting head 112 between tasks. In other words, the mounting point to be moved after one component is mounted is determined by the mounting order of the components in which an increase in useless tact due to the movement of the multi-mounting head 112 is suppressed.

2.8 Return Optimization Method A third specific algorithm for optimizing the mounting order (S521) in FIG. 19 is a return optimization method. This processing corresponds to step S325 in FIG.
This method focuses on the return trajectory of the multi-mounting head 112 that moves to pick up the component of the next task after completing the component mounting of one task in one task group, and configures the task group. This is a method for optimizing the arrangement of tasks (order in task units).
FIG. 32 is a diagram for explaining a procedure for optimizing the order of tasks by the return optimization method. Here, the movement locus (mounting path) of the multi-mounting head 112 that moves between the board and the component supply unit when ten tasks are arranged in the component supply units 115a and 115b on the Z axis is indicated by an arrow line. It is shown.
Here, circles indicate representative positions of the multi-mounting head 112. That is, the circle on the board indicates the position (final mounting point) of the multi-mounting head 112 immediately after the mounting of the last component in one task, and the circle on the Z-axis indicates the first in each of the 20 tasks. The position of the multi-mounting head 112 when sucking a component (hereinafter, referred to as “suction position”) is shown. It should be noted that the numerical values attached to the circles are numbers for distinguishing each suction position (task).

[First step]
In this step, the implementation path is drawn according to the following rules.
(i) Return to the suction position at the shortest distance from the final mounting point of each task, that is, minimize the return locus.
(ii) The mounting path is sequentially drawn starting from the first suction position. Since one suction position corresponds to one task, the final mounting point corresponding to the suction position is uniquely specified. In FIG. 32, a mounting path connecting the suction position and the final mounting point is depicted in the order of 1 → 5 → 14 → 2 → 8 → 3 → 17 → 12 → 16 → 1.
(iii) After returning to the first suction position (the first suction position), this is set as the shortest traveling partial path 1.
(iv) Next, a suction position not included in the shortest traveling partial path found so far is searched. In FIG. 32, the fourth suction position is found.
(v) Return to (ii) above, and repeat until there are no more unused suction positions. In FIG. 32, five shortest circular partial paths are drawn.
By such a first step, the order of the suction positions, that is, the order of the tasks, which minimizes the return trajectory of the multi-mounting head 112 when starting from the specific suction position is determined.

[Second step]
Next, for each of the shortest traveling partial paths drawn in the first step, it is specified which suction position should be started. Specifically, the return trajectory of the multi-mounting head 112, which is moved to start the next shortest cyclic partial path after completing the mounting of all components belonging to one shortest cyclic partial path, is minimized. The first suction position in the shortest cyclic partial path and the order of the shortest cyclic partial paths are determined.

As a result, the execution order of the tasks is determined so that the return trajectory of the multi-mount head 112 between the tasks is shortened for all the tasks constituting one task group.
Although FIG. 32 shows the mounting path in the task group in which the 20 suction positions are different, as shown in FIG. 33, the same applies to a task group including a plurality of suction positions at the same position. Can be optimized. At this time, there is a degree of freedom in selecting the final mounting point corresponding to the same position of the suction position. What is necessary is just to select a point and create the shortest traveling partial route.
As described above, the random selection method and intersection resolution method enable optimization of (i) the implementation order within a task, and (ii) optimization of the implementation order taking into account all tasks, without changing the task form. On the other hand, by the return optimization method, after all the tasks are fixed (that is, with the members of each task fixed), the order of the tasks is optimized.

2.9 Optimization for general-purpose parts General-purpose parts have a wide variety of component sizes, nozzles, component recognition cameras, and supply forms (tapes and trays). It is possible. Here, a method of searching for an optimal state while efficiently changing the state of the task is employed. This processing corresponds to Step S321 in FIG.

The evaluation index for performing the optimization is the mounting time. Therefore, a mounting time simulation that accurately simulates the operation time of the component mounter 100 is mounted. The optimization algorithm for general parts is as follows.
(1) Setting of the number of loops As a practical problem, it is not possible to evaluate all combinations, so that an end condition is set in advance. If the mounting time does not decrease even if the predetermined loop number process is continued, the optimization process is terminated.
(2) Initial state generation First, an initial state is generated for all general-purpose components. The initial state is a summary of all mounting points of the general-purpose component on a task basis, and any state may be used as long as all of the constraint conditions of the component mounter 100 are satisfied.
(3) Change of state Search for the optimal task state while changing the task state. Procedures for changing the state include:
-Swap two implementation points that exist in different tasks,
-Swap the mounting order of two mounting points in the same task,
・ Swap two component tapes,
Etc. Here, in order to flexibly change the state of the task, replacement with an empty mounting point is also possible. For example, the process of moving the mounting point of a certain task to a task that has room can be a replacement between the former mounting point and the latter mounting point. By repeating this process, the number of tasks can be reduced.

Whether or not to adopt the state after the change is determined by whether or not the wearing time is reduced. However, if the state where the wearing time is constantly reduced is adopted, the state is caught by the local minimum. Therefore, a state in which the mounting time increases with a certain probability is adopted.
Hereinafter, specific contents of optimization for general-purpose components will be described.
FIG. 34A is a flowchart showing a procedure for optimizing a mounting order of general-purpose components by the general-purpose component optimizing unit 316b, and FIG. 34B explains an approach for searching for an optimal solution by the optimization. FIG. 7 is a diagram (a diagram showing tacts of all possible states).
As shown in FIG. 34A, the general-purpose component optimization unit 316b generates an initial state X for all components (general-purpose components) belonging to the component groups G [6] to G [9]. (S550), by optimizing the initial state X by the hill-climbing method by the optimization engine unit 316c to determine the optimum state Xopt (S551), and then to the initial state X by the multi-canonical method. By causing the optimization engine unit 316c to execute the optimization, the optimal state Xopt obtained in step S551 is updated (S552). Finally, the updated optimal state Xopt is again optimized by the hill-climbing method. The optimization engine unit 316c executes the optimization to update the optimal state Xopt obtained in step S552 (S553).

As described above, the optimization by the multi-canonical method (S552) for searching for the optimal solution at the global starting point is inserted in the middle of the hill-climbing method (S551 and S553) for surely finding the local optimal solution. Therefore, the search for a state that is locally optimal but not globally optimal (such as the state 1) shown in FIG. 34 (b)) is avoided, and the global optimal state ( The state 5)) shown in FIG. 34B is obtained.
FIG. 35 is a flowchart showing a detailed procedure of optimization (S551, S553) by the hill-climbing method shown in FIG. That is, the optimization engine unit 316c, which has been notified of the initial state X and the end condition, generates the initial state X (S560), and repeats the inner loop until the outer loop end condition is satisfied (S561). (S562 to S568). Here, the outer loop end condition is a condition for confirming that there is no more optimal solution. For example, by changing (searching) all kinds of parameters that cause a state change, etc. The inner loop end condition is that a certain range of parameter is changed (searched) for one type of parameter.

In the inner loop, the optimization engine unit 316c first generates a state candidate Xtmp using one selected by the general-purpose component optimization unit 316b from nine types of state changes described later (S563, S564), and If the state Xtmp has feasibility (feasibility) described later (S565) and the tact of the candidate state Xtmp is smaller than the tact of the immediately preceding state (S566, S567), the state and the tact are updated. (S568).
As a result, a locally optimum state is deterministically obtained.
FIG. 36 is a flowchart showing a detailed procedure of optimization (S552) by the multi-canonical method shown in FIG. In this drawing, the bin number is a number indicating each section (bin) obtained by equally dividing the horizontal axis (all possible states) shown in FIG. H [i] indicates that the candidate state Xtmp belonging to the bin with the bin number i is selected (S576, S577), that the candidate state Xtmp has feasibility (S578), and that the entropy is reduced. It is a variable that stores the total number of times determined (S579 to S581).

As can be seen by comparing the flowchart shown in this figure with the one using the hill-climbing method shown in FIG. 35, a series of steps of generating a state candidate Xtmp based on the state X and determining whether to accept it or not. These processes are common in that the above process is repeated. The difference is in the method of determining acceptance. In the hill-climbing method shown in FIG. 35, when the tact of the state candidate Xtmp is smaller than the state X (determinately), it is accepted. In the multicanonical method shown in No. 36, the state candidate Xtmp is stochastically received with reference to the entropy in the tact (S580 to S582).

Here, in order to explain the nine types of state changes and the feasibility in the flowcharts shown in FIGS. 35 and 36, first, an intermediate expression used by the general-purpose component optimization unit 316b will be described. In order to facilitate optimization, the general-purpose component optimization unit 316b introduces the following three types as intermediate expressions of the Z-axis array, stores a state using those expressions, and stores the state in the optimization engine unit 316c. Or instruct.
(i) Gorder [i] (i = 1, ..., L)
This is a variable that specifies the priority order when arranging the input L component groups (task group TG [i] (i = 1,..., L) on the Z axis. When i! = J, Gorder [i]! = Gorder [j].
(ii) block [i] (i = 1, ..., L)
A variable that specifies whether to place the task group TG [i] (i = 1,..., L) in the left or right Z block (the component supply units 115a and 115b). Take the symbol value on the right.
(iii) Corder [i] [j] (i = 1, ..., L, j = 1, ..., M [i])
A number that specifies the arrangement order of the component tapes j (= 1,..., M [i]) belonging to the task group TG [i] (i = 1,. 1 to M [i] are taken as values. j! If k = Corder [i] [j]! = Corder [i] [k]. When Corder [i] [j] <Corder [i] [k], there is a relationship of “Z number of component tape j <Z number of component tape k”.
FIG. 37 shows an example of an intermediate expression used by the general-purpose component optimization unit 316b. Table 460 shows a specific example of the intermediate expression used by general-purpose component optimization section 316b, and tables 461 to 464 show the meaning (conversion to Z-axis array) of the intermediate expression shown in table 460. Specifically, the Z-axis array indicated by the intermediate representation shown in the table 460 is specified through the following conversion.

First, Gorder [i] = 1, that is, the task group TG [2] having the highest priority in determining the Z-axis arrangement is arranged (table 461). Since TG [2] is block [2] = “right”, TG [2] is arranged left-justified near the component recognition camera 116 of the right block. At this time, the component cassette 114 storing a total of M [i = 2] = 6 component tapes j (i = 1,..., 6) belonging to TG [2] is Corder [i = 2]. ] [j] is placed left-aligned to the component recognition camera in the right block so that the young one in [j] is on the left.
Next, TG [4] with Gorder [i] = 2 is arranged (table 462). Since block [4] = “left”, the block is arranged right-aligned to the component recognition camera in the left block. At this time, the component cassette 114 storing a total of M [i = 4] = three component tapes j (i = 1,..., 3) has a smaller Corder [i = 2] [j]. It is arranged right-aligned to the component recognition camera so as to be on the left.

Similarly, TG [3] where Gorder [i] = 3 and TG [1] where Gorder [i] = 4 may be arranged in this order (tables 463 and 464).
Next, nine types of state changes to be selected by the general-purpose component optimization unit 316b (step S564 in FIG. 35 and step S577 in FIG. 36) will be described. It is as follows.
(1) Two mounting points of the same general-purpose component group are randomly selected, and their task numbers and head numbers (positions of the suction nozzles 112a to 112b in the multi-mounting head 112) are swapped.
(2) Randomly select two mounting points in the same task and swap their mounting order.
(3) Two task groups (two general-purpose parts groups) are randomly selected, and their Gorders are swapped.
(4) One task group (one general-purpose part group) is selected at random, and the value (“left” or “right”) of the block is changed.
(5) Two-component tapes of the same task group are randomly selected, and their Corders are swapped.
(6) In the same task group, consecutive Corder subsections are randomly selected and shifted.
(7) In the same task group, a continuous Corder partial section is randomly selected, and the Corder is changed so that the corresponding component tape is arranged on the Z axis according to the average X coordinate value of the mounting point.
(8) One task is selected at random, and the head number is changed based on the Z number of the mounting point of the task.
(9) The operation mode (direct mode, shuttle mode) of the tray component that can be operated in the shuttle mode is randomly changed.

Here, the “tray component operation mode” means that the tray components are supplied using an elevator (having a plurality of steps) built in the tray supply unit 117 (moved to a position where the multi-mounting head 112 can be sucked and placed. The "direct mode" is a system in which components are placed directly on one tray, and the "shuttle mode" is a system in which the components are taken out from a plurality of trays by reciprocating movement using a shuttle conveyor 118 and collected. This is a method in which a plurality of components are arranged in a line and inserted. Various types of information on these operation modes are included in the mounting apparatus information 307c and affect the time required to move necessary components to a predetermined position.
As for the feasibility check (step S565 in FIG. 35 and step S578 in FIG. 36) by the general-purpose component optimization unit 316b, the state Xtmp is possible only when the following five check items are simultaneously satisfied. Is considered a solution.
(1) In each task, the Z number of the mounting point in the direct mode is the same level. That is, in the direct mode, it is considered that only tray components placed on one stage can be supplied at the same time.
(2) In each task, there is no interference between component points at the time of suction. In other words, depending on the shapes of two components that are sucked adjacent to each other, the components may come into contact with each other.
(3) In each task, the mounting point can be picked up (the set of the head number and the Z number of the mounting point is appropriate). That is, it is taken into consideration that not all of the suction nozzles mounted on the multi-mounting head 112 can be moved to any of the 96 component cassettes 114 (the components can be suctioned).
(4) In each task, a mounting point can be mounted (a set of a head number and a coordinate value of the mounting point is appropriate). It is taken into account that not all the suction nozzles constituting the multi-mounting head 112 can be moved to all places on the substrate.
(5) The arrangement of the suction nozzles in the nozzle station 119 can be determined so that the suction nozzle patterns of all tasks in all task groups can be realized. That is, it is taken into consideration that there are restrictions on the arrangement position, the number, and the like of the replacement suction nozzles that can be arranged in the nozzle station 119.
(6) The multiple mounting heads 112 and the components on the Z axis are arranged at the same pitch. That is, it is confirmed that components (or component cassettes) that can be simultaneously picked up by the multi-mounting head 112 are arranged on the Z axis.

  As described above, since the general-purpose component optimization unit 316b performs not only local (local) optimization but also optimization using a probabilistic search (step S550 in FIG. 34A). S553), the problem that the local minimum is calculated as the optimal solution is avoided.

3 Operation of the optimization device (detailed)
Next, the operation of the above-described optimization device 300 will be described in more detail. That is, detailed contents of the individual optimization algorithms described above and the operation of the optimization device 300 under various restrictions will be described.

3.1 "Mowing method"
The “pruning method” is an algorithm that compensates for the disadvantage of the “task group generation method” described above (steps S320a to S320d in FIG. 14). Hereinafter, the details of the “pruning method” will be described while clarifying the problem in the “task group generation method”.

3.1.1 Overview of “Task Group Generation Method” The basic concept of the optimization algorithm for small parts in the “task group generation method” is as shown in FIG. As for all the mounted components, `` collect n component tapes with the same number of components, and simultaneously pick up one point at a time from these n component tapes, create an n-point simultaneous suction task '' . In the target component mounter in the present embodiment, n is 10 (or 4).

FIG. 38 is a component histogram for describing the task group generation method. The horizontal axis indicates the Z axis (component cassette, component type), and the vertical axis indicates the total number of components belonging to the component type.
However, in the above-described algorithm, actually, not only component tapes with the same number of components are used, but component tapes with the same number of components are created by component division.
Even in such a case, when the number of collected n component tapes is not uniform (components A to J in FIG. 38), a component tape for compensating the variation is created, and the n component tapes are used. to add. The maximum number of added component tapes is (n-1) (the left part 506 in FIG. 38).
A set of n to n + (n-1) component tapes thus created is referred to as a “task group” (a task is generated by picking up components from those component tapes. Naming that focuses on).

Usually, a plurality of task groups are generated. The number of task groups depends on the total number of component types. There may be only one task group.
Arrangement of cassettes on the Z axis is performed in units of task groups.

3.1.2 Task of “Task Group Generation Method” The algorithm of the “task group generation method” has the following problems.
(1) Since the task groups are arranged on the Z axis, the task groups cannot be arranged unless the Z axis is at least 10 or less. Therefore, an unused portion may be generated on the Z axis.
(2) The degree of freedom in arranging the task groups is low, and it is difficult to move the component types (component tapes and cassettes) between the front sub-equipment 110 and the rear sub-equipment 120. It was difficult to adjust the balance of mounting time.
(3) Parts are divided for each task group, and a cassette for storing the component tape generated by the parts division is used.Therefore, considering all task groups, there is a tendency that many cassettes to be used for the parts division are required. .

Such issues include the number of component tapes that make up the task group (10 to 19 for a 10 nozzle head) and the number of component tapes that can be placed on the Z-axis (up to 48 single cassettes, double cassettes). Up to 96) on the same order.
Therefore, the degree of freedom in arranging the task groups on the Z axis is low. For example, if the maximum number of component tapes that can be arranged on the Z-axis is about ten times the number of component tapes that make up the task group, it is considered that the degree of freedom is reduced.

3.1.3 "Mowing method"
The “pruning method” is composed of three processes of “part histogram preparation processing” (step S320a in FIG. 14), “pruning processing” (step S320c in FIG. 14), and “core processing” (step S320d in FIG. 14). ing. These processes are devised based on the problem in the “task group generation method”. In the following description, the number of nozzles on the head is n.
(1) Component histogram creation processing (step S320a in FIG. 14)
The component histogram creation process is a process of creating a histogram (component histogram) in which component tapes are arranged in descending order of the number of components, and the component histogram is a prerequisite for the “cutting process”.
In the "task group generation method", component tapes are divided into a plurality of groups called task groups, whereas in the "cutting method", component tapes are grouped into one group called a component histogram.
Since the component histogram can be divided into component tape units and the divided ones can be arranged in the front sub-equipment 110 and the rear sub-equipment 120, compared to the “task group generation method”, component movement in smaller units can be achieved. It is possible.

(2) Mowing process (step S320c in FIG. 14)
The pruning process is a process of generating a suction pattern from a component histogram, and picks up mounting points one by one for each of n component tapes from the side of the component histogram having the smaller remaining number of components, and performs n-point simultaneous suction. It is based on generating a suction pattern.
As a result of the trimming process, there is a component tape in which mounting points that are not sucked remain. This component tape is referred to as “core component tape”. Also, the component cassette containing the core component tape is referred to as a “core cassette”.
The number of core component tapes is always (n-1) or less, no matter how many component tapes make up the initial component histogram.
The advantage of the pruning process is that the problem of performing component division for all component tapes that make up the component histogram and generating an n-point simultaneous suction task , A problem of generating an n-point simultaneous suction task ".
For parts other than the core component tape in the component histogram, n-point simultaneous suction has already been realized, so only for the core component tape, component division may be performed so that an n-point simultaneous suction pattern can be realized. This processing is called “core processing”.

(3) Core processing (Step S320d in FIG. 14)
The “core processing” is an extension of the idea of “creating a component tape that complements the shortage of mounting points in order to realize simultaneous n-point suction” in the “task group generation method”.
Since the number of core component tapes is 1 to (n-1), the number of component tapes (complementary component tapes) that complement the shortage of mounting points is (n-1) to 1.
In the “task group generation method”, a complementary component tape is required for each task group. On the other hand, in the "reaping method", since there is only one group of component tapes, and only a maximum of (n-1) complementary component tapes are required for the group, the method is used more than the "task group generation method". The number of cassettes is small.
In the “task group generation method”, when each component tape is divided into components by the maximum number of divisions, the component tape having the largest number of components is determined, and the same number of suction patterns of n-point simultaneous suction as the number of components are generated.
On the other hand, in the “core processing”, the total number of components of the core component tape is obtained, and the number of suction patterns of n-point simultaneous suction is estimated from a value obtained by dividing the total by n.

3.1.4 Optimization of Small Parts by "Mowing Method" Next, a description will be given of a typical small component optimizing process using the "Mowing Method" having the above advantages.
FIG. 39 is a flowchart of the small part optimizing process (corresponding to step S320 in FIG. 14) by the “reaping method”. One of the indices for optimizing small components is to minimize the number of times the multi-mounting head 112 performs suction up and down operations (hereinafter, referred to as “number of suction up and down operations”) during component suction, and Is to minimize the head moving distance. That is, a component suction pattern that maximizes simultaneous picking of 10 points is determined (S331), and mounting point data that minimizes the moving distance of the multi-mounting head 112 is assigned (S335).

(1) Determination of the component suction pattern (S331 in FIG. 39)
Determining the component suction pattern involves determining the arrangement of component tapes and the order of suction by the multi-mounting head 112, that is, setting the target component tapes in the component supply units 115a and 115b in any order. That is, it is also necessary to determine the order in which the multi-mounting head 112 should be sucked to the set component tape group.
(I) Creation of a component histogram (S332 in FIG. 39)
Each electronic component is sorted by the number of components, and a component histogram is created. The horizontal axis is the arrangement of component tapes (“Z arrangement”), and this histogram represents an image in which components are set in the component supply units 115a and 115b. Since all small parts are 8 mm taping, 10 points can be picked up at the same time. Therefore, whether or not simultaneous suction is possible can be easily determined by looking at the connection of the component histograms in the Z-axis direction. FIG. 40A shows a component histogram 500 in which the number of component tapes is 21, the minimum number of components is 1, and the maximum number of components is 15.
(Ii) Mowing processing (S333 in FIG. 39)
In the component histogram 500 shown in FIG. 40A, a component in which ten consecutive components are arranged in the Z-axis direction from the right end where the number of components is small is searched for. This process corresponds to a process of trimming the component histogram 500 in order from the bottom in units of 10 components, and is referred to herein as a “milling process”. As a result, as shown in FIG. 40B, four 10-point simultaneous suction tasks 500a to 500d are generated.
(Iii) Core processing (S334 in FIG. 39)
When the pruned 10-point simultaneous suction tasks 500a to 500d are removed from the part histogram 500 shown in FIG. 40B, a part histogram 501 having a narrow bottom remains as shown in FIG. The remaining component histogram 501 is called a “core”. Since the spread of the core 501 in the Z-axis direction is less than 10, a 10-point simultaneous suction task cannot be generated from the core 501 as it is. Therefore, “core processing” is performed to cut down the core 501 and generate a 10-point simultaneous suction task.

First, the number of components constituting the core 501 is counted, and a target is set. Since there are a total of 36 mounting points in the component histogram 501 shown in FIG. 41, three 10-point simultaneous suction tasks and one 6-point simultaneous suction task are created.
When trying to mow the core 501 shown in FIG. 41 in units of 10 parts, three types of components are insufficient at the lowest stage of the core 501, and five types of components are insufficient at the second stage from the bottom. In the third stage from the bottom, six types of components are insufficient. When the core 501 is trimmed until the number of trimmed components reaches 36, the pattern 501b shown in FIG. 41 is completed. If a part is assigned to this pattern 501b, a target task can be generated. The number of components included in the pattern 501b is equal to the number of components of the pattern 501a existing above the fifth row of the component histogram. Therefore, the components of the pattern 501a may be cut out for each component tape, and may be vertically embedded in the pattern 501b.

As shown in FIG. 41, since the component 1 has 11 components left in the pattern 501a, it is divided into small vertical bars of 4 + 4 + 2 + 1 and embedded in the pattern 501b in order. If the components 2 and 3 are embedded in the pattern 501b without being divided, the core processing is completed.
When the results of the pruning process and the core process are combined, the component histogram becomes a component histogram 504 in FIG. The component histogram 504 is obtained by combining the task group 503 generated by the pruning process and the task group 502 generated by the core process. The component histogram 504 is an ideal component suction pattern, and all components can be efficiently sucked by seven simultaneous suctions of ten points and one simultaneous suction of six points.

(2) Assignment of mounting point data (S335 in FIG. 39)
Assignment of mounting point data starts with a task that includes components with a small number of components. In the component histogram 504 shown in FIG. 42, mounting points are assigned from tasks including the component 21 having the number of components of one. In task 1, since the number of components of the seven components from component 15 to component 21 is 1, the existing data may be assigned as it is as the mounting point data. Since the number of parts 14 is two, a problem arises in which of the two mounting point data is selected. In this case, after mounting the mounting point of the component 15 already determined, mounting point data that minimizes the movement of the multi-mounting head 112 is selected.

  However, since the component 15 is mounted by the mounting head H4 and the component 14 is mounted by the mounting head H3, the mounting point must be selected in consideration of the offset between the mounting heads H4 and H3. The same concept is used for selecting the mounting point of the component 13. For example, if the component 14a is selected as the mounting point data from the component 14a and the component 14b, the distance from the component 14a is calculated, and the mounting point of the component 13 is selected. Thereafter, similarly, the mounting points of the component 12 are selected, and all the mounting points of the task 1 are determined.

3.1.5 Related Individual Processing The “mowing method” is processing for generating a task (accurately, a suction pattern) from a component type of a component group classified into small components.
The details are as described in the following individual processing.
・ "Mowing method"
-"Small component task generation processing"

3.2 "Intersection Resolution Method"
The “intersection elimination method” is an algorithm that compensates for the disadvantage of the “greedy method”. This processing corresponds to step S324 in FIG.
Hereinafter, the details of the “intersection elimination method” will be described while clarifying the issues in the “greedy method”.

3.2.1 Overview of "greedy method" When assigning mounting points to tasks, mounting points are selected from component types so that the distance between the mounting points mounted by each nozzle is minimized. When calculating the distance, the pitch between nozzles is taken into account.
This mounting point selection method is a method classified as a “greedy method”. This processing corresponds to Step S320e in FIG.
In the "greedy method", even if the distance between the mounting points for a certain task is minimized, the mounting points are not selected in consideration of the distance between the mounting points of other tasks. Is not always optimal.

3.2.2 Problem of "greedy method" When assigning mounting points to a suction pattern by the "greedy method", particularly, the mounting path diagram on the upper side of FIG. 43 (the mounting points arranged at the corresponding board positions for each task) This is a problem in the case of a mounting path as shown in FIG.
FIG. 43 shows a case where there are three tasks with the number of mounting points of five. In FIG. 43, circles indicate mounting points, and arrows indicate mounting paths (order). The subscript of the mounting point indicates the component type. For example, A1, A2, and A3 are three mounting points belonging to the component type A. Also, the mounting points connected by the arrows of the same color constitute one task.
First, in the state “before intersection cancellation” on the upper side of FIG. 43, the mounting point where the component type B1 exists is selected as the mounting point closest to the mounting point where the component type A1 exists, and the mounting point closest to the component type B1 is selected. As a point, a mounting point where the component type C2 exists, instead of the component type C1, is selected. This is because in the "greedy method", the closest mounting point is selected as the next mounting point to be mounted.
Further, when the mounting point is selected by repeatedly applying the “greedy method”, as a result, as shown in the state “before intersection cancellation” in the upper part of FIG. The path connecting the mounting points where the type C1 exists intersects the path connecting the other mounting points.

3.2.3 "Crossing solution"
If a human decides a mounting route, a task in which the mounting paths do not intersect should be created, as in the case of “after intersection is resolved” on the lower side of FIG.
Therefore, after selecting the mounting points by the "greedy method", it is sufficient to find a place where the paths intersect, and to perform a process for eliminating it. This process is the "intersection elimination method".
As a result, the state becomes “after intersection is resolved” on the lower side of FIG. 43, and it can be expected that the total distance of the route is smaller than before the intersection is resolved.
Specifically, in the example of FIG. 43, it is possible to create a task in which the mounting path is shortened by rearranging the mounting path by selecting and exchanging two of the component types B1 to B3 and repeating the selection. it can.
Actually, it is necessary to consider the interval between the nozzles, but it is omitted here because the purpose is to show the concept. The details of the “intersection solving method” are as described in the individual processing described later.

3.2.4 Related Individual Processing The "intersection elimination method" is a process of finding a location where the mounting paths intersect after the selection of mounting points by the "greedy method" and eliminating it. As a result, it is expected that the total distance of the mounting paths will be smaller than before the intersection of the mounting paths is eliminated.

The details are as described in the following individual processing.
・ "Intersection elimination method"

3.3 "Return optimization method"
Hereinafter, the details of the “return optimization method” will be described while clarifying the conception process. This processing corresponds to step S325 in FIG.

3.3.1 Examination of Component Mounting Operation As shown in FIG. 44, the operation of mounting components is divided into the following three steps from a macro viewpoint.
(1) Component adsorption → Component recognition camera
(2) Recognition → component placement
(3) Component mounting → Next component suction ... "Return"

3.3.2 Necessity of optimizing the “return” process First, in the above-mentioned step (1), optimization is performed by arranging the component tape with a large number of components on the Z axis close to the component recognition camera. Be converted to
Next, regarding the above (2), it is considered that the distance is almost constant, and is not an object of optimization. Because the position of the component recognition camera and the board are fixed, the amount of movement of the head on the board at the time of mounting is considerably smaller than the length of the Z axis, and if all mounting points are at the center of the board Because it is possible.
However, with respect to the above (3), the “return” process is almost the same as the distance of the above (2), and optimization is possible. That is, by optimizing this process, a reduction in mounting time can be expected.

3.3.3 "Return optimization method"
We have devised an optimization algorithm for the "return" process in (3) above.
The basic idea of this optimization algorithm is to "find the unimplemented task at the position on the Z axis that can return in the shortest distance from the coordinates of the final mounting point of a task, and then implement it next It is a task. " For example, in the figure, when examining the distance from the final mounting point, task B is shorter than task A, and therefore the task to be mounted next is task B.

3.3.4 Related Individual Processing The operation of mounting components is broken down into the following three steps from a macro perspective. (1) Component adsorption → Component recognition camera
(2) Recognition → component placement
(3) Component mounting → Next component suction ... "Return"
The “return optimization method” optimizes the moving distance of the head in (3), and can reduce the mounting time.
The details are as described in the following individual processing.
・ "Return optimization method"

3.4 Arrangement Fixing Process 3.4.1 Outline In some cases, a user designates a Z number for arranging a plurality of component types. This specifies the arrangement of the component types on the Z axis, and is called “fixed arrangement”.
On the other hand, in the optimization algorithm, the arrangement of the component tapes on the Z axis is also an object to be optimized, and therefore, it is necessary to realize an optimization algorithm that takes into account the arrangement fixed by the user.
However, it is considered that the variation in fixing the arrangement by the user becomes very large.

Even if some fixed sequence variations are assumed at the algorithm design stage and an optimization algorithm corresponding to them can be devised, it is not always possible to cope with unexpected fixed sequence variations. This is because the algorithm tends to be specialized in an assumed variation of array fixing, and there is a risk that an unexpected array fixing has no effect.
Furthermore, even if the algorithm can be modified to correspond to an unexpected fixed sequence variation, the addition of an exceptional algorithm will reduce the readability of the program and may cause problems in maintenance. .
Therefore, as shown in FIG. 45, the following method is adopted as the most reliable and safe method. FIG. 45 is a component histogram showing an outline of optimization under the constraint of fixed array.
(1) Prepare a temporary Z-axis (temporary Z-axis) and optimize the arrangement of component types on the temporary Z-axis without considering the arrangement fixation. That is, an ideal component tape array is created (a component histogram is created with priority given to simultaneous suction).
(2) Move the component tape from the temporary Z axis to the actual Z axis (actual Z axis). At this time, in consideration of the arrangement, the component tapes to be arranged are arranged first.
(3) Next, the component tape not to be fixed in arrangement is moved from the temporary Z axis to the actual Z axis. At this time, component tapes that are not to be arrayed and fixed are arranged in the gaps between the arrayed and fixed component tapes.

Finally, a suction pattern is generated from the component tapes on the actual Z-axis by a cutting process.
According to this method, a single algorithm can cope with any arrangement fixed by the user.
Further, the algorithm for fixing the arrangement devised this time corresponds to the arrangement fixation specified by the user by breaking the ideal arrangement of component tapes generated by the algorithm under the condition that there is no arrangement fixation.
Therefore, it is possible to compare the mounting time between the case of using the ideal component tape arrangement and the case of using the component tape arrangement with fixed arrangement.
This provides the user with information to compare the operational advantage of the fixed array with the ease of model switching and the short implementation time without the fixed array, and to review their trade-offs. To do.

3.4.2 Related Individual Processing In the “fixed array”, the user designates a Z number for arranging a plurality of component tapes. Therefore, in the optimization algorithm, the arrangement of component tapes on the Z-axis is also an object to be optimized, and therefore, it is necessary to realize an optimization algorithm that takes into consideration the arrangement fixed by the user.

The array fixing algorithm devised this time corresponds to the user-specified array fixing by breaking the ideal component tape array generated by the algorithm under the condition that there is no array fixing.
The details are as described in the following individual processing.
・ "Overall flow (start from histogram)"
・ "Relationship between fixed parts and" mountain "in cassette block"
-"Fixed array: Judgment of availability of fixed destination"
・ "About double cassette alignment"
・ "Fixed array of double cassettes (supplement)"

3.5 Support for LL size board 3.5.1. Overview
The LL size board is a board that is larger in the transport direction than a normal board that has no restrictions on the mounting area. Therefore, as shown in FIG. 46, the LL size substrate has a mounting area (LL restriction area) in which components can be mounted only by a specific head (nozzle).

In addition, these heads cannot pick up components from component tapes (cassettes) arranged in a certain range of Z numbers.
Therefore, as shown in FIG. 47, the LL constraint is avoided by the following two methods.
(1) Replacement of component tape on Z axis
(2) Change of suction method The process of (1) above is a process of arranging the component tape including the mounting point of the LL restriction area in the range of the Z axis that can be suctioned by the head capable of mounting the mounting point of the LL restriction area. . When component tapes are arranged at all Z numbers on the Z axis, component tapes are replaced.
In the process (2), first, the component histogram including the mounting points existing in the LL restriction area is virtually divided into the following two component histograms.
・ Parts histogram composed of mounting points existing in the LL restricted area ・ Parts histogram composed of mounting points not present in the LL restricted area The results are combined into one task.

3.5.2 Replacement of component tape on Z axis
(1) The heads 1 to 6 cannot mount components in the LL restricted area.
(2) The heads 7 to 10 can mount components also in the LL restriction area.
(3) Due to mechanical limitations, the range of Z that can be attracted to each head is limited.
(4) If there is a component tape having the mounting point of the LL restriction area in Z = 1 to 11, the component tape is present in the Z = 12 or higher and does not have the mounting point of the LL restriction area. Replace with

3.5.3 Change of adsorption method
(1) The mounting points in each Z are divided into “mounting points in the LL restriction area” and “mounting points not in the LL restriction area”. It only separates the handling after processing and does not divide the parts.
(2) Divide the head into heads 1 to 6 and heads 7 to 10, and virtually consider two heads, 6 heads and 4 heads.
(3) For mounting points that are not in the LL constraint area, pruning is performed with six heads, and a six-point task is created.
(4) For the mounting points in the LL constraint area, pruning is performed with four heads, and a four-point task is created.
(5) A 6-point task and a 4-point task are combined into a 10-point task.

3.5.4 Related individual processing
In order to support LL size substrates, it is necessary to change the suction method and replace the component tapes on the Z axis, and for each of them, we created two algorithms.
The details are as described in the following individual processing.
・ "LL constraint: Change of adsorption method (1)"
In order to support the LL size board, the mounting points are divided into those that exist in the LL restricted area and those that do not exist. Is absorbed by the heads 1 to 6.
The components are sequentially sucked from the component tape on the base side of the “mountain”, but in the case of the left block, from the Z with a larger Z number than the Z range sucked by the heads 1 to 6, that is, in the direction approaching the camera The head is sucked by the heads 1 to 4 while moving. The same applies to the right block.

・ "LL constraint: Change of adsorption method (2)"
After the mounting points that do not exist in the LL restriction area are sequentially sucked by the heads 1 to 6 from the component tape on the base side of the “mountain”, the mounting points that similarly exist in the LL restriction area are also referred to as “mounting points”. The heads 1 to 6 sequentially pick up the components from the component tape on the base side of "."
That is, unlike “LL restriction: change of suction method (1)”, suction is not always performed while moving in the direction approaching the camera direction.
・ "LL constraint: Replacement of component tape on Z axis (1)"
For component tapes with Z numbers 1 to 11, search for component tapes that include mounting points with X coordinates greater than 400 mm and replace them with component tapes that do not include mounting points with X coordinates greater than 400 mm.
・ "LL constraint: Replacement of component tape on Z axis (2)"
The handling of the X coordinate of the mounting point is made finer than the above “LL constraint: replacement of component tapes on the Z axis (1)”.

3.6 Support for XL size board 3.6.1 Overview
The XL size board is a board that is larger in size in the direction orthogonal to the transport direction than a normal board that has no restrictions on the mounting area. Therefore, as shown in FIG. 46, the XL-size board has a mounting area in which components can be mounted only in specific equipment (the front sub-equipment 110 or the rear sub-equipment 120). FIG. 46 is a diagram showing a restricted area on the board of a special size (XL, LL) (an area that cannot be mounted because the head cannot move).

The XL size board is composed of the following three mounting areas.
・ Area where parts can be mounted only in front sub-equipment 110 ・ Area where parts can be mounted only in rear sub-equipment 120 ・ Area where parts can be mounted in both front sub-equipment 110 and rear sub-equipment 120 Further identification as well as LL size board There are regions where components can be mounted only with the head (nozzle).
Based on the restrictions shown in FIG. 46, the correspondence to the XL size substrate will be described below.
(1) Assignment to front sub-equipment 110 / rear sub-equipment 120 by mounting point coordinates
(2) Component division by mounting point coordinates
(3) Initial distribution using an area that can be mounted in both the front sub facility 110 and the rear sub facility 120
(4) Avoidance of LL constraint Details are as described in the individual processing described later.

3.6.2 Related individual processing
Judgment for the XL size board is made by determining whether the sub-equipment that can be mounted for each mounting point is the front sub-equipment 110 or the rear sub-equipment 120, and assigning the mounting points to the front sub-equipment 110 and the rear sub-equipment 120. By dividing, it was realized.

Note that the restrictions of the XL-size substrate include the restrictions of the LL-size substrate, so the processing corresponding to the XL-size substrate includes the processing corresponding to the LL-size substrate.
The details are as described in the following individual processing.
・ "XL constraint"

3.7 Load Balance Processing 3.7.1 Overview Load balance processing is processing for adjusting the balance between the front sub-equipment 110 and the rear sub-equipment 120 using the load level as an index in the initial distribution processing. This processing corresponds to Step S314b in FIG.
3.7.2 Level of Balance Adjustment Method As a method of adjusting the balance, parts are moved between the front sub-equipment 110 and the rear sub-equipment 120. There are two levels for moving parts.
(1) `` Mountain '' unit
(2) Component tape unit The `` mountain '' means a group of component tapes created as a result of optimization, that is, a component tape group arranged in a certain order or a component histogram corresponding to the component tape group. I do. It is close to the concept of the task group described above.

The level of component movement to be executed differs between the load balance processing and the line balance processing.
Load balance processing: "mountain", component tape Line balance processing: "mountain", component tape, mounting point The load level calculation currently used in the load balance processing is the value of the load level for tasks composed of general-purpose components Is not accurate. Therefore, in the load balancing process, it is determined that the effect is small even if the unit of the component movement is made fine, and the component movement is not performed in the unit of the mounting point.
The details are as described in the individual processing described later.

3.7.3 Related Individual Processing “Load balancing processing” is processing for adjusting the balance between the load levels of the front sub-equipment 110 and the rear sub-equipment 120. This is necessary when the component tape is distributed to the front sub-equipment 110 and the rear sub-equipment 120.

First, component tapes are packed and arranged in order from the front sub-equipment 110, and component tapes that cannot be arranged in the front sub-equipment 110 are arranged in the rear sub-equipment 120.
With this as an initial state, the load balance between the front sub-equipment 110 and the rear sub-equipment 120 is calculated, and the component tapes arranged in the front sub-equipment 110 are sequentially ordered until the difference in the load balance becomes OK. Go to
The method of calculating the load balance of each sub-equipment is as described in “Operation of the optimization device (outline)”.
The details are as described in the following individual processing.
・ "Load level balance adjustment (" mountain "unit)"
・ "Load level balance adjustment (part tape unit)"

3.8 Line balance processing 3.8.1 Overview Line balance processing is processing for adjusting the balance between the front sub-equipment 110 and the rear sub-equipment 120 using a mounting time as an index after a task is generated. This processing corresponds to step S323 in FIG. The only difference between the line balance processing and the load balance processing is that the balance indices are different, and the processing is similar to each other.
3.8.2 Level of Balance Adjustment Method As a method of adjusting the balance, parts are moved between the front sub-equipment 110 and the rear sub-equipment 120. There are three levels for moving parts.
(1) `` Mountain '' unit
(2) Component tape unit
(3) Mounting point unit The line balancing process differs from the load balancing process in that components are moved in mounting point units.

3.8.3 Related Individual Processing “Line balance processing” is processing for adjusting the balance between the mounting times of the front sub-equipment 110 and the rear sub-equipment 120. After generating tasks for the front sub-equipment 110 and the rear sub-equipment 120, calculate the mounting time of each sub-equipment with a tact simulator and move the parts from the sub-equipment with a long mounting time to the sub-equipment with a short mounting time Thereby, the balance of the mounting time of the front sub-equipment 110 and the rear sub-equipment 120 is adjusted. Although there is a difference in the balance index and the like, the processing is similar to the above-described load balancing processing.

The details are as described in the following individual processing.
"Process of moving a mountain from front sub-equipment 110 to rear sub-equipment 120"
"Process of moving component tape from front sub-equipment 110 to rear sub-equipment 120"
"Process of moving mounting point from front sub-equipment 110 to rear sub-equipment 120"
・ "Swap processing in line balance processing"

3.9. Details of individual processing by the optimizer 3.9.1 "Pruning method"
Tasks are generated by the following procedure.
(1) Create a part histogram 510 (FIG. 48).
(2) Pruning is performed on the component histogram 510 to leave a core portion (FIG. 49).
In this drawing, a frame surrounded by a square is a suction pattern of simultaneous suction of 10 points.
(3) The cut portion 511a (FIG. 50 (a)) and the core portion 511b (FIG. 50 (b)) are separated.
(4) The template 512 is allocated to the core portion 511b (FIG. 51).
In this drawing, black squares (mounting points) surrounded by a square frame indicate mounting points that could not be covered by the template, and these were used to complement the left side 513 of the template (points indicated by “*”). use.
(5) The mounting point 514 that complements the left side of the template is determined (FIG. 52).
(6) Complement the left side 513 of the template (FIG. 53).
In this figure, white squares indicate mounting points used for complementation, black squares surrounded by square frames indicate mounting points not used for complementation, and “*” surrounded by square frames. Indicates an implementation point that could not be complemented.
(7) The “mountain” 515 is recreated for the core part and the part complemented by the template (FIG. 54).
(8) The task 511a created by pruning in the above (2) is also recreated in the form of a “mountain” 516 (FIG. 55).
(9) The “mountain” 516 obtained by cutting and the “mountain” 515 of the core portion are combined to obtain the “mountain” 517 (FIG. 56).
(10) The entire “mountain” 517 is trimmed to obtain a suction pattern 518 (FIG. 57).

In this drawing, the twenty-fourth task (task 24) shows that the number of times the head moves up and down during suction (the number of times the head moves up and down) is three.
(11) If there is no restriction at all, it is arranged on the Z axis as it is (FIG. 58).
The case where the constraint is considered is as follows (from (12)).
(12) A task (a group of components surrounded by a square frame) is generated by the “reaping method” (FIG. 59).
Here, processing of the core portion is performed. However, at this stage, the maximum number of divisions, cassette resources, and the number of usable Z numbers are not considered.
In this example, since cassette numbers 1 to 6 are divided,
Cassette number = 1: Part A
Cassette number = 2: Part B
Cassette number = 3: Part C
Cassette number = 4: Part D
Cassette number = 5: Part E
Cassette number = 6: Part F
The state of division is, for example, since the part A is divided into five parts, A1, A2, A3, A4, A5
Is expressed as The same applies to the parts B, C, D, E, and F. Other parts are represented by black squares.
(13) Consider the maximum number of divisions and optimize the number of cassette divisions (FIG. 60).
Here, assuming that the maximum cassette division number of the component A is 4, the cassette division number of the component A is optimized.
Since the part A is divided into five parts, one of A2 to A5 is integrated into A1 to A5. At this time, if a component having the minimum number of components is selected from A2 to A5, the number of tasks affected by the integration is minimized.
In this case, since the number of parts of A5 is the minimum (three), A5 is selected and those A5 are distributed to A1 to A4. As a result, since the position where A5 is located is vacant, F2, E2 and D2 on the left side of A5 are moved right by one.
(14) The cassette arrangement after such optimization is the suction pattern 518b shown in FIG.
Here, in the tasks 21 to 22, the number of times of up and down suction is two.
(15) Then, as shown in a diagram 518c, the number of cassettes used is optimized (FIG. 62).
Here, it is assumed that the number of cassettes used is one more than the cassette resource.
Among the components A2 to 4, B2, C2, D2, E2, and F2, those having the minimum number of components are selected and integrated. Specifically, F2 with the minimum number of parts (one piece) is selected and integrated into F1.
(16) The cassette arrangement after such optimization is as shown in the suction pattern 518d in FIG.
It can be seen that the number of cassettes has decreased by one.
(17) Then, as shown in a diagram 518e, the number of occupied Z-axes is optimized, that is, the range of usable Z-axis is considered (FIG. 64).
Here, it is assumed that the number of Z axis used is one more than the Z axis empty.
Among the components A2 to 4, B2, C2, D2, and E2, those having the minimum number of components are selected and integrated. Specifically, E2 having the minimum number of components (two) is selected and integrated into E1.
(18) The cassette arrangement after such optimization is as shown in the suction pattern 518f shown in FIG.
Here, the task 24 has the same number of times of up-and-down suction, which is still four, but the task 23 has three times of up-and-down suction.
(19) Arrange on the Z axis.
Here, as shown in diagram 518g, it is assumed that B1 is a component that is originally fixed to Z number = 15 (FIG. 66).
(20) First, the fixed cassette 519 is arranged on the Z axis (FIG. 67).
(21) The non-fixed cassette is arranged on the Z axis. The result is a suction pattern 520 (FIG. 68).
At this time, the non-fixed cassettes are arranged on the Z axis in the cassette arrangement order determined in (19) above so as to avoid the fixed cassettes.
(22) Return to the shape of “mountain” 521 (FIG. 69).
(23) The task is generated again by the “pruning method”, and the suction pattern 522 is obtained (FIG. 70).
However, processing of the core portion is not performed. Here, the task 24 has three times of ups and downs, the tasks 22 to 23 have two times of ups and downs, and the tasks 17 to 19 have two times of ups and downs.

3.9.2 Cassette division by parallelogram The method of cassette division using a parallelogram template for the core component is as follows.
(1) Here, it is assumed that the total number of target core components 525 is 30 (the upper part of FIG. 71). That is, three tasks of the ten-point suction are created.
(2) First, since the number of cassettes is 9, a parallelogram (template) 523 corresponding to the number of cassettes is created (right in the middle part of FIG. 71). Note that the right end of each step of the parallelogram 526 is a character (A to I) indicating the type of a component when a component is assigned to the parallelogram 526 in the case of 10-point cut × 9. .
(3) Attention is paid to the first stage (lowest stage) 525a of the target component 525, and since the right end is “I”, this is replaced with the stage of the parallelogram 526 (here, the right end is the same character (“I”)). , At the bottom of the parallelogram 526 (at the bottom of FIG. 71).
(4) Similarly, paying attention to the second stage 525b of the target component 525, since the right end is “F”, this is replaced with the stage of the parallelogram 526 (here, the right end is the same character (“F”)). (The fourth row of the parallelogram 526) (the upper row in FIG. 72).
(5) Similarly, paying attention to the third stage 525c of the target part 525, since the right end is “C”, this is replaced with the stage of the parallelogram 526 (here, the right end is the same character (“C”)). (7th stage of the parallelogram 526) (the middle stage in FIG. 72).
(6) Since there is no more stage where the rightmost character matches, the remaining component 525c is arranged at an empty position (“X”) in each of the arranged stages (first, fourth, and seventh stages) (FIG. 72 lower).
(7) At that time, the components 525e and 525f having the larger number of remaining components are allocated (upper and middle rows in FIG. 73).
(8) If the number of remaining components is the same, the component 525g is allocated in the order of the character of the component (lower part in FIG. 73).
(9) According to the above rules, all the remaining components 525h to 525k are placed on the template 526 (the upper part of FIGS. 74 and 75).
(10) As a result of placing all the components on the template 526, the first, fourth and seventh stages of the template 526 are filled with the components (the middle stage of FIG. 75). , The cassette division is completed (the lower part of FIG. 75).

3.9.3 Cassette division by rectangle The method of cassette division using a rectangular template for the core component is as follows.
(1) A rectangular template (here, a template having a width of 10 and a height of 3) 528 is applied to the lower part of the 30 target core components 527 (the upper part of FIG. 76).
(2) An area to be complemented (open square) 528a is arranged on the left side of the complemented area (middle of FIG. 76).
(3) The components 527a and 527b having a large number of remaining components are placed in the complementary area 528a of the template (the lower part of FIG. 76 and the upper part of FIG. 77).
(4) If the number of remaining components is the same, the components 527c are allocated in the order of the characters of the components (the middle part of FIG. 77).
(5) According to the above rules, all remaining components 527d to 527d are placed on the template 528a (the lower part of FIG. 77, FIGS. 78 and 79). When the arrangement is completed, the cassette division is completed.

3.9.4 Core Processing Method with a Given Number of Cassettes Basic core processing is performed to create an ideal “mountain” shape, and then a complementary cassette is compressed and stored in a given cassette resource.
At the time of performing the core processing, a process of allocating the number of components to the complementary cassette is performed so that a given number of complementary cassettes can be formed, and finally, the type of the component remaining in the core portion and originally used for component division ( It is considered that a method in which the number of parts of the component tape) is evenly distributed to the same component type is also possible.
As for the double cassette, since the core remains on the odd-numbered Z number, a complementary cassette can be made in the same manner as in the core processing of the single cassette. In this case, the complementary cassette uses only the odd side (for the odd Z number) of the double cassette. The processing for compressing the cassette may be performed in the same manner as for the single cassette.
In particular,
(1) Core processing is performed on the core to create an ideal “mountain”.
At this time, a complementary cassette is made.
(2) Obtain the number N of complementary cassettes.
(3) The number N of complementary cassettes is compared with the given number M of cassettes.
(4) If N ≦ M, the process ends.
The return value is N.
In the core processing, there is a case where it is not necessary to use the given number of cassettes, so N is used as the return value.
Since the maximum number of complementary cassettes is nine, it is meaningless to provide more than ten cassettes.
The return value N manages the cassette resources.
(5) If N> M, compress one cassette.
(5.1) The cassette C having the smallest number of parts is searched from the “mountain”.
(5.2) A cassette D having the same component tape as that of the cassette C is searched from the “mountain”.
There may be a plurality of cassettes D.
Cassette C is not included in cassette D.
(5.3) Evenly distribute the number of components of the cassette C to the cassette D.
If the distribution is not even, the number of parts in the cassette is increased toward the core of the “mountain”.
For example, if the number of components of the cassette C is 5 and there are three cassettes D, the cassette C is divided into 2, 2, and 1 and distributed to 2, 2, and 1 in order from the cassette on the core side of “mountain”.
(6) Subtract 1 from the number N of complementary cassettes.
(7) Return to (3).

3.9.5 Task generation processing of small parts The correspondence between nozzle numbers and Z numbers is determined, and processing for generating a suction pattern for each task is performed.
The correspondence between the nozzle and the mounting point is determined by the "greedy method".
A suction pattern is generated by scanning from the “base” side of “mountain”. Therefore, the scanning direction of the head and the Z-axis are opposite in the left block in which “the base” exists on the side with the smaller Z number and the right block in which the “base” exists on the side with the larger Z number. Is the same process.
In the case of the double cassette, the number of all components of the component tape on the even Z number side is allocated to the suction pattern, and then the number of components of the component tape on the odd Z number side is allocated to the suction pattern. If the number of suction points of the last task created from the component tapes existing on the even Z number side is less than 10, the less than 10 is adsorbed from the component tapes existing on the odd Z number.
・ Points in programming In the processing described below, in order to determine whether the component tape placed on the real Z axis is the component tape to be sucked, whether the component tape belongs to the “mountain” to be processed Has been determined. Therefore, it is convenient to prepare information for identifying a “mountain” such as a “mountain” number as an attribute of the component tape and to set the information in advance. Since two or more “mountains” may be created from one component group, it is better not to use the component group number for identifying “mountains”.
In case of left block ("mountain" of single cassette)
(8) Set 1 to the task number t.
(9) The total number of mounting points belonging to the component tapes constituting the “mountain” is obtained, and the total number of mounting points is determined.
(9.1) If the total number of mounting points is zero, perform the following processing.
(9.1.1) Proceed to (15).
Since there is no “mountain” with no mounting points, it is considered an error.
(10) Among the nozzles of the task with the task number t, the nozzle with the smallest nozzle number is found from the nozzles not associated with the Z number, and the nozzle number is set to Nvac.
The nozzle numbers are 1 to 10.
If there is no nozzle associated with the Z number, Nvac is 1.
(10.1) When a Z number is associated with all nozzles, the following processing is performed.
(10.1.1) Proceed to (13).
The process will proceed to the generation of the suction pattern of the next task.
The number of suction points for this task is 10.
(11) With respect to the Z number at which the component tape constituting the “mountain” exists, the smallest Z number is obtained from the Z numbers that can be suctioned by the nozzle of the nozzle number Nvac, and is set as Zvac.
The Z number is “odd number in the range of 1 to 48” in the case of the front sub-equipment 110.
The Z number is “odd number in the range of 97 to 144” for the rear sub-equipment 120.
(11.1) If no such Z number is found, perform the following processing
(11.1.1) Proceed to (13).
The process proceeds to the generation of the suction pattern of the next task, and this task has the number of suction points of less than 10.
For example, when the component tape exists only at Z = 1, the suction can be performed only by the nozzle 1 and there is no component tape that can be suctioned by the nozzles 2 to 10, so that the Zvac cannot be determined.
(12) If the total number of mounting points is positive and Nvac is 10 or less, the following processing is performed.
(12.1) When the Z number is not associated with the nozzle whose nozzle number is Nvac, and the component tape existing in Zvac belongs to this “mountain”, the following processing is performed.
(12.1.1) Zvac is associated with the nozzle whose nozzle number is Nvac.
(12.1.2) One is subtracted from the number of mounting points of the component tape existing in the Zvac.
(12.1.3) 1 is subtracted from the total number of mounting points.
For example, in the case of suctioning in the form of “missing teeth” in the first suction, in the second suction, the immediately adjacent nozzle is not necessarily empty, so this condition determination (first half) was included.
In addition, since there is a possibility that a component tape (such as a component tape of a fixed cassette) irrelevant to the “mountain” may be present in the middle of the “mountain”, the condition determination (the latter half) is included.
(12.2) One is added to Nvac.
(12.3) Add 2 to Zvac.
(12.4) Return to (12).
(13) Add 1 to the task number.
(14) Return to (10).
(15) The suction pattern generation processing ends.
In case of right block ("mountain" of single cassette)
(16) Set 1 to the task number t.
(17) The total number of mounting points belonging to the component tapes constituting this “mountain” is determined, and the total is determined as the total number of mounting points.
(17.1) If the total number of mounting points is zero, perform the following processing.
(17.1.1) Proceed to (23).
Since there is no “mountain” with no mounting points, it is considered an error.
(18) Among the nozzles of the task with the task number t, the nozzle with the largest nozzle number is found from the nozzles not associated with the Z number, and the nozzle number is set to Nvac.
The nozzle numbers are 1 to 10.
If there is no nozzle associated with the Z number, Nvac is 10.
(18.1) If all nozzles are associated with a Z number, the following processing is performed.
(18.1.1) Proceed to (21).
The process will proceed to the generation of the suction pattern of the next task.
The number of suction points for this task is 10.
(19) With respect to the Z number in which the component tape forming the “mountain” exists, the largest Z number is obtained from the Z numbers that can be suctioned by the nozzle of the nozzle number Nvac, and is set as Zvac.
The Z number is “odd number in the range of 49 to 96” for the front sub-equipment 110.
The Z number is “odd number in the range of 145 to 192” for the rear sub-equipment 120.
(19.1) If no such Z number is found, perform the following processing
(19.1.1) Proceed to (21).
The process proceeds to the generation of the suction pattern of the next task, and this task has the number of suction points of less than 10.
For example, when the component tape exists only at Z = 1, the suction can be performed only by the nozzle 1 and there is no component tape that can be suctioned by the nozzles 2 to 10, so that the Zvac cannot be determined.

(20) When the total number of mounting points is positive and Nvac is 1 or more, the following processing is performed.
(20.1) When the Z number is not associated with the nozzle having the nozzle number of Nvac and the component tape existing in Zvac belongs to this “mountain”, the following processing is performed.
(20.1.1) The nozzle with the nozzle number Nvac is associated with Zvac.
(20.1.2) One is subtracted from the number of mounting points of the component tape existing in the Zvac.
(20.1.3) Subtract 1 from the total number of mounting points.
For example, in the case of suctioning in the form of “missing teeth” in the first suction, in the second suction, the immediately adjacent nozzle is not necessarily empty, so this condition determination (first half) was included.
In addition, since there is a possibility that a component tape (such as a component tape of a fixed cassette) irrelevant to the “mountain” may be present in the middle of the “mountain”, the condition determination (the latter half) is included.
(20.2) Subtract 1 from Nvac.
(20.3) Subtract 2 from Zvac.
(20.4) Return to (20).
(21) Add 1 to the task number.
(22) Return to (18).
(23) The suction pattern generation process ends.
Left block ("mountain" of double cassette)
(24) Suction is performed on the even Z number side of the double cassette in the same manner as in the above-described case of the left block (“mountain” of the single cassette).
However, the only difference is that the suction operation is performed not from the odd Z number but from the even Z number.
(25) If the last task among the tasks sucked from the even Z number side of the double cassette is a task with less than 10 points, the task is set to the initial value of the task when sucking the odd Z number side of the double cassette. I do.
In the last task, suction is performed in order from nozzle 1, and the side with the larger nozzle number is empty. If this is used as the initial value of the task when sucking the odd-numbered Z-number side as it is, for example, the suction from around Z = 1 cannot be performed. Therefore, the mounting point that has already been sucked is moved to the nozzle with the larger nozzle number so that the nozzle with the smaller nozzle number becomes empty.
(26) Suction is performed on the odd Z number side of the double cassette in the same manner as in the case of the above “right block (“ mountain ”of a single cassette)”.
However, the only difference is that the suction operation is performed not from the odd Z number but from the even Z number.
If the number of suction points of the last task is less than 10 as a result of sucking the Z number on the even side of the double cassette, the difference is that the first task of suction on the odd Z number side of the double cassette has the initial value. .
Right block ("mountain" of double cassette)
(27) Suction is performed on the even-numbered Z-number side of the double cassette in the same manner as in the case of “right block (“ mountain ”of single cassette)” described above.
However, the only difference is that the suction operation is performed not from the odd Z number but from the even Z number.
(28) If the last task among the tasks sucked from the even Z number side of the double cassette is a task with less than 10 points, the task is set to the initial value of the task when sucking the odd Z number side of the double cassette. I do.
In the last task, suction is performed in order from the nozzle 10, and the side with the smaller nozzle number is empty. If this is used as the initial value of the task when the odd Z number side is sucked as it is, for example, the suction from around Z = 96 cannot be performed. Therefore, the mounting point that has already been sucked is moved to the nozzle with the smaller nozzle number so that the nozzle with the larger nozzle number becomes empty.
(29) Suction is performed on the odd Z number side of the double cassette in the same manner as in the above-mentioned “in the case of the right block (“ mountain ”of the single cassette)”.
However, the only difference is that the suction operation is performed not from the odd Z number but from the even Z number.
If the number of suction points of the last task is less than 10 as a result of sucking the Z number on the even side of the double cassette, the difference is that the first task of suction on the odd Z number side of the double cassette has the initial value. .

3.9.6 "Intersection Resolution Method"
In the "intersection eliminating method", optimization of mounting point allocation is performed after all suction patterns are determined, mounting points are allocated to each suction pattern using the "greedy method" (+ HC method), etc., and provisional tasks are determined. This is one of the algorithms that perform

FIG. 80 (a) shows a mounting path diagram (mounting path diagram determined by the greedy method) 530a before applying the intersection elimination method, and FIG. 80 (b) shows a mounting path diagram after applying the intersection elimination method. 530b. As shown in this figure, this algorithm reduces the useless crossing points of the moving trajectory of the head.
If the mounting point of the task to be processed is caught by the head limitation on the LL board or XL board, the message `` All mounting points of the partial task to be rearranged have a head number satisfying head1 = head2. Only in the case of "." In other cases, if the "intersection elimination method" is enforced, the head may not reach the mounting point with an extremely high probability.
FIG. 81 (a) is a mounting path diagram for explaining an algorithm of the intersection resolving method, and FIG. 81 (b) is a diagram showing an example of one intersection (intersection by line A and line B) by four mounting points. It is. The specific algorithm is as follows.
(0) Calculate the amount of movement of the head when hitting the mounting point of each task, and find the sum for all tasks
(1) 1 is substituted for the Z coordinate and the cut point at which the mounting point is rearranged.
(2) Assign 1 to task 1 to be rearranged (task1 = 1)
(3) Assign (task1 + 1) to task 2 to be reassigned (task2 = task1 + 1). (4) Find the head number (head1, head2) corresponding to the cutpoint for each task
(5) Are the two head numbers both appropriate?
(5.1) If it is not appropriate (there is no mounting point corresponding to the specified Z coordinate), go to (13)
(5.2) If appropriate, go to (6)
(6) Calculate the amount of movement of the head when hitting the mounting point of each task, and find the sum (olength)
(7) Rearrange the partial tasks on the left side of the cutpoint
(8) Calculate the amount of movement of the head when hitting the mounting point of each task, and find the sum (nlengthL)
(9) Rearrange the partial tasks on the right side of the cutpoint
(10) Calculate the amount of movement of the head when hitting the mounting point of each task, and find the sum (nlengthR)
(11) Compare the three movement amounts, olength, nlengthL, nlengthR, and find the smallest one. (12) Adopt the task that gives the smallest movement amount as a new task
(13) Increment task 2 (task2 = task2 + 1)
(14) Compare task 2 with the number of tasks
(14.1) If task 2 does not exceed the number of tasks, return to (4)
(14.2) In other cases, go to (15)
(15) Increment task 1 (task1 = task1 + 1)
(16) Compare task 1 with the number of tasks
(16.1) If task 1 does not exceed the number of tasks, return to (3)
(16.2) In other cases, go to (17)
(17) Increment the cut point (coutpoint = cutpoint + 1)
(18) Compare cutting point with maximum Z coordinate
(18.1) If the cutting point does not exceed the maximum Z coordinate, go to (2)
(18.2) In other cases, go to (19)
(19) Calculate the amount of head movement when hitting the mounting point of each task, and find the sum of the amount of head movement for all tasks
(20) Check whether the total amount of head movement has decreased
(20.1) If decreasing, go to (0)
(20.2) Termination in Other Cases FIG. 82 is a mounting path diagram showing an application example of the intersection elimination method using such an algorithm, and FIG. FIG. 531a, and FIG. 82 (b) shows a mounting path diagram 531b after application. It can be seen that the number of locations where the mounting paths cross is reduced and the total mounting path is shortened.

3.9.7 "Return optimization method"
The “return optimization method” is an algorithm that optimizes the mounting order of tasks after the allocation of mounting points to all tasks is determined.

The details are as follows.
(1) Find the X coordinate of the final mounting point of each task
(2) Create a task number list (up []) arranged in descending order of the X coordinate of the final mounting point
(3) Find the maximum Z coordinate of the component tape for each task (the maximum value of the Z coordinate taken by head No. 10 during suction)
(4) Create a task number list (point []. Task) arranged in descending order of the maximum Z coordinate
(5) Assign the task with the largest X coordinate of the final mounting point to the mounting order 1
(6) As the next task to be implemented, assign the task with the largest maximum Z coordinate among the remaining tasks.
(7) Are there any tasks whose mounting order is not determined?
(7.1) If remaining, go to (8)
(7.2) In other cases, go to (10)
(8) Of the remaining tasks, assign the one with the largest X coordinate of the final mounting point as the task to be mounted next
(9) Are there any tasks whose mounting order is not determined?
(9.1) If remaining, go to (6)
(9.2) In other cases, go to (9)
(10) Calculate the amount of head movement when hitting the mounting point of each task, and find the sum of the amount of head movement for all tasks
(11) Assign 1 to task 1 that changes the mounting order (task1 = 1)
(12) Assign (task1 + 1) to task 2 for which the mounting order is to be changed (task2 = task1 + 1)
(13) Calculate the amount of movement of the head when hitting the mounting point of each task, and find the total (olength) for all tasks
(14) The task to be implemented next to task 1 is implemented next to task 2, and the task to be implemented next to task 2 is implemented next to task 1 to determine the implementation order of a new task.
(15) Calculate the amount of movement of the head when hitting the mounting point of each task, and find the total (nlength) for all tasks
(16) Compare the two movement amounts, olength and nlength, and find the minimum one
(17) Adopt the mounting order that gives the minimum movement as the new mounting order
(18) Increment task 2 (task2 = task2 + 1)
(19) Compare task 2 with the number of tasks
(19.1) If task 2 does not exceed the number of tasks, return to (11)
(19.2) In other cases, go to (19)
(20) Increment task 1 (task1 = task1 + 1)
(21) Compare the number of tasks with task 1 (task1)
(21.1) If task 1 does not exceed the number of tasks, return to (10)
(21.2) In other cases, go to (21)
(22) Calculate the amount of head movement when hitting the mounting point of each task, and find the sum of the amount of head movement for all tasks
(23) Check whether the total amount of head movement has decreased
(23.1) If decreasing, go to (0)
(23.2) In other cases, termination As described above, this algorithm is largely composed of the following two parts. (Part 1)
(i) As shown in FIG. 83, a suction position (task) located at the shortest distance from the final mounting point of each task is found (solid arrow in the figure). FIG. 83 is a view showing the "return" operation in FIG. 44, in which the final mounting point (circle in a rectangle) on the board and the position on the Z-axis of the component cassette to be next sucked (lined in a row) Circles 1 to 19) are shown.
(ii) The mounting path is sequentially written starting from the first suction position (dotted arrow in the figure).
(iii) When the route returns to the first suction position, the route is set as the shortest cyclic partial route 1.
(iv) Search for a suction position that is not included in the shortest traveling partial path found so far ("4" in the example shown in FIG. 83).
(v) Return to (ii) above.
As a result, in the example shown in FIG. 83, there are five shortest traveling routes.

(Part 2)
It is determined from which picking position the mounting order can be optimized for a plurality of shortest cyclic routes when mounting is started. This is no problem since it is implemented from the right. It is good if you do not return.
FIG. 84 (a) is a diagram showing the "return" operation when there are a plurality of mounting points in the same component cassette, and FIG. 84 (b) shows the head when this "return optimization method" is applied. Is a simulation result showing the return trajectory of the trajectory, and it can be seen that the useless cross (left figure) in the trajectory 532a before application is reduced like the trajectory 532b after application.

3.9.8 Overall flow (start from histogram)
(1) Create a component group from the mounting point data.
(2) Make a “mountain” for each part group of small parts.
(2.1) The component tapes are classified into the following three types according to the cassette to be used.
1. Component tape using single cassette
2. Component tape using double cassette (feed pitch 2 mm)
3. Component tape using double cassette (feed pitch 4mm)
(2.2) Make a “mountain” on the temporary Z axis for the component tape that uses a single cassette.
(2.2.1) Create a part histogram on the temporary Z axis.
Arrange the component tapes in descending order of the number of components.
The component tape having the largest number of components is arranged at Z = 1.
(2.2.2) Let N be the number of component tapes that make up the component histogram.
(2.2.3) Convert from the temporary Z axis to the actual Z axis.
The component tapes from Z = 1 to N on the temporary Z axis are arranged in that order at odd Z numbers in the range of Z = 1 to 2N on the real Z axis.
(2.3) For a component tape using a double cassette having a feed pitch of 2 mm, a “peak” is formed on the temporary Z axis.
(2.3.1) Create a part histogram on the temporary Z axis.
Arrange the component tapes in descending order of the number of components.
The component tape having the largest number of components is arranged at Z = 1.
(2.3.2) Let N be the number of component tapes that make up the component histogram.
(2.3.3) The value obtained by dividing N by 2 (rounded up below the decimal point) is defined as M.
(2.3.4) Prepare M double cassettes.
(2.3.5) Prepare a second temporary Z axis.
(2.3.6) The M double cassettes are arranged so as to be spaced from Z = 1 to N on the second temporary Z axis.
(2.3.7) The component tapes from Z = 1 to M on the temporary Z axis are arranged at odd Z numbers of Z = 1, 3, 5,..., N−1 on the second temporary Z axis.
It will be placed on the odd side of the double cassette.
(2.3.8) Component tapes from Z = (M + 1) to N on the temporary Z axis are arranged at even Z numbers Z = 2, 4, 6,..., N on the second temporary Z axis.
It will be placed on the even side of the double cassette.
When N is an odd number, the double cassette arranged at Z = (N−1, N) on the second temporary Z axis becomes empty on the even number side, but is left as it is.
(2.3.9) The second temporary Z axis is set as the temporary Z axis again.
(2.4) For a component tape using a double cassette with a feed pitch of 4 mm, create a “mountain” on the temporary Z axis.
Except for the difference in the feed pitch, the process is the same as the above-described process of “creating a“ mountain ”on the temporary Z-axis for a component tape using a double cassette with a feed pitch of 2 mm”.
(2.5) Merge the component histograms of the double cassette with the feed pitch of 2 mm and 4 mm.
(2.5.1) The “mountain” of the double cassette with the feed pitch of 2 mm and the “mountain” of the double cassette with the feed pitch of 4 mm are arranged on the same temporary Z axis.
The “mountain” of the double cassette with the feed pitch of 2 mm is arranged from Z = 1, and subsequently, the “mountain” of the double cassette with the feed pitch of 4 mm is arranged.
Since the cassettes are rearranged in the next process, the arrangement order may be reversed.
(2.5.2) Rearrange the double cassettes on the temporary Z axis in descending order of the number of component tapes on the odd Z number side.
A double cassette having a component tape having the largest number of components is arranged at Z = 1.
Do not break the pair of double cassettes.
A "mountain" is formed in which double cassettes having a feed pitch of 2 mm and 4 mm are mixed.
Looking at the number of components of the odd-numbered component tape, the histogram becomes a monotonically decreasing histogram.
Looking at the number of components of the component tape having an even Z number, the histogram may not be monotonically decreasing.

(3) All “mountains” are [forcefully] arranged on the real Z-axis.
The “mountains” are packed and arranged from the front sub-equipment 110, and it is checked whether or not all the “mountains” are completely on the real Z-axis.
They are arranged in the order of component groups in “mountain” units.
The “mountain” extending over the front and rear sub-equipment 120 is divided and distributed to the front and rear sub-equipment 120.
In the small parts, one part group is divided into a “mountain using a single cassette” and a “mountain using a double cassette”. Some parts groups have only one “mountain”.
When one component group is divided into a “mountain using a single cassette” and a “mountain using a double cassette” for small components, each is treated as an independent “mountain”.
It is assumed that general-purpose parts are “mountain” in parts group units.
It is assumed that general-purpose components are divided as specified by the user.
・ Placement rules Small parts are available in single and double cassettes. Considering the adjacent condition, the arrangement order is set so that the single cassette and the double cassette are hardly adjacent to each other.
1) A double cassette is placed in the front sub-equipment 110.
(i) From the Z number of the A block = (47, 48), the empty blocks are searched for and arranged in ascending order of the Z number.
(ii) When there is no more space in the A block, move to the Z number of the B block = (95, 96), and search for and arrange a space in ascending Z number order.
2) Place a single cassette in the front sub-equipment 110.
(i) From the Z number of the B block = 49, the empty blocks are searched for and arranged in descending order of the Z number.
(ii) When there is no more free space in the B block, move to the Z number of the A block = 1, and search for and arrange a free space in descending order of the Z number.
3) A double cassette is placed in the rear sub-equipment 120.
(i) From the Z number of the C block = (143,144), empty is searched for and arranged in ascending order of the Z number.
(ii) When there is no more space in the C block, move to the Z number of the D block = (191, 192), and search for and arrange a space in ascending Z number order.
4) A single cassette is placed in the rear sub-equipment 120.
(i) Starting from the Z number = 145 of the D block, empty areas are searched for and arranged in descending order of the Z number.
(ii) When there is no more free space in the D block, move to the Z number = 97 of the C block and search for and arrange the free space in descending order of the Z number.
If there are component tapes whose arrangement is to be fixed, those component tapes are arranged at the Z number of the fixing destination, and then component tapes which are not the arrangement fixation target are arranged.
The fixing of the arrangement of the double cassette will be described in detail in "About the fixing of the arrangement of the double cassette".
(3.1) The part group number is represented by n, and n = 0.
(3.2) When n is larger than the maximum value of the part group number, the process proceeds to (3.7).
(3.3) If a single cassette “mountain” belonging to the component group n exists, the following processing is performed.
(3.3.1) It is arranged in the front sub-equipment 110.
(3.3.2) As a result, if it cannot be placed on the front sub-equipment 110, the “mountain” is divided in units of component tapes, and component tapes that cannot be placed on the front sub-equipment 110 are arranged on the rear sub-equipment 120.
(3.3.3) As a result, if it cannot be fully loaded on the rear sub-equipment 120, an error is generated.
When arranging “mountains” of small parts, the above arrangement rules are followed.
(3.4) If there is a “mountain” of a double cassette belonging to the component group n, the following processing is performed.
(3.4.1) It is arranged in the front sub-equipment 110.
(3.4.2) As a result, if it does not fit on the front sub-equipment 110, the “mountain” is divided in units of component tapes, and component tapes that do not fit on the front sub-equipment 110 are arranged on the rear sub-equipment 120.
(3.4.3) As a result, if it cannot be fully loaded on the rear sub-equipment 120, an error is generated.
When arranging “mountains” of small parts, the above arrangement rules are followed.
(3.5) Add 1 to n.
(3.6) Return to (3.2).
(3.7) The state of the “mountain” of the front sub-equipment 110 and the rear sub-equipment 120 is stored.
In the most packed state, all "mountains" have been placed.

(4) “mountains” are arranged in order from the front sub-equipment 110.
The initial state of the arrangement of the “mountains” when the balance between the front sub-equipment 110 and the rear sub-equipment 120 is adjusted based on the load level is created.
From the front sub-equipment 110 in the order of the front sub-equipment 110 → the rear sub-equipment 120, the parts are arranged in ascending order of “mountains” in the component group, and the arrangement state of the “mountains” is set as an initial state.
If there is a component tape to be array-fixed, the component tape to be array-fixed is arranged at the Z number of the fixing destination, and then a component tape not to be array-fixed is arranged.
When the component tape to be fixed in arrangement and the “mountain” to which it belongs are arranged in the same block, the component tape to be fixed in arrangement is included in the “mountain” to form one “mountain”, The "Mowing method" is applied to the "mountain".
When the component tape to be fixed in arrangement and the “mountain” to which the component tape belongs are arranged in different blocks, separate “mountains” are set, and the “cutting method” is applied to each “mountain”.
(4.1) The part group number is represented by n, and n = 0.
(4.2) If n is larger than the maximum value of the part group number, proceed to (4.8).
(4.3) If there is a “mountain” of a single cassette belonging to the component group n, the following processing is performed.
(4.3.1) It is arranged in the front sub-equipment 110.
(4.3.2) As a result, if it does not fit on the front sub-equipment 110, the “mountain” is divided in units of component tapes, and component tapes that cannot be put on the front sub-equipment 110 are arranged on the rear sub-equipment 120.
(4.3.3) As a result, if it cannot be fully loaded on the rear sub-equipment 120, an error is generated.
The “mountain” is placed on the left or right block where the Z space is larger.
If the left and right blocks have the same Z space, they are placed in the right block.
If there is a vacancy in Z of the left block, but the “mountain” does not fit, the “mountain” is divided into two parts tape units and arranged in the left and right blocks.
(4.4) If there is a “mountain” of the double cassette belonging to the component group n, the following processing is performed.
(4.4.1) It is arranged in the front sub-equipment 110.
(4.4.2) As a result, if it cannot be placed on the front sub-equipment 110, the “mountain” is divided in units of component tapes, and component tapes that cannot be placed on the front sub-equipment 110 are arranged on the rear sub-equipment 120.
(4.4.3) As a result, if it cannot be fully loaded on the rear sub-equipment 120, an error is generated.
The “mountain” is placed on the left or right block where the Z space is larger.
If the left and right blocks have the same Z space, they are placed in the right block.
If there is a vacancy in Z of the left block, but the “mountain” does not fit, the “mountain” is divided into two parts tape units and arranged in the left and right blocks.
(4.5) The “mountains” of the front sub-equipment 110 and the rear sub-equipment 120 are rearranged using the load level.
For each block, the “mountains” are rearranged in order of the load level such that the “mountains” having a large load level are closer to the camera (sensor).
(4.6) Add 1 to n.
(4.7) Return to (4.2).
(4.8) The state of the “mountain” of the front sub-equipment 110 and the rear sub-equipment 120 is stored.
(5) Use "load level" to balance front and rear.
(5.1) Perform “load level balance adjustment processing (“ mountain ”unit movement)”.
The details are as described in “Load level balance adjustment processing (“ mountain ”unit)” described later.
In the "load level balance adjustment processing (" mountain "unit)", the load level balance is finally adjusted for each mounting point.
(6) Apply the "Mowing Method" to small parts.
This is in line with the flow of cassette division processing in the current product version.
(6.1) The mowing process is performed for each “mountain”, and the core portion is left.
(6.1.1) In the case of "mountain" of a single cassette The mowing is performed in the order of odd-numbered Z numbers (Z = large to small).
When the simultaneous suction of 10 points cannot be performed, the mowing process ends.
(6.1.2) In the case of “mountain” of the double cassette The mowing is performed in the order of even Z number (Z = large → small) → odd Z number (Z = large → small).
If at least one component number remains in the even-numbered Z number, the trimming process is performed starting from that.
For example, if only one point can be adsorbed on the even Z number side, the remaining nine points are adsorbed on the odd Z number side.
When 10 points cannot be simultaneously picked up on the odd Z number side, the mowing process is terminated.
The core portion remains on the odd Z number side.
(6.2) A flag is provided for “mountain”.
The initial value of the flag is set to TRUE.
(6.3) The state of the “mountain” of the front sub-equipment 110 and the rear sub-equipment 120 is stored.
(6.4) The state of the cassette resource is stored.
(6.5) Search for “mountain” M having the highest core portion from “mountains” whose flag is TRUE.
(6.5.1) If "mountain" M is not found, proceed to (7).
That is, the core processing for all “mountains” has been completed.
(6.6) It is checked whether one cassette of the same type as the cassette type K used by the “mountain” M remains in the resource.
(6.7) If remaining, perform the following processing.
(6.7.1) The core processing is performed by adding only one cassette type K to the number of cassettes used by the “mountain” M.
Refer to the attached sheet "Core processing method with given number of cassettes".
(6.7.2) If the core height does not change, return to (6.6).
(6.7.3) If the height of the core becomes low, proceed to (6.9).
(6.8) If not left, the following processing is performed.
(6.8.1) The state of the “mountain” of the front sub-equipment 110 and the rear sub-equipment 120 is returned to one previous state.
(6.8.2) Return the state of the cassette resource to the previous state.
(6.8.3) Set the flag of “mountain” M to FLASE.
(6.8.4) Return to (6.3).
To find the next highest “mountain” in the core,
(6.9) All “mountains” are arranged on the real Z axis.
(6.10) If it can be arranged, return to (6.1).
(6.11) If placement is not possible, perform the following processing.
(6.11.1) The state of the “mountain” of the front sub-equipment 110 and the rear sub-equipment 120 is returned to one previous state.
(6.11.2) Return the state of the cassette resource to the previous state.
(6.11.3) The flag of “mountain” M is set to FLASE.
(6.11.4) Return to (6.3).
(7) Generate tasks for small parts.
(7.1) Perform “small component task generation processing”.
The details are as described in “Small component task generation processing” described later.
(8) Optimize general-purpose components.
(9) Use the mounting time to balance before and after.
(9.1) “Process of moving a mountain from the front sub-equipment 110 to the rear sub-equipment 120” is performed.
The details are as described later in “Process of moving a mountain from front sub-equipment 110 to rear sub-equipment 120”.

3.9.9 Positional relationship between fixed components and “mountains” in cassette block “mountains” on the temporary Z axis are component tapes that are subject to array fixation and component tapes that are not subject to array fixation It is composed of
The component tape for which the arrangement is fixed is called a “fixed component tape”.
Component tapes that are not subject to array fixing are referred to as “non-fixed component tapes”.
A cassette block may be simply referred to as a “block”.
The left cassette block is called a "left block", and the right cassette block is called a "right block".
The Z number for fixing the fixed component tape is called a “fixing destination”.
A plurality of component tapes are created from a certain component tape (a component group of a certain component type) by component division, the component tapes are stored in a “cassette”, and the cassette is arranged on the Z axis.
When the component division is not performed for a certain component tape (a component group of a certain component type), the number of divisions is considered to be 1, and one component tape is created from the component tape (the component group of the component type). Think like that.

(10) Count the number of fixed destinations existing in the right block and set it as NR.
Only the fixing destinations related to the fixed component tapes belonging to this “mountain” are counted.
There may be a plurality of fixed component tapes belonging to this “mountain”.
There may be a plurality of fixing destinations of one component tape.
(11) Count the number of fixed destinations existing in the left block and set it as NL.
Count in the same way as for the right block.
(12) If NR> NL, perform the following processing.
This is the case where the right block is often fixed.
(12.1) Place the “mountain” in the right block.
See below for the process of placing “mountains” in blocks.
The details are as described in “fixed array: determination of availability of fixed destination” below.
(12.2) If it cannot be placed in the right block, place it in the left block.
This is a case where another “mountain” is already arranged in the right block, and there is no Z space enough to arrange the “mountain”.
As a result, the fixed component tape exists in the right block and the “mountain” exists in the left block, but the suction operation across the left and right blocks is not performed. The fixed component tape in the right block and the “mountain” in the left block are treated as separate “mountains”.
(12.2.1) If it cannot be arranged in the left block, the “mountain” is divided into two parts in component tape units and arranged in the left and right blocks.
Since the “mountain” is divided into two, “mountain” existing in the same block as the fixed component tape and “mountain” existing in a block different from the fixed component tape are formed.
The “mountain” existing in the same block as the fixed component tape is treated as one “mountain” (histogram) on the temporary Z axis in the “cutting method”.

(13) When NR = NL, the following processing is performed.
This is a case in which the right and left blocks are fixed at the same number.
(13.1) The “mountain” is arranged on the left or right block where the Z space is larger.
(13.2) When the vacancy of Z in the left and right blocks is the same, the “mountain” is arranged in the right block.
(13.3) If it cannot be placed in the right block, place it in the left block.
This is a case where another “mountain” is already arranged in the right block, and there is no Z space enough to arrange the “mountain”.
As a result, the fixed component tape exists in the right block and the “mountain” exists in the left block, but the suction operation across the left and right blocks is not performed. The fixed component tape in the right block and the “mountain” in the left block are treated as separate “mountains”.
(13.3.1) If the “mountain” cannot be arranged in the left block, the “mountain” is divided into two parts in component tape units and arranged in the left and right blocks.
Since the “mountain” is divided into two, “mountain” existing in the same block as the fixed component tape and “mountain” existing in a block different from the fixed component tape are formed.
The “mountain” existing in the same block as the fixed component tape is treated as one “mountain” (histogram) on the temporary Z axis in the “cutting method”.
(14) If NR <NL, perform the following processing.
This is the case where there are many fixed destinations for the left block.
(14.1) Place the “mountain” on the left block.
(14.2) If it cannot be placed in the left block, place it in the right block.
In this case, another “mountain” is already arranged in the left block, and there is no Z space enough to arrange the “mountain”.
As a result, although the fixed component tape exists in the left block and the “mountain” exists in the right block, the suction operation across the left and right blocks is not performed. The fixed component tape in the left block and the “mountain” in the right block are treated as separate “mountains”.
(14.2.1) If it cannot be arranged in the left block, the “mountain” is divided into two parts in component tape units and arranged in the left and right blocks.
Since the “mountain” is divided into two, “mountain” existing in the same block as the fixed component tape and “mountain” existing in a block different from the fixed component tape are formed.
The “mountain” existing in the same block as the fixed component tape is treated as one “mountain” (histogram) on the temporary Z axis in the “cutting method”.

3.9.10 Fixed array: Judgment of availability of fixed destination The maximum dividable number of component tapes from which fixed component tapes are based is ND.
NT is the number of component tapes created from the component tapes by the "cutting method" (core processing). NT ≦ ND.
Let NZ be the number of fixed parts in the block related to the component tape.
In particular,
(1) The following processing is performed on the component tapes constituting the “mountain” in order from one end of the “mountain”.
(1.1) Select one component tape.
(1.2) If NT ≦ (ND−NZ) for the component tape, the following processing is performed.
(1.2.1) The component tapes (NT) constituting the “mountain” are arranged on the Z-axis without using any fixing point related to the component tape.
Arrange the component tapes along the shape of the “mountain”.
As a result, the component tape may be placed at the fixing destination, but it is still possible.
(1.2.2) A component tape of the component type is placed at the fixing destination.
The board to be optimized does not suck the component from the fixing destination, but it is considered to be sucked by another board and is arranged as specified by the user.
(1.3) If NT> (ND−NZ) for the component tape, the following processing is performed.
(1.3.1) Of the component tapes that are made from the component tapes and that make up the “mountain”, {NT− (ND−NZ)} component tapes are arranged at the fixing destination from the one with the smaller number of components.
As the fixed point, select a fixed point close to “mountain” on the real Z axis.
(1.3.2) The remaining component tapes are arranged on the Z-axis without using any fixed parts related to the component tapes.
As a result, the component tape may be arranged at the fixing destination, but that is also possible.
(1.4) Return to (1.1).

3.9.11 Regarding sequence fixing of double cassette The optimization for the sequence fixing restriction for double cassette is as follows.
(1) For a component tape using a double cassette having a feed pitch of 2 mm, a “peak” is formed on the temporary Z axis (FIG. 85). That is, the component histogram 535 arranged in the descending order of the number of components is cut at an intermediate point (return position) and turned back, and the components are combined such that the front half and the rear half thereof are alternately exchanged to obtain the component histogram 536. (Create a pair by folding).
(2) Similarly, a “peak” is formed on the temporary Z-axis for a component tape using a double cassette having a feed pitch of 4 mm (FIG. 86). In other words, the component histogram 537 arranged in the descending order of the number of components is cut at an intermediate point (return position) and turned back, and the components are combined such that the front half and the rear half are alternately replaced to obtain the component histogram 538. (Create a pair by folding).
(3) The component histograms 536 and 538 of the double cassette having a feed pitch of 2 mm and 4 mm are fused to obtain a component histogram 539 (FIG. 87). That is, while maintaining the double cut pair, the component tapes on the odd Z number side are rearranged in descending order of the number of components.
(4) Separation into an odd Z number and histogram 539a (FIG. 88 (a)) and an even Z number histogram 539b (FIG. 88 (b)).
(5) When there is no restriction on the fixed array, these histograms 539a and 539b may be arranged as they are on the actual Z axis (FIGS. 89 (a) and (b)).
(6) If there is a constraint on the fixed arrangement (the odd-numbered Z-numbered parts A to C shown in FIG. 90 (a) and the even-numbered Z-numbered parts D and E shown in FIG. The target part) is as follows.
(7) The parts whose arrangement is to be fixed are extracted for each of the odd-numbered Z-number and the even-numbered Z-number in units of double cassettes containing them, and placed on the right end (FIGS. 91 (a) and (b)).
(8) For only the odd-numbered side, the non-fixed component tape 540 is returned on the real Z-axis (FIG. 92 (a)). The even-numbered side is left as it is (FIG. 92 (b)).
(9) The gaps of the “mountains” are reduced to obtain component histograms 541a and 541b on the odd and even sides (FIGS. 93A and 93B).
At this time, the gap on the odd-numbered “mountain” can be narrowed by the unit of the double cassette (FIG. 93 (a)), but the “mountain” on the even-numbered side matches the “mountain” 541a on the odd-numbered side. In some cases, a gap may remain (FIG. 93B).
(10) The component tapes on the even-numbered side are rearranged for each feed pitch to obtain a component histogram 541c (FIG. 94B). The odd side is left as it is (FIG. 94 (a)).
Specifically, on the even-numbered side, for a component tape having a feed pitch of 2 mm, a component tape existing on the actual Z axis and a component tape that is removed together with the component tape to be arrayed and fixed are not combined and fixed. And rearrange them in the descending order of the number of parts, and store them on the even side of a double cassette having a feed pitch of 2 mm.
The same process is performed for the component tape having the feed pitch of 4 mm on the even-numbered side as the component tape having the feed pitch of 2 mm.
As a result, the double cassettes (43, 44), (45, 46), (47, 48) become unnecessary.

3.9.12 LL restriction: Change of adsorption method (1)
(2) The same number of flags as Z (Z number) are provided, and Z and the flags are associated one-to-one.
(3) Perform the following processing for the mounting points in the left block.
(3.1) The following processing is performed on the component tapes arranged in Z.
-If no component tape is placed in Z, set the flag to FALSE.
-If there is no mounting point included in the LL restriction area, set the flag to FALSE.
・ If there is a mounting point included in the LL restriction area, set the flag to TRUE.
(3.2) Count the number of mounting points that are not included in the LL constraint area and set it as Nf (f means free).
(3.3) Count the number of mounting points included in the LL restricted area and set it as Nr (r means restricted).
(3.4) If either Nf or Nr is not zero, the following processing is repeated.
(3.4.1) When both Nf and Nr are not zero
(i) For the mounting points not included in the LL restriction area, trimming of six-point suction is performed, and allocation is performed to the heads 1 to 6 in the order of the Z number.
The number of the mounted mounting points is Pf.
Suction is performed a plurality of times so that heads 1 to 6 can be fully loaded.
Let Zmax be the largest Z number among the Z numbers where the picked-up mounting points existed.
(ii) Subtract Pf from Nf.
(iii) Pruning of four-point suction is performed for mounting points that are present in Z that is greater than Zmax and that are included in the LL restriction area, and are allocated to the heads 7 to 10 in the order of Z numbers.
Let Pr be the number of mounted mounting points.
Suction is performed a plurality of times so that the heads 7 to 10 can be fully loaded.
(iv) Subtract Pr from Nr.
(3.4.2) When Nf is not zero and Nr is zero
(i) For all mounting points, 10 points are cut off and assigned to heads 1 to 10 in the order of Z numbers.
The number of the mounted mounting points is Pf.
Suction is performed a plurality of times so that the heads 1 to 10 can be fully loaded.
(ii) The number of mounting points Pf sucked is subtracted from Nf.
(3.4.3) When Nf is zero and Nr is not zero
(ii) For all the mounting points, trimming of four-point suction is performed, and allocation is performed to the heads 7 to 10 in the order of Z numbers.
Let Pr be the number of mounted mounting points.
Suction is performed a plurality of times so that the heads 7 to 10 can be fully loaded.
The heads 1 to 6 do not adsorb.
(ii) The number Pr of mounted mounting points is subtracted from Nr.
(3.4.4) When Nf and Nr are Zero The processing for the left block ends.

(4) Perform the following processing for the mounting points in the right block.
(4.1) The following processing is performed on the component tapes arranged in Z.
-If no component tape is placed in Z, set the flag to FALSE.
-If there is no mounting point included in the LL restriction area, set the flag to FALSE.
・ If there is a mounting point included in the LL restriction area, set the flag to TRUE.
(4.2) Count the number of mounting points not included in the LL constraint area and set it as Nf (f means free)
(4.3) Count the number of mounting points included in the LL restricted area and set it as Nr (r means restricted).
(4.4) If either Nf or Nr is not zero, the following processing is repeated.
(4.4.1) When both Nf and Nr are not zero
(i) For the mounting points included in the LL restriction area, the four points are trimmed and the heads 7 to 10 are allocated in the order of the Z number.
Let Pr be the number of mounted mounting points.
Suction is performed a plurality of times so that the heads 7 to 10 can be fully loaded.
The largest Z number among the Z numbers where the sucked mounting points exist is defined as Zmin.
(ii) Subtract Pr from Nr.
(iii) For the mounting points that are present in Z smaller than Zmin and are not included in the LL restriction area, the six points are trimmed by suction and assigned to the heads 1 to 6 in the Z number order.
The number of the mounted mounting points is Pf.
Suction is performed a plurality of times so that heads 1 to 6 can be fully loaded.
(iv) Subtract Pf from Nf.
(4.4.2) When Nf is not zero and Nr is zero
(i) For all mounting points, 10 points are cut off and assigned to heads 1 to 10 in the order of Z numbers.
The number of the mounted mounting points is Pf.
Suction is performed a plurality of times so that the heads 1 to 10 can be fully loaded.
(ii) The number of mounted mounting points Pf is subtracted from Nf.
(4.4.3) When Nf is zero and Nr is not zero
(i) The four points are trimmed for all mounting points and assigned to the heads 7 to 10 in the order of Z numbers.
Let Pr be the number of mounted mounting points.
Suction is performed a plurality of times so that the heads 7 to 10 can be fully loaded.
The heads 1 to 6 do not adsorb.
(ii) The number Pr of mounted mounting points is subtracted from Nr.
(4.4.4) When Nf and Nr are Zero The processing for the right block ends.
(5) End

3.9.13 LL restriction: Change of adsorption method (2)
(1) The same number of flags as Z (Z number) are provided, and Z and the flags are associated one-to-one.
(2) Perform the following processing for the mounting points in the left block.
(2.1) The following processing is performed on the component tapes arranged in Z.
-If no component tape is placed in Z, set the flag to FALSE.
-If there is no mounting point included in the LL restriction area, set the flag to FALSE.
・ If there is a mounting point included in the LL restriction area, set the flag to TRUE.
(2.2) Count the number of mounting points that are not included in the LL constraint area and set it as Nf (f means free).
(2.3) Count the number of mounting points included in the LL restricted area and set it as Nr (r means restricted).
(2.4) If either Nf or Nr is not zero, the following processing is repeated.
(2.4.1) When both Nf and Nr are not zero
(i) For the mounting points that are not included in the LL restriction area, trimming of six-point suction is performed, and allocation is performed to the heads 1 to 6 in the order of the Z number.
The number of the mounted mounting points is Pf.
Suction is performed a plurality of times so that heads 1 to 6 can be fully loaded.
Let Zf be the minimum Z number among the Z numbers where the sucked mounting points existed.
(ii) Subtract Pf from Nf.
(iii) Pruning of four-point suction is performed on the mounting points included in the LL restriction area, and assigned to the heads 7 to 10 in the order of the Z numbers.
Let Pr be the number of mounted mounting points.
Suction is performed a plurality of times so that the heads 7 to 10 can be fully loaded.
Let Zr be the smallest Z number among the Z numbers where the picked-up mounting points existed.
(iv) Subtract Pr from Nr.
(v) If Zf ≦ Zr, the NC data is arranged in the order of heads 1 to 6 and heads 7 to 10.
Heads 1 to 6 and heads 7 to 10 are sucked in this order.
The suction order matches the mounting order, and the mounting order is the order of the NC data.
(vi) If Zf> Zr, the NC data is arranged in the order of heads 7 to 10 and heads 1 to 6.
Suction is performed in the order of heads 7 to 10 and heads 1 to 6.
The suction order matches the mounting order, and the mounting order is the order of the NC data.
(2.4.2) When Nf is not zero and Nr is zero
(i) For all mounting points, 10 points are cut off and assigned to heads 1 to 10 in the order of Z numbers.
The number of the mounted mounting points is Pf.
Suction is performed a plurality of times so that the heads 1 to 10 can be fully loaded.
(ii) The number of mounting points Pf sucked is subtracted from Nf.
(2.4.3) When Nf is zero and Nr is not zero
(i) The four points are trimmed for all mounting points and assigned to the heads 7 to 10 in the order of Z numbers.
Let Pr be the number of mounted mounting points.
Suction is performed a plurality of times so that the heads 7 to 10 can be fully loaded.
There may be many 4-point suction tasks.
(ii) The number Pr of mounted mounting points is subtracted from Nr.
(2.4.4) When Nf and Nr are Zero The processing for the left block ends.
(3) Perform the following processing for the mounting points in the right block.
(3.1) The following processing is performed on the component tapes arranged in Z.
-If no component tape is placed in Z, set the flag to FALSE.
-If there is no mounting point included in the LL restriction area, set the flag to FALSE.
・ If there is a mounting point included in the LL restriction area, set the flag to TRUE.
(3.2) Count the number of mounting points not included in the LL constraint area and set it as Nf (f means free)
(3.3) Count the number of mounting points included in the LL restricted area and set it as Nr (r means restricted).
(3.4) If either Nf or Nr is not zero, the following processing is repeated.
(3.4.1) When both Nf and Nr are not zero
(i) For the mounting points included in the LL restriction area, the four points are trimmed and the heads 7 to 10 are allocated in the order of the Z number.
Let Pr be the number of mounted mounting points.
Suction is performed a plurality of times so that the heads 7 to 10 can be fully loaded.
Let Zr be the smallest Z number among the Z numbers where the picked-up mounting points existed.
(ii) Subtract Pr from Nr.
(iii) For the mounting points that are not included in the LL restriction area, trimming of the six-point suction is performed, and allocation is performed to the heads 1 to 6 in the order of the Z numbers.
The number of the mounted mounting points is Pf.
Suction is performed a plurality of times so that heads 1 to 6 can be fully loaded.
Let Zf be the minimum Z number among the Z numbers where the sucked mounting points existed.
(iv) Subtract Pf from Nf.
(v) If Zf ≦ Zr, the NC data is arranged in the order of heads 7 to 10 and heads 1 to 6.
Suction is performed in the order of heads 7 to 10 and heads 1 to 6.
The suction order matches the mounting order, and the mounting order is the order of the NC data.
(vi) If Zf> Zr, the NC data is arranged in the order of heads 1 to 6 and heads 7 to 10.
Heads 1 to 6 and heads 7 to 10 are sucked in this order.
The suction order matches the mounting order, and the mounting order is the order of the NC data.
(3.4.2) When Nf is not zero and Nr is zero
(i) For all mounting points, 10 points are cut off and assigned to heads 1 to 10 in the order of Z numbers.
The number of the mounted mounting points is Pf.
Suction is performed a plurality of times so that the heads 1 to 10 can be fully loaded.
(ii) The number of mounted mounting points Pf is subtracted from Nf.
(3.4.3) When Nf is zero and Nr is not zero
(i) The four points are trimmed for all mounting points and assigned to the heads 7 to 10 in the order of Z numbers.
Let Pr be the number of mounted mounting points.
Suction is performed a plurality of times so that the heads 7 to 10 can be fully loaded.
There may be many 4-point suction tasks.
(ii) The number Pr of mounted mounting points is subtracted from Nr.
(3.4.4) When Nf and Nr are Zero The processing for the right block ends.
(4) End

3.9.14 LL constraint: Replacement of component tape on Z axis (1)
(1) By this stage, it is assumed that all mountains have been determined by the "Mowing Method".
(2) For the A block, the following processing is performed for the positions Z = 1 to 11.
(2.1) The component tape present at the position Z is defined as the component tape K, and the maximum value Xmax of the X coordinate of the mounting point belonging to the component tape K is determined.
If no component tape exists at the position Z, Xmax = 0.
(2.2) When Xmax ≦ 400.0 [mm] (When component tape K does not have a mounting point existing in the LL restriction area)
(2.2.1) Do nothing.
The maximum value of the X coordinate of the mounting point that can be mounted by the nozzle 1 is 400.0 [mm].
(2.3) When Xmax> 400.0 [mm] (when component tape K has a mounting point that exists in the LL restricted area)
(2.3.1) A component M including the component tape K is configured, and among component tapes existing after Z = 12, there is no mounting point in the LL restriction area, and the component tape K and the number of components are Find a nearby component tape and replace it with component tape K.
In the case of component tapes using a double cassette, it is necessary that the feed pitches match.
(2.3.2) If not found, from the component tapes that exist in the A block and are not the mountain M, and that do not have the mounting point in the LL constraint area, from the component tapes that exist after Z = 12 Then, find a component tape with the smallest number of components and replace it.
There is a case where the component tape is replaced with a component tape of a different component group.
In the case of component tapes using a double cassette, it is necessary that the feed pitches match.
(2.3.3) If the component tape is not found, a component tape having no mounting point in the LL restriction area and having the minimum number of components is found from the component tapes constituting the mountain existing in the B block, and replaced with the component tape.
There is a case where the component tape is replaced with a component tape of a different component group.
For one task, there is a case where it is adsorbed from both the A block and the B block.
In the case of component tapes using a double cassette, it is necessary that the feed pitches match.
(2.3.4) If not found, implementation is not possible.
(3) End.
3.9.15 LL constraint: Replacement of component tape on Z axis (2)
(1) By this stage, it is assumed that all mountains have been determined by the "Mowing Method".
(2) Create a task.
(3) The correspondence between the head number of each task and the position Z is checked, and for each of the positions Z, the minimum value of the head number for picking up the mounting point is determined therefrom.
(4) For the A block, the following processing is performed for the positions Z = 1 to 11.
(4.1) The component tape at the position Z is defined as the component tape K, and the maximum value Xmax of the X coordinate of the mounting point belonging to the component tape K is determined.
If no component tape exists at the position Z, Xmax = 0.
(4.2) Let Xh be the maximum value of the X-coordinate that can be mounted with the minimum head number, which picks up the mounting point from position Z.
(4.3) When Xmax ≦ Xh (when component tape K does not have a mounting point existing in the LL restricted area)
(4.3.1) Do nothing.
(4.4) When Xmax> Xh (when component tape K has a mounting point existing in the LL restricted area)
(4.4.1) Construct a mountain M including the component tape K, and, among component tapes existing after Z = 12, do not have a mounting point in the LL constraint area, and the component tape K and the number of components are Find a nearby component tape and replace it with component tape K.
In the case of component tapes using a double cassette, it is necessary that the feed pitches match.
(4.4.2) If not found, if there is no mounting point in the LL constraint area from among component tapes that are present in the A block and are not the mountain M, and that constitute the mountain that is present after Z = 12 Then, find a component tape with the smallest number of components and replace it.
There is a case where the component tape is replaced with a component tape of a different component group.
In the case of component tapes using a double cassette, it is necessary that the feed pitches match.
(4.4.3) If the component tape is not found, a component tape having no mounting point in the LL restriction area and having the minimum number of components is found from the component tapes constituting the mountain existing in the block B, and replaced with the component tape.
In some cases, m may be replaced with a component tape of a different component group.
For one task, there is a case where it is adsorbed from both the A block and the B block.
In the case of component tapes using a double cassette, it is necessary that the feed pitches match.
(4.4.4) If not found, implementation is not possible.
(5) End.

3.9.16 Support for XL size substrates (XL restrictions)
The XL constraint is avoided by the following method.
(1) Assignment to front sub-equipment 110 / rear sub-equipment 120 by mounting point coordinates
(2) Component division by mounting point coordinates
(3) Initial distribution using an area that can be mounted in both the front sub facility 110 and the rear sub facility 120
(4) Avoidance of LL constraint Specifically, it is as follows.
(1) Assignment to front sub-equipment 110 / rear sub-equipment 120 by mounting point coordinates Now, assignment to the front sub-equipment 110 / rear sub-equipment 120 by mounting point coordinates is a table shown in FIG.
(2) Component division by mounting point coordinates
(2.1) There are the following three types depending on the mounting point coordinates of the component tape.
(i) Allocate the component tape to the front sub-equipment 110.
(ii) Allocate the component tape to the rear sub-equipment 120.
(iii) The component tape is divided and assigned to the front sub-equipment 110 and the rear sub-equipment 120.
(2.2) In the case of the above (3), component division is required. Instead of distributing the number of components to the front sub-equipment 110 / rear sub-equipment 120, the mounting points themselves are allocated to the front sub-equipment 110 / rear sub-equipment 120.
(3) Initial distribution using an area that can be mounted in both the front sub facility 110 and the rear sub facility 120
(3.1) The component tapes corresponding to the areas 1) and 2) shown in FIG.
(3.1.1) The load level is calculated for each component tape corresponding to the areas 1) and 2), and the total is set as the load level of the preceding sub-equipment 110.
(3.2) The component tapes corresponding to the areas 6) and 7) are distributed to the rear sub-equipment 120.
(3.2.1) The load level is calculated for each component tape corresponding to the areas 6) and 7), and the total is set as the load level of the preceding sub-equipment 110.
(3.3) The component tapes corresponding to the areas 4), 5), and 6) are arranged in the Z of the front sub-equipment 110 as many as can be placed in the order of the component groups and in ascending order of the number of components.
(3.3.1) The load level of the placed component tape is calculated and added to the load level of the front sub-equipment 110.
Among the component tapes corresponding to (3.4) areas 4), 5) and 6), component tapes that could not be arranged in the front sub-equipment 110 are arranged in Z of the rear sub-equipment 120.
(3.4.1) The load level of the placed component tape is calculated and added to the load level of the rear sub-equipment 120.
If it cannot be placed in the rear sub-equipment 120, an error is generated.
(3.5) When (load level of front sub-equipment 110) <(load level of rear sub-equipment 120)
(3.5.1) The balance will not improve any more, so I will end.
(3.6) If (load level of front sub-equipment 110)> (load level of rear sub-equipment 120), the following processing is repeated.
(3.6.1) In the component tapes corresponding to the areas 4) 5) and 6) in the front sub-equipment 110, the component tape having the largest component group number and the minimum number of components is transferred to the rear sub-equipment 120. send.

If the data cannot be sent to the rear sub-equipment 120 (= there is no Z free space in the rear sub-equipment 120), the balance is not improved any more, so the processing is terminated.
(3.6.2) Recalculate the load level of the front sub-equipment 110 and the load level of the rear sub-equipment 120.
(4) Avoid LL constraint
(4.1) Regions 2) and 5) in the previous sub-equipment 110 are LL-constrained regions, so that processing corresponding to LL constraints is performed.
(4.2) Regions 3) and 6) in the rear sub-equipment 120 are LL-constrained regions, so that processing corresponding to LL constraints is performed.

3.9.17 Load level balance adjustment processing ("mountain" unit)
The features are as follows.
(i) The state in which the load level of the front sub-equipment 110 is higher than the load level of the rear sub-equipment 120 is set as an initial state, and the load level is moved from the front sub-equipment 110 to the rear sub-equipment 120 in units of "mountains". Adjust the balance.
(ii) For the “mountain” existing on the balance point, the load level balance is adjusted for each component tape. The details are as described later in “Load level balance adjustment processing (unit of component tape) (A)”.
The specific procedure is as follows.
(1) Flag all mountains.
The initial value of the flag is TRUE.
(2) When the flags of all the mountains arranged in the preceding sub-equipment 110 are FALSE, the following processing is performed.
(2.1) Proceed to (15).
This would be the case if all the mountains that were located in the front sub-equipment 110 have been moved to the rear sub-equipment 120 (this should not be possible).
(3) The current arrangement state of the peaks of the front sub-equipment 110 and the rear sub-equipment 120 is stored.
(4) The process of selecting a mountain M to be moved is performed as follows.
(4.1) The maximum value of the component group number of the component tape constituting the peak arranged in the preceding sub-equipment 110 is determined, and is set as PGmax.
(4.2) If the flags of all the ridges composed of a single cassette or a double cassette containing the component tape having the PGmax of the component group number are FALSE, the following processing is performed.
(4.2.1) The process of moving the mountain from the front sub-equipment 110 to the rear sub-equipment 120 ends.
Since no mountain to be moved remains, the process of moving the mountain from the front sub-equipment 110 to the rear sub-equipment 120 ends.
The line is not always balanced.
(4.3) Both a `` mountain composed of a single cassette containing a component tape with a part number of PGmax '' and a `` mountain composed of a double cassette containing a component tape with a part number of PGmax '' exist If
(4.3.1) A mountain composed of a single cassette is defined as a mountain M.
(4.4) Only one of the `` mountain consisting of a single cassette containing component tapes with part number PGmax '' and the `` mountain consisting of double cassettes containing component tapes with part number PGmax '' If there
(4.4.1) The mountain is designated as mountain M.
(5) The mountain M is removed from the mountains arranged in the front sub-equipment 110, and the remaining mountains are rearranged. (6) The mountain M is added to the mountains arranged in the rear sub-equipment 120, and the mountains are rearranged.
(7) If the front sub-equipment 110 or the rear sub-equipment 120 cannot satisfy the nozzle-related restrictions, the following processing is performed.
(7.1) The arrangement state of the mountain of the front sub-equipment 110 and the rear sub-equipment 120 is returned to the stored state.
(7.2) Set the flag of the mountain M to FALSE.
This mountain M is no longer a target for movement.
(7.3) Proceed to (14).
(8) In the front sub-equipment 110 or the rear sub-equipment 120, if the peaks cannot be completely placed on the Z axis, the following processing is performed.
(8.1) The arrangement state of the peaks of the front sub-equipment 110 and the rear sub-equipment 120 is returned to the stored state.
(8.2) Proceed to (15).
Since there is only a mountain M that can be moved, the mountain M is divided into component tape units and distributed to the front sub-equipment 110 and the rear sub-equipment 120 to try to improve the line balance.
Since the mountain M is not always on the line balance point, there is a possibility that the line balance cannot be perfected even if the line balance can be improved.
(9) The load level of the front sub-equipment 110 is calculated.
(9.1) Calculate the load level for small parts.
(9.2) Calculate the load level for general parts.
(9.3) The load level of the small component and the load level of the general-purpose component are added to obtain the load level of the front sub-equipment 110.
(10) The load level of the post-sub equipment 120 is calculated.
(10.1) Calculate the load level for small parts.
(10.2) Calculate the load level for general parts.
(10.3) The load level of the small component and the load level of the general-purpose component are added to make the load level of the rear sub-equipment 120.
(11) When the mounting time of the front sub-equipment 110 matches the load level of the rear sub-equipment 120, the following processing is performed.
(11.1) Proceed to (15).
The load levels of the front sub-equipment 110 and the rear sub-equipment 120 are completely balanced.
(12) When the load level of the front sub-equipment 110 is smaller than the load level of the rear sub-equipment 120, the following processing is performed.
(12.1) The arrangement state of the peaks of the front sub-equipment 110 and the rear sub-equipment 120 is returned to the stored state.
(12.2) “Load level balance adjustment processing (part tape unit)” is performed on the mountain M.
For the “mountain” existing on the balance point, the load level balance is adjusted for each component tape. The details are as described later in “Load level balance adjustment processing (unit of component tape) (A)”.
(12.3) Proceed to (15).
Mountain M is now on the line balance point.
The state is returned to the state where the mountain M is arranged in the front sub-equipment 110.
Thereafter, an attempt is made to improve the line balance by dividing the mountain M into component tape units and distributing them to the front sub-equipment 110 and the rear sub-equipment 120.
(13) When the load level of the front sub-equipment 110 is longer than the load level of the rear sub-equipment 120, the following processing is performed.
(13.1) The flag of the mountain M is set to FALSE.
It is assumed that mountain M has been moved.
(13.2) Proceed to (14).
Further, movement is performed in units of mountains.
(14) Return to (2) above.
(15) The “load level balance adjustment processing (in“ mountain ”units)” ends.

3.9.18 Load level balance adjustment processing (part tape)
The features are as follows.
(i) The state in which the load level of the front sub-equipment 110 is higher than the load level of the rear sub-equipment 120 is set as an initial state, and the load level balance is moved from the front sub-equipment 110 to the rear sub-equipment 120 in units of component tape. adjust.
(ii) Since the accuracy of the load level is not good, the load level balance is not adjusted for each mounting point.
The specific procedure is as follows.
(1) A flag is provided on the component tape constituting the mountain M.
The initial value of the flag is TRUE.
(2) Create a part (seed) list for the mountain M.
(3) If the flags of all component tapes in the component list are FALSE, the following processing is performed.
(3.1) Proceed to (13).
The “load level balance adjustment processing (part tape unit)” ends.
(4) The arrangement state of the peaks of the front sub-equipment 110 and the rear sub-equipment 120 is stored.
(5) Select a component tape K having the minimum number of components from component tapes whose flag is TRUE in the component list.
(6) Allocate the component tape K to the rear sub-equipment 120.
(7) Allocate component tapes that remain in the component list, for which the flag is TRUE and are not allocated to the front sub-device 110 or the rear sub-device 120, to the front sub-device 110.
Mountains other than the mountain M are allocated to the front sub-equipment 110 or the rear sub-equipment 120.
(8) The load level is calculated for the front sub-equipment 110.
(8.1) Calculate the load level of small parts.
(8.2) Calculate the load level of general parts.
(8.3) The load level of the small component and the load level of the general-purpose component are added to make the load level of the front sub-equipment 110.
(9) The load level is calculated for the rear sub-equipment 120.
(9.1) Calculate the load level of small parts.
(9.2) Calculate the load level of general parts.
(9.3) The load level of the small parts and the load level of the general-purpose parts are added to make the load level of the rear sub-equipment 120.
(10) When the load level of the front sub-equipment 110 and the load level of the rear sub-equipment 120 are the same, the following processing is performed.
(10.1) Proceed to (13).
This means that the load levels of the front sub-equipment 110 and the rear sub-equipment 120 are completely balanced.
(11) When the load level of the front sub-equipment 110 is lower than the load level of the rear sub-equipment 120, the following processing is performed.
(11.1) The flag of the component tape K is set to FALSE.
The component tape K has been moved.
(11.2) Proceed to (13).
By moving the component tape K from the front sub-equipment 110 to the rear sub-equipment 120, the load level of the rear sub-equipment 120 becomes higher than that of the front sub-equipment 110. finish.
(12) If the load level of the front sub-equipment 110 is higher than the load level of the rear sub-equipment 120, the following processing is performed.
(12.1) The flag of the component tape K is set to FALSE.
The component tape K has been moved.
(12.2) Return to (3).
Further, the movement is performed in component tape units.
(13) The “load level balance adjustment processing (in units of component tape)” ends.

3.9.19 Processing to move mountain from front sub-equipment to rear sub-equipment
(1) Flag all mountains.
The initial value of the flag is TRUE.
(2) When the flags of all the mountains arranged in the preceding sub-equipment 110 are FALSE, the following processing is performed.
(2.1) Proceed to (16).
This would be the case if all the mountains that were located in the front sub-equipment 110 have been moved to the rear sub-equipment 120 (this should not be possible).
(3) The current arrangement state of the peaks of the front sub-equipment 110 and the rear sub-equipment 120 is stored.
(4) The process of selecting a mountain M to be moved is performed as follows.
(4.1) The maximum value of the component group number of the component tape constituting the peak arranged in the preceding sub-equipment 110 is determined, and is set as PGmax.
(4.2) If the flags of all the ridges composed of a single cassette or a double cassette containing the component tape having the PGmax of the component group number are FALSE, the following processing is performed.
(4.2.1) The process of moving the mountain from the front sub-equipment 110 to the rear sub-equipment 120 ends.
Since no mountain to be moved remains, the process of moving the mountain from the front sub-equipment 110 to the rear sub-equipment 120 ends.
The line is not always balanced.
(4.3) Both a `` mountain composed of a single cassette containing a component tape with a part number of PGmax '' and a `` mountain composed of a double cassette containing a component tape with a part number of PGmax '' exist If
(4.3.1) A mountain composed of a single cassette is defined as a mountain M.
(4.4) Only one of the `` mountain consisting of a single cassette containing component tapes with part number PGmax '' and the `` mountain consisting of double cassettes containing component tapes with part number PGmax '' If there
(4.4.1) The mountain is designated as mountain M.
(5) The mountain M is removed from the mountains arranged in the front sub-equipment 110, and the remaining mountains are rearranged. (6) The mountain M is added to the mountains arranged in the rear sub-equipment 120, and the mountains are rearranged.
(7) If the front sub-equipment 110 or the rear sub-equipment 120 cannot satisfy the nozzle-related restrictions, the following processing is performed.
(7.1) The arrangement state of the mountain of the front sub-equipment 110 and the rear sub-equipment 120 is returned to the stored state.
(7.2) Set the flag of the mountain M to FALSE.
This mountain M is no longer a target for movement.
(7.3) Proceed to (15).
(8) In the front sub-equipment 110 or the rear sub-equipment 120, if the peaks cannot be completely placed on the Z axis, the following processing is performed.
(8.1) The arrangement state of the peaks of the front sub-equipment 110 and the rear sub-equipment 120 is returned to the stored state.
(8.2) Proceed to (16).
Since there is only a mountain M that can be moved, the mountain M is divided into component tape units and distributed to the front sub-equipment 110 and the rear sub-equipment 120 to try to improve the line balance.
Since the mountain M is not always on the line balance point, there is a possibility that the line balance cannot be perfected even if the line balance can be improved.

(9) A task is generated for the front sub-equipment 110.
(9.1) Generate tasks for small parts.
(9.2) Generate tasks for general parts.
(10) A task is generated for the rear sub-equipment 120.
(10.1) Generate tasks for small parts.
(10.2) Create a task for general parts.
(11) The mounting time is calculated for the front sub-equipment 110 and the rear sub-equipment 120.
This is a case where mountains are arranged in both the front sub-equipment 110 and the rear sub-equipment 120.
(12) If the mounting time of the front sub-equipment 110 matches the mounting time of the rear sub-equipment 120, the following processing is performed.
(12.1) Proceed to (16).
This means that the front sub-equipment 110 and the rear sub-equipment 120 are completely balanced.
(13) If the mounting time of the front sub-equipment 110 is shorter than the mounting time of the rear sub-equipment 120, the following processing is performed.
(13.1) The arrangement state of the mountain of the front sub-equipment 110 and the rear sub-equipment 120 is returned to the stored state.
(13.2) A process of moving a component tape from the front sub-equipment 110 to the rear sub-equipment 120 is performed on the mountain M.
(13.3) Proceed to (16).
Mountain M is now on the line balance point.
The mountain M is returned to the state where it is arranged in the front sub-equipment 110.
Thereafter, an attempt is made to improve the line balance by dividing the mountain M into component tape units and distributing them to the front sub-equipment 110 and the rear sub-equipment 120.
(14) When the mounting time of the front sub-equipment 110 is longer than the mounting time of the rear sub-equipment 120, the following processing is performed.
(14.1) The flag of the mountain M is set to FALSE.
(14.2) Proceed to (15).
This is a case where it is necessary to move further from the front sub-equipment 110 to the rear sub-equipment 120.
(15) Return to (2) above.
(16) The “processing of moving the mountain from the front sub-equipment 110 to the rear sub-equipment 120” ends.

3.9.20 Processing to move component tape from front sub-equipment to rear sub-equipment The features are as follows.
(i) A state in which the mounting time of the front sub-equipment 110 is longer than the mounting time of the rear sub-equipment 120 is set as an initial state, and the mounting time is balanced by moving from the front sub-equipment 110 to the rear sub-equipment 120 on a component tape basis. To adjust.
(ii) The number of moving component tapes is not small. Many component tapes move to the rear sub-equipment 120.
Component tapes may be placed in the front sub-equipment 110 and the rear sub-equipment 120. Perform component division.
(iii) Balance is good.
The specific procedure is as follows.
(1) A flag is provided on the component tape constituting the mountain M.
The initial value of the flag is TRUE.
(2) Create a part (seed) list for the mountain M.
(3) If the flags of all component tapes in the component list are FALSE, the following processing is performed.
(3.1) Proceed to (14).
"Process of moving component tape from front sub-equipment 110 to rear sub-equipment 120" is ended.
(4) The arrangement state of the peaks of the front sub-equipment 110 and the rear sub-equipment 120 is stored.
(5) Select a component tape K having the minimum number of components from component tapes whose flag is TRUE in the component list.
(6) Allocate the component tape K to the rear sub-equipment 120.
(7) Allocate component tapes that remain in the component list, for which the flag is TRUE and are not allocated to the front sub-device 110 or the rear sub-device 120, to the front sub-device 110.
Mountains other than the mountain M are allocated to the front sub-equipment 110 or the rear sub-equipment 120.
(8) A task is generated for the preceding sub-equipment 110.
(8.1) Create a task for small parts.
Parts are divided by the core processing.
(8.2) Create a task for general parts.
Parts are divided as specified by the user.
(9) A task is generated for the rear sub-equipment 120.
(9.1) Generate a task for small parts.
Parts are divided by the core processing.
(9.2) Generate a task for general parts.
Parts are divided as specified by the user.
(10) The mounting time of the front sub-equipment 110 and the mounting time of the rear sub-equipment 120 are calculated.
(11) If the mounting time of the front sub-equipment 110 and the mounting time of the rear sub-equipment 120 are the same, the following processing is performed.
(11.1) Proceed to (14).
This means that the front sub-equipment 110 and the rear sub-equipment 120 are completely balanced.
(12) If the mounting time of the front sub-equipment 110 is shorter than the mounting time of the rear sub-equipment 120, the following processing is performed.
(12.1) The flag of the component tape K is set to FALSE.
It is assumed that the component tape K has been moved.
(12.2) The process of moving the mounting point from the front sub-equipment 110 to the rear sub-equipment 120 is performed on the component tape K.
By moving the component tape K from the front sub-equipment 110 to the rear sub-equipment 120, the mounting time of the rear sub-equipment 120 is longer than that of the front sub-equipment 110. The line is distributed to the rear sub-equipment 120 to improve the line balance.
(12.3) Proceed to (14).
(13) If the mounting time of the front sub-equipment 110 is longer than the mounting time of the rear sub-equipment 120, the following processing is performed.
(13.1) The flag of the component tape K is set to FALSE.
It is assumed that the component tape K has been moved.
(13.2) Return to (3).
Further, the movement is performed in component tape units.
(14) The “processing of moving the component tape from the front sub-equipment 110 to the rear sub-equipment 120” ends.

3.9.21 Processing for moving mounting point from front sub-equipment to rear sub-equipment Processing for dividing component tape K in units of mounting points and distributing it to front sub-equipment 110 and rear sub-equipment 120 is performed as follows.
(1) Arrange the mounting points in ascending y-coordinate order.
(1.1) If the y-coordinate is the same, arrange them in ascending order of x-coordinate.
This is called an implementation point list.
When the component tape K is divided for each mounting point and is divided into the front sub-equipment 110 and the rear sub-equipment 120, there is a possibility that only one component tape K is arranged in the front sub-equipment 110 or the rear sub-equipment 120. . In such a case, it is considered advantageous that a collection of mounting points close to each other is advantageous. Therefore, the mounting points are rearranged by the coordinates here.
If the “greedy method” is applied commonly to the same component tapes in the front sub-equipment 110 and the rear sub-equipment 120, this rearrangement is unnecessary. This rearrangement is effective if the “greedy method” is applied independently to the front sub-equipment 110 and the rear sub-equipment 120.
(2) Set 1 to a value n indicating the number of mounting points allocated to the preceding sub-equipment 110.
(3) If n is greater than the number of mounting points of the component tape K, the following processing is performed.
(3.1) Proceed to (12).
The “processing of moving the mounting point from the front sub-equipment 110 to the rear sub-equipment 120” ends.
(4) Assign the nth mounting point from the top of the mounting point list to the preceding sub-equipment 110.
(5) Assign the (n + 1) th to last mounting points in the mounting point list to the rear sub-equipment 120.
(6) A task is generated for the front sub-equipment 110.
(6.1) Generate a task for small parts.
Parts are divided by the core processing.
(6.2) Generate a task for general parts.
Parts are divided as specified by the user.
(7) A task is generated for the rear sub-equipment 120.
(7.1) Generate a task for small parts.
Parts are divided by the core processing.
(7.2) Generate a task for general parts.
Parts are divided as specified by the user.
(8) The mounting time of the front sub-equipment 110 and the mounting time of the rear sub-equipment 120 are calculated.
(9) When the mounting time of the front sub-equipment 110 and the mounting time of the rear sub-equipment 120 are the same, the following processing is performed.
(9.1) Proceed to (12).
The “processing of moving the mounting point from the front sub-equipment 110 to the rear sub-equipment 120” ends.
This means that the front sub-equipment 110 and the rear sub-equipment 120 are completely balanced.
(10) If the mounting time of the front sub-equipment 110 is shorter than the mounting time of the rear sub-equipment 120, the following processing is performed.
(10.1) Proceed to (12).
The “processing of moving the mounting point from the front sub-equipment 110 to the rear sub-equipment 120” ends.
The balance between the front sub-equipment 110 and the rear sub-equipment 120 is much better, but not perfect.
(11) If the mounting time of the front sub-equipment 110 is longer than the mounting time of the rear sub-equipment 120, the following processing is performed.
(11.1) Add 1 to n.
(11.2) Return to (3).
The mounting point is further moved from the front sub-equipment 110 to the rear sub-equipment 120.
(12) The “processing of moving the mounting point from the front sub-equipment 110 to the rear sub-equipment 120” ends.

3.9.22 Swap Process in Line Balancing Process Next, the line balancing process (swap process) when there is no free space in the Z axis of the movement destination will be described in comparison with the case where there is free space in the Z axis.
FIGS. 95 (a) and (b) are explanatory diagrams showing an example of mounting time of the front sub-equipment 110 and the rear sub-equipment 120 when there is an empty space on the Z axis, and a line balance process at that time. 95 (c) and (d) are explanatory diagrams showing an example of the mounting time of the front sub-equipment 110 and the rear sub-equipment 120 when there is no free space on the Z axis, and the line balance processing (swap processing) at that time. is there.
When there is an empty space on the Z axis, as shown in FIGS. 95 (a) and (b), the movement process is the same as the above (3.9.19 to 3.9.21). In this example, In order to eliminate the difference between the two mounting times, a balance is achieved by moving the part 545 for 7.5 seconds from the front sub-equipment 110 to the rear sub-equipment 120.
On the other hand, when there is no free space on the Z axis, as shown in FIGS. 95 (c) and (d), the component 547 having a large number of components allocated to the front sub-equipment 110 and the rear sub-equipment 120 are allocated. The component 546 having a small number of components is swapped in units of component cassettes (component tapes). As a result, the mounting time corresponding to the difference in the number of parts moves from the front sub-equipment 110 to the rear sub-equipment 120, and the mounting time is leveled.

3.9.23 "Mowing method" of double cassette
The "cutting method" for double cassettes is as follows.
(1) For a component tape using a double cassette having a feed pitch of 2 mm, a “mountain” is formed on the temporary Z axis (FIG. 96). That is, the component histogram 550 arranged in the descending order of the number of components is cut at an intermediate point (return position) and turned back, and the front half and the rear half are alternately replaced with each other (the pair is turned back to form a pair). ), A component histogram 551 is obtained.
(2) Similarly, a “peak” is formed on the temporary Z axis for a component tape using a double cassette with a feed pitch of 4 mm (FIG. 97). In other words, the component histogram 552 arranged in the descending order of the number of components is cut at an intermediate point (return position) and turned back, and the front half and the rear half are alternately replaced (the pair is turned back to form a pair). By doing so, a part histogram 553 is obtained.
(3) The component histograms 551 and 553 of the double cassette having the feed pitches of 2 mm and 4 mm are fused to obtain a component histogram 554 (FIG. 98). That is, while maintaining the double cut pair, the component tapes on the odd Z number side are rearranged in descending order of the number of components.
(4) The histogram is divided into an odd-numbered histogram 554a (FIG. 99 (a)) and an even-numbered histogram 554b (FIG. 99 (b)).
(5) In each of the histograms 554a and 554b, a suction pattern of simultaneous suction of 10 points is created by trimming from a component tape having a small number of components (FIGS. 100 (a) and 100 (b)). As a result, core portions 555a and 555b remain in the respective histograms.
(6) Complementary patterns 556a and 556b are created for each of the odd-numbered core portion 555a and the even-numbered core portion 555b (FIGS. 101A and 101B). That is, the number of mounting points of the core portion is 92 points on the odd side and 12 points on the even side, which is 104 points in total. Therefore, ten 10-point tasks are created (one 4-point task remains).
Here, since the number of parts of the core 555b on the even side is 3 at the maximum, three 10-point tasks are created on the even side and the remaining tasks are created on the odd side.
(7) The odd-numbered and even-numbered complementary component tapes 557a and 557b are arranged (FIGS. 102A and 102B). In this drawing, the complementary component tape is indicated by “*” on the odd number side, and is indicated by “#” on the even number side.
As shown in this figure, the number of complementary component tapes on the odd and even sides may not match.
(8) An even-numbered complementary component tape 557b is superimposed on an odd-numbered complementary component tape 557a to form one complementary component tape 5558 (FIGS. 103 (a) and 103 (b)).
(9) A mounting point is assigned to the synthesized complementary component tape 558 (FIGS. 104A and 104B).
At this time, the complementary component tape synthesized on the odd and even sides is composed of only one component tape. Therefore, when the synthesis is released (divided) and complementary tapes on the odd and even sides are produced, they always have the same feed pitch, so that they can be stored in a double cassette as a pair.
(10) Divide the synthesized complementary component tape into odd-numbered sides 558a and even-numbered sides 558b (FIGS. 105 (a) and (b)).
(11) Adsorption patterns 559a and 559b are created for the odd and even histograms, respectively (FIGS. 106 (a) and (b)).
By adopting such a cassette arrangement, when two component tapes are stored in the double cassette, the constraint that only component tapes having the same feed pitch must be stored is satisfied, and a small suction pattern ( (The frequency of simultaneous adsorption is high).

3.9.24 Nozzle Replacement Algorithm As shown in FIG. 11, the types of nozzles that can be suctioned are limited depending on the type of component. Therefore, when picking up a component, the multi-mounting head 112 needs to previously mount a nozzle of a type corresponding to the component tape to be sucked (replace the nozzle at the nozzle station).
Therefore, in the optimization, it is necessary to determine the arrangement of the component tapes so as to suppress the frequency of nozzle replacement. The algorithm for this ("nozzle replacement algorithm") is as follows.
FIG. 107 is a diagram for explaining an algorithm of nozzle replacement. FIG. 107 (a) is a table showing the type (number of usable nozzles) of the target component and the number of components. ) Is a component histogram showing the processing process. Here, the numerical values given to the components in FIG. 107 (b) indicate the nozzle numbers, the arrows indicate the process of creating a suction pattern by component division, and the numerical values surrounded by circles indicate the suction patterns. In this case, the application is based on the application of the “reaping method”. In particular,
1) First, those parts that cannot be divided into 10 units under “adjacent conditions” because they are large parts are excluded from the target. Here, the “adjacent condition” is a spatial clearance to be secured when a component is sucked, moved, or mounted by a head, and a space to be secured to avoid contact between components during mounting. Margin.
2) Arrange in the order of the number of parts for each nozzle.
In this case, since the type of component (the number of usable nozzles) and the number of components are as shown in FIG. 107A, the components are arranged in the left five columns in FIG. 107B.
3) Create a frame for the number of tasks from the total number of parts.
In this example, since the total number of parts is 67, 70 frames are created.
4) Break the peak from the parts with a large number of parts to fill 10 nozzles.
The specific rules are as follows.
-Pack from the top so as to enter the frame from the one with the largest number of parts (here, from part number 5).
・ At this time, the number of handheld nozzles is the same as the maximum division constraint. That is, the division is performed under this constraint.
5) Finally, try to fit within the frame.
As a result, the number of tasks falls within the minimum task.
6) Since the above procedure is an optimization taking into account the nozzle configuration, the configuration of the nozzle arrangement and task order, including large parts, will be reviewed next.
Specifically, for a large component, processing such as interposing is performed in the task configuration determined above.
7) In this example, by changing the order of the tasks, nozzle change is performed only once (only during (6) → 7)), and nozzle replacement occurs.

3.10 Screen Display Example Next, the function of the user interface of the optimization device 300 will be described. In other words, based on the optimization program stored in the optimization program storage unit 305, the screen display example and the input unit 303 displayed on the display unit 302 by the arithmetic control unit 301 are required for the optimization device 300 to interact with the user. The following description focuses on the parameters obtained from the user via the Internet.

3.10.1 Main Screen On this screen, as shown in FIG. 108, the optimizing device 300 displays information on the state of optimization and the type program. The meaning (optimization) of each display item (hereinafter, items enclosed in []) and items that can be selected from the pop-up menu displayed when the display item is selected (hereinafter, items with * attached) The processing of the device 300) is as follows.
1) Menu [File]
* Open Obtains the selection of the type program (such as the mounting point data 307a to be optimized) and various libraries (such as the component library 307b) from the user, and reads the selected type program. The read results (type program name, number of mounting points, component type, equipment information, optimization information) are displayed on the main screen.
* Overwrite save If "Yes" is pressed in the overwrite confirmation message, the optimized type program is overwritten and saved.
* Save As Displays the Save As screen and saves the optimized product program with the entered save file name.
* Close Closes the selected product program.
* Termination of optimization Terminate the application.
[optimisation]
* Optimization Optimizes the read type program information, executes a simulation of the optimization result, and displays the result on the main screen. This is because various resources and optimization conditions can be set before the optimization is performed.
* Stop Stops optimization.
* Detailed optimization information Displays the detailed optimization information screen.
[Configuration]
Set optimization resources and optimization conditions.
-Resource * Cassette number setting Displays the cassette number setting screen. On the other hand, the user can input the number of cassettes usable in the present equipment.
* Part division number setting Displays the part division number setting screen. On the other hand, the user can specify the number of component divisions for simultaneous suction.
* Nozzle number setting Displays the nozzle number setting screen. On the other hand, the user can input the number of nozzles that can be used in the facility.
* Nozzle station selection Displays the nozzle station selection screen. On the other hand, the user can input the plate ID of the nozzle station that can be used in the present facility.
・ Optimization condition * Option setting Displays the option setting screen. On the other hand, the user can set optional specifications and optimization conditions of the facility.
* Z axis information Displays the Z axis information screen. The characteristics of the components arranged on each Z axis are displayed.
* Nozzle station information Displays the nozzle station information screen. Displays the nozzle station information of this equipment.
[printing]
The optimization information, the resource information, and the like are printed on a printer or the like included in the optimization device 300.
* Optimization detail information Prints the optimization detail information.
* Z-axis information Executes printing of Z-axis information.
* Nozzle station information Prints nozzle station information.
* Cassette number information Prints the cassette number information.
* Parts division number information Prints the parts division number information.
* Nozzle number information Prints nozzle number information.
* Nozzle station selection information Executes printing of nozzle station selection information.
[help]
Manage screen version and help.
* Help Start Help.
* Version information Displays version information.
2) Optimization information Information before and after optimization is displayed for each sub-equipment (“1st stage” and “2nd stage” in the figure).
* Mounting time (seconds)
The simulation result before / after optimization is displayed.
* Optimization rate (%)
Displays the mounting time before / after optimization as a ratio (%).
<Calculation formula> (Mounting time after optimization / Mounting time before optimization) * 100
* CPH (point)
Displays the number of mounting points per hour.

<Calculation formula> (number of mounting points / mounting time) * 3600 (seconds)
* Number of tasks Displays the number of tasks.
3) Equipment information Equipment information is displayed for each sub-equipment ("1st stage" and "2nd stage" in the figure).
* Head type Displays the head type of the front / rear sub equipment. (10 heads)
* Camera Displays the camera status of the front / rear sub equipment. (2D sensor, 2D + 3D sensor)
* Tray Displays the tray status of the front / rear sub equipment. (Hand tray, elevator tray)
* Number of mounting points The number of mounting points for the front / rear sub equipment in the product program is displayed.
* Part type Displays the number of component types for the front / rear sub-equipment in the product program.
4) Type program information Displays information on the currently selected type program.
* Type program name The type program name currently selected is displayed.
* Number of mounting points Displays the number of mounting points in the product program.
* Part type Displays the number of component types in the product program.
5) Optimize button Optimizes the read type program information, executes a simulation of the optimization result, and displays the result on the main screen. However, before performing the optimization, it is necessary to set various resources and optimization conditions.
6) Detailed optimization information button Displays the detailed optimization information screen.
7) Exit button Terminates the application.

3.10.2 Open Screen On this screen, as shown in FIG. 109, the optimizing device 300 can open a kind program by designating a kind program and various libraries.
1) Type program list Displays a list of type programs (file name, creation date, update date, capacity).
2) Type program search After input of the type program (excluding the leading P), the type program can be searched by pressing the search button. In addition, since a prefix search is performed on the input characters, it is not necessary to input all the program names.
3) Library selection Various registered libraries are displayed.
* Parts library Displays the registered part library name. The initials begin with "L". This parts library corresponds to the parts library 307b shown in FIG.
* Supply library Displays the registered supply library name. Note that the initials begin with “Y”. This supply library is one piece of information constituting the mounting apparatus information 307c shown in FIG. 9 and holds information on the specifications of the component supply units 115a and 115b, the component cassette, the tray supply unit 117, and the tray. I have.
* Mark library Displays the registered mark library name. Note that the initials begin with "B". This mark library is one of the pieces of information constituting the mounting apparatus information 307c shown in FIG. 9, and is a shape of the recognition mark printed on the board used for positioning the multi-mounting head 112 with respect to the board. Holds information about
* Nozzle library Displays the registered nozzle library name. The initials begin with "V". This nozzle library is one piece of information constituting the mounting apparatus information 307c shown in FIG. 9, and holds information on the shape and the like of various suction nozzles.
4) Open button Opens the specified product program in the selected library. When the user double-clicks on the type program list, the same processing as that performed by the open button is performed.
5) Cancel button Returns to the main screen.

3.10.3 Optimization Detail Information Screen In this screen, as shown in FIG. 110, the optimization device 300 performs optimization details for each sub-equipment (“1st stage” and “2nd stage” in the figure). Display information.
1) Type program information Displays information of the currently selected type program.
* Type program name The type program name currently selected is displayed.
* Number of mounting points Displays the number of mounting points in the product program.
* Part type Displays the number of component types in the product program.
2) Detailed optimization information Information before and after optimization is displayed for each sub-equipment (“1st stage” and “2nd stage” in the figure).
* Mounting time (seconds)
The simulation result before / after optimization is displayed.
* Optimization rate (%)
Displays the mounting time before / after optimization as a ratio (%).
<Calculation formula> (Mounting time after optimization / Mounting time before optimization) * 100
* CPH (point)
Displays the number of mounting points per hour.
<Calculation formula> (number of mounting points / mounting time) * 3600 (seconds)
* Number of tasks Displays the number of tasks.
* Nozzle replacement frequency Displays the number of times nozzle replacement is performed.

* Nozzle replacement time Displays the total time required for nozzle replacement.
* Number of times of adsorption The number of times of adsorption is displayed.
* Adsorption time Displays the total time required for adsorption.
* Number of scans Displays the number of scans.
* Scan time Displays the total time required for scanning.
3) Adsorption number information The number of times of adsorption of 1 to 10 points before / after optimization is displayed for each sub-equipment (“1st stage” and “2nd stage” in the figure).

4) Equipment information Equipment information is displayed for each sub-equipment ("1st stage" and "2nd stage" in the figure).
* Head type Displays the head type of the front / rear sub equipment. (10 heads)
* Camera Displays the camera status of the front / rear sub equipment. (2D sensor, 2D + 3D sensor)
* Tray Displays the tray status of the front / rear sub equipment. (Hand tray, elevator tray)
* Number of mounting points The number of mounting points for the front / rear sub equipment in the product program is displayed.
* Part type Displays the number of component types for the front / rear sub-equipment in the product program.
5) Print button Executes printing of detailed optimization information.
6) Cancel button Exits the detailed optimization information screen and returns to the main screen.

3.10.4 Cassette Number Setting Screen In this screen, as shown in FIG. 111, the optimizing device 300 performs display of cassette number information / setting of the maximum number according to a user's instruction.
1) Cassette number information Displays the cassette number information. To confirm the adjacent condition of the cassette, the user sets the supply code of the parts library.
* Supply code Displays the supply code of the cassette.
Example)
First character: Type (E: embossed P: paper)
The second and third characters: Cassette width (08: 8 mm width)
4th and 5th characters: Feed pitch (04: 4mm pitch)
6th character: Drive system (C: cylinder)
7th character: Cassette type (W: W cassette)
* Current number Displays the number of cassettes currently used.
* Maximum number Displays the maximum number of cassettes that can be used with this equipment.
2) Print button Executes printing of cassette number information.
3) OK button Saves the currently displayed maximum number and closes the cassette number setting screen. 4) Cancel button Exits the cassette number setting screen and returns to the main screen. However, the maximum number is not saved.
5) Maximum number input area The user can input the maximum number by double-clicking the maximum number area.

3.10.5 Parts Division Number Setting Screen On this screen, as shown in FIG. 112, the optimization device 300 performs display of parts division information / setting of the maximum number of divisions according to a user's instruction.
1) Parts division number information Displays the parts division number information.
* Part name Displays the part name used in the product program. The user can input the part name of the type program in order to efficiently perform the part division.
* Number of mounting points Displays the number of mounting points for each component.
* Current division number Displays the current division number for each part.
* Maximum number of divisions Displays the maximum number of divisions for each part. Note that, by default at startup, the current division number is displayed.
2) Print button Executes printing of the number of parts division information.
3) OK button Saves the currently displayed maximum number of divisions and exits the part division number setting screen. 4) Cancel button Exits the part division number setting screen and returns to the main screen. However, the maximum number of divisions is not stored.
5) Maximum division number input area The user can input the maximum division number by double-clicking the area with the maximum division number. Note that the maximum division number is valid only while the application is running. At the next startup, the current division number is the default display of the maximum division number.
Sort display * When the component name or the number of mounting points is clicked, the display is sorted.

3.10.6 Nozzle Number Setting Screen In this screen, as shown in FIG. 113, the optimization device 300 performs display of nozzle number information / setting of the maximum number according to a user's instruction.
1) Nozzle number information Displays nozzle number information.
* Nozzle shape code Displays all nozzle shape codes in the nozzle library.
* Nozzle type Displays the nozzle library number (1 to 99).
* Current number Displays the number currently used.
* Maximum number Displays the maximum number that can be used.
2) Print button Executes printing of nozzle number information.
3) OK button Save the currently displayed maximum number of nozzles and exit the nozzle number setting screen.
4) Cancel button Exits the nozzle number setting screen and returns to the main screen. However, the maximum number is not stored.
5) Maximum number input area The user can input the maximum number by double-clicking the maximum number area.

3.10.7 Nozzle Station Selection Screen In this screen, as shown in FIG. 114, the optimization device 300 performs display of nozzle station selection information / nozzle station selection according to a user's instruction.
1) Nozzle plate ID
The validity / invalidity of the nozzle plate ID can be set for each sub-equipment (“1st stage” and “2nd stage” in the figure). A plurality of IDs other than gray display can be selected.
By moving the cursor, it is possible to switch to the display of the nozzle station diagram of the ID on the cursor. Note that the display is switched even when the check box is not selected.
2) Nozzle station diagram Displays the nozzle station diagram on the cursor.
3) Print button Executes printing of nozzle station selection information.
4) OK button Saves the selected nozzle plate ID and exits the nozzle station selection screen.
5) Cancel button Exits the nozzle station selection screen and returns to the main screen. However, the storage of the nozzle plate ID is not performed.

3.10.8 Option Setting Screen On this screen, as shown in FIG. 115, the optimization device 300 sets the equipment option / optimization level according to the user's instruction.
1) Equipment setting Equipment options can be set.
-XL constraint XL constraint can be set. (Valid or invalid)
・ Z axis speed TA
The speed of the Z axis TA can be set. (Normal or low speed)
・ Z axis speed TB
The speed of the Z axis TB can be set. (Normal or low speed)
-Rear cassette part 180 ° rotation Rear cassette part 180 ° rotation can be set. (Invalid or valid)
-Rear tray part 180 ° rotation Rear tray part 180 ° rotation can be set. (Invalid or valid)
-Rear hand tray component 180 ° rotation Rear hand tray component 180 ° rotation can be set. (Invalid or valid)
-Leading shuttle control Leading shuttle control can be set. (Invalid or valid)
-Preceding suction control Preceding suction control can be set. (Invalid or valid)
・ Substrate stopper position (front)
The substrate stopper position of the front sub-equipment 110 can be set. (Lower left or upper left or lower right or upper right)
・ Substrate stopper position (rear)
The substrate stopper position of the rear sub-equipment 120 can be set. (Lower left or upper left or lower right or upper right)
・ Hand tray (front)
The manual tray of the front sub-equipment 110 can be set. (Invalid or valid)
・ Hand tray (rear)
A manual tray for the rear sub-equipment 120 can be set. (Invalid or valid)

2) Prohibition of sorting before and after sorting By checking this item, sorting before and after can be prohibited.

・ Front
Optimization is performed only for the front sub-equipment 110.
・ Rear
Optimization is performed only on the rear sub-equipment 120.
・ Both
Optimization is performed in the front and rear sub-equipment 120. Note that if the sorting is prohibited, the F / R fixed setting can be performed on the Z-axis information screen.
3) Optimization level setting The optimization execution level can be set in the range of 1 to 5 (simple to detailed) (the default level is 4).
4) Collection conveyor setting Collection conveyors for the first and second stages can be set.
Not set: No Use recovery conveyor (small): Small Use recovery conveyor (large): Large
5) OK button Saves the currently set options (equipment options, optimization level, fore-aft sorting, collection conveyor), and exits the option setting screen.
6) Cancel button Exit the option setting screen and return to the main screen. However, equipment options, optimization levels, sorting out before and after, and collection conveyors are not stored.
7) Algorithm setting The optimization algorithm can be set. (1 or 2)
・ Algorithm 1
Optimize with small parts algorithm.
・ Algorithm 2
Optimize small parts with general part algorithm.
8) Equipment information Displays equipment information.
-Equipment direction Displays the equipment direction. (Forward flow or reverse flow)
・ Transport standard Displays the transport standard. (Front or back)
• Transfer speed Displays the transfer speed.

3.10.9 Z-Axis Information Screen In this screen, as shown in FIG. 116, the optimization device 300 displays information on components set on the Z-axis in accordance with a user's instruction.
1) Z axis information Displays Z axis information.
* Part name Displays the part name set on the ZNo.
* Number of components Displays the number of components (mounting points) set on the ZNo.
* Shape code Displays the component shape code set on the ZNo.
* Nozzle Displays the used nozzle number (same as the nozzle type on the nozzle number setting screen) of the component set on the ZNo.
* Camera Displays the part recognition camera (2DS, 2DL, 3DS, 3DL) of the part set on the ZNo.
* Speed Displays the head speed XY (1 to 8) of the component set on the ZNo.
* Supply code Displays the supply code of the part set on the ZNo.
* W designation It is necessary to designate S (single) or W (double) for each part name.
* Shuttle disabled If the component set on the ZNo is a tray component and can be supplied by shuttle, it can be set to disabled (not performed). Note that a check box is not displayed for a part that cannot be supplied by the shuttle even if it is a tray part.
* F / R fixed Make settings so that parts set on the ZNo do not move between sub-equipments by optimization. In addition, it becomes available only when prohibition of sorting before and after is checked on the option setting screen. If no data is displayed after the ZNo, it indicates that no component is set on the Z axis.
2) Switching before / after optimization Switches the Z-axis information before / after optimization. However, if the optimization is not performed, the display after the optimization cannot be performed.
3) Print button Executes printing of Z axis information.
4) OK button Saves the Z-axis information (W designation, shuttle disabled) and exits the Z-axis information screen. However, the Z-axis information after optimization cannot be edited. The OK button is grayed out.
5) Cancel button Exits the Z axis information screen and returns to the main screen. However, the Z-axis information is not stored.

3.10.10 Nozzle Station Information Screen In this screen, as shown in FIG. 117, the optimization device 300 displays the nozzle station information of the facility according to the user's instruction.
1) Nozzle plate ID
The nozzle plate ID for each sub-equipment (“1st stage” and “2nd stage” in the figure) is displayed.
2) Nozzle station information Displays nozzle station information.
* No
Displays the station No.
* Nozzle shape code Displays the nozzle shape code on the nozzle station.
3) Switch before / after optimization Switch nozzle station information before / after optimization. However, if the optimization is not performed, the display after the optimization cannot be performed.
4) Print button Executes printing of nozzle station information.
5) Cancel button Exits the nozzle station information screen and returns to the main screen.

4 Operation of Optimizer (Application)
Next, an applied operation of the above-described optimization device 300 will be described. That is, a part in which the optimization algorithm described above is improved and its function is extended will be described.

4.1 Optimization of small parts 4.1.1 Optimization of Z arrangement without dividing parts The suction pattern 504 shown in FIG. 42 is an optimized suction pattern that maximizes productivity, The condition is that it can be divided into For example, it is necessary to prepare as many as five tapes for component 1 (component tapes with black square marks). This results in an increase in outgoing parts, which may not be acceptable for some users. That is, in a situation where only one component tape can be used (prepared) for one type of component, it cannot be applied.

Therefore, there is a need for an algorithm for determining a suction pattern that can be applied when parts cannot be divided. Hereinafter, the algorithm will be described.
FIG. 118 is a flowchart showing a processing procedure of an algorithm for determining an efficient suction pattern (Z array) without dividing the components.
First, all the target component tapes are sorted in descending order of the number of components, and numbers (i = 1 to N) are assigned in descending order (S600). Then, component tapes are taken out from the arrangement in descending order of the number of components, and rearranged as follows (S601 to S607).
First, the component tape of No. 1 is taken out and placed on the Z axis (S601). Next, the process of arranging the component tapes with the numbers 2 and thereafter (i = 2 to N) on either the right end or the left end on the Z axis is repeated (S602 to S607). That is, the component tapes of Nos. 2 to 15 (Yes in S605) are arranged on the Z axis in the order of right end, right end, left end (S604 to S606), and the component tapes of No. 16 and thereafter are (No in S603). ), The arrangement at the right end is repeated (S606).

The Z array obtained by such rearrangement is a target suction pattern, that is, a suction pattern with a small number of times of suction up and down.
FIG. 119 shows an arrangement of component tapes for explaining the processing procedure of the flowchart shown in FIG. 118. That is, the upper diagram shows the component tape arrangement 600 after all target component tapes are sorted on the temporary Z-axis in descending order of the number of components. A component tape array 601 after component tapes are taken out in order and rearranged on the Z axis is shown. With respect to the component tapes of Nos. 2 to 15, the right end, the left end, the right end, the right end, the left end,...

  FIG. 120 to FIG. 123 are diagrams for explaining the level of optimization by this optimization algorithm. That is, FIG. 120 shows a component histogram 605 in which component tapes are simply arranged in the descending order of the number of components (from right to left), and FIG. 121 shows a suction up / down frequency pattern 606 when the component histogram 605 is cut off. On the other hand, FIG. 122 shows a component histogram 607 rearranged in the procedure shown in FIG. 118, and FIG. 123 shows a suction up / down frequency pattern 608 when the component histogram 607 is cut off.

In FIG. 121 and FIG. 123, the horizontal axis indicates the arrangement of component tapes (temporary Z axis, Z axis), the left vertical axis indicates the number of times of up / down suction, the right vertical axis indicates the number of tasks, and is surrounded by a square frame. The part group indicates one task (a group of parts to be simultaneously picked up).
As can be understood by comparing the suction up / down frequency patterns shown in FIGS. 121 and 123, the number of tasks is not changed (13) by the rearrangement by the optimization algorithm. Has decreased from 31 times to 25 times. This is because, by rearranging the component tapes according to the procedure shown in FIG. 118, a part of the component histogram 605 (the component tapes with component tape numbers 3, 6, 9, 12, and 15) shown in FIG. This is because it has been moved to the component histogram 607a shown in FIG.
That is, the component histogram 607 shown in FIG. 122 has a triangular shape having two slopes (one is steeper than the other). This shape is close to an ideal shape after optimization by the core processing (for example, a histogram in which the space in the component histogram 504 shown in FIG. 42 is arranged by packing the space downward). Therefore, compared to the component histogram 605 shown in FIG. 120, it can be said that the component tape has a higher optimization level.

4.1.2 Optimization in Distribution Processing to Left and Right Blocks In the initial distribution processing, component tapes are distributed to the front sub-equipment 110 and the rear sub-equipment 120, and then, within each sub-equipment, to the component group to which each component tape belongs. Based on this, the component tape is distributed to one of the left block 115a and the right block 115b of the component supply unit.

At this time, in the conventional method, the component cassettes are arranged in one block without a gap depending on the state of distribution to the left and right blocks 115a and 115b. In other words, there may be a problem that the component tape is not divided, that is, the core processing is not performed on the component histogram at all. For this reason, the number of times of up and down suction increases, and the tact time becomes longer.
For example, even though the left block 115a has enough space, the component cassette is arranged in the right block 115b without a gap, and as a result, the component tape is not divided at all in the right block 115b. appear. In particular, this may occur in the case of a histogram having a large number of component tapes, such as a histogram extending over the left and right blocks 115a and 115b.
Therefore, in such a case, a mountain having the lowest priority is found from among the mountains allocated to the empty blocks, and the component cassette of that mountain is moved to the other empty block, so that a new one is obtained. What free space is available, which enables core processing for mountains that were not possible.

FIG. 124 is a flowchart showing the procedure of the mountain distribution process to the left and right blocks 115a and 115b. Now, with respect to a certain mountain, a state in which core processing cannot be performed (hereinafter, this state is referred to as “block overflow”) has occurred because there is not enough space in a component cassette in a block on which the mountain is placed. And
At this time, first, of all the mountains distributed to the left and right blocks 115a and 115b, a mountain that is divided (straddling) and distributed to the left and right blocks 115a and 115b as a low priority mountain, or The peak having the lowest height (the number of parts) of the core portion is specified (S620).
When the component tapes placed on the block where “block overflow” has occurred among the component tapes that constitute the specified mountain are moved to the other block in ascending order of the number of components, the block It is determined whether or not core processing can be performed on the mountains that have been sorted (S621).

As a result, only when it is determined that it is possible, the component tapes are moved by a necessary amount (S622), and then the trimming process and the core process are executed for all the mountains that can be subjected to the core process (S623).
Finally, it is examined whether or not the component tapes that have not been moved remain on the mountain where the component tapes have been moved, and whether those component tapes can be moved to the other block (S624). As a result, if it is determined that the tape can be moved, the remaining component tapes are also moved to the other block (S625).
FIG. 125 shows a state of the process according to the flowchart shown in FIG. 124, that is, a state of moving a mountain between blocks. Here, a state is shown in which mountains 620, 621, 622a, and 622b that have been divided and assigned to the left and right are to be moved.

Note that the peaks 620, 621, 622a, and 622b are represented as the outer shapes of the component histogram. Also, the peaks 620, 621, 622a, and 622b are higher on the center side between the left block 115a and the right block 115b because the component recognition camera 116 is placed near the center and the components are sucked. Since the multi-mounting head 112 needs to pass over the component recognition camera 116, the component tape having a large number of components is arranged near the center in order to reduce the total moving distance of the multi-mounting head 112. It is.
FIG. 125 (a) shows an initial allocation state of three peaks 620, 621, 622a and 622b allocated to a certain sub-equipment to the left and right blocks 115a and 115b. The right block 115b is a block that is “block overflowing”, in which a mountain 620 and a part 622b of the mountain divided and allocated to the left and right blocks 115a and 115b are arranged. On the other hand, the left block 115a is a block that is not “block overflow”, and includes a mountain 621 and a remaining part 622a of the mountain allocated and divided.

FIG. 125 (b) shows a state where a part 622c of the mountain 622b is moved from the right block 115b to the left block 115a in order to secure a space enough to allow core processing of the mountain 620. .
FIG. 125 (c) shows a state in which the mowing process and the core process are performed on each of the peaks 620 and 621. The shape of the peaks 620 and 621 has changed to a triangle having a steep slope and a gentle slope.
FIG. 125 (d) shows a state where the remaining portion 622d of the mountain that has been divided and moved is also moved from the right block 115b to the left block 115a.
FIG. 126 shows a state of processing in another case according to the flowchart shown in FIG. 124, that is, a state in which the core portion is the lowest mountain to be moved. It shows the same state as the movement of the mountain shown in FIG. 125 except that the (state) of the mountain to be moved is different. That is, of the three peaks 625, 626, and 627 (FIG. 126 (a)), the core moves a part 627a of the lowest peak 627 from the right block 115b to the left block 115a (FIG. 126). (B)), the trimming process and the core process are performed on the peaks 625 and 626 (FIG. 126 (c)). Finally, the remaining portion 627b of the split and moved peak is also moved from the right block 115b to the left block 115a to move the peak. 627a (FIG. 126 (d)).

  As described above, by moving a component cassette (component tape) from a block having no empty Z number to a block having an empty space, an empty space of the Z number is created, and core processing is performed using the empty space. Until then, the part division that has been abandoned becomes possible. That is, by examining the movement of the component tape beyond the block, the core processing that has been abandoned can be performed, an ideal suction pattern can be generated, and the number of times of up and down suction can be reduced.

4.1.3 Estimation of the Number of Double Cassettes Used One component group (component tape group “mountain”) to be mounted is given, and when the core treatment for this is completed, each component tape is moved to the Z axis (component cassette). ). The same applies to the case where a component cassette (double cassette) capable of storing two component tapes at the same time is used. However, when a double cassette is used, all component tapes are paired to form one double cassette. However, there are cases where component tapes to be paired are fixed, and the number of double cassettes required for assignment becomes a problem.

Therefore, when all component tapes constituting a given “mountain” are assigned to the Z-axis for a double cassette, the required number of double cassettes is calculated in advance based on NC data and the like (estimated). ) A method was devised.
FIG. 127 is a flowchart illustrating a processing procedure of an algorithm for estimating the number of double cassettes used.
First, the total number N of target component tapes is specified (S640).
Next, when all the component tapes to be fixed are classified into four types of groups A to D shown in FIG. 128, the numbers Na, Nb, Nc, and Nd of the component tapes belonging to each group are specified ( S641 to S644). That is,
(i) Number Na of component tapes belonging to group A:
The number Na of pairs that are paired with component tapes of the same component group (always an even number),
(ii) Number of component tapes belonging to group B Nb:
A part group number (a series of numbers assigned to identify each part group) of a part tape that is paired with a part tape of a different part group is higher than the part group number of the different part group. The number Nb of small things,
(iii) Number of component tapes Nc belonging to group C:
The number Nc of pairs which are paired with the component tapes of the different component groups and whose component group numbers of the component tapes of this mountain are larger than the component group numbers of the different component groups,
(iv) Number of component tapes Nd belonging to group D:
The number Nd of parts without a pair of component tapes,
Is counted.

Finally, the required number Nw of double cassettes to be obtained is calculated according to the following formula (S645).
Nw = Na / 2 + Nb + Nd + ceil ((N-Na-Nb-Nc-2Nd) / 2)
Here, ceil (x) means a minimum integer value equal to or greater than a real number x.
The basis of this calculation formula is as follows.
The right side of the calculation formula is the number of double cassettes necessary to store the component tape to be fixed (first to third items) and the number of double cassettes necessary to store the component tape to be non-fixed (item 4). Of the total.
The first term Na / 2 on the right side is the number of double cassettes necessary to store the component tapes of group A.

The second term Nb on the right side is the number of double cassettes necessary to store the component tapes of the group B and the component tapes of a different (larger group number) component group to be paired with these component tapes. As described above, when two component tapes of different component groups are stored in one double cassette, it is calculated that a double cassette is required for a component group having a small group number. The number Nc of double cassettes required for storing the tape is not counted (not added to the right side of the calculation formula).
The third term Nd on the right side is the number of double cassettes necessary to store the component tapes of group D (and component tapes to be unfixed to be paired with these component tapes).
The fourth term ceil ((N-Na-Nb-Nc-2Nd) / 2) on the right-hand side is for a case where a part (Nd) of the component tape to be fixed is stored in a pair with the component tape of group D. , The number of double cassettes necessary to store the component tapes to be fixed. If a part (Nd) of the non-fixed component tapes is not stored as a pair with the component tapes of group D, the number is ceil ((N-Na-Nb-Nc-Nd) / 2. ).

From the above, the required number of double cassettes is the sum of the first to fourth terms on the right side of the above calculation formula.
FIG. 129 shows a calculation example of the required number of double cassettes. FIG. 129 (a) shows the arrangement a to z of the target component tapes, FIG. 129 (b) shows a breakdown of those component tapes, and FIG. 129 (c) shows that each component tape is stored in a double cassette. FIG. 129 (d) shows a formula for calculating the required number of double cassettes.
As can be seen from FIG. 129, the required number of double cassettes can be calculated for any arrangement of component tapes by the above calculation formula.
The above equation can be simplified as follows by rearranging it.
Nw = Na / 2 + Nb + Nd + ceil ((N-Na-Nb-Nc-2Nd) / 2)
= Ceil (Na / 2 + Nb + Nd + (N-Na-Nb-Nc-2Nd) / 2)
= Ceil ((N + Nb-Nc) / 2)

4.1.4 Pair fixing of double cassettes A double cassette can store two component tapes with a tape width of 8 mm. However, component tapes for different feed pitches cannot be mixed and stored. For this reason, when optimizing small components for a double cassette, a pair is created by folding the component histogram created for each feed pitch at half of the total number of component tapes, and then the corresponding feed pitch double It will be stored in a cassette.

However, there are cases where two component tapes cannot be stored in a double cassette in a free combination, that is, a pair of component tapes is fixed due to circumstances at a production site or the like. Therefore, when a component tape group including component tapes to be fixed in pairs and including component tapes having different feed pitches is given, how to arrange the components in the double cassette becomes a problem.
Therefore, an algorithm for optimizing the double cassette in consideration of the pairing restriction, that is, an algorithm for determining the Z arrangement of the component tapes suitable for the trimming process has been devised.
FIG. 130 is a flowchart showing the processing procedure of the Z-array optimization algorithm in consideration of the double cassette pair fixing. Here, it is assumed that a component tape using a double cassette having a feed pitch of 2 mm and 4 mm is included.

First, component tapes to be fixed in pairs are separated (S660). Specifically, for each component tape group having the same feed pitch, component tapes that are not to be fixed in pairs and component tapes that are to be fixed in pairs are divided.
Then, with respect to the component tape using the double cassette having the feed pitch of 2 mm, a “mountain” is formed on the temporary Z axis (S661). More specifically, pairs of component tapes that are not the target of pair fixing are formed in the same manner as in the conventional algorithm (the above-described folding method), and component tapes that are the target of pair fixing are directly paired.
Similarly, for a component tape using a double cassette with a feed pitch of 4 mm, a “mountain” is formed on the temporary Z axis (S662). Specifically, pairs of component tapes that are not the target of pair fixing are formed in the same manner as in the conventional algorithm, and component tapes that are the target of pair fixing are directly paired.

Finally, the component histograms of the double cassettes having the feed pitches of 2 mm and 4 mm are merged (S663). At this time, a double cassette fixed to a pair is also included. Specifically, the double cassettes created in steps S661 and S662 are put together, and the double cassettes are rearranged in descending order of the number of parts on the odd-numbered side of the double cassette.
FIGS. 131 to 134 show specific examples of the processing in steps S660 to S663 in FIG.
FIG. 131 shows the process in step S660 of FIG. Here, FIG. 131A shows a state in which a component tape group having a feed pitch of 2 mm is divided into a component histogram 660 of component tapes not to be fixed in pairs and component tapes 661a and 661b to be fixed in pairs. Is shown. Similarly, FIG. 131 (b) shows a state in which a component tape group having a feed pitch of 4 mm is divided into a component histogram 665 of component tapes not to be fixed in pairs and component tapes 666a and 666b to be fixed in pairs. Is shown.

FIG. 132 shows the process in step S661 of FIG. Here, a return position (dotted line) 661c is shown in the component histogram 660 of FIG. 132 (a). FIG. 132 (b) shows the component histogram 662 after the component histogram 660 has been turned back at the turnback position. Here, the term “folding” refers to combining the first half and the second half separated at the folding position such that the component tapes are alternately arranged while maintaining their respective arrangement orders.
FIG. 133 shows the process in step S662 of FIG. A return position (dotted line) 665c is shown in the component histogram 665 of FIG. 133 (a). FIG. 133 (b) shows the component histogram 667 after the component histogram 665 has been turned back at the turnback position.

  FIG. 134 shows the processing in step S663 of FIG. Here, FIG. 134 (a) shows a state in which the component tapes obtained in steps S662 and S663 of FIG. 130 are arranged in a line (on the temporary Z axis). A part histogram 662, a part histogram 661a and 661b having a feed pitch of 2 mm as a fixed target, a part histogram 667 having a feed pitch of 4 mm not a fixed pair, and a 4 mm feed pitch of a fixed pair. This shows a state where the component histograms 666a and 666b are arranged on the temporary Z axis. FIG. 134 (b) rearranges the order of the number of component tapes on the odd-numbered Z-number side in descending order (dotted line in the diagram) while maintaining the pair of double cassettes in the sequence shown in FIG. 134 (a). The state is shown. This Z arrangement is a sequence of component tapes finally required.

As can be seen from the Z arrangement shown in FIG. 134 (b), the arrangement of the component tapes is an arrangement suitable for the trimming process while maintaining the pair of double cassettes fixed. That is, if attention is paid only to the component tapes located at only the odd Z number (or only the even Z number) that the multi-mounting head 112 is sucked by one suction up and down operation, those component tapes have a large number of components. It is arranged in order.
4.1.5 Optimization Algorithm Considering NG Head If an NG head occurs during operation of the component mounter 100, the mounting of the component is continued after minimizing the influence of the NG head. Need to be done. Here, the "NG head" is a mounting head in a state in which components can no longer be sucked.

So, under the following assumptions:
(i) Even if an NG head occurs during operation, the arrangement of the component cassettes (component tapes) on the Z axis is not changed.
(ii) Since the suction is performed without using the NG head, the suction pattern changes.
(iii) Mounting points that can only be picked up by the mounting head with the head number of the NG head are not mounted.
Under such a condition, a method of generating a suction pattern using only normal mounting heads other than NG heads has been devised. The “head number” is a number (1 to 10 from the left) for identifying each mounting head constituting the multi-mounting head 112.

Specifically, for an array of component tapes (“mountain”) created as if there is no NG head, components are not sucked from the component tape corresponding to the position of the NG head, and On the other hand, a mowing process is performed to generate a suction pattern, so that the NG head can be handled.
At this time, priority is given to maximizing the number of components per task, even if the number of times of suction and drop per task becomes two or more. In other words, the components are sucked up and down to suck the components until the components are fully loaded on the multi-mounting head 112 (the state where the components are sucked by all the normal mounting heads), and then the components are mounted on the board. I decided.
FIG. 135 is a flowchart showing the processing procedure of the optimization algorithm considering the NG head.

First, for a given component histogram, as many components as possible are picked up by one suction up / down operation using only a normal mounting head excluding the NG head (S680). As a result, if the multi-mounting head 112 is not in the full loaded state and the components to be suctioned are left (No in S681), the multi-mounting head 112 becomes fully loaded or until the components to be suctioned are exhausted. By moving the mounting head 112 and performing the suction up and down operation again (S680), the operation of sucking the component to the empty mounting head is repeated (S680, S681).
If the multi-mounting head 112 is full, or if all the components have been sucked (Yes in S681), the suction in one task is completed, and the multi-mounting head 112 moves to the board 20 and mounts the components. (S682).

The above processing (S680 to S682) is repeated until there is no more target component (S683). As a result, even in a situation where an NG head is generated, a component suction pattern in which the full mounting state of the multi-mount head 112 is prioritized is completed. That is, component mounting can be completed with a small number of tasks.
FIGS. 136 to 138 are diagrams for comparing and explaining the suction patterns when there is no NG head and when there is an NG head. FIG. 136 shows a target component histogram 680.

FIG. 137 shows a suction pattern (due to the trimming process and the core process) 681 for the component histogram 680 shown in FIG. 136 when there is no NG head. On the other hand, FIG. 138 shows a suction pattern 685 for the component histogram 680 shown in FIG. 136 when the mounting head H2 of the head number 2 is an NG head.
The suction patterns 681 and 685 shown in FIGS. 137 and 138 are patterns obtained when the component tapes A, B, and C are divided according to the core processing with respect to the component histogram 680 shown in FIG. It is. The left vertical axis indicates the number of times of suction up and down (integrated value), and the right vertical axis indicates the number of tasks. The component group surrounded by a rectangular frame indicates one task. However, in FIG. 138, the second and ninth tasks are shown separately in two rectangular frames 687a and 687b and 688a and 688b, respectively.

  In FIG. 138, for example, in the second task, a total of two components are sucked to the mounting heads H1 and H10 by the first suction and up / down operation 687a, and to the mounting heads H3 to H9 by the second suction and up / down operation 687b. A total of seven components are sucked, and as a result, a total of nine components are sucked by nine mounting heads excluding the mounting head H2, and are in a fully loaded state.

As can be seen by comparing FIG. 137 and FIG. 138, when the NG head is considered, the number of times of up and down suction greatly increases from 16 to 24 as compared with the normal case where the NG head is not considered. The number has increased only slightly from 13 to 14. As a result, the optimization considering the NG head is realized.
In the case of a substrate having a size equal to or larger than the LL size, there is a region that can be mounted only by the heads 7 to 10, so that the optimization of the NG head for such a substrate becomes a problem. In the conventional algorithm, the heads 1 to 10 are divided into two head groups, heads 1 to 6 and heads 7 to 10, and a suction pattern is generated separately for each head group. In this case, it is possible to cope with the same problem by performing suction without using the NG head for each head group. However, since there is a Z number that can be picked up only by a specific mounting head, replacement of the component cassette is required.

4.2 Simultaneous Optimization of Multiple NC Data Some users who use the component mounter 100 may change a plurality of boards without changing the position and arrangement of the component cassettes set in the component mounter 100. Sometimes you want to produce in a short time. Therefore, it is necessary to determine an optimal arrangement of component cassettes and the like that can be used in common for mounting a plurality of substrates and that reduces the total time required to complete mounting all of the plurality of substrates. . That is, an algorithm for optimizing the component mounting order for a plurality of NC data is required. Hereinafter, the optimization algorithm will be described.

The basic principle of the optimization algorithm is as follows. That is, the condition under which the trimming process operates as expected is that the component cassettes are arranged in the order of the number of components of each component tape. Therefore, for each board, the correlation coefficient between the order of the number of components and the component cassette array is calculated, the component cassette array that maximizes this is determined, and after that, optimization using the algorithm described above is performed. I do.
FIG. 139 is a flowchart showing the overall processing procedure when optimizing a plurality of NC data simultaneously. First, with respect to a plurality of given NC data (S700), it is examined whether there is NC data having a certain similarity (S701), and if there is a similarity, the mounting points of the NC data are matched. This process is repeated for all NC data (S700 to S703).

Here, the determination of the similarity is performed by expressing each NC data by a vector having the number of components for each component type as a component (a vector having the component type as a component element and the number of components of each component type as the size of each component element). It is assumed that there is similarity when the direction cosine (cos θ) between two vectors is larger than a predetermined threshold. That is,
If cos θ> threshold, it is assumed that the two NC data are similar.
The direction cosine is considered to be an index indicating how common the component types included in the two NC data are.
Then, for one or more pieces of NC data that have undergone such a combining process, the optimization of the Z arrangement for each NC data is repeated in ascending order of the number of boards produced (S704). At this time, if the NC data to be optimized includes a component tape for which the Z arrangement has already been determined, the component tape is removed and the Z arrangement of the component tape is performed using a normal method such as pruning. To determine.

  As described above, when many of the component types included in the NC data are common, the plurality of NC data are optimized as one NC data. Otherwise, the NC data is individually optimized.

Next, a specific algorithm for optimizing a plurality of NC data as one NC data, that is, a simultaneous optimization algorithm for a plurality of NC data will be described.
Note that the optimization target here is a Z array. There are three main targets of optimization: (i) Z array, (ii) implementation path in task, and (iii) task order. However, multiple NC data are optimized at the same time. In such a case, it is necessary to make the Z arrangement of the components common, so that it is most important to optimize the Z arrangement. The other two (ii) and (iii) can be processed by optimizing the task generated by the pruning process for the determined Z array.

By the way, the Z arrangement expected in the pruning process for optimizing individual NC data is a “Z arrangement in which component tapes are arranged in descending order of the number of components” after core processing. Therefore, the optimization algorithm here is an algorithm that determines a common Z array that best satisfies such conditions for individual NC data.
FIG. 140 is a flowchart showing a processing procedure for simultaneously optimizing the Z array of a plurality of NC data. First, an initial Z array is determined using one of three methods called a “ranking method”, a “total number method”, and a “production number method” (S710).

  Here, the “ranking method” is a method of arranging component tapes in the descending order of the average value of the number of parts in each component tape, and the “total number method” is a method in which the total number of components of each component tape is This is a method of arranging component tapes in descending order. The "production number method" is a method of arranging component tapes by giving priority to the NC data having the largest number of components, and the other NC data by the "total number method". It is a method of arranging. Which of these three methods is to be adopted depends on a predetermined criterion known by simulation, for example, if the number of NC data is less than 5, a "production number method" is adopted. decide.

Then, after determining the initial Z array, optimization is performed by stochastic search (S711). In other words, the Z array is changed at random, and if the average simultaneous suction number (the average value of the number of parts simultaneously sucked per task) increases, it is adopted, and if not, it is not adopted (the change is undone). ) Is repeated. For example, a state transition of (i) pulling out one component tape in the Z arrangement, (ii) filling the empty space created by pulling out with left justification, and (iii) inserting the pulled out component tape into another position. Try and adopt if optimization level improves. By repeating such state transition and evaluation, the optimization level can be gradually improved.
FIG. 141 is a diagram illustrating a specific example illustrating three types of methods used for determining an initial Z array. Here, for convenience of explanation, an example is shown in which an initial Z arrangement is determined for three pieces of NC data 1 to 3 using five types of component tapes A to E (or a part thereof).

FIG. 141A shows the number of components (numerical value) for each component tape used in the NC data and the order (parentheses) when the component tapes are arranged in descending order of the number of components for the three NC data 1 to 3. ), The number of boards produced using the NC data, the average rank of each component tape, and the total number of components.
FIG. 141 (b) shows an initial Z determined by the “ranking method”, “total member method”, and “production number method” for the three NC data 1 to 3 shown in FIG. 141 (a). The result of the sequence is shown.
According to the “ranking method”, (i) the ranking of each component tape is determined for each NC data, (ii) the average value of the ranking of each component tape is determined, and (iii) the components are ranked in descending order of the determined average value. The arrangement of the tapes is referred to as an initial Z arrangement. Therefore, the rank “CABED” is determined based on the value of the “average rank” shown in FIG. 141 (a), as shown in FIG. 141 (b).

According to the “total member number method”, (i) the total number of components is calculated for each component tape, and (ii) the component tapes arranged in descending order of the calculated total are defined as an initial Z array. Therefore, based on the value of the “total number of parts” shown in FIG. 141A, the rank “ACBDE” is determined as shown in FIG. 141B.
Further, according to the “production number method”, (i) the NC data with the largest number of productions is specified, and (ii) the component tape used in the specified NC data is optimized, so that the component tape is placed on the Z axis. The tapes are arranged and fixed. (Ii) With respect to other component tapes, the arrangement of the component tapes obtained by arranging them on the vacant Z-axis is used as the initial Z arrangement according to the “total number of components”. Accordingly, here, the order “ABC” is first determined in accordance with the order of the NC data 2 based on the value of “production number” shown in FIG. The order “DE” is determined according to the order.

A description will be given of a result of evaluating the above simultaneous optimization algorithm for a plurality of NC data by simulation.
In the simulation, as the component distribution, the following distribution is used in consideration of the general property of NC data, in which the number of small components is large, and the number of components gradually decreases as the component size increases.
The average number n (part) of parts used for the part
n (part) = C / part
(Part is a series of component tape numbers, C is a constant),
For each NC data A, this is the noise plus
n (part, A) = (random number of C / part width) ± (random number of (C / 3) / part width)
And
The number of NC data is given by a random number from 1 to 20.
As a simulation method,
(i) Determine the number of NC data to be optimized with random numbers.
(ii) Determine the number of parts for each NC data
(iii) Find the initial Z array by the above three methods
(iv) This is a procedure of stochastically searching for the optimal Z array that minimizes the number of suctions according to the above-described state transition rule.
What has been found as a result of the simulation under such conditions is as follows.
(i) Regardless of any of the three methods relating to the initial Z array, as the number of NC data increases, the number of parts that can be simultaneously picked up gradually decreases.
(ii) When the number of NC data is small, the number of simultaneous adsorptions is larger in the initial Z array according to the “production number method”, and as the number of NC data increases, the number of simultaneous adsorptions is larger in the “total member number method”. To go.
(iii) The above-mentioned state transition was performed 1000 times and optimization was performed. As a result, the number of adsorptions was slightly improved by more than 10% (the number of adsorptions was reduced by 10%).

  From the above simulation results, it is considered that the “production number method” is the best when the number of NC data is less than 5, and the “total number method” is the best when the NC data is 5 or more.

4.3 Optimization of general-purpose parts (introduction of rule base)
Heretofore, the optimization algorithm for general-purpose parts has been an algorithm based on a stochastic search, as described in 2.9 “Optimization for general-purpose parts”. In other words, general-purpose components are subject to various restrictions such as the type of nozzle that can be picked up, and so on. Then, the quality of the state was evaluated by the mounting time, and the state was changed stochastically to search for a state in which the mounting time became shorter.

However, in the optimization by such a stochastic search, if the initial state given to the optimization processing is poor, the time required for the optimization tends to be extremely long. In fact, according to the conventional algorithm, the task generated as the initial state is not a preferable initial state in terms of the suction operation. For example, although component tapes are continuously placed in the Z-axis direction and ten components can be sucked up and down ten times, the component tapes are arranged at the same Z number by performing ten times of up and down operations. In this case, a mounting order in which 10 parts are picked up from the initial state is set as the initial state.
Therefore, in order to speed up the optimization process for general-purpose parts, we considered an algorithm that considers nozzle restrictions and generates an optimal initial state based on certain rules and optimizes the nozzle replacement operation. . Hereinafter, four types of the algorithm (“stealing method”, “task division”, “task fusion”, and “task replacement”) will be described.

4.3.1 Interception Method The “interception method” is an algorithm for generating an initial task (a task sequence corresponding to the initial state) to be given to an optimization method based on a stochastic search. As the name implies, this is a method of searching for a component that can be picked up in the Z-axis direction, and is similar to the above-described pruning process developed as an optimization algorithm for small components.

FIG. 142 is a flowchart showing a processing procedure of an initial task generation algorithm by the “stealing method”. It is roughly divided into the first half of processing for arranging component tapes on the Z axis (S720 to S722) and the second half of processing for repeating task generation (S723 to S726).
Specifically, in the first half of the process, first, a component histogram in which component tapes are arranged in descending order of the number of components for each component group is created for the target general-purpose component (S720).

Next, each created component histogram is divided into component histograms for each nozzle type (S721). That is, from the component histogram created for each component group, the process of extracting all component tapes sucked by nozzles of the same type and arranging the component tapes in descending order of the number of components is repeated for all nozzle types included in the component histogram. .
Then, the obtained component histogram for each nozzle type is packed from the inside of the left and right blocks 115a and 115b by just packing component tapes one by one and arranged on the Z axis (S722).

  In the latter half of the process, a task is generated by taking out (preempting) components while scanning in the Z-axis direction for each component group with respect to the component histogram obtained in the first half of the process (S723 to S726). The process (S724) is repeated upward from the bottom of each component histogram until there are no more components (S725). The obtained task sequence becomes a target initial task.

  Note that as the order of the component histograms to be scanned, the order with the smaller number of nozzle resources is prioritized. In addition, even if nozzles of different types are mixed, priority is given to maximizing the number of components constituting each task. For example, when two nozzles of the type M and eight nozzles of the type S are mounted on the multi-mounting head 112, after extracting two components from the component histogram of the nozzle type M, the nozzle type Eight parts are extracted from the part histogram of S, and one task is completed.

FIG. 143 is a diagram illustrating a specific example of the first half process (S720 to S722) in the flowchart illustrated in FIG. 142.
FIG. 143 (a) shows a component histogram for each component group generated in step S720 of FIG. 142. Here, two component histograms 720 and 721 are shown.
FIG. 143 (b) shows a component histogram for each nozzle type generated in step S721 of FIG. 142. Here, the component histogram 720 is divided into component histograms 720a and 720b, and the component histogram 721 is divided into component histograms 721a and 721b.
FIG. 143 (c) shows the part histogram arranged on the Z axis in step S722 of FIG. 142. Here, component histograms 720a and 721a are arranged in the right block 115b, and component histograms 720b and 721b are arranged in the left block 115a.

FIG. 144 is a diagram illustrating a specific example of the second half processing (S723 to S726) in the flowchart illustrated in FIG. 142.
FIG. 144A shows the scanning direction and the scanning order (numerical values 1 to 13) in step S724 of FIG. Here, two target component histograms, that is, a component histogram 725 of the nozzle type S and a component histogram 726 of the nozzle type M are shown. Note that there is no restriction on the number of nozzle resources for any type of nozzle.
FIG. 144 (b) shows how tasks are generated in step S724 of FIG. 142. Here, a state in which one task is generated by the eight components located at the bottom of the component histogram 725 and the two components located at the second stage from the bottom is shown.

  FIG. 144 (c) illustrates tasks (suction patterns 1 to 5) repeatedly generated in the latter half of the process (S723 to S726) in FIG. That is, the numerical values described in the components of the component histograms 725 and 756 are the task numbers (order). Here, task 3 of task number 3 is a task in which components belonging to component histogram 725 and components belonging to component histogram 726 are mixed, that is, nozzles of different types are used.

  FIG. 144D illustrates the task sequence 727 generated in the latter half of the process (S723 to S726) in FIG. 142, that is, the initial task 727 finally generated. Note that components belonging to the component histogram 725 are surrounded by thin lines, and components belonging to the component histogram 726 are surrounded by thick lines. The head of the task sequence (the one with the earliest mounting order) is the task 1 placed at the bottom.

FIG. 144E shows the nozzle pattern of the initial task 727 shown in FIG. 144D. Here, the “nozzle pattern” is a pattern indicating the type of nozzle to be used in association with the position of the mounting head (the mounting position on the multi-mounting head 112) for the target task row. Here, it is shown that the type S and the type M are mixed in the third task.
FIG. 145 is a diagram showing the effect of such a “trailing method”. Here, for convenience of description, the number of times of up and down suction is compared between the initial task 731 generated by the algorithm up to now and the initial task 732 generated by this “interception method” for one component histogram 730. I have.
Note that components surrounded by a thick line indicate that they belong to the same task, and the numerical value in the component indicates a mounting head number that sucks the component.

  According to the conventional method, as shown in the upper part of FIG. 145, the component histogram 730 includes four tasks 731, 732a and b, 733a and b, and 734. And On the other hand, according to the “straightening method”, the component histogram 730 includes four tasks 735 to 738 as shown in the lower part of FIG. is there.

4.3.2 Task Splitting By optimizing nozzle replacement, the mounting time of general-purpose components can be significantly reduced. However, the nozzle replacement cannot be directly controlled using the NC data, and the main body of the component mounter 100 automatically performs the nozzle replacement according to the component type described in the NC data. Therefore, in order to optimize the replacement operation during the nozzle replacement, the type of parts included in the task is changed and the indirect optimization is performed, and the task optimization and the nozzle replacement operation are performed simultaneously. Need, this is not practical.

Therefore, we decided to create a task organization that would reduce the number of nozzle replacements and reduce the overall mounting time.
One method of task organization is “task division”. Specifically, it examines the nozzle pattern used by each task in the initial task, focuses on the task where nozzle replacement is performed before and after the task, and configures the component type that can be picked up with the nozzle pattern before nozzle replacement The task is divided into tasks to be performed and tasks composed of component types that can be picked up by the nozzle pattern after the nozzle replacement, and reconfigured into tasks in which useless nozzle replacement is not performed.
FIG. 146 is a flowchart illustrating a processing procedure of an algorithm for optimizing a nozzle replacement operation by “task division”. First, it is determined whether or not there is a task using two or more types of nozzles by examining a nozzle pattern for a task sequence (or an initial task) to be optimized (S740).

As a result, when such a task does not exist (No in S740), it is determined that “task division” is unnecessary, and the process ends.
On the other hand, when such tasks exist (Yes in S740), the tasks are separated so that the number of nozzle types is one (S741). Then, one of the tasks obtained by the separation (a task using a nozzle different from the immediately preceding task) is a series of tasks having the same nozzle (hereinafter, a series of tasks using the same type of nozzle is referred to as a “task set”). (S742).
As a result, a task in which different types of nozzles are mixed disappears, and all tasks are knitted only by parts using the same type of nozzle.
FIG. 147 is a diagram showing a specific example of the processing in the flowchart shown in FIG. 146. Here, an example is shown in which “task division” is performed on an initial task generated by the “stealing method”.

FIG. 147 (a) shows a task sequence 740 to be subjected to “task division”, and here is equal to the task sequence 727 shown in FIG. 144 (d). Here, two types of nozzles are mixed in the task 3 forming the task row 740. Therefore, before and after the task 3, the nozzle needs to be replaced. That is, when performing task 3 after completing task 2, and when performing task 4 after completing task 3, nozzle replacement from type S to M is required.
FIG. 147 (b) shows how tasks are separated in step S741 of FIG. 146. Here, the task 3 is divided into a task 741 including only the nozzle type S and a task 742 including only the nozzle type M.
FIG. 147 (c) shows how the task is moved in step S742 of FIG. 146. Here, the separately generated task 742 has been moved to the end of the task set 743.

FIG. 147 (d) is a nozzle pattern corresponding to the task row in FIG. 147 (c). Tasks containing different types of nozzles have disappeared, and the original initial task is now composed of three tasks containing only type S nozzles and three tasks containing only type M nozzles.
As can be seen from the nozzle pattern after such “task division”, this task row requires only nozzle replacement (from type S to M) when task 3 is completed and task 4 is executed. The number of nozzle changes required one time has been reduced to one.

4.3.3 Task Fusion According to the “task division” described above, the number of tasks increases, although the number of nozzle replacements decreases. Therefore, there may be a case where the optimization level is not sufficient as a whole.

Therefore, the “task fusion” that suppresses the increase in the total number of tasks by fusing the increased tasks with other tasks is performed.
FIG. 148 is a flowchart illustrating a processing procedure of an optimization algorithm based on “task fusion”. First, for the task sequence to be optimized, it is determined for each task set whether there is a set of tasks that can be merged, that is, whether there is a set of tasks whose mounting head positions do not overlap. (S750). Specifically, for each mounting head (for the same position on the multi-mounting head 112), the AND of the nozzle mounting (the logical product "1" when the nozzle mounting is necessary and "0" when the nozzle mounting is unnecessary) is calculated. A task group is searched for such that the values of all nozzles become 0 when taken.
As a result, when there is no such set of tasks (No in S750), it is determined that “task fusion” is impossible, and the process ends.

On the other hand, if such a set of tasks exists (Yes in S750), the tasks are merged (S751). Specifically, the tasks are combined into one task by connecting the tasks while maintaining the position of the nozzle.
FIG. 149 is a diagram showing a specific example of the processing in the flowchart shown in FIG. 148. Here, an example is shown in which “task fusion” is performed on a task sequence generated by “task division” shown in FIG. 147.
FIG. 149 (a) shows a task sequence to be subjected to “task fusion”, and here, it is equivalent to that shown in FIG. 147 (c). Here, since the task 5 and the task 742 belong to the same task set and include only components that are sucked at different positions on the multi-mount head 112, it is determined that the tasks 5 and 742 can be merged. Is done.

FIG. 149 (b) shows how the tasks are merged in step S751 of FIG. 148. Here, a state is shown in which task 5 and task 742 are fused while maintaining the position of the component.
FIG. 149 (c) is a nozzle pattern corresponding to the task row in FIG. 149 (b). As can be seen from comparison with the nozzle pattern shown in FIG. 147 (d), the task including the type M nozzle is reduced by one in the nozzle pattern shown in FIG. 149 (c). Thereby, the total number of tasks can be reduced without increasing the number of times of nozzle replacement, and the optimization level can be further improved. Specifically, the number of nozzle replacements has been reduced from two to one.

4.3.4 Task Swap The nozzle replacement operation has been optimized by the above-mentioned “task division” and “task fusion”. However, the optimization is based on one target task sequence (in many cases, one task sequence). Optimization within the part group). In other words, there is a possibility that useless nozzle replacement has occurred in relation to the task rows located before and after that.

FIG. 150 shows a specific example. Here, the nozzle patterns 760 and 761 of two component groups 1 and 2 in which the nozzle replacement operation is optimized for each component group by “task division” or “task fusion” are shown. The nozzle pattern 760 of the part group 1 includes a type S task set 760a and a type M task set 760b arranged in this order, and the nozzle pattern 761 of the part group 2 includes a type M task set 761a and a type S task set 761a. Of task sets 761b are arranged in this order.
As can be seen from the nozzle pattern shown in this figure, useless nozzle replacement has occurred between the component groups. For example, focusing on the nozzle of type S, the nozzle is attached to the multi-mounting head 112 to execute the task set 760a, and is removed from the multi-mounting head 112 to execute the following task set 760b. There is a waste that it has to be mounted on the multi-mounting head 112 again in order to execute. That is, since the nozzle type changes from S to M to S to M, a total of three nozzle replacements are required.

Therefore, in order to eliminate useless nozzle replacement between component groups, "task replacement" for replacing task sets belonging to each component group within the component group is performed.
FIG. 151 is a flowchart illustrating the processing procedure of the optimization algorithm based on “task replacement”. Here, the optimal arrangement of task sets is determined by the so-called brute force method.
Specifically, first, permutations of all possible task sets are specified for all of the plurality of component groups to be optimized (S760). At this time, each task set can be moved only within the component group so that the arrangement in the component group is not changed.
Then, for each permutation (S761 to S763), the number of nozzle replacements in the nozzle pattern is calculated (S762), and the optimal solution for obtaining the nozzle pattern (permutation of the task set) having the smallest number of nozzle replacements is determined (S764).

FIG. 152 shows an example of a nozzle pattern obtained by optimization based on “task replacement”. This nozzle pattern is obtained by “task replacement” with respect to the nozzle pattern shown in FIG. 150, and the order of the two task sets 760a and 760b constituting the part group 1 is replaced.
According to the nozzle pattern shown in this figure, since the nozzle type only changes from M to S to M, a total of two nozzle replacements are sufficient. In other words, three nozzle replacements have been reduced to two due to “task replacement”.

4.4 Optimization Considering Nozzle Constraint Here, a method for dealing with the case where the position of the nozzle placed at the nozzle station 119 is fixed and optimization of small parts when the number of nozzles is less than 10 are described. Will be described.

4.4.1 Correspondence to the case where the nozzle arrangement on the nozzle station is fixed In the case of optimizing a plurality of NC data simultaneously, the nozzle arrangement on the nozzle station 119 should be different for each NC data. For example, the “nozzle arrangement on the nozzle station 119” may be fixed due to the reason that is not allowed, and may become one of the constraints in the optimization.

When the nozzle arrangement on the nozzle station 119 is fixed, the nozzle arrangement is specified based on the mechanical restrictions of the component mounter 100 (the movable range of the multi-mount head 112 on the nozzle station 119 and the component supply units 115a and 115b). It may happen that the specified component cannot be sucked by the nozzle that has been set.
Therefore, when the nozzle arrangement on the nozzle station 119 is given, a feasible solution check (determination as to whether or not there is a possible component mounting order) is performed. That is, when the nozzle arrangement on the nozzle station 119 and the Z arrangement of the component tapes are given, it is determined whether all the target components can be sucked by the corresponding nozzles.
FIG. 153 is a diagram for describing restrictions as a premise thereof, and is a diagram illustrating restrictions on nozzle replacement based on the movable range of the multi-mounting head 112 on the nozzle station 119. Here, the relative positional relationship between the multi-mounting head 112 and the nozzle station 119 when the multi-mounting head 112 is located at the rightmost end (a front view of the multi-mounting head 112 and the nozzle station 119), and the nozzle station 119 A plan view and a table 770 showing the relationship between the nozzle arrangement and a replaceable (wearable) mounting head (a symbol ○ means replaceable) are shown.

In the left direction, the multi-mounting head 112 can move beyond the position of the nozzle station 119, and there is no restriction on nozzle replacement.
As can be seen from this figure, the mounting heads to which the nozzles n1 to n4 placed in the first to fourth rows counted from the leftmost end on the nozzle station 119 can be mounted are the mounting heads H1 to H10, and there are no restrictions. . However, there are restrictions on the other nozzles n5 to n10. That is, the mounting heads that can mount the nozzles n5 placed in the fifth row are limited to the mounting heads H2 to H10, and the mounting heads that can mount the nozzles n6 placed in the sixth row are limited to the mounting heads H3 to H10. ..., the mounting heads to which the nozzles n10 placed in the tenth row can be mounted are limited to the mounting heads H7 to H10.
FIG. 154 is a diagram for describing another restriction, and is a diagram illustrating a restriction on component suction based on the movable range of the multi-mounting head 112 on the component supply units 115a and 115b (corresponding to FIG. 5). Is an example that differs in details from the contents of FIG. 5).

FIG. 154 (a) shows the relative positional relationship between the multi-mounting head 112 and the component supply unit 115a when the multi-mounting head 112 is located at the leftmost end. Here, a series of numerical values written on the component supply unit 115a is a Z number. FIG. 154 (b) shows the relative positional relationship between the multi-mounting head 112 and the component supply unit 115b when the multi-mounting head 112 is located at the rightmost end. FIG. 154 (c) is a table showing mounting heads (marked with ○) and mounting heads that cannot be accessed (marked with “x”) for each Z number.
As can be seen from this drawing, not all mounting heads can access component tapes with Z numbers 1 to 17 and 86 to 96. That is, the mounting head H1 is the only mounting head that can access the component tape of the Z number 1, the mounting heads H1 and H2 are the mounting heads that can access the component tape of the Z number 2,..., Z number 17 The mounting heads that can access the component tape of No. Z are mounting heads H1 to H9, and the mounting heads that can access the component tape of Z No. 86 are mounting heads H2 to H10. The mounting heads that can be accessed are the mounting heads H7 to H10.

It is to be noted that the reason why such an access restriction occurs near both the left and right ends of the Z axis is due to the design of the component mounter 100. In other words, the priority is given to increasing the number of component tapes that can be set in the component supply units 115a and 115b rather than to avoid such restrictions.
As can be seen from FIGS. 153 and 154 described above, when the nozzle arrangement on the nozzle station 119 is given, a possible solution check is to determine whether the mounting nozzle can access a component arranged near the left end of the Z axis. It is only necessary to consider whether or not. The mounting head that can access the vicinity of the left end of the Z axis is limited to the mounting head with the lower head number (FIG. 154 (c)), and the mounting head with the lower head number sucks all the nozzles on the nozzle station 119. This is because it is not always possible (FIG. 153 (c)).
On the other hand, such a study is not necessary for the vicinity of the right end of the Z axis. At least, the mounting head H10 can suck all the nozzles on the nozzle station 119 (FIG. 153 (c)) and can access the position of the maximum Z number 96 (FIG. 154 (c)). This is because there is no restriction due to the nozzle arrangement on the nozzle station 119.

FIG. 155 is a flowchart showing a processing procedure for checking a feasible solution when the nozzle arrangement on the nozzle station is given.
First, the smallest Z number PZmin (Ntype, Z array) for each nozzle type Ntype is specified from the given Z array of component tapes (S780). For example, among the component tapes using the type S nozzle, the Z number of the component tape placed at the left end is specified.

Next, the following processing is repeated for each nozzle type Ntype from the given nozzle arrangement NP on the nozzle station 119 (S781 to S785).
First, the minimum head number Hmin (Ntype, NP) capable of sucking the nozzle of the nozzle type Ntype is specified (S782). For example, when the nozzle of the type S is placed in the sixth row counting from the leftmost row on the nozzle station 119, by referring to the table shown in FIG. The minimum number of the mounting head capable of sucking the nozzle can be specified as “3”.

Next, the minimum Z coordinate NZmin (Ntype, NP) that can be reached by the nozzle of the nozzle type Ntype is specified from the head number Hmin (Ntype, NP) that has just been specified (S783). For example, when the head number Hmin (Ntype, NP) is “3”, the minimum Z coordinate NZmin (Ntype, NP) that can be reached by this nozzle can be obtained by referring to the table shown in FIG. ) Can be specified as “4”.
Next, it is determined whether or not the currently specified Z coordinate NZmin (Ntype, NP) is equal to or less than the Z number PZmin (Ntype, Z array) specified in step S780 for the nozzle type Ntype (S784). That is, for the Ntype,
NZmin (Ntype, P) ≤ PZmin (Ntype, Z array)
Is determined.

This means that, for a nozzle of this nozzle type Ntype, the minimum accessible Z coordinate NZmin (Ntype, NP) determined by the nozzle arrangement on the nozzle station 119 is the minimum Z number PZmin (Ntype, Ntype, NZ) determined from the given Z array. (Z arrangement) or less, that is, with respect to the movement of the multi-mounting head 112 to the left, it is determined whether or not the nozzle can suck all the components to be sucked.
As a result, when the determination in step S784 is established for all the nozzle types Ntype, it is determined that a possible solution exists for the given nozzle arrangement and Z arrangement (S786). It is determined that no possible solution exists (S787).
By performing such a feasible solution check at the time of configuring the initial task or at the time of updating the state, it is possible to perform optimization taking into account the effect of fixing the nozzle arrangement on the nozzle station.

4.4.2 Optimization of small parts when less than 10 nozzles are used The multi-mounting head 112 can simultaneously pick up to 10 parts at the same time. It is assumed that ten nozzles are mounted on the multi-mounting head 112. However, at a production site, a situation occurs in which the number of nozzles that can be used in a certain mounter 100 is less than ten. Even in such a situation, the multi-mounting head 112 can change the position of the nozzle to be mounted or the like by exchanging the nozzle on the nozzle station 119. If at least one nozzle of any type is prepared, a component placed at any position on the Z-axis can be sucked, and mounting of all components can be completed.

However, since nozzle replacement is a time-consuming operation, a small component having a large number of mounting points requires a mounting order that minimizes nozzle replacement.
Therefore, as the optimization of the small parts when the number of used nozzles is less than 10, the following optimization is performed based on the algorithm for the small parts so far such as the mowing process so that the nozzle replacement is minimized. I'm supposed to.
Now, assuming that the number of nozzles to be used is n (<10), the following two nozzle patterns are prepared for the nozzle patterns, and only these two nozzle patterns (in some cases, only one of the nozzle patterns) are used. To mount all the small parts.
(i) Nozzle pattern 1
Pattern with nozzles attached to head numbers 1 to n
(ii) Nozzle pattern 2
Patterns with nozzles attached to head numbers (10-n + 1) to 10 FIG. 156 shows an example of two nozzle patterns prepared when the number of used nozzles is six. Nozzle pattern 1 is a state in which nozzles are mounted only on six mounting heads of head numbers 1 to 6, and nozzle pattern 2 is a state in which nozzles are mounted only on six mounting heads of head numbers 5 to 10. State.

FIG. 157 is a flowchart illustrating the timing of nozzle replacement. Here, according to the position (left / right block, Z number) of the given component tape in the Z arrangement, which of the two types of nozzle patterns 1 and 2 is used, and at what timing the nozzle pattern is exchanged Is shown.
With respect to the component tapes arranged in the left block (left in S800), if at least one component tape is arranged at the position of Z numbers 1 to 17 (Yes in S801), the nozzle pattern 1 is used to set Z. Components are sucked in the directions of numbers 1 to 48 (S802), and when there are no more components to be sucked at the positions of Z numbers 1 to 17 (No in S801), the nozzle pattern of the multi-mount head 112 is changed from 1 to 1. 2, and the remaining components are sucked in the direction from Z numbers 18 to 48 (S803).

On the other hand, when no component tapes are arranged at the positions of the Z numbers 1 to 17 in the left block (No in S801), the components are sucked in the direction of the Z numbers 18 to 48 using the nozzle pattern 2 from the beginning. (S803).
Also, for the component tapes arranged in the right block (right in S800), all components are sucked from the beginning with the nozzle pattern 2 (S804).

The timing of such nozzle replacement is as follows. That is, as can be seen from the head numbers of the mounting heads that can be accessed for each Z number shown in FIG. 154 (c), the mounting head H10 with the head number 10 can access the positions where the Z numbers are 18 to 96. Therefore, as long as the number n of the used nozzles is 1 or more, the component tapes of Z numbers 18 to 96 can always be sucked by the nozzle pattern 2. On the other hand, as for the component tapes of Z numbers 1 to 17, at least the mounting head H1 of head number 1 can access those component tapes. The seventeen component tapes can always be sucked. Then, the nozzle pattern 2 that can access many Z numbers is preferentially used. As a result, it is possible to use any type of Z arrangement using only two types of nozzle patterns and with a small number of nozzle replacements.
Note that the suction pattern may be generated by performing a trimming process or the like on the given component histogram in units of n points instead of the unit of 10 points so far, thereby generating the suction pattern.

As described above, the optimization of the component mounting order according to the present invention has been described based on the embodiment, but the present invention is not limited to this embodiment.
For example, the optimization device 300 has been used to generate optimal NC data for downloading to each of the component mounters 100 and 200 having a specific configuration. Needless to say, it can be used to determine the configuration of the production line required to satisfy the required specifications. The mounting point data of the substrate to be produced and the mounting device information of the modeled virtual electronic component mounting system are given to the optimizing device 300, and whether or not the obtained optimum state (line tact) satisfies the required specifications. Judge it.

More specifically, this optimization device 300 may be used as (i) a mounting device design, for example, by changing the number of nozzles of the multi-mounting head 112 from 4 nozzle heads → 10 nozzle heads → 8 nozzle heads, By changing the component cassette from 21.5 mm to 22 m, changing the pitch of the component cassette (Z-axis pitch), or changing the position of the component recognition camera, the most efficient (high productivity) ) Used to determine a head, etc .; (ii) used to determine which production line (or mounting device) of a plurality of production lines should produce a target board; Calculation of what productivity (how many substrates can be produced per hour) if any options (number and types of component cassettes and nozzles) are equipped as sales and sales tools It can be used for.
Further, the optimizing device 300 is a device independent of the component mounters 100 and 200, but may be built in the component mounters 100 and 200.

The state optimizing unit 316 optimizes the small parts belonging to the parts groups G [1] to G [5] and the general parts belonging to the parts groups G [6] to G [9] by different search approaches. However, the present invention is not limited to such a classification or approach.
In the intersection solving method, the mounting order is changed so as to eliminate the intersection between the polygonal lines (paths) connecting the mounting points of the components of the two tasks. However, you may try to swap paths that do not intersect. This is because the tact can be shortened by exchanging the paths of the tasks that do not intersect.
As an industrial applicability, the component mounting order optimizing method according to the present invention can be used in an optimizing device that optimizes a component mounting order by a component mounter that mounts electronic components on a printed circuit board. it can. Not only as a controller for a component mounter installed and used on the production line, but also at the stage before constructing the production line, estimating the relationship between the configuration / specification of the component mounter to be introduced and the component mounting time. It can also be used as a simulation / evaluation tool used for this purpose.

5. Explanation of terms The meanings of the main terms used in the present embodiment are listed below.
Component mounting system:
A system that includes an optimization device and a component mounting machine.
Optimizer:
Equipment that optimizes the order in which components are mounted. Specifically, in order to produce a board in a short tact time (mounting time), an optimal arrangement of component cassettes in the component mounter (in which position (the Z axis) the component cassettes containing the component tapes are placed in the component mounter) Or the order of picking up and mounting components by the multi-mounting head (from which component cassette to pick up components and to which mounting point on the board).
Component mounting machine:
A production robot that picks up components from a component cassette and mounts them on a board using a multi-mounting head according to the NC data after optimization. Some types have multiple sub-equipment.
Sub-equipment:
An apparatus (mounting unit) that includes one multi-mounting head and a plurality of component cassettes, and executes component mounting on a board independently of (in parallel with) other sub-equipment.
Single cassette:
A type of component cassette in which only one component tape is loaded.
Double cassette:
A type of component cassette that can be loaded with up to two component tapes. However, it is limited to component tapes having the same feed pitch.
Z axis:
A coordinate axis (or its coordinate value) for specifying the arrangement position of the component cassettes to be mounted for each component mounter (or sub-equipment when sub-equipment is provided).
Parts type:
Types of electronic components such as resistors and capacitors. Each component type is associated with component information (electrical characteristics, shape, number of components, maximum number of divisions, cassette type, etc.).
Parts tape:
Multiple parts of the same part type arranged on a tape. In the optimization processing, it refers to data that specifies a set of components belonging to the same component type (a plurality of these components arranged on a virtual tape). In some cases, a component group (or one component tape) belonging to one component type is divided into a plurality of component tapes by a process called “component division”. The number of tapes after division is called the number of divisions.
Implementation point:
A coordinate point on the board where components should be mounted. Components of the same component type may be mounted at different mounting points. The total number of components (mounting points) arranged on component tapes of the same component type matches the number of components of the component type (total number of components to be mounted).
Parts histogram:
A columnar graph drawn on coordinates with the component tape (component type) on the horizontal axis and the number of components on the vertical axis. The optimization eventually maps to an array of component cassettes.
core:
The component histogram, in which component tapes are arranged in descending order of component count, is trimmed by the "pruning method" using the suction pattern of simultaneous n-point suction, and the remaining components are called "core components". The component tapes and component cassettes containing these are called "core component tapes" and "core cassettes", respectively.
Reap:
This is a process for removing the suction pattern of n-point simultaneous suction from components having the smallest number of components in the component histogram in which component tapes are arranged in descending order of the number of components.
task:
A single mounting operation (suction / movement / mounting) in a series of mounting operations in which a multi-mounting head picks up / moves a component and mounts it on a board.
Suction pattern:
A diagram showing, by task unit, components that the multi-mounting head (simultaneously) picks up for one or more tasks, or a group of those components.
Task group:
A group of related tasks from the viewpoint of simultaneous picking of parts. By collecting n parts tapes with the same number of parts and creating a task group aiming to collect tasks that can simultaneously pick up n points so that n parts can be picked up one by one from those n part tapes at the same time The optimization method for determining the arrangement of the component cassettes is called a “task group generation method”.
Mountain:
A collection of component tapes whose arrangement is determined by optimization, or a corresponding component histogram of those component tapes. The component histogram that has been optimized by the “pruning method” has a “mountain” shape having a gentle slope and a steep slope. The generated “mountain” may be further optimized.
Line balance:
The level of leveling in the distribution of tact for each component mounter (or sub-equipment if sub-equipment is provided). The process of determining the component mounting order so as to equalize the tact distribution is called "line balance process".

  The present invention provides a component mounting order optimizing device that determines an optimal component mounting order for a component mounter that mounts electronic components on a board such as a printed wiring board, and also includes a controller that determines a component mounting order. It can be used as a component mounter or component mounter system, or as a stand-alone simulator (component mount order optimization tool) that determines and evaluates the component mounting order without being connected to the component mounter. it can.

1 is an external view illustrating a configuration of an entire component mounting system according to the present invention. FIG. 2 is a plan view showing a main configuration of a component mounting machine in the component mounting system. FIG. 3 is a schematic diagram illustrating a positional relationship between a work head and a component cassette of the component mounter. (A) shows a configuration example of a total of four component supply units included in each of two mounting units provided in the component mounting machine, and (b) shows the number of mounted various component cassettes and the position on the Z axis in the configuration. FIG. It is a figure and a table showing an example of a position (Z-axis) of a component supply part to which 10 nozzle heads can adsorb. FIG. 3 is a diagram illustrating examples of various chip-type electronic components to be mounted. FIG. 4 is a diagram illustrating an example of a carrier tape containing components and a supply reel thereof. FIG. 4 is a diagram illustrating an example of a component cassette in which taping electronic components are mounted. FIG. 3 is a block diagram illustrating a hardware configuration of the optimization device. FIG. 10 is a diagram illustrating an example of contents of mounting point data illustrated in FIG. 9. FIG. 10 is a diagram illustrating a content example of a component library illustrated in FIG. 9. FIG. 10 is a diagram illustrating an example of contents of mounting device information illustrated in FIG. 9; FIG. 3 is a functional block diagram illustrating a configuration of an optimization device. FIG. 10 is a module configuration diagram illustrating a functional configuration of an optimization program illustrated in FIG. 9. (A) is a figure for demonstrating the component group which a component group production | generation part produces | generates, (b) is a figure which shows the example of the component table produced | generated in the process of production | generation of a component group by a component group production | generation part. is there. FIG. 11 is a diagram illustrating a state of a process of assigning task groups to sub-equipments by a first LBM unit of a line balance optimizing unit. 9 shows a tact distribution before the line balance is optimized by the second LBM unit of the line balance optimizing unit, a state of movement of the task group by the optimization, and a tact distribution after the optimization. It is a flowchart which shows the optimization procedure of the line balance by the 2nd LBM part of a line balance optimization part. It is a flowchart which shows the schematic procedure of the mounting order optimization of a small component by the small component optimization part of a state optimization part. It is a figure for explaining a suction pattern. 5 shows a component histogram of a component for which a suction pattern is to be generated by the task group generation method, and a suction pattern generated from the component histogram. FIG. 9 shows a non-arranged portion in the component histogram and a suction pattern generated from an unarranged portion of the component histogram. 9 is a component histogram of a component for which a suction pattern is generated by a pruning method. FIG. 24 is a diagram illustrating a state in which components are picked up (cutted) in units of 10 components from the component histogram illustrated in FIG. 23. 25 is a component histogram for components left after the pruning illustrated in FIG. 24. FIG. 26 is a diagram illustrating a state in which a diagram according to the task group generation method is generated for the component histogram illustrated in FIG. 25. It is a suction pattern for the component type for which the Z-axis is determined by the trimming method. 28 is a component histogram corresponding to the suction pattern shown in FIG. 27 (reconstructed without changing the Z axis). It is a flowchart which shows the procedure of optimization of the mounting order of components by the random selection method. It is a state in which two mounting points are exchanged by a random selection method. It is a figure showing signs that a mounting order of parts is optimized by an intersection elimination method. FIG. 11 is a diagram illustrating a movement locus (mounting path) of the work head generated when optimizing the order of tasks by the return locus method. FIG. 11 is a diagram illustrating a movement locus of the working head generated by the return locus method when a plurality of suction patterns at the same position are included. (A) is a flowchart showing a procedure for optimizing a mounting order of general-purpose components by a general-purpose component optimizing unit, and (b) is a state vs. state for explaining an optimal solution search approach by the optimization. It is a figure showing the relation of tact. FIG. 35 is a flowchart showing a detailed procedure of optimization (steps S551 and S553) by the hill-climbing method shown in FIG. 35 is a flowchart showing a detailed procedure of optimization (step S552) by the multi-canonical method shown in FIG. 34 (a). (A) shows a specific example of the intermediate expression used by the general-purpose component optimization unit, and (b) to (e) show the meaning (conversion to the Z-axis array) of the intermediate expression shown in (a). It is. 6 is a component histogram for explaining the concept of optimization by the “task group method”. It is a flowchart of the optimization processing with respect to a small part. It is a component histogram which shows a mode of a pruning process. 6 is a component histogram showing a state of core processing. 9 is a component histogram after a pruning process and a core process are performed. FIG. 3 is an implementation path diagram for explaining the concept of optimization by the “intersection elimination method”. It is a figure showing movement of a work head for explaining the concept of "return optimization." 9 is a component histogram showing an outline of optimization under constraints on array fixation. FIG. 9 is a diagram illustrating a restricted area based on a limit of a movable range of the work head when mounting a component on the LL size board and the XL size board. 9 is a component histogram for explaining the concept of optimization for an LL size substrate. 9 is a component histogram for explaining a step (1) of optimization by the “pruning method”. 9 is a component histogram for explaining the step (2). 6 is a component histogram for explaining the step (3). 6 is a component histogram for explaining the same step (4). 6 is a component histogram for explaining the step (5). 9 is a component histogram for explaining the step (6). 6 is a component histogram for explaining the step (7). 6 is a component histogram for explaining the step (8). 9 is a component histogram for explaining the step (9). 6 is a component histogram for explaining the same step (10). 6 is a component histogram for explaining the same step (11). 9 is a component histogram for explaining the same step (12). 9 is a component histogram for explaining the same step (13). 9 is a component histogram for explaining the same step (14). 9 is a component histogram for explaining the same step (15). 6 is a component histogram for explaining the same step (16). It is a component histogram for explaining the same step (17). It is a component histogram for explaining the same step (18). It is a parts histogram for explaining the same step (19). 6 is a component histogram for explaining the step (20). 6 is a component histogram for explaining the same step (21). It is a component histogram for explaining the same step (22). It is a component histogram for explaining the same step (23). 9 is a component histogram for explaining steps (1) to (3) of optimization by cassette division using a parallelogram template. 6 is a component histogram for explaining the same steps (4) to (6). 9 is a component histogram for explaining the same steps (7) to (8). 9 is a component histogram for explaining a part of the step (9). 9 is a component histogram for explaining a remaining step (10) of the same step (9). 9 is a component histogram for explaining steps (1) to (3) of optimization by cassette division using a rectangular template. 6 is a component histogram for explaining the same steps (3) to (5). 6 is a component histogram for explaining a part of the step (5). 9 is a component histogram for explaining the remaining part of the step (5). FIG. 4 is a mounting path diagram illustrating the concept of optimization by the “intersection elimination method”. FIG. 4 is a mounting path diagram for explaining an algorithm of “intersection resolving method”. FIG. 14 is a mounting path diagram showing an application example of optimization by the “intersection eliminating method”. It is a movement locus diagram of the working head which shows the concept of optimization by "return optimization method". (A) is a diagram showing a “return” operation when there are a plurality of mounting points in the same component cassette, and (b) shows a return locus of a head when a “return optimization method” is applied. It is a simulation result figure. 9 is a component histogram for explaining the optimization step (1) under the constraint of fixed array for double cassettes. 9 is a component histogram for explaining the step (2). 6 is a component histogram for explaining the step (3). 6 is a component histogram for explaining the same step (4). 6 is a component histogram for explaining the step (5). 9 is a component histogram for explaining the step (6). 6 is a component histogram for explaining the step (7). 6 is a component histogram for explaining the step (8). 9 is a component histogram for explaining the step (9). 6 is a component histogram for explaining the same step (10). (A) and (b) are explanatory diagrams showing an example of the mounting time of the front sub-equipment and the rear sub-equipment when there is an empty space on the Z axis, and the line balance processing at that time. d) is an explanatory diagram showing an example of the mounting time of the front sub-equipment and the rear sub-equipment when there is no free space on the Z axis, and the line balance processing (swap processing) at that time. 9 is a component histogram for explaining a step (1) of optimization by a “reaping method” for a double cassette. 9 is a component histogram for explaining the step (2). 6 is a component histogram for explaining the step (3). 6 is a component histogram for explaining the same step (4). 6 is a component histogram for explaining the step (5). 9 is a component histogram for explaining the step (6). 6 is a component histogram for explaining the step (7). 6 is a component histogram for explaining the step (8). 9 is a component histogram for explaining the step (9). 6 is a component histogram for explaining the same step (10). 6 is a component histogram for explaining the same step (11). It is a figure for explaining an algorithm of a nozzle exchange. (A) is a table which shows the kind (number of nozzles which can be used) and the number of objects of a target, and (b) is a parts histogram which shows a processing process. It is. It is a figure showing the example of a display of a "main screen." It is a figure showing the example of a display of "". It is a figure showing the example of a display of an "open screen." FIG. 14 is a diagram illustrating a display example of a “cassette number setting screen”. FIG. 14 is a diagram illustrating a display example of a “part division setting screen”. FIG. 9 is a diagram illustrating a display example of a “nozzle number setting screen”. It is a figure showing the example of a display of a "nozzle station selection screen." FIG. 14 is a diagram illustrating a display example of an “option setting screen”. It is a figure showing the example of a display of a "Z-axis information screen." It is a figure showing the example of a display of a "nozzle station information screen." It is a flowchart which shows the process sequence of the algorithm which determines an adsorption | suction pattern (Z arrangement | sequence) efficiently without dividing a component. FIG. 118 illustrates an arrangement of component tapes for describing the processing procedure in the flowchart illustrated in FIG. 118. FIG. 118 is a diagram for explaining the level of optimization by the optimization algorithm shown in FIG. 118, and shows a component histogram in which component tapes are simply arranged in descending order of the number of members (from right to left). FIG. 120 shows a pattern of the number of times of up and down suction when the component histogram shown in FIG. 120 is cut off. 118 shows the component histograms rearranged in the procedure shown in FIG. 118. FIG. 123 illustrates a suction up / down frequency pattern when the component histogram illustrated in FIG. 122 is trimmed. It is a flowchart which shows the procedure of the distribution process of a mountain to right and left block. 124 shows the state of the process according to the flowchart shown in FIG. 124. FIG. 124 shows the state of processing in another case according to the flowchart shown in FIG. 124. It is a flowchart which shows the processing procedure of the algorithm which estimates the used number of double cassettes. The breakdown of the component tape belonging to a certain component group is shown. A calculation example of the required number of double cassettes is shown. It is a flowchart which shows the processing procedure of the optimization algorithm of Z arrangement | sequence which considered the fixed pair of the double cassette. FIG. 140 illustrates the state of the process in step S660 in FIG. 130. FIG. 140 illustrates the state of the process in step S661 in FIG. 130. FIG. 140 illustrates the state of the process in step S662 in FIG. 130. FIG. 140 illustrates the state of the process in step S664 in FIG. 130. It is a flowchart which shows the processing procedure of the optimization algorithm in consideration of NG head. FIG. 10 is a diagram for comparing and explaining respective suction patterns when there is no NG head and when there is an NG head, and shows a target component histogram; 136 shows a suction pattern (by the trimming process and the core process) for the component histogram shown in FIG. 136 when there is no NG head. FIG. 136 shows a suction pattern for the component histogram shown in FIG. 136 when the mounting head with the head number 2 is an NG head. 11 is a flowchart illustrating an overall processing procedure when simultaneously optimizing a plurality of NC data. It is a flowchart which shows the processing procedure which optimizes the Z arrangement | sequence of several NC data simultaneously. FIG. 9 is a diagram illustrating a specific example illustrating three types of methods used for determining an initial Z array. It is a flowchart which shows the processing procedure of the generation algorithm of the initial task by "stealing method". FIG. 142 is a diagram illustrating a specific example of the first half process (S720 to S722) in the flowchart illustrated in FIG. 142. FIG. 142 is a diagram illustrating a specific example of the second half process (S723 to S726) in the flowchart illustrated in FIG. 142. It is a figure showing the effect of the optimization by the “stealing method”. It is a flowchart which shows the processing procedure of the optimization algorithm of the nozzle replacement operation by "task division". FIG. 149 is a diagram illustrating a specific example of the processing in the flowchart illustrated in FIG. 146. It is a flowchart which shows the processing procedure of the optimization algorithm by "task fusion". FIG. 149 is a diagram illustrating a specific example of the processing in the flowchart illustrated in FIG. 148. An example of a nozzle pattern before optimization by “task replacement” is shown. It is a flowchart which shows the processing procedure of the optimization algorithm by "task replacement". An example of a nozzle pattern obtained by optimization by "task replacement" is shown. FIG. 7 is a diagram illustrating restrictions on nozzle replacement based on a movable range of a multi-mounting head on a nozzle station. FIG. 9 is a diagram illustrating restrictions on component suction based on the movable range of the multi-mounting head on the feeder supply unit. It is a flowchart which shows the processing procedure at the time of checking a possible solution when the nozzle arrangement on a nozzle station is given. An example of two nozzle patterns prepared when the number of used nozzles is six is shown. 157 is a flowchart illustrating timing of nozzle replacement when a component is sucked using the nozzle pattern illustrated in FIG. 156.

Explanation of reference numerals

Reference Signs List 10 component mounting system 20 circuit board 100 component mounter 110 mounting unit 112 work head 112a to 112b suction nozzle 113 XY robot 114 component cassette 115a, b, 225a, b component supply unit 116 recognition camera 117 tray supply unit 118 shuttle conveyor 119 nozzle Station 120 Mounting unit 300 Optimization device 301 Operation control unit 302 Display unit 303 Input unit 304 Memory unit 305 Optimization program storage unit 306 Communication I / F unit 307 Database unit 307a Mounting point data 307b Parts library 307c Mounting device information 314 Parts group Generation unit 315 Line balance optimization unit 315a First LBM unit 315b Second LBM unit 315c Third LBM unit 316 State optimization unit 316a Section 316b general components optimization unit 316c optimizes the engine unit 316c optimization engine unit

Claims (8)

From a row of component cassettes containing components, a component group is sucked by a mounting head capable of sucking a maximum of n (≧ 2) components, and the mounting head is moved by an XY robot and mounted on a board. A component mounter,
In the case where the task is a component group mounted by a series of one operation of mounting the component group on the board after sucking and moving the component group by the mounting head,
When the mounting head finishes mounting all the components belonging to one task on the board, among the unmounted tasks, the next task is to mount the task that moves the minimum distance for picking up the component. And a component mounter. It is intended for a component mounter that has a row of component cassettes that store component tapes that make a group of components of the same type into one component tape, and a mounting head that picks up components from the row and mounts them on a board. A method of optimizing the order of mounting components by executing a computer,
In the task group consisting of a sequence of a plurality of tasks each of which is a component mounted by one series of operations in a series of operations of suction, movement, and mounting of components by the mounting head, all tasks belonging to a certain task When the position of the mounting head immediately after mounting the component on the board is the final mounting point of the task, and the position of the mounting head when picking up a component belonging to a certain task is the suction position of the task,
Any one of the first tasks belonging to the task group is selected, the final mounting point of the first task is specified, and of all the tasks except the first task, the task adsorption position is the first task. Specifying the second task closest to the final mounting point, specifying the final mounting point of the second task and specifying the third task in the same manner, the first task is specified again. Repeating, until the first task is identified again, a shortest cyclic partial path identifying step of identifying a group of those tasks as the shortest cyclic partial path;
A repeating step of repeating the shortest cyclic partial route specifying step until all tasks belonging to the task group are specified;
And arranging the shortest cyclic partial paths specified by the repetition in the repetition step. The component mounting order optimizing method is further configured such that the distance that the mounting head has to move in order to start the next shortest cyclic partial path after completing the mounting of all components belonging to a certain shortest cyclic partial path is minimized. And determining the order of the first task in each of the shortest traveling partial paths and the shortest traveling partial paths, and changing the order of the tasks so that the determined tasks are arranged. The described component mounting order optimization method. It is intended for a component mounter that has a row of component cassettes that store component tapes that make a group of components of the same type into one component tape, and a mounting head that picks up components from the row and mounts them on a board. A method of optimizing the order of mounting components by executing a computer,
A task group generating step of generating, by the mounting head, a task group consisting of a sequence of a plurality of tasks each of which is a component mounted by one series of operations in a series of operations of picking up, moving, and mounting components; ,
For each of the generated task groups, the order of the tasks in the task group is changed so that the time required for mounting all the components that compose the task group is minimized, and the mounting order of the components corresponding to the obtained task order And a task replacement step that optimizes
The position of the mounting head immediately after all the components belonging to a certain task have been mounted on the board is the final mounting point of the task, and the position of the mounting head when picking up the components belonging to a certain task is the suction position of the task. If you do
In the task replacement step,
Any one of the first tasks belonging to the task group is selected, the final mounting point of the first task is specified, and of all the tasks except the first task, the task adsorption position is the first task. Specifying the second task closest to the final mounting point, specifying the final mounting point of the second task and specifying the third task in the same manner, the first task is specified again. Repeating, until the first task is identified again, a shortest cyclic partial path identifying step of identifying a group of those tasks as the shortest cyclic partial path;
A repeating step of repeating the shortest cyclic partial route specifying step until all tasks belonging to the task group are specified;
And arranging the shortest cyclic partial paths specified by the repetition in the repetition step. It is intended for a component mounter that has a row of component cassettes that store component tapes that make a group of components of the same type into one component tape, and a mounting head that picks up components from the row and mounts them on a board. A mounting order optimizing device that optimizes a mounting order of components by executing a computer,
In the task group consisting of a sequence of a plurality of tasks each of which is a component mounted by one series of operations in a series of operations of suction, movement, and mounting of components by the mounting head, all tasks belonging to a certain task When the position of the mounting head immediately after mounting the component on the board is the final mounting point of the task, and the position of the mounting head when picking up a component belonging to a certain task is the suction position of the task,
Any one of the first tasks belonging to the task group is selected, the final mounting point of the first task is specified, and of all the tasks except the first task, the task adsorption position is the first task. Specifying the second task closest to the final mounting point, specifying the final mounting point of the second task and specifying the third task in the same manner, the first task is specified again. Until the first task is identified again, the shortest cyclic partial path identifying means for identifying a group of those tasks as the shortest cyclic partial path;
Repeating means for repeating the processing by the shortest cyclic partial route specifying means until all tasks belonging to the task group are specified;
An arrangement unit for arranging the shortest cyclic partial paths specified by the repetition by the repetition unit. It is intended for a component mounter that has a row of component cassettes that store component tapes that make a group of components of the same type into one component tape, and a mounting head that picks up components from the row and mounts them on a board. A program for optimizing the order of mounting components by executing a computer,
A non-transitory computer-readable storage medium storing a program for causing a computer to execute steps included in the component mounting order optimizing method according to claim 2.   A computer-readable recording medium on which the program according to claim 6 is recorded. What is claimed is: 1. A component mounting machine comprising: a row of component cassettes storing component tapes each having a group of components of the same type as one component tape; and a mounting head that sucks components from the row and mounts them on a substrate. ,
A component mounter that mounts components in a component mounting order optimized by the component mounting order optimizing method according to any one of claims 2 to 4.

Images