JP7489618B2 - Placement support method, trained model generation method, program, placement support system, and work system - Google Patents

Description

The present disclosure generally relates to a placement assistance method, a method for generating a trained model, a program, a placement assistance system, and a work system. More specifically, the present disclosure relates to a placement assistance method, a method for generating a trained model, a program, a placement assistance system, and a work system that assist in determining the placement of multiple components at multiple supply units for a component mounting system that mounts multiple components supplied to multiple supply units on a target object.

Patent Document 1 describes an operation analysis device that monitors a component mounting line to prevent a decrease in the line mounting tact time or line operating rate, and analyzes and removes the causes of the decrease, thereby maintaining high production efficiency of the entire line.

In this operation analysis device, in order to eliminate the cause of the line mounting tact, which is the mounting tact actual value of the bottleneck component mounter that has the largest mounting tact actual value in the component mounting line, not reaching the target value, the arrangement of component supply devices or the component mounting order are optimized to reduce tact loss. As a result, if the cause is, for example, an imbalance in the mounting tact actual values of each component mounter, the allocation of components to each mounter can be corrected to reduce tact loss and maintain the line mounting tact at the target value.

JP 2002-111298 A

However, even if Patent Document 1 corrects the allocation of parts to each mounting machine, it does not clearly disclose how to allocate parts to each mounting machine, and improvements are required regarding the placement of parts (including allocation).

The present disclosure has been made in consideration of the above-mentioned reasons, and aims to provide a placement assistance method capable of proposing improved part placement, a method for generating a trained model, a program, a placement assistance system, and an operation system.

A placement assistance method according to one aspect of the present disclosure is a method for assisting in the determination of placement of multiple components in multiple supply units in a component mounting system, and includes an acquisition step and a proposal step. The component mounting system is a system that generates a product by mounting the multiple components supplied to the multiple supply units on an object. The acquisition step is a step of acquiring mounting data related to the multiple components included in the product. The proposal step is a step of proposing, using a trained model, a placement of the multiple components for the multiple supply units from the mounting data so as to shorten the mounting time required to generate the product in the component mounting system. The trained model is a model generated in correspondence with the characteristics of at least one of the multiple supply units.

A method for generating a trained model according to one aspect of the present disclosure generates a trained model to be used in the placement assistance method by associating it with the characteristics of at least one of the multiple supply units.

A program relating to one aspect of the present disclosure is a program for causing one or more processors to execute the placement assistance method or the method for generating the trained model.

A placement assistance system according to one aspect of the present disclosure is a system for assisting in determining the placement of multiple components at multiple supply units in a component mounting system, and includes an acquisition unit and a proposal unit. The component mounting system is a system that generates a product by mounting the multiple components supplied to the multiple supply units on a target object. The acquisition unit acquires mounting data related to the multiple components included in the product. The proposal unit uses a trained model to propose a placement of the multiple components for the multiple supply units from the mounting data so as to shorten the mounting time required to generate the product in the component mounting system. The trained model is a model associated with the characteristics of at least one of the multiple supply units.

A working system according to one aspect of the present disclosure includes the placement assistance system and the component mounting system. The component mounting system is used to place the multiple components proposed by the placement assistance system to generate the product.

FIG. 1 is a schematic diagram showing the appearance of a component mounting system to which a placement support method according to a first embodiment is applied. FIG. 2 is a conceptual diagram showing the placement of components in a supply unit of the component mounting system according to the placement support method. FIG. 3 is a block diagram showing an overview of a work system including the placement assistance system according to the first embodiment. 4A and 4B are conceptual diagrams showing how the placement of a plurality of components in a supply section is proposed by the placement support method. FIG. 5 is a flowchart showing an example of the placement support method. FIG. 6 is an explanatory diagram showing a schematic diagram of the allocation of parts in a multi-stage manner according to the placement support method. FIG. 7 is an explanatory diagram that shows a schematic view of rearrangement according to the above-mentioned arrangement support method.

(Embodiment 1)
Hereinafter, the placement assistance method, the program, the placement assistance system 20 (see FIG. 3), and the work system 200 (see FIG. 3) according to this embodiment will be described with reference to FIGS. 1 to 7. FIG.

1, the placement assistance method according to this embodiment is a placement assistance method that supports the placement of multiple components P1 in multiple supply units 3 in a component mounting system 100. The component mounting system 100 has multiple supply units 3 for supplying the components P1 used in the component mounting system 100, and generates a product P3 by mounting the multiple components P1 supplied to these multiple supply units 3 on a target object P2.

That is, as shown in FIG. 1, the component mounting system 100 includes at least one mounting machine (mounting device) 10 for mounting a component P1 on an object P2. The mounting machine 10 has a capture unit 41 (see FIG. 1) for capturing the component P1. The capture unit 41 is, for example, a suction nozzle, and captures (holds) the component P1 in a state in which it can be released (i.e., released from capture). The mounting machine 10 captures the component P1 supplied to the supply unit 3 with the capture unit 41, moves the capture unit 41 with the component P1 captured, and releases the component P1 above the object P2, thereby mounting the component P1 on the mounting surface of the object P2.

In this embodiment, as an example, the component mounting system 100 is a mounting line configured by connecting multiple mounting machines 10 (four in the example of FIG. 1). In other words, the component mounting system 100 includes multiple mounting machines 10 that are connected to each other to configure a mounting line. In such a component mounting system 100 (mounting line), for example, multiple mounting machines 10 sequentially mount components P1 on one target object P2, ultimately generating a product P3 on which multiple components P1 are mounted.

Each of the multiple mounting machines 10 has at least one supply unit 3 and mounts the component P1 supplied to the supply unit 3. That is, each mounting machine 10 mounts one or more components P1 supplied to its own supply unit 3 onto the target object P2. In the example of FIG. 1, each of the multiple mounting machines 10 has multiple (here, two) supply units 3. In this embodiment, since the component mounting system 100 includes multiple mounting machines 10, even if each of the multiple mounting machines 10 has one supply unit 3, there will be multiple supply units 3 when viewed as the component mounting system 100 as a whole.

The component mounting system 100 is used in manufacturing operations for various products such as electronic devices, automobiles, clothing, food, pharmaceuticals, and crafts in facilities such as factories, laboratories, offices, and educational facilities.

In such a component mounting system 100, there are many patterns for arranging the multiple components P1 in the multiple supply units 3 provided in the multiple mounting machines 10. In other words, since it is possible to select which of the multiple mounting machines 10 the multiple components P1 should be arranged in, the number of arrangement patterns for the multiple components P1 in the multiple supply units 3 becomes enormous if the number of components P1 is, for example, several hundred. The mounting time required to generate the product P3 in the component mounting system 100 varies greatly depending on the arrangement of the multiple components P1 in the multiple supply units 3. For example, the mounting time also varies greatly depending on how the multiple components P1 are distributed to the multiple mounting machines 10. Thus, although the arrangement of the multiple components P1 in the multiple supply units 3 is an important factor in the mounting time of the component mounting system 100, the task of finding an appropriate arrangement is very difficult because there are many arrangement patterns.

Therefore, the placement assistance method according to this embodiment assists in the task of determining the placement of multiple components P1 in multiple supply units 3 in the component mounting system 100 as described above. This placement assistance method makes it possible to find a more appropriate placement more easily (in a short time), and, for example, makes it possible to determine the placement of multiple components P1 in multiple supply units 3 without the assistance of an expert or the like.

In other words, the placement assistance method according to this embodiment is a method for proposing a placement of multiple components P1 in multiple supply units 3 for the component mounting system 100. The component mounting system 100 generates a product P3 by mounting multiple components P1 supplied to the multiple supply units 3 on a target object P2. Here, the placement assistance method according to this embodiment further employs the following three functions to enable the proposal of an improved placement of components P1.

The first function is to propose a placement using the trained model M1 (see FIG. 3). To realize the first function, the placement support method has an acquisition step and a proposal step. The acquisition step is a step of acquiring mounting data D1 (see FIG. 3). The mounting data D1 is data regarding multiple components P1 included in the product P3. The proposal step is a step of proposing a placement of multiple components P1 for multiple supply units 3 from the mounting data D1 using the trained model M1 so as to shorten the mounting time. The trained model M1 is a model generated in correspondence with the characteristics of at least one supply unit 3 out of the multiple supply units 3. The mounting time is the time required for the component mounting system 100 to generate the product P3.

In this way, in the first function, the placement support method uses the trained model M1 in the proposal step to propose the placement of multiple components P1 for multiple supply units 3. In other words, in the proposal step, rather than proposing a uniquely determined placement, it is possible to more easily find an appropriate placement by, for example, realizing the skill of "deciding" the placement as an expert does in the trained model M1. In particular, since the trained model M1 is a model generated in correspondence with the characteristics of at least one of the multiple supply units 3, the placement of multiple components P1 is proposed, for example, taking into account the performance or specifications of the mounting machine 10. Therefore, this placement support method makes it possible to propose an improved placement of components P1.

The second function is the allocation of parts P1 in a multi-stage manner. To achieve the second function, the placement support method has a primary proposal step and a secondary proposal step. The primary proposal step is a step of proposing the placement of multiple parts P1 for multiple primary groups G11, G12 (see FIG. 6). The secondary proposal step is a step of proposing the placement of multiple parts P1 for multiple secondary groups G21 to G24 when each of the multiple primary groups G11, G12 is classified into multiple secondary groups G21 to G24 (see FIG. 6) after the primary proposal step. Each of the multiple secondary groups G21 to G24 includes at least one supply unit 3 out of the multiple supply units 3.

Here, the primary proposal step and the secondary proposal step in the second function are included in the proposal step in the first function. In other words, the proposal step includes the primary proposal step and the secondary proposal step.

Thus, in the second function, the proposal for the placement of multiple parts P1 is divided into a primary proposal step in which parts P1 are placed in multiple primary groups G11 and G12, and a secondary proposal step in which parts P1 are placed in multiple smaller secondary groups G21 to G24. In short, the placement of multiple parts P1 is achieved not in a single stage process but in a hierarchically set multi-stage process, so that at each stage, the number of options for deriving an appropriate placement of multiple parts P1 can be kept small. As a result, the process for proposing the placement of multiple parts P1 at each stage can be simplified. Therefore, this placement support method makes it possible to propose an improved placement of parts P1.

The third function is rearrangement. To realize the third function, the placement support method has a proposal step, a specification step, and a rearrangement step. The proposal step is a step of proposing the placement of multiple components P1 for multiple supply units 3. The specification step is a step of specifying a certain supply unit 3 among the multiple supply units 3 as a target supply unit. The supply unit 3 specified as the target supply unit is a supply unit 3 that satisfies a specific condition regarding the time required to generate the product P3 in the component mounting system 100 when the placement of the multiple components P1 proposed in the proposal step is adopted. The rearrangement step is a step of proposing the placement of multiple components P1 for the target supply unit and the reference supply unit. The reference supply unit consists of at least one supply unit 3 other than the target supply unit among the multiple supply units 3.

In this way, in the third function, even if the placement of the multiple components P1 proposed in the proposal step is not optimal, a new placement of the multiple components P1 can be proposed in the rearrangement step. Moreover, the supply unit 3 to be rearranged in the rearrangement step is a target supply unit among the multiple supply units 3 that satisfies a specific condition regarding the time required to generate the product P3 in the component mounting system 100 when the placement proposed in the proposal step is adopted. In other words, in the rearrangement step, it is possible to re-propose a placement of the supply unit 3 that satisfies a certain condition regarding the time required to generate the product P3 in the component mounting system 100, particularly among the placements proposed in the proposal step. Therefore, this placement support method makes it possible to propose an improved placement of the components P1.

In addition, the placement assistance method according to this embodiment is executed by a placement assistance system 20, as an example. In other words, the placement assistance system 20 is one aspect for realizing the placement assistance method described above. That is, the placement assistance system 20 assists in determining the placement of multiple components P1 in multiple supply units 3 for the component mounting system 100. As described above, the component mounting system 100 is a system that generates a product P3 by mounting multiple components P1 supplied to multiple supply units 3 on an object P2, and in this embodiment, as an example, is a mounting line.

Here, the placement assistance system 20 mainly comprises a computer system having one or more processors and one or more memories. In other words, the placement assistance method according to this embodiment is used on a computer system (placement assistance system 20). In other words, the placement assistance method can also be embodied as a program. The program according to this embodiment is a program for causing one or more processors to execute the placement assistance method according to this embodiment.

Furthermore, the work system 200 according to this embodiment includes a placement assistance system 20 and a component mounting system 100. The component mounting system 100 is a system that is used to place multiple components P1 proposed by the placement assistance system 20 and generates a product P3.

(2) Details The placement assistance method, the program, the placement assistance system 20, and the work system 200 according to this embodiment will be described in detail below.

(2.1) Premise In this embodiment, a case will be described in which the component mounting system 100 is used in the manufacture of electronic devices in a factory. A typical electronic device has various circuit blocks, such as a power supply circuit and a control circuit. In the manufacture of these circuit blocks, as an example, a solder application process, a mounting process, and a soldering process are performed in this order. In the solder application process, cream solder is applied (or printed) to a board (including a printed wiring board). In the mounting process, a component P1 (including an electronic component) is mounted (mounted) on the board. In the soldering process, for example, the board on which the component P1 is mounted is heated in a reflow furnace to melt the cream solder and perform soldering.

In the mounting process, the component mounting system 100 mounts multiple components P1 (electronic components) on a substrate, which is the target object P2. As a result, the component mounting system 100 generates a product P3 consisting of a substrate (target object P2) on which multiple components P1 are mounted. In other words, the product P3 includes the target object P2 (here, the substrate) and multiple components P1 mounted on the target object P2.

In this embodiment, as an example, a case will be described in which the component mounting system 100 is used to mount a component (component P1) using surface mount technology (SMT: Surface Mount Technology). That is, the component as component P1 is a surface mount component (SMD: Surface Mount Device) and is mounted by being placed on the surface (mounting surface) of the substrate as the object P2. However, this is not limited to this example, and the component mounting system 100 may also be used to mount a component (component P1) using insertion mount technology (IMT: Insertion Mount Technology). In this case, the component as component P1 is a component for insertion mounting having a lead terminal, and is mounted on the surface (mounting surface) of the substrate (object P2) by inserting the lead terminal into a hole in the substrate as the object P2.

In addition, the "supply unit" in this disclosure is a portion (location) of the component mounting system 100 for supplying the components P1 used in the component mounting system 100. The component mounting system 100 has multiple supply units 3, and performs mounting work using the components P1 supplied to each supply unit 3. For example, the supply unit 3 is configured to be able to mount a tape feeder for supplying the components P1. The tape feeder is capable of supplying the multiple components P1 from the supply unit 3 to the component mounting system 100 by attaching a reel of carrier tape containing multiple components P1. This type of supply unit 3 includes multiple slots to which tape feeders can be attached, and multiple tape feeders are attached to these multiple slots to supply multiple components P1 to the component mounting system 100 (mounting machine 10). In this embodiment, as an example, it is assumed that the supply unit 3 is a "table" consisting of a collection of multiple slots. The "table" referred to here is a virtual table and does not have an actual entity.

In the following, as an example, three mutually orthogonal axes, the X-axis, the Y-axis, and the Z-axis, are set, and the axes parallel to the surface of the substrate, which is the object P2, are the "X-axis" and "Y-axis", and the axis parallel to the thickness direction of the substrate is the "Z" axis. In particular, the "X-axis" is an axis along the direction in which the multiple mounting machines 10 in the component mounting system 100 are arranged. Furthermore, the side of the capture unit 41 as seen from the substrate, which is the object P2, is defined as the positive direction of the Z-axis (also referred to as "above"). In addition, the state as seen from the positive direction of the Z-axis (above) is also referred to as "planar view" below. The X-axis, Y-axis, and Z-axis are all virtual axes, and the arrows indicating "X", "Y", and "Z" in the drawings are merely indicated for the purpose of explanation, and none of them have any substance. In addition, these directions are not intended to limit the directions when the component mounting system 100 is used.

In addition, the proposal of the "placement" of the parts P1 in this disclosure includes proposals of various items related to the placement of the parts P1 in the multiple supply units 3, and includes, for example, both the allocation of the parts P1 and the proposal of the arrangement of the parts P1. The allocation of the parts P1 means dividing the parts P1 into multiple parts and allocating one or more parts P1 after the division to one of the multiple supply units 3 (tables in this embodiment). In other words, when multiple parts P1 are allocated to one supply unit 3 in the allocation, the order of arrangement of the multiple parts P1 is not specified. On the other hand, the proposal of the arrangement of the multiple parts P1 means dividing the parts P1 into multiple parts and allocating one or more parts P1 after the division to one of the multiple supply units 3 (tables in this embodiment), and even specifying the order (arrangement) within each supply unit 3. Therefore, when the supply unit 3 is a "table" consisting of a set of multiple slots as in this embodiment, the arrangement of the parts P1 specifies which parts P1 correspond to which slots of the multiple slots included in one supply unit 3 (table). In other words, in the proposed arrangement, when multiple parts P1 are allocated to one supply unit 3, even the order in which these multiple parts P1 are arranged is specified.

In this embodiment, as an example, it is assumed that the "placement" proposal of part P1 made by the placement support method or placement support system 20 is only the allocation of part P1 out of the proposals for the allocation and arrangement of part P1. That is, in the placement support method or placement support system 20 according to this embodiment, a plurality of parts P1 are divided into a plurality of parts, and one or more parts P1 after the division are assigned to one of a plurality of supply units 3, but the order in which the one or more parts P1 are arranged after the division is not specified.

In addition, the aspect of "proposing" the placement in this disclosure includes both an aspect in which the derivation result of the placement of component P1 is output to a person or an external system, and an aspect in which the placement of multiple components P1 in multiple supply units 3 is actually determined according to the derivation result of the placement of component P1. According to the former aspect, the person or external system that receives the derivation result of the placement of component P1 from the placement assistance method or placement assistance system 20 merely receives a proposal (suggestion) of the placement of component P1 and decides whether or not to adopt this derivation result. According to the latter aspect, the derivation result of the placement of component P1 from the placement assistance method or placement assistance system 20 is adopted as is by the component mounting system 100 without the intervention of a person or an external system.

In addition, the "characteristics" of the supply unit 3 in this disclosure include, for example, characteristics of the mounting machine 10 in which the supply unit 3 is provided. The characteristics of the mounting machine 10 include, for example, the type, performance, characteristics, or specifications of the mounting machine 10, such as the speed (maximum mounting speed), the number of capture units 41 (number of nozzles), the number of supply units 3, the type of capture unit 41, the number of slots, the number of mounting heads 4, and the model. The "number of capture units 41" means the maximum number of capture units 41 that can be attached to one mounting head 4. The "type of capture unit 41" includes, for example, a suction nozzle that adsorbs the component P1 and a mechanical chuck that clamps (picks) the component P1. The "number of slots" means the number of slots in one supply unit 3 (or one mounting machine 10), in other words, the maximum number of tape feeders that can be attached to one supply unit 3 (or one mounting machine 10).

In addition, the "speed" of the mounting machine 10 in this disclosure means the maximum mounting speed of the mounting machine 10, that is, the maximum speed for mounting the component P1 in the mounting machine 10. In other words, the mounting machine 10 operates to capture the component P1 with the capture unit 41, and to move the capture unit 41 while it has captured the component P1, so that the component P1 is mounted on the target P2. Therefore, the speed (maximum mounting speed) of the mounting machine 10 varies depending on, for example, the speed at which the capture unit 41 captures and releases (i.e., releases the capture) the component P1, and the movement speed of the capture unit 41. However, in this embodiment, the speed is classified into a number of speed ranges, such as a multi-function machine (low speed machine), a medium speed machine, and a high speed machine, and if there is a slight difference in speed, all of them are considered to be in the same speed range, such as being considered a "high speed machine."

Furthermore, the "specific condition" in this disclosure is a specific condition that is set with respect to the time required to generate the product P3 in the component mounting system 100. As an example, a specific condition may be that the time required to generate the product P3 in the component mounting system 100 (mounting time) is the longest or shortest among the multiple supply units 3.

(2.2) Configuration of Component Mounting System Next, the configuration of a component mounting system 100 in which the placement support method according to this embodiment is used will be described with reference to FIGS. 1 and 2. FIG.

As described above, the component mounting system 100 according to this embodiment includes multiple (four in this embodiment) mounting machines 10 that make up a mounting line. The multiple mounting machines 10 are arranged in a line along the X-axis. When distinguishing between the multiple mounting machines 10, each of the multiple mounting machines 10 is also referred to as the first mounting machine 11, the second mounting machine 12, the third mounting machine 13, and the fourth mounting machine 14, in order from the negative side of the X-axis, as shown in FIG. 1.

In the component mounting system 100 according to the present embodiment, the object P2 moves through the multiple mounting machines 10 in the order of the first mounting machine 11 at the front, the second mounting machine 12, the third mounting machine 13, and the fourth mounting machine 14 at the rear. In other words, the object P2 passes through the multiple mounting machines 10 in sequence while moving in the positive direction of the X-axis. In the component mounting system 100, while the object P2 passes through the multiple mounting machines 10, each mounting machine 10 mounts multiple components P1 on the object P2. In this way, the component mounting system 100 introduces the substrate, which is the object P2, from the first mounting machine 11 side, mounts multiple components P1 (electronic components) on the object P2 using the multiple mounting machines 10, and then discharges the product P3 from the fourth mounting machine 14 side.

Therefore, the component mounting system 100 can generate a product P3 on which multiple components P1 are mounted by sequentially mounting components P1 using multiple mounting machines 10. The "mounting time" in this disclosure is the time required for the component mounting system 100 to generate the product P3, and is, for example, the time from when one target object P2 is introduced into the component mounting system 100 to when it is discharged from the component mounting system 100 as the product P3. In other words, the "mounting time" is the time from when a single board, which is the target object P2, is introduced into the first mounting machine 11 to when multiple components P1 are mounted on this board (target object P2) and it is discharged from the fourth mounting machine 14 as the product P3.

In this embodiment, each of the multiple mounting machines 10 has a common basic configuration. Therefore, unless otherwise specified, the following describes the configuration of only the first mounting machine 11, but the other mounting machines (the second mounting machine 12, the third mounting machine 13, and the fourth mounting machine 14) also have a basically similar configuration.

The mounting machine 10 (first mounting machine 11) has at least one supply unit 3 (table) and at least one mounting head 4. In this embodiment, the mounting machine 10 has two supply units 3. The two supply units 3 are located at both ends of the mounting machine 10 in the Y-axis direction. The mounting machine 10 also has two mounting heads 4. The two mounting heads 4 are arranged on either side of the Y-axis direction when viewed from the center line of the mounting machine 10 in the Y-axis direction, so that they correspond one-to-one to the two supply units 3, respectively.

In other words, the mounting machine 10 has a structure that is approximately symmetrical in the Y-axis direction, with the mounting heads 4 and the supply units 3 disposed on either side of the center line in the Y-axis direction. Therefore, the mounting machine 10 drives the two mounting heads 4 individually, so that the two mounting heads 4 mount the components P1 supplied to the two supply units 3 on the target object P2 introduced between the two supply units 3.

Thus, in this embodiment, one mounting machine 10 has two mounting heads 4, but since two supply units 3 (tables) are also provided, one mounting head 4 corresponds to one supply unit 3. In other words, each supply unit 3 corresponds to only one mounting head 4, and the "number of mounting heads 4" included in the characteristics of the supply unit 3 is "1".

The mounting head 4 has one or more capture portions 41. In this embodiment, the mounting head 4 in the first mounting machine 11 has multiple (for example, 16) capture portions 41 (see FIG. 1). With the capture portion 41 capturing a component P1 (component), the mounting head 4 moves the capture portion 41 closer to the target object P2 (board) and mounts the component P1 on the mounting surface T21 of the target object P2.

The supply unit 3 is supplied with a component P1 that is captured by the capture unit 41 of the mounting head 4. In this embodiment, as described above, the supply unit 3 is a "table" consisting of a collection of multiple slots, each of which can accommodate a tape feeder. Therefore, multiple tape feeders can be attached to the supply unit 3, up to the number of slots. The mounting head 4 captures the component P1 (component) from the tape feeder attached to the supply unit 3 with the capture unit 41.

In addition to the above configuration, the mounting machine 10 may further include a drive device, a transport device, a control device, a communication unit, etc. However, the drive device, transport device, control device, communication unit, etc. are not essential components of the mounting machine 10. In Figures 1 and 2, illustrations of components of the mounting machine 10 other than the supply unit 3 and the mounting head 4 are appropriately omitted.

The drive device is a device that moves the mounting head 4. In this embodiment, the drive device moves the mounting head 4 within the X-Y plane. The "X-Y plane" here is a plane that includes the X-axis and Y-axis and is perpendicular to the Z-axis. In other words, the drive device moves the mounting head 4 in the X-axis and Y-axis directions.

The conveying device is a device that conveys a substrate as the object P2. The conveying device is realized, for example, by a belt conveyor. The conveying device conveys the object P2 (substrate) along the X-axis. The conveying device conveys the object P2 to a mounting space at least below the mounting head 4, that is, facing the capture unit 41 in the Z-axis direction. The conveying device then stops the object P2 in the mounting space until the mounting head 4 has completed mounting of the component P1 (component) on the object P2 (substrate).

The control device controls each part of the component mounting system 100. The control device is mainly composed of a microcontroller having one or more processors and one or more memories. In other words, the functions of the control device are realized by the processor of the microcontroller executing a program recorded in the memory of the microcontroller. The program may be pre-recorded in the memory, may be provided via a telecommunications line such as the Internet, or may be recorded on a non-temporary recording medium such as a memory card and provided.

The communication unit is configured to communicate with a higher-level system directly or indirectly via a network or a repeater, etc. This allows the mounting machine 10 to send and receive data between the higher-level system and the mounting machine 10.

The mounting machine 10 configured as described above is used with tape feeders, each of which has a reel of carrier tape containing a plurality of components P1, attached to each of a plurality of slots in the supply unit 3. As a result, each of the plurality of slots in the supply unit 3 is sequentially supplied with components P1 taken out of the carrier tape by the tape feeder. In this state, the mounting machine 10 performs a basic operation of capturing components P1 supplied to the supply unit 3 by the tape feeder with the capturing unit 41 of the mounting head 4, and mounting (placing) them at the target position on the target object P2. The mounting machine 10 repeatedly performs such basic operations.

Here, if the mounting head 4 has multiple (16, for example) capture units 41, the mounting machine 10 can mount multiple components P1 with one basic operation. In other words, the mounting machine 10 can capture multiple components P1 collectively with the multiple capture units 41 in the mounting head 4 in one basic operation, and therefore can mount multiple components P1 collectively. Furthermore, a mounting machine 10 equipped with multiple (two, for example) mounting heads 4 can mount multiple components P1 simultaneously with these multiple mounting heads 4 by driving the multiple mounting heads 4 individually.

In this embodiment, the multiple mounting machines 10 included in the component mounting system 100 have some different characteristics, as described below. Here, the first mounting machine 11 and the second mounting machine 12 have the same characteristics, and the third mounting machine 13 and the fourth mounting machine 14 each have a different characteristic from the first mounting machine 11 (or the second mounting machine 12).

That is, since there are many types, performance, characteristics, or specifications of mounting machines 10, the multiple mounting machines 10 constituting the component mounting system 100 may have different characteristics. Specifically, the mounting machines 10 may differ in, for example, speed (maximum mounting speed), number of capture units 41 (number of nozzles), number of supply units 3, type of capture unit 41, number of slots, number and model of mounting heads 4, etc.

In this embodiment, the different characteristics of the multiple (here, four) mounting machines 10 include speed, number of capture units 41, type and model of capture units 41, etc. As an example, with regard to "speed," the first mounting machine 11 and second mounting machine 12 are "high-speed machines," whereas the third mounting machine 13 is a "medium-speed machine," and the fourth mounting machine 14 is a "multi-function machine." With regard to the "number of capture units 41," the first mounting machine 11 and second mounting machine 12 have "16," whereas the third mounting machine 13 has "8," and the fourth mounting machine 14 has "3."

On the other hand, common features of multiple (here, four) mounting machines 10 include the number of supply units 3, the number of slots, and the number of mounting heads 4. As an example, the "number of supply units 3" is "2" for all mounting machines 10, as described above. Similarly, as an example, the "number of mounting heads 4" is "2" for all mounting machines 10, as described above. In addition, it is assumed that the "number of slots" is "60" for all mounting machines 10.

However, since the "number of capture units 41" represents the number of capture units 41 in one mounting head 4, a mounting machine 10 having two mounting heads 4 will have twice as many capture units 41 as described above. For example, the number of capture units 41 in the entire first mounting machine 11 is "32". Similarly, since the "number of slots" represents the number of slots in one supply unit 3, a mounting machine 10 having two supply units 3 will have twice as many slots as described above. For example, the number of slots in the entire first mounting machine 11 is "120". In addition, the characteristics of the mounting machine 10 as described above may differ even if the model of the mounting machine 10 is the same, for example, when the detailed specifications of the mounting machine 10 are different, or when performance has deteriorated due to aging, etc.

As described above, in this embodiment, the component mounting system 100 includes four mounting machines 10, and each mounting machine 10 includes two supply units 3, so that the component mounting system 100 has eight supply units 3 as a whole. When distinguishing between these eight supply units 3, each of the multiple supply units 3 is also called the first supply unit 31, the second supply unit 32, the third supply unit 33, the fourth supply unit 34, the fifth supply unit 35, the sixth supply unit 36, the seventh supply unit 37, and the eighth supply unit 38, as shown in FIG. 2. The first supply unit 31 and the second supply unit 32 are provided in the first mounting machine 11, and the third supply unit 33 and the fourth supply unit 34 are provided in the second mounting machine 12. The fifth supply unit 35 and the sixth supply unit 36 are provided in the third mounting machine 13, and the seventh supply unit 37 and the eighth supply unit 38 are provided in the fourth mounting machine 14.

(2.3) Configuration of Placement Support System Next, the configuration of the placement support system 20 according to this embodiment will be described with reference to FIG.

As described above, the placement assistance system 20 is a system that assists in determining the placement of multiple components P1 in multiple supply units 3 for the component mounting system 100. In this embodiment, as described above, the component mounting system 100 as a whole has eight supply units 3, the first supply unit 31 to the eighth supply unit 38. Therefore, the placement assistance system 20 of this embodiment is used to determine the placement of multiple components P1 in these eight supply units 3.

As shown in FIG. 3, the placement assistance system 20 includes an acquisition unit 21, a proposal unit 22, an identification unit 23, a relocation unit 24, an output unit 25, a simulation unit 26, and a model storage unit 27. In this embodiment, the placement assistance system 20 is mainly configured as a computer system having one or more processors and one or more memories, as described above. Of the placement assistance system 20, the functions of at least the acquisition unit 21, the proposal unit 22, the identification unit 23, the relocation unit 24, the output unit 25, and the simulation unit 26 are embodied by one or more processors executing programs.

The acquisition unit 21 acquires mounting data D1. The mounting data D1 is data related to the multiple components P1 included in the product P3. In other words, the mounting data D1 is data related to all of the multiple components P1 mounted by the component mounting system 100. As an example, when "200" components P1 are mounted by the component mounting system 100, the mounting data D1 includes data related to each of these "200" components P1. The mounting data D1 includes at least data for distinguishing the multiple components P1, such as the type, product number, specifications, manufacturer, size, number of terminals, terminal arrangement, and circuit constants for each of the multiple components P1.

The suggestion unit 22 proposes an arrangement of multiple components P1 for multiple supply units 3. In this embodiment, the suggestion unit 22 uses the trained model M1 to propose an arrangement of multiple components P1 for multiple supply units 3 from the mounting data D1 so as to shorten the mounting time. The trained model M1 is a model generated in association with the characteristics of at least one of the multiple supply units 3. The mounting time is the time required for the component mounting system 100 to generate the product P3. The trained model M1 will be explained in detail in the "(3.2) Trained Model" section.

Here, the placement assistance system 20 (placement assistance method) according to this embodiment employs a multi-stage allocation (placement proposal) of the part P1. Therefore, in this embodiment, in order to accommodate the multi-stage allocation, the proposal unit 22 includes multiple processing blocks. Specifically, the proposal unit 22 includes a primary proposal unit 221, a secondary proposal unit 222, and a tertiary proposal unit 223.

The primary proposal unit 221 proposes the placement of the multiple parts P1 for the multiple primary groups G11, G12. The secondary proposal unit 222 proposes the placement of the multiple parts P1 for the multiple secondary groups G21 to G24 when each of the multiple primary groups G11, G12 is classified into multiple secondary groups G21 to G24 that are further divided. Each of the multiple secondary groups G21 to G24 includes at least one supply unit 3 out of the multiple supply units 3. The tertiary proposal unit 223 proposes the placement of the multiple parts P1 for the multiple tertiary groups G31 to G38 when each of the multiple secondary groups G21 to G24 is classified into multiple tertiary groups G31 to G38 (see FIG. 6) that are further divided. Each of the multiple tertiary groups G31 to G38 includes at least one supply unit 3 out of the multiple supply units 3. The multi-stage method will be described in detail in the section "(3.3) Multi-stage method".

When the arrangement of the multiple components P1 proposed by the suggestion unit 22 is adopted, the identification unit 23 identifies one of the multiple supply units 3 as a target supply unit. Here, the supply unit 3 identified as the target supply unit is a supply unit 3 that satisfies a specific condition regarding the time required to generate the product P3 in the component mounting system 100. In other words, the identification unit 23 identifies a supply unit 3 that satisfies a specific condition regarding the time required to generate the product P3 in the component mounting system 100 as a "target supply unit." The identification of the target supply unit will be explained in detail in the "(3.4) Rearrangement" section.

The rearrangement unit 24 proposes an arrangement of multiple components P1 to the target supply unit and the reference supply unit. Here, the reference supply unit consists of at least one supply unit 3 other than the target supply unit among the multiple supply units 3. Therefore, the identification unit 23 essentially extracts a supply unit 3 as a "target supply unit" and a supply unit 3 as a "reference supply unit" from among the multiple supply units 3. In other words, the identification unit 23 identifies at least one supply unit 3 among the multiple supply units 3 that does not satisfy a specific condition with respect to the time required to generate the product P3 in the component mounting system 100 as the "target supply unit."

The rearrangement unit 24 then rearranges the multiple parts P1 for the target supply unit and reference supply unit that have been extracted (identified) from among the multiple supply units 3. In other words, the rearrangement unit 24 proposes a new arrangement in place of the arrangement (allocation) of the multiple parts P1 for the target supply unit and reference supply unit that was once proposed by the proposal unit 22. In this embodiment, the rearrangement unit 24 resets the arrangement of the multiple parts P1 for the target supply unit and reference supply unit, and then proposes a new arrangement of the multiple parts P1. Rearrangement will be explained in more detail in the "(3.4) Rearrangement" section.

The output unit 25 outputs the placement data D2 generated by at least one of the proposal unit 22 and the rearrangement unit 24. In this disclosure, "placement data" is data representing the placement of multiple components P1 with respect to multiple supply units 3 proposed by at least one of the proposal unit 22 and the rearrangement unit 24. In other words, the placement data D2 is data representing the allocation of all of the multiple components P1 mounted by the component mounting system 100 to one of the supply units 3. As an example, when "200" components P1 are mounted by the component mounting system 100, the mounting data D1 includes data specifying which of the multiple (here, eight) supply units 3 each of these "200" components P1 will be placed on.

The output unit 25 may output the placement data D2 in the form of, for example, output to the simulation unit 26, transmission to the component mounting system 100 or other systems, display, audio output, recording (writing) to a non-temporary recording medium, and printing (printing out). In this embodiment, as an example, the output unit 25 directly or indirectly outputs the placement data D2 to the simulation unit 26. Furthermore, in this embodiment, as an example, the output unit 25 directly or indirectly outputs (transmits) the placement data D2 to the component mounting system 100. This enables the component mounting system 100 to mount multiple components P1 using the placement of multiple components P1 defined in the placement data D2. In addition, for example, according to a display or audio output mode, the output unit 25 can output the placement data D2 to a person such as a worker.

The simulation unit 26 simulates the operation of mounting multiple components P1 in the component mounting system 100. Here, the simulation unit 26 simulates the operation of the component mounting system 100 in the case where the arrangement of multiple components P1 proposed by at least one of the proposal unit 22 and the rearrangement unit 24 is adopted. In this embodiment, the simulation unit 26 acquires the arrangement data D2 from the output unit 25 and performs a simulation based on this arrangement data D2. At this time, the simulation unit 26 performs the simulation using device data representing the configuration of the component mounting system 100 in addition to the arrangement data D2. The "device data" here includes the characteristics of the supply unit 3, that is, the characteristics of the mounting machine 10 in which the supply unit 3 is provided.

The simulation unit 26 determines at least the "mounting time" required to generate the product P3 in the component mounting system 100. In particular, in this embodiment, the simulation unit 26 determines the mounting time for each of the multiple supply units 3 included in the component mounting system 100. In other words, the simulation unit 26 determines, as a simulation result, the mounting time required to mount the component P1 for each supply unit 3 for the component mounting system 100 including multiple supply units 3. The simulation results in the simulation unit 26 are output, for example, from the output unit 25 in any manner. In this embodiment, the simulation results in the simulation unit 26 are output at least to the identification unit 23 and used to identify the target supply unit in the identification unit 23.

The model storage unit 27 stores the trained model M1. The trained model M1 stored in the model storage unit 27 is used at least for proposing the placement of the component P1 in the proposal unit 22. In this embodiment, the trained model M1 stored in the model storage unit 27 is used not only for proposing the placement of the component P1 in the proposal unit 22 but also for proposing the placement of the component P1 in the rearrangement unit 24. As will be explained in detail in the "(3.2) Trained Model" section, the trained model M1 is generated in advance. Moreover, the trained model M1 is a model generated in association with the characteristics of at least one of the multiple supply units 3 in the component mounting system 100. Here, since the component mounting system 100 includes multiple supply units 3 having different characteristics, there are multiple types of trained models M1. Therefore, the model storage unit 27 is configured to be able to store multiple (types) of trained models M1.

In addition to the above configuration, the placement assistance system 20 may further include a communication unit and a user interface, etc. However, the communication unit and the user interface, etc. are not essential components of the placement assistance system 20. In Figure 3, the illustration of components of the placement assistance system 20 other than the acquisition unit 21, proposal unit 22, identification unit 23, relocation unit 24, output unit 25, simulation unit 26 and model storage unit 27 has been appropriately omitted.

The communication unit is configured to communicate with the component mounting system 100 and/or the higher-level system directly or indirectly via a network or a repeater, etc. This enables the placement assistance system 20 to exchange data with the component mounting system 100 and/or the higher-level system.

The user interface includes, for example, a touch panel display, and accepts user operations and presents (displays) information to the user. The user interface is not limited to a touch panel display, and may include input devices such as a keyboard, a pointing device, a mechanical switch, or a gesture sensor. The user interface may also include a voice input/output unit instead of or in addition to the touch panel display.

As described above, the placement assistance system 20 according to this embodiment constitutes the work system 200 together with the component mounting system 100. That is, the work system 200 according to this embodiment includes the placement assistance system 20 and the component mounting system 100, as shown in FIG. 3. The component mounting system 100 is a system that is used in the placement of multiple components P1 proposed by the placement assistance system 20, and generates a product P3.

(3) Placement Support Method The placement support method according to this embodiment will be described in more detail below. In this embodiment, as described above, the placement support method employs three functions, namely, placement proposal using the trained model M1, allocation of the parts P1 in a multi-stage manner, and rearrangement. Therefore, the placement support method according to this embodiment will be described below in four parts: a basic mode, placement proposal using the trained model M1, allocation of the parts P1 in a multi-stage manner, and rearrangement.

In the following, as an example, a case will be described in which 240 components P1 are allocated to the following eight supply units 3. That is, as described above in this embodiment, the component mounting system 100 has a total of four mounting machines 10, namely, a first mounting machine 11, a second mounting machine 12, a third mounting machine 13, and a fourth mounting machine 14. Furthermore, since each of these four mounting machines 10 has two supply units 3, when viewed as a whole, there are a total of eight supply units 3 to which multiple (here, 240) components P1 are allocated.

Here, it is assumed that the first mounting machine 11 and the second mounting machine 12 each have a characteristic that the number of capture units 41 is "16". Therefore, the first supply unit 31 and the second supply unit 32 provided in the first mounting machine 11, and the third supply unit 33 and the fourth supply unit 34 provided in the second mounting machine 12 each have a characteristic that the number of capture units 41 is "16". On the other hand, it is assumed that the third mounting machine 13 each has a characteristic that the number of capture units 41 is "8". Therefore, the fifth supply unit 35 and the sixth supply unit 36 provided in the third mounting machine 13 each have a characteristic that the number of capture units 41 is "8". Furthermore, it is assumed that the fourth mounting machine 14 each has a characteristic that the number of capture units 41 is "3". Therefore, the seventh supply unit 37 and the eighth supply unit 38 provided in the fourth mounting machine 14 each have a characteristic that the number of capture units 41 is "3".

However, these conditions are merely a specific example to explain the placement support method and are not intended to limit the scope of application of the placement support method.

(3.1) Basic Mode First, a basic mode of the placement support method according to the present embodiment will be described with reference to Fig. 4A, Fig. 4B, and Fig. 5. Fig. 4A and Fig. 4B are conceptual diagrams showing how the placement support method proposes placement of a plurality of components P1 in a plurality of supply units 3.

In the component mounting system 100 having multiple (here, eight) supply units 3 as in this embodiment, there are many patterns of arrangement of multiple components P1 in the multiple supply units 3. In particular, in this embodiment, as described above, the number of slots in each supply unit 3 is "60", so that the component mounting system 100 as a whole has 480 (= 60 x 8) slots. Therefore, the component mounting system 100 can mount up to 480 components P1, and if the number of components P1 actually mounted is several hundred, the arrangement patterns of the multiple components P1 in the multiple supply units 3 will be enormous. Moreover, the mounting time required to generate the product P3 in the component mounting system 100 varies greatly depending on the arrangement of the multiple components P1 in the multiple supply units 3.

In other words, even if the configuration of the component mounting system 100 is the same, if the arrangement of the multiple components P1 in the multiple supply units 3 is different, each component P1 will be mounted on a different mounting machine 10, and the time required to mount each component P1 will vary greatly. Therefore, for example, even if the mounting time required to generate one product P3 is shortened, it is very difficult to find an appropriate arrangement pattern for the multiple components P1 from a huge number of patterns. Therefore, the arrangement support method of this embodiment supports the task of determining the arrangement of the multiple components P1 in the multiple supply units 3 in the component mounting system 100, making it possible to find a more appropriate arrangement more easily (in a short time).

Here, as described above, in the placement assistance method according to this embodiment, the proposal for the "placement" of part P1 is only the allocation of part P1 out of the proposals for the allocation and arrangement of part P1. In other words, in this embodiment, multiple parts P1 are divided into multiple parts, and one or more parts P1 after the division are assigned to one of multiple supply units 3, but the order in which the one or more parts P1 are arranged after the division is not specified.

That is, the placement support method according to the present embodiment has a basic aspect of allocating a plurality of unplaced parts P1 to two or more supply units 3, as shown in FIG. 4A and FIG. 4B. In FIG. 4A and FIG. 4B, the unplaced space Sp0 marked with "unplaced" is a virtual space that typically represents a state in which the part P1 has not been allocated to any of the supply units 3, that is, a state in which the part P1 has not been allocated. In other words, in FIG. 4A, the part P1 present in the unplaced space Sp0 is in a state in which the part P1 has not been allocated to any of the supply units 3. On the other hand, in FIG. 4A and FIG. 4B, the space marked with "supply unit" is a virtual space that typically represents a state in which the part P1 has been allocated to any of the supply units 3, that is, a state in which the part P1 has been allocated. In other words, in FIG. 4B, the part P1 present in the space marked with the reference symbol "31" is in a state in which the part P1 has been allocated (allocated) to the first supply unit 31. Similarly, in FIG. 4B, the part P1 present in the space marked with the reference symbol "32" is in a state in which the part P1 has been allocated (allocated) to the second supply unit 32.

In the placement support method according to the present embodiment, as shown in FIG. 4A, the initial state is set to a state in which all of the multiple components P1 mounted by the component mounting system 100 are not placed. In other words, in the placement support method, first, all of the multiple components P1 are not assigned to any of the supply units 3, thereby realizing the initial state in which all of the multiple components P1 are present in the non-placed space Sp0, as shown in FIG. 4A. When the initial state is realized, the placement support method next selects the non-placed components P1 one by one, and determines which supply unit 3 the selected components P1 should be assigned (allocated). In other words, in the placement support method, the multiple components P1 are assigned to one of the supply units 3 one by one, thereby sequentially determining the allocation destinations of the multiple components P1.

Then, all of the multiple components P1 mounted by the component mounting system 100 are distributed (assigned) to one of the supply units 3, that is, there are no more unplaced components P1, and processing according to the basic aspect of the placement support method is completed. In short, according to the basic aspect of the placement support method, as shown in FIG. 4B, there are no more components P1 present in the unplaced space Sp0, and all of the multiple components P1 are distributed (assigned) to one of the supply units 3.

Specifically, for example, the multiple parts P1 are allocated by associating part identification information for identifying each of the multiple parts P1 with supply unit identification information for identifying each of the multiple supply units 3 on a one-to-one basis. As an example, for a part P1 that is not allocated to any supply unit 3, the part identification information is not associated with supply unit identification information, or the supply unit identification information corresponding to the part identification information is null. Also, for a part P1 allocated to the first supply unit 31, the part identification information is associated with the supply unit identification information of the first supply unit 31. Then, when the part identification information and supply unit identification information are associated with each of the multiple parts P1, a state in which all of the multiple parts P1 are allocated to any of the supply units 3 is realized. Data representing the correspondence between the part identification information and the supply unit identification information in a state in which all of the multiple parts P1 are allocated becomes the arrangement data D2. In other words, the placement data D2 represents the correspondence between each of the multiple parts P1 and the supply unit identification information, and thus represents the allocation of each of the multiple parts P1 to one of the supply units 3.

In the placement support method according to this embodiment, the learned model M1 is used to propose the placement of the component P1 according to this basic aspect. In other words, in the basic aspect of the placement support method, the learned model M1 is used to propose the placement of the multiple components P1 for the multiple supply units 3. The learned model M1 is a model generated in correspondence with the characteristics of at least one of the multiple supply units 3. In the placement support method according to this embodiment, the learned model M1 is used to propose the placement of the multiple components P1 for the multiple supply units 3 so as to shorten the mounting time required to generate the product P3 in the component mounting system 100. As will be explained in more detail in the "(3.2) Learned Model" section, the learned model M1 is generated by reinforcement learning.

As shown in Fig. 5, the placement support method according to this embodiment is broadly divided into a processing part S100 for temporary placement and a processing part S200 for rearrangement. In both the processing part S100 for temporary placement and the processing part S200 for rearrangement, the placement support method proposes the allocation of multiple parts P1 according to the basic aspects described above. In other words, in this embodiment, the trained model M1 is used in both the processing part S100 for temporary placement and the processing part S200 for rearrangement.

That is, in the processing part S100 for temporary placement, the multiple components P1 are temporarily allocated to multiple supply units 3. However, if at least some of the components P1 are allocated to an inappropriate supply unit 3, this may lead to problems such as an increased mounting time in the component mounting system 100. Therefore, in the processing part S200 for rearrangement, a more appropriate placement (allocation) is proposed by re-allocating the multiple components P1 that were temporarily allocated.

In addition, in the processing part S100 for temporary placement, when multiple parts P1 are allocated to multiple supply units 3, if the number of supply units 3 increases, the search space for performing reinforcement learning of the trained model M1 becomes wider, making it difficult to generate the trained model M1. Therefore, the placement support method according to this embodiment adopts a multi-stage allocation of parts P1 in the processing part S100 for temporary placement. In other words, according to the multi-stage allocation of parts P1, even if multiple parts P1 are allocated to two groups at each stage, for example, the multiple parts P1 can be ultimately allocated to four or more groups. As a result, it is possible to narrow the search space for performing reinforcement learning of the trained model M1 while using the trained model M1 in the allocation at each stage.

As described above, according to the placement assistance method of this embodiment, it is possible to determine the placement of multiple parts P1 at multiple supply units 3, for example, without the assistance of an expert, etc., and it is possible to propose improved placements of parts P1.

In addition, the flowchart shown in Figure 5 is merely an example, and the order of processing may be changed as appropriate, and processing may be added or deleted as appropriate.

(3.2) Trained Model Next, the proposal of the placement of multiple components P1 using the trained model M1 in the placement assistance method according to this embodiment will be described in more detail with reference to FIGS. 5 and 6.

The placement support method includes an acquisition step and a proposal step. The acquisition step is a step of acquiring mounting data D1, and corresponds to process S101 in the flowchart shown in FIG. 5. The acquisition step is executed by the acquisition unit 21 of the placement support system 20. The mounting data D1 is data on multiple components P1 included in the product P3. The proposal step is a step of proposing, using the trained model M1, the placement of multiple components P1 to multiple supply units 3 from the mounting data D1 so as to shorten the mounting time. The proposal step is executed by the proposal unit 22 of the placement support system 20. In this embodiment, a multi-stage method of allocating components P1 is adopted, so in the flowchart shown in FIG. 5, the processes S102 to S10m of allocation from the "1st stage" to the "Nth stage" correspond to the proposal step.

In this embodiment, the generation (pre-learning) of the trained model M1 is performed by the placement support system 20. As described above, reinforcement learning is used to generate the trained model M1. That is, the placement support system 20 uses, for example, the proposal unit 22 as a reinforcement learning learner. In this disclosure, "reinforcement learning" is a method of machine learning technology that is a framework for automatically acquiring rules from data.

In machine learning technology, aside from "reinforcement learning," a widely used technique is "supervised learning." "Supervised learning" is a technique for learning based on data in which the learning target numerical values are clearly defined. Therefore, in "supervised learning," a large amount of data (learning data) with a clearly defined target is required for learning. However, in the technical field of the component mounting system 100, the target numerical values for the arrangement of multiple components P1 in multiple supply units 3 are not clear, and it is difficult to collect a large amount of learning data for learning.

In contrast, "reinforcement learning" is a method in which an "action" is performed in a certain "environment", the "situation" that changes as a result of the "action" is observed, and a high "reward" is given to the "action" that produces a good result. Then, "reinforcement learning" learns how to select the optimal "action" that maximizes the "reward" through a series of these "actions". Therefore, "reinforcement learning" does not require the preparation of learning data with a clearly defined goal, as in "supervised learning", and it is possible to generate a trained model M1 that has learned how to "select" the optimal "action". In other words, "reinforcement learning" learns how to "act" by repeating trial and error in a given "environment" and evaluating "actions" from the reward series observed, even when the solution space is too wide and the actual correct answer is unknown.

Therefore, reinforcement learning is particularly effective in generating a trained model M1 in cases where the correct answer is not clear and optimization over a huge number of combinations is required, such as a trained model M1 for proposing the placement of multiple parts P1 in multiple supply units 3.

More specifically, in reinforcement learning, the mounting time when a certain arrangement is adopted for multiple components P1 is calculated, and the "reward" is determined based on how close this mounting time is to the ideal mounting time (the mounting time according to the specifications of the mounting machine 10). In this embodiment, the mounting time is calculated by the simulation unit 26 as an example. In reinforcement learning, the larger the search space becomes, the more times trial and error is repeated to obtain a good "reward", and the longer it takes to search for the optimal solution. Therefore, in the placement support method according to this embodiment, a multi-stage method of allocating components P1 is adopted to reduce the size of the search space in reinforcement learning, and the multi-stage method will be described in detail in the "(3.3) Multi-stage method" section.

In addition, in the component mounting system 100, for example, the characteristics and number of mounting machines 10, as well as the type and number of components P1 to be mounted, differ depending on the specifications of the product P3. However, if reinforcement learning is performed each time to generate a trained model M1 in accordance with the product P3, there is a concern that it will take time from when the specifications of the product P3 are determined until the trained model M1 is generated. Therefore, in this embodiment, the trained model M1 is generated by "pre-learning", which is learning before the specifications of the product P3 are determined. As a result, in the placement support method of this embodiment, once the specifications of the product P3 are determined, it is possible to immediately start proposing the placement of the components P1 using the trained model M1 generated by pre-learning.

In addition, in this embodiment, the learning device (e.g., the proposal unit 22) creates placement data D2 in a reinforcement learning algorithm, and the created placement data D2 is output from the output unit 25 to the simulation unit 26. As a result, the simulation unit 26 performs a simulation based on the placement data D2 and calculates the "implementation time". The learning device (e.g., the proposal unit 22) converts the implementation time thus calculated into a ratio to the ideal implementation time, and performs reinforcement learning using the calculated ratio as a "reward". Then, reinforcement learning is performed by repeating the above-mentioned series of processes many times to obtain a better "reward". In this embodiment, as an example, deep reinforcement learning (DQN: Deep Q-Network) that adds the performance of feature extraction of deep learning to perform reinforcement learning is used.

In this embodiment, the trained model M1 is a model generated in correspondence with the characteristics of at least one of the multiple supply units 3 in the component mounting system 100. In other words, the characteristics of the supply unit 3 are reflected in the trained model M1, and the trained model M1 is generated for each characteristic of the supply unit 3. Therefore, if the component mounting system 100 includes multiple supply units 3 having different characteristics, multiple types of trained models M1 will be generated.

In particular, in this embodiment, the trained model M1 is associated with the characteristics of the supply unit 3 to which the part P1 is to be allocated (assigned) in the proposed placement of the part P1 using the trained model M1. Therefore, for example, even when the part P1 is allocated to two allocation destinations as shown in FIG. 6, various trained models M11 to M17 are used depending on the allocation destination. The trained models M11 to M17 used in FIG. 6 are explained in the section "(3.3) Multi-stage method". In this disclosure, when there is no particular distinction between these multiple trained models M11 to M17, each of the multiple trained models M11 to M17 is simply referred to as "trained model M1".

Here, for example, if the two supply units 3 to which parts are to be allocated have the same characteristic (referred to as the "first characteristic"), the trained model M1 is generated in correspondence with the first characteristic. In other words, the trained model M1 includes a model generated in correspondence with the characteristic (first characteristic) of one of the multiple supply units 3. As an example, if the two supply units 3 to which parts are to be allocated both have the characteristic that the number of capture units 41 is "16", the trained model M1 for allocating part P1 reflects the characteristic that the number of capture units 41 is "16".

On the other hand, for example, if the two supply units 3 to which parts are to be allocated have different characteristics (referred to as a "first characteristic" and a "second characteristic"), the trained model M1 will be generated in correspondence with both the first characteristic and the second characteristic. In other words, the trained model M1 includes a model generated in correspondence with the characteristics of two or more supply units 3 out of the multiple supply units 3. As an example, if the two supply units 3 to which parts are to be allocated have the characteristic of "16" or "8" as the number of capture units 41, respectively, the trained model M1 for allocating part P1 will reflect the two characteristics of "16" and "8" as the number of capture units 41.

In addition, the trained model M1 includes models generated in association with one or more feature groups when the features of the multiple supply units 3 are classified into one or more feature groups. For example, the speed (maximum mounting speed) of the mounting machine 10 can be classified into multiple speed ranges, such as a multi-function machine (low speed machine), a medium speed machine, and a high speed machine, and can be considered to be in the same speed range, such as a "high speed machine" if there is a slight difference in speed. In other words, if the feature of the supply unit 3 is the "speed" of the mounting machine 10, the "speed range" such as a multi-function machine (low speed machine), a medium speed machine, and a high speed machine becomes a feature group including multiple features (speeds). In this way, when a feature group including multiple different features is set, the trained model M1 also includes models generated in association with the feature group, rather than individual features. Since a feature group is a set of multiple features, the trained model M1 generated in association with the feature group also corresponds to the trained model M1 generated in association with at least one feature of the supply unit 3.

In addition, the placement support method according to this embodiment further includes an update step for updating the learned model M1. That is, the learned model M1 generated by machine learning (reinforcement learning in this embodiment) is not fixed, but can be updated appropriately by the update step. Specifically, a learner (e.g., the proposal unit 22) updates the learned model M1 stored in the model storage unit 27 periodically or irregularly. The update of the learned model M1 may be realized, for example, by transfer learning that uses knowledge acquired in the past for current learning, or by replacing it with a new learned model M1. If the learned model M1 can be updated, it is also possible to generate an appropriate learned model M1 according to the user's request during the operation phase of the placement support method (placement support system 20).

In addition, in this embodiment, the trained model M1 includes a model generated in association with information about the product P3 generated by the component mounting system 100. The information about the product P3 includes, for example, the technical field to which the product P3 belongs, the lot size of the product P3, or the price range of the product P3. The component mounting system 100 is used to generate products P3 in various technical fields, such as the technical field of electronic devices or the technical field of automobiles. Therefore, due to circumstances specific to each technical field (industry), even if the product P3 is similar, the optimal arrangement of the components P1 may differ. Therefore, in this embodiment, the trained model M1 is generated in association with information about the product P3 so as to reflect not only the characteristics of at least one supply unit 3 but also circumstances specific to the product P3.

In addition, in the placement support method according to the present embodiment, when placing multiple components P1 on multiple supply units 3, a constraint is attached that specifies whether a specific component P1 can be placed on a specific supply unit 3. That is, depending on the supply unit 3, it may be impossible to place a specific component P1 due to constraints on the specifications of the mounting machine 10 or constraints on the size of the component P1. In preparation for such a case, it is possible to attach a constraint. By attaching a constraint, the specific supply unit 3 is excluded from the allocation target for a specific component P1 that is specified in the constraint as being impossible to place on a specific supply unit 3. As an example, in a supply unit 3 where a tape feeder cannot be attached and only a tray can be installed, it is possible to specify in the constraint that it is impossible to place a component P1 supplied by a tape feeder.

Furthermore, in this embodiment, the determination of whether or not a specific part P1 can be placed on a specific supply unit 3 based on the constraint conditions is made separately from the proposal of the placement of multiple parts P1 on multiple supply units 3 using the trained model M1. That is, the constraint on the placement of part P1 based on the constraint conditions is made, for example, before or after the proposal of the placement of part P1 using the trained model M1. As an example, if the constraint on the placement of part P1 is made based on the constraint conditions before the proposal of the placement of part P1 using the trained model M1, the options for deriving an appropriate placement of multiple parts P1 can be reduced.

As described above, in the placement assistance method according to this embodiment, when placing multiple components P1, the learned model M1 is used to propose the placement of multiple components P1 for multiple supply units 3 from the mounting data D1 so as to shorten the mounting time (proposal step). Therefore, in the proposal step, rather than proposing a uniquely determined placement, it is possible to more easily find an appropriate placement by, for example, realizing the skill of "deciding" the placement as an expert does with the learned model M1.

(3.3) Multi-Stage Method Next, among the placement support methods according to this embodiment, the allocation of the components P1 in the multi-stage method will be described in more detail with reference to FIGS.

The placement support method has a primary proposal step and a secondary proposal step. The primary proposal step is a step of proposing the placement (allocation) of multiple parts P1 to multiple primary groups G11 and G12, and corresponds to process S102 in the flowchart shown in FIG. 5. The primary proposal step is executed by the primary proposal section 221 of the proposal section 22 in the placement support system 20. The secondary proposal step is a step of proposing the placement of multiple parts P1 to multiple secondary groups G21 to G24 after the primary proposal step, and corresponds to process S103 in the flowchart shown in FIG. 5. The secondary proposal step is executed by the secondary proposal section 222 of the proposal section 22 in the placement support system 20. That is, in the flowchart shown in FIG. 5, the processes S102 to S10m of allocation of the "1st tier" to the "Nth tier", which correspond to the proposal step, include the primary proposal step and the secondary proposal step.

Here, the placement support method according to this embodiment further includes a tertiary proposal step. The tertiary proposal step is a step of proposing placement of the multiple components P1 for the multiple tertiary groups G31 to G38 when each of the multiple secondary groups G21 to G24 is classified into multiple tertiary groups G31 to G38 after the secondary proposal step. Each of the multiple tertiary groups G31 to G38 includes at least one supply unit 3 out of the multiple supply units 3. In short, according to the tertiary proposal step, each of the multiple secondary groups G21 to G24 is classified into multiple more finely divided tertiary groups G31 to G38, and placement of the multiple components P1 for these multiple tertiary groups G31 to G38 is proposed.

That is, in this embodiment, as shown in FIG. 5, the placement of multiple parts P1 is realized not in a two-stage process but in a hierarchically set N-stage process (where N is a natural number equal to or greater than "3"). Therefore, even after the second-stage allocation (S103) is performed as the secondary proposal step, the third stage and subsequent stages are allocated, making it possible to realize a more detailed allocation. In particular, in this embodiment, it is assumed that multiple parts P1 are allocated to two groups at each stage and ultimately allocated to eight supply units 3, so that the allocation is performed in three stages. Therefore, the tertiary proposal step corresponds to the final stage in the proposal step, that is, the process S10m (where m is "4") in the flowchart shown in FIG. 5. The tertiary proposal step is executed by the tertiary proposal unit 223 of the proposal unit 22 in the placement support system 20.

Figure 6 is an explanatory diagram that shows a schematic diagram of the allocation of parts P1 in a multi-stage method. In this embodiment, the proposal for the placement of multiple parts P1 is divided into at least a primary proposal step in which parts P1 are placed in multiple primary groups G11 and G12, and a secondary proposal step in which parts P1 are placed in multiple smaller secondary groups G21 to G24. In this way, the placement of multiple parts P1 is realized not in a single-stage process but in a hierarchically set multi-stage process, so that the number of options for deriving an appropriate placement of multiple parts P1 can be reduced at each stage.

6, the multiple primary groups G11 and G12 correspond to groups after classification when the non-placed space Sp0, in which multiple parts P1 are not assigned to any supply unit 3, i.e., are present, is classified into two or more groups. Each of the multiple primary groups G11 and G12 includes at least one supply unit 3 out of the multiple supply units 3.

6, the primary group G11 (represented as "Group 1" in the figure) includes a first supply unit 31, a second supply unit 32, a third supply unit 33, and a fourth supply unit 34. The primary group G12 (represented as "Group 2" in the figure) includes a fifth supply unit 35, a sixth supply unit 36, a seventh supply unit 37, and an eighth supply unit 38.

In the primary proposal step, the multiple parts P1 that are in an unplaced state and exist in the unplaced space Sp0 are assigned to multiple primary groups G11 and G12. As a result, the multiple (here, 240) parts P1 are assigned to one of the multiple primary groups G11 and G12.

6, the multiple secondary groups G21 to G24 correspond to the groups after classification when each of the multiple primary groups G11, G12 is further classified into two or more groups. Here, it is assumed that each of the two primary groups G11, G12 is further classified into two secondary groups to generate a total of four secondary groups G21 to G24. Each of the multiple secondary groups G21 to G24 includes at least one supply unit 3 out of the multiple supply units 3.

Specifically, as shown in FIG. 6, the secondary group G21 (labeled as "Group 1_1" in the figure) and the secondary group G22 (labeled as "Group 1_2" in the figure) are located under the primary group G11. In other words, the primary group G11 is classified into two secondary groups G21 and G22. The secondary group G23 (labeled as "Group 2_1" in the figure) and the secondary group G24 (labeled as "Group 2_2" in the figure) are located under the primary group G12. In other words, the primary group G12 is classified into two secondary groups G23 and G24. The secondary group G21 includes the first supply unit 31 and the second supply unit 32, and the secondary group G22 includes the third supply unit 33 and the fourth supply unit 34. The secondary group G23 includes the fifth supply unit 35 and the sixth supply unit 36, and the secondary group G24 includes the seventh supply unit 37 and the eighth supply unit 38.

In the secondary proposal step, the multiple parts P1 included in the primary group G11 are assigned to multiple secondary groups G21 and G22, and the multiple parts P1 included in the primary group G12 are assigned to multiple secondary groups G23 and G24. As a result, the multiple parts P1 (240 in this case) are assigned to one of the multiple secondary groups G21 to G24.

6, the multiple tertiary groups G31 to G38 correspond to the groups after classification when each of the multiple secondary groups G21 to G24 is further classified into two or more groups. Here, it is assumed that each of the four secondary groups G21 to G24 is further classified into two tertiary groups, generating a total of eight tertiary groups G31 to G38. Each of the multiple tertiary groups G31 to G38 includes at least one supply unit 3 out of the multiple supply units 3.

Specifically, as shown in FIG. 6, the tertiary groups G31 and G32 are located under the secondary group G21, and the tertiary groups G33 and G34 are located under the secondary group G22. In other words, the secondary group G21 is classified into two tertiary groups G31 and G32, and the secondary group G22 is classified into two tertiary groups G33 and G34. The tertiary groups G35 and G36 are located under the secondary group G23, and the tertiary groups G37 and G38 are located under the secondary group G24. In other words, the secondary group G23 is classified into two tertiary groups G35 and G36, and the secondary group G24 is classified into two tertiary groups G37 and G38. 6, the multiple tertiary groups G31 to G38 are the final stages of the multi-stage system, and therefore each of the multiple tertiary groups G31 to G38 includes one supply unit 3. That is, in this embodiment, it is assumed that the parts P1 will be ultimately distributed to eight supply units 3, and therefore the multiple (here, eight) tertiary groups G31 to G38 that are the final stages correspond one-to-one to the first to eighth supply units 31 to 38, respectively.

In the tertiary proposal step, the multiple parts P1 included in the secondary group G21 are allocated to multiple tertiary groups G31 and G32, and the multiple parts P1 included in the secondary group G22 are allocated to multiple tertiary groups G33 and G34. Furthermore, in the tertiary proposal step, the multiple parts P1 included in the secondary group G23 are allocated to multiple tertiary groups G35 and G36, and the multiple parts P1 included in the secondary group G24 are allocated to multiple tertiary groups G37 and G38. As a result, the multiple parts P1 (240 in this case) are allocated to one of the multiple tertiary groups G31 to G38, that is, the first to eighth supply units 31 to 38.

Here, in the placement assistance method according to this embodiment, as shown in FIG. 6, the learned model M1 is used when allocating each stage. That is, in at least one of the primary proposal step and the secondary proposal step, the placement of the multiple parts P1 is proposed using the learned model M1. As explained in the "(3.2) Learned model" section, the learned model M1 is a model generated in correspondence with the characteristics of at least one of the multiple supply units 3. In this embodiment, the learned model M1 is used in each of the primary proposal step, the secondary proposal step, and the tertiary proposal step.

That is, in the primary proposal step of allocating the multiple components P1 to the multiple primary groups G11 and G12, the trained model M11 is used to propose the placement of the multiple components P1 for the multiple primary groups G11 and G12 from the mounting data D1 so as to shorten the mounting time. In the secondary proposal step of allocating the multiple components P1 to the multiple secondary groups G21 to G24, the trained models M12 and M13 are used to propose the placement of the multiple components P1 for the multiple secondary groups G21 to G24 from the mounting data D1 so as to shorten the mounting time. In the tertiary proposal step of allocating the multiple components P1 to the multiple tertiary groups G31 to G38, the trained models M14 to M17 are used to propose the placement of the multiple components P1 for the multiple tertiary groups G31 to G38 from the mounting data D1 so as to shorten the mounting time.

Here, as is clear from FIG. 6, the trained model M1 used differs depending on whether it is the first proposal step, the second proposal step, or the third proposal step. In other words, in this embodiment, as described above, various trained models M11 to M17 are used depending on the allocation destination of the part P1. For example, the trained model M11 used in the first proposal step is different from the trained models M12 and M13 used in the second proposal step. As a result, in this embodiment, the algorithm for proposing the placement of multiple parts P1 differs between the first proposal step and the second proposal step. Here, "different algorithms" means that the trained model M1 used is different in this embodiment. Furthermore, in this embodiment, an algorithm (trained model M1) different from that of the first proposal step and the second proposal step is also adopted in the third proposal step.

In addition, in this embodiment, various trained models M11 to M17 are used for allocation of parts P1 in each stage depending on the allocation destination of parts P1. For example, in the secondary proposal step, trained model M12 is used for allocating parts P1 included in primary group G11 to secondary groups G21 and G22. In contrast, in the secondary proposal step, trained model M13 is used for allocating parts P1 included in primary group G12 to secondary groups G23 and G24. Similarly, in the tertiary proposal step, trained model M14 is used for allocating parts P1 included in secondary group G21 to tertiary groups G31 and G32. In contrast, in the tertiary proposal step, trained model M17 is used for allocating parts P1 included in secondary group G24 to tertiary groups G37 and G38.

As a result, in the placement assistance method of this embodiment, seven trained models M11 to M17 are used, as shown in Figure 6.

Specifically, the trained model M11 used in the primary proposal step is a model for allocating multiple parts P1 to be allocated to the first to eighth supply units 31-38 into two primary groups G11, G12. For the trained model M11, the allocation destinations include the first to fourth supply units 31-34, each of which has 16 capture units 41, the fifth and sixth supply units 35, 36, each of which has 8 capture units 41, and the seventh and eighth supply units 37, 38, each of which has 3 capture units 41. Therefore, the trained model M11 is a model generated in correspondence with the three features of 16, 8, and 3 as the number of capture units 41, so as to reflect these three features.

The trained model M12 used in the secondary proposal step is a model for allocating multiple parts P1 to be allocated to the first to fourth supply units 31 to 34 into two secondary groups G21, G22. For the trained model M12, the allocation destinations include the first to fourth supply units 31 to 34, which have 16 capture units 41 in number. Therefore, the trained model M12 is a model generated in association with the feature of 16 as the number of capture units 41, so as to reflect one feature of 16 as the number of capture units 41.

On the other hand, the trained model M13 used in the secondary proposal step is a model for allocating multiple parts P1 to be allocated to the fifth to eighth supply units 35 to 38 into two secondary groups G23, G24. For the trained model M13, the allocation destinations include the fifth and sixth supply units 35, 36, which have eight capture units 41, and the seventh and eighth supply units 37, 38, which have three capture units 41. Therefore, the trained model M13 is a model generated in correspondence with the two features of eight and three as the number of capture units 41, so as to reflect these two features.

The trained model M14 used in the tertiary proposal step is a model for allocating multiple parts P1 to be allocated to the first and second supply units 31 and 32 into two tertiary groups G31 and G32. For the trained model M14, the allocation destinations include the first and second supply units 31 and 32, which have 16 capture units 41 in number. Therefore, the trained model M14 is a model generated in association with the feature of 16 as the number of capture units 41, so as to reflect one feature of 16 as the number of capture units 41.

The trained model M15 used in the tertiary proposal step is a model for allocating multiple parts P1 to be allocated to the third and fourth supply units 33 and 34 into two tertiary groups G33 and G34. For the trained model M15, the allocation destinations include the third and fourth supply units 33 and 34, which have 16 capture units 41 in number. Therefore, the trained model M15 is a model generated in association with the feature of 16 as the number of capture units 41, so as to reflect one feature of 16 as the number of capture units 41.

The trained model M16 used in the tertiary proposal step is a model for allocating multiple parts P1 to be allocated to the fifth and sixth supply units 35 and 36 into two third groups G35 and G36. For the trained model M16, the allocation destinations include the fifth and sixth supply units 35 and 36, which have eight capture units 41 in number. Therefore, the trained model M16 is a model generated in association with the feature of eight as the number of capture units 41, so as to reflect one feature of eight as the number of capture units 41.

The trained model M17 used in the tertiary proposal step is a model for allocating multiple parts P1 to be allocated to the seventh and eighth supply units 37 and 38 into two tertiary groups G37 and G38. For the trained model M17, the allocation destinations include the seventh and eighth supply units 37 and 38, which have three capture units 41 in number. Therefore, the trained model M17 is a model generated in association with the feature of three as the number of capture units 41, so as to reflect one feature of three as the number of capture units 41.

The multiple (here, seven) trained models M11 to M17 as described above are generated by performing reinforcement learning individually. That is, reinforcement learning is performed in advance for each of the multiple stages (here, three stages) set up hierarchically, and trained models M11 to M17 as described above are generated. By linking the multiple trained models M11 to M17 generated in this way, a multi-stage architecture as shown in Figure 6 is constructed.

As described above, in the placement support method according to this embodiment, the placement of multiple parts P1 is realized not in a single step, but in a multi-step (here, three-step) process set up hierarchically. This makes it possible to reduce the number of options for deriving an appropriate placement of multiple parts P1 at each step, compared to a multi-selection method in which multiple parts P1 are allocated in a single step. As a result, the search space for performing reinforcement learning of the trained model M1 can be narrowed. For example, when 240 parts are allocated to the above-mentioned eight supply units 3, the search space is assumed to be "10240" in the one-step multi-selection method. In contrast, in the three-step multi-step method, even under the same conditions, the search space for the first proposal step is "1080", the search space for the second proposal step is "1072", and the search space for the third proposal step is "1036". As a result, the search space for allocating 240 parts to the eight supply units 3 is "1080 + 1072 + 1036," which is significantly narrower than the "10240" required for the multiple-selection method, and the size of the search space can be reduced to a size that allows for actual learning.

Furthermore, the above description is merely one example of the placement support method according to this embodiment, and for example, placement of multiple parts P1 may be realized in a two-stage process, or a four-stage or more process, rather than a three-stage process. When allocation is performed in two stages, the tertiary proposal step is omitted, and secondary groups G21 to G24 are each associated one-to-one with multiple supply units 3. On the other hand, when allocation is performed in four stages, multiple quaternary groups are generated further below (subordinate to) the tertiary groups G31 to G38 to which multiple parts P1 are allocated in the tertiary proposal step. When allocation is performed in five stages, multiple quintic groups are generated further below (subordinate to) the quaternary groups.

(3.4) Rearrangement Next, rearrangement, which is one of the arrangement assistance methods according to this embodiment, will be described in more detail with reference to FIG. 5 and FIG.

The placement support method includes a proposal step, a specification step, and a rearrangement step. The proposal step is a step of proposing the placement of multiple components P1 for multiple supply units 3, and corresponds to steps S102 to S10m in the flowchart shown in FIG. 5. The proposal step is executed by the proposal unit 22 of the placement support system 20. The specification step is a step of specifying a certain supply unit 3 among multiple supply units 3 as a target supply unit, and corresponds to step S201 in the flowchart shown in FIG. 5. The specification step is executed by the specification unit 23 of the placement support system 20. The supply unit 3 specified as the target supply unit is a supply unit 3 that satisfies a specific condition regarding mounting time when the placement of multiple components P1 proposed in the proposal step is adopted. The rearrangement step is a step of proposing the placement of multiple components P1 for the target supply unit and the reference supply unit, and corresponds to steps S202 to S204 in the flowchart shown in FIG. The rearrangement step is executed by the rearrangement unit 24 of the placement support system 20. The reference supply unit includes at least one supply unit 3 other than the target supply unit among the plurality of supply units 3 .

In this embodiment, additional allocation (S203) is repeated a specified number of times in the rearrangement steps (S202-S204) for the parts P1 that have been placed in the target supply unit. The "additional allocation" here refers to a process of redoing the placement (allocation) of some parts P1 to multiple supply units 3 based on the placement (allocation) of multiple parts P1 to multiple supply units 3 proposed in the temporary placement (S100). Therefore, the "specified number of times" that additional allocation is repeated is a number determined according to the number of parts P1 that are the target of additional allocation, etc.

That is, in the processing part S100 for temporary placement, when multiple components P1 are allocated to multiple supply units 3, if at least some of the components P1 are inappropriately placed (allocated), the desired result may not be obtained as is. For example, if a certain component P1 is allocated to an inappropriate supply unit 3 in temporary placement (S100), it may take time to mount this component P1, which may result in an inconvenience of longer mounting time for the component mounting system 100 as a whole. Therefore, in the placement support method according to this embodiment, the placement of the component P1 is not finalized in the proposal step, but the proposal for placement in the proposal step is made a "temporary placement", and the placement of the multiple components P1 can be changed (modified) in the subsequent relocation step.

Moreover, the subject of rearrangement in the rearrangement step is a supply unit 3 that satisfies a specific condition regarding mounting time, that is, a component P1 that has been placed in a target supply unit. Furthermore, the subject of rearrangement in the rearrangement step is not only a target supply unit that satisfies a specific condition, but also includes a component P1 that has been placed in a reference supply unit, which is a supply unit 3 other than the target supply unit. Therefore, it is possible to shuffle multiple components P1 that have been placed in different supply units 3 (target supply unit and reference supply unit) in the temporary arrangement. Therefore, even if a certain component P1 is allocated to an inappropriate supply unit 3 in the temporary arrangement, by rearranging the component P1 in the rearrangement step, inconveniences such as longer mounting time can be eliminated.

In addition, in the placement support method according to this embodiment, as shown in FIG. 5, the identification step (S201) and the rearrangement step (S202 to S204) are repeatedly performed to search for a placement of the component P1 that improves the mounting time. In this embodiment, "improving the mounting time" means that the mounting time of the component mounting system 100 as a whole is shortened. In other words, as shown in FIG. 5, the placement support method determines whether the mounting time has improved after the rearrangement step (S202 to S204) (S205). Therefore, the placement data D2 created in the rearrangement step is output from the output unit 25 to the simulation unit 26. As a result, the simulation unit 26 performs a simulation based on the placement data D2 and calculates the "mounting time" after the rearrangement. In the process S205, it is determined whether the mounting time has improved based on the mounting time thus obtained.

Here, in the placement assistance method, if the mounting time has not improved (S205: No), the placement of component P1 changed in the rearrangement step is returned to the state before the rearrangement step, i.e., the "original component placement" (S206). On the other hand, if the mounting time has improved (S205: Yes), process S206 is skipped. In other words, if the mounting time has not improved as a result of the rearrangement, the placement of component P1 obtained by the rearrangement is adopted only if the mounting time is improved (shortened) by returning to the state before the rearrangement.

Then, the placement support method determines whether or not there has been a movement of the component P1 from the state before the rearrangement step (S207). If there has been a movement of any of the components P1 in the rearrangement step (S207: Yes), the process returns to S201. On the other hand, if there has been no movement of any of the components P1 in the rearrangement step (S207: No), the process part S200 for rearrangement is terminated. That is, as long as the placement of the component P1 can be changed (corrected) in the rearrangement step, there is a movement of any of the components P1 in the rearrangement step (S207: Yes), so the identification step (S201) and the rearrangement step (S202 to S204) are repeatedly executed. In other words, the identification step (S201) and the rearrangement step (S202 to S204) are repeatedly executed until there are no patterns of change (correction) of the placement of the component P1 that improve the mounting time in the rearrangement step.

Figure 7 is an explanatory diagram that shows a schematic representation of the state of rearrangement. Figure 7 conceptually shows the state of rearrangement of multiple components P1 that have been arranged in the fourth supply unit 34 and the sixth supply unit 36 out of the first to eighth supply units 31 to 38, based on the arrangement (allocation) of multiple components P1 to multiple supply units 3 proposed in the provisional arrangement. Here, as an example, we take the case where the fourth supply unit 34 is the "target supply unit" and the sixth supply unit 36 is the "reference supply unit."

Here, in this embodiment, as an example, the supply unit 3 that is a bottleneck in terms of mounting time is set as the target supply unit. In other words, the component mounting system 100 sequentially mounts multiple components P1 arranged in multiple supply units 3, and among the multiple supply units 3, the supply unit 3 that takes the longest time to mount the component P1 becomes the bottleneck in terms of mounting time. In other words, in this embodiment, the specific condition includes that the time required to generate the product P3 in the component mounting system 100 (mounting time) is the longest among the multiple supply units 3. In the example of FIG. 7, when the arrangement proposed in the tentative arrangement is adopted in the component mounting system 100, the time required to mount the component P1 arranged in the fourth supply unit 34 is the longest among the multiple supply units 3, so the fourth supply unit 34 is extracted as the target supply unit.

In this way, in the identification step, among the multiple supply units 3, a supply unit 3 that satisfies a specific condition regarding mounting time is identified as a target supply unit. Therefore, in this embodiment, the placement data D2 created in the temporary placement is output from the output unit 25 to the simulation unit 26. As a result, the simulation unit 26 performs a simulation based on the placement data D2 and calculates the "mounting time". The identification unit 23 extracts a supply unit 3 that satisfies the specific condition (target supply unit) based on the mounting time obtained in this way.

On the other hand, the reference supply unit is composed of any supply unit 3 other than the target supply unit (here, the fourth supply unit 34) among the multiple supply units 3. In this embodiment, it is assumed that any one of the multiple supply units 3 other than the target supply unit (here, the fourth supply unit 34) is selected as the reference supply unit in sequence. The example of FIG. 7 shows the state of rearrangement when the sixth supply unit 36 of the multiple supply units 3 is selected as the reference supply unit. For example, when the rearrangement shown in FIG. 7 is completed, the seventh supply unit 37 of the multiple supply units 3 is next selected as the reference supply unit, and rearrangement is performed.

In addition, in the rearrangement step, as shown in FIG. 7, the multiple parts P1 arranged in the target supply unit (fourth supply unit 34) and the reference supply unit (sixth supply unit 36) are temporarily moved to the unarranged space Sp0, and then reassigned to the target supply unit and the reference supply unit. In other words, the multiple parts P1 arranged in the target supply unit and the reference supply unit in the tentative arrangement are once returned to a state in which they are not assigned to any supply unit 3, that is, an unarranged state, and then reassigned to the target supply unit and the reference supply unit. In other words, in the rearrangement step, the arrangement of the multiple parts P1 is reset for the target supply unit and the reference supply unit, and then a new arrangement of the multiple parts P1 is proposed.

In addition, in the placement assistance method according to this embodiment, as shown in FIG. 7, the learned model M1 is also used in the placement (allocation) of the multiple parts P1 in the rearrangement. In this embodiment, the placement of the multiple parts P1 is proposed using the learned model M1 in at least one of the proposal step and the rearrangement step. As explained in the "(3.2) Trained model" section, the trained model M1 is a model generated in association with the characteristics of at least one of the multiple supply units 3. In particular, in this embodiment, the trained model M1 is used in both the proposal step and the rearrangement step.

Here, in this embodiment, as described above, various trained models M1 are used depending on the allocation destination of the part P1. Therefore, in rearrangement in which the fourth supply unit 34 as the target supply unit and the sixth supply unit 36 as the reference supply unit are the allocation destinations, a trained model M18 other than the trained models M11 to M17 used in the provisional arrangement (proposal step) is used. As a result, in this embodiment, the algorithms for proposing the arrangement of the multiple parts P1 are different between the proposal step and the rearrangement step.

Specifically, the trained model M18 used in the rearrangement step is a model for allocating multiple parts P1 to be allocated to two supply units 3 to the fourth supply unit 34 and the sixth supply unit 36. For the trained model M18, the allocation destinations include the fourth supply unit 34, which has 16 capture units 41, and the sixth supply unit 36, which has 8 capture units 41. Therefore, the trained model M18 is a model generated in correspondence with the two features of 16 and 8 as the number of capture units 41, so as to reflect these two features.

Also, in the rearrangement step, similar to the proposal step, the arrangement of the multiple components P1 for the multiple supply units 3 is proposed so as to shorten the mounting time. That is, in the rearrangement step, the arrangement of the multiple components P1 for the target supply unit and the reference supply unit is proposed so as to shorten the mounting time required to generate the product P3 in the component mounting system 100.

Furthermore, in this embodiment, in the rearrangement step, as in the proposal step, when multiple parts P1 are arranged on multiple supply units 3, a constraint condition is attached that specifies whether or not a specific part P1 can be arranged on a specific supply unit 3. As explained in detail in the "(3.2) Trained model" section, by attaching the constraint condition, for a specific part P1 that is specified in the constraint condition as being impossible to arrange on a specific supply unit 3, the specific supply unit 3 is excluded from the rearrangement (allocation) target.

In the placement support method according to this embodiment, as described above, the identification step (S201) and the rearrangement step (S202 to S204) are repeatedly executed until all the placement change patterns for component P1 that improve the mounting time in the rearrangement step are exhausted. Therefore, if, as a result of rearrangement of a supply unit 3 (target supply unit) that satisfies a specific condition (becomes a bottleneck) with respect to mounting time, another supply unit 3 satisfies the specific condition, the rearrangement step is repeated with the other supply unit 3 as the target supply unit. On the other hand, if, as a result of rearrangement of a supply unit 3 that satisfies a specific condition (becomes a bottleneck) with respect to mounting time, the same supply unit 3 continues to satisfy the specific condition, it is determined that all the placement change patterns for component P1 have been exhausted, and the processing part S200 for rearrangement is terminated.

As described above, the placement assistance method according to this embodiment further includes a rearrangement step (S202-S204) in addition to the acquisition step (S101) and proposal step (S102-S10m). The rearrangement step is a step that proposes the placement of multiple components P1 for multiple supply units 3 after the proposal step. In particular, in this embodiment, the trained model M1 generated by reinforcement learning is also used in the rearrangement step, and the placement of components P1 can be re-proposed so as to shorten the mounting time compared to the placement proposed in the proposal step.

Furthermore, since the placement assistance method according to this embodiment employs a multi-stage approach in the proposal step, the rearrangement step is a step of proposing the placement of multiple components P1 after the secondary proposal step. In the rearrangement step, the placement of multiple components P1 is proposed at least for the supply unit 3 that requires the longest time among the multiple supply units 3 to generate the product P3 in the component mounting system.

(4) Modifications The first embodiment is merely one of various embodiments of the present disclosure. Various modifications of the first embodiment are possible depending on the design, etc., as long as the object of the present disclosure can be achieved. In addition, all drawings referred to in this disclosure are schematic diagrams, and the ratios of the sizes and thicknesses of the components in the drawings do not necessarily reflect the actual dimensional ratios. In addition, functions similar to the placement support method according to the first embodiment may be embodied in a placement support system 20, a (computer) program, or a non-transitory recording medium on which a program is recorded. The program according to one aspect is a program for causing one or more processors to execute the placement support method according to the first embodiment.

Below, we list some variations of the first embodiment. The variations described below can be combined as appropriate.

The placement assistance system 20 in the present disclosure includes a computer system. The computer system is mainly composed of a processor and a memory as hardware. The processor executes a program recorded in the memory of the computer system to realize the function of the placement assistance system 20 in the present disclosure. The program may be pre-recorded in the memory of the computer system, provided through an electric communication line, or recorded and provided in a non-transitory recording medium such as a memory card, an optical disk, or a hard disk drive that can be read by the computer system. The processor of the computer system is composed of one or more electronic circuits including a semiconductor integrated circuit (IC) or a large-scale integrated circuit (LSI). The integrated circuits such as IC or LSI referred to here are called differently depending on the degree of integration, and include integrated circuits called system LSI, VLSI (Very Large Scale Integration), or ULSI (Ultra Large Scale Integration). Furthermore, a field-programmable gate array (FPGA) that is programmed after the manufacture of the LSI, or a logic device that can reconfigure the connection relationship inside the LSI or reconfigure the circuit partition inside the LSI, can also be adopted as a processor. The electronic circuits may be integrated in one chip or distributed among multiple chips. The chips may be integrated in one device or distributed among multiple devices. The computer system referred to here includes a microcontroller having one or more processors and one or more memories. Thus, the microcontroller is also composed of one or more electronic circuits including a semiconductor integrated circuit or a large-scale integrated circuit.

In addition, it is not a required configuration of the placement assistance system 20 that multiple functions in the placement assistance system 20 are concentrated in one housing. The components of the placement assistance system 20 may be distributed across multiple housings. Furthermore, at least some of the functions of the placement assistance system 20 may be realized by the cloud (cloud computing) or the like.

Examples of functions that can be realized by distributing them across two or more systems, such as the cloud (or server) and the edge, include the following:

The first function is a function for generating a trained model M1. That is, the trained model M1 can be generated in one system, and the deployment support method (proposal step, etc.) using this trained model M1 can be performed in another system. As an example, the trained model M1 can be generated in the cloud (or server), and the deployment support method (proposal step, etc.) using this trained model M1 can be performed on an edge. In this case, in one system such as a cloud, a trained model generation method is embodied in which a trained model used in the deployment support method is generated in association with the characteristics of at least one supply unit 3 out of multiple supply units 3.

The second function is the function of allocating the components P1 in a multi-stage manner. That is, the first proposal step can be performed in one system, and the second proposal step and subsequent steps (including the third proposal step, etc.) can be performed in another system. As an example, the first proposal step can be performed in the cloud (or a server), and the second proposal step and subsequent steps (including the third proposal step, etc.) can be performed in an edge. In this case, the placement assistance system 20 is embodied in a system such as an edge, and only the second proposal step and subsequent steps (including the third proposal step, etc.) are performed. In short, the placement assistance system 20 assists in the determination of the placement of the multiple components P1 in the multiple supply units 3 with respect to the component mounting system 100. The component mounting system 100 generates a product P3 by mounting the multiple components P1 supplied to the multiple supply units 3 on the target P2. The placement assistance system 20 does not include a primary proposal unit 221, but includes a secondary proposal unit 222. The secondary proposal unit 222 proposes the placement of the multiple components P1 for the multiple secondary groups G21 to G24 when each of the multiple primary groups G11, G12 is classified into multiple secondary groups G21 to G24. Each of the multiple secondary groups G21 to G24 includes at least one supply unit 3 among the multiple supply units 3. The secondary proposal unit 222 receives a proposal for the placement of the multiple components P1 for the multiple primary groups G11, G12 and proposes the placement of the multiple components P1 for the multiple secondary groups G21 to G24. Here, the proposal for the placement of the multiple components P1 for the multiple primary groups G11, G12 is made by another system (such as a cloud or a server) that is separate from the placement support system 20 and includes the primary proposal unit 221. In other words, the placement support system 20 receives a proposal for the placement of the multiple components P1 for the multiple primary groups G11, G12 from the other system and proposes the placement of the multiple components P1 for the multiple secondary groups G21 to G24.

The third function is a rearrangement function. That is, the proposal step can be performed in one system, and the identification step and rearrangement step can be performed in another system. As an example, the proposal step can be performed on an edge, and the identification step and rearrangement step can be performed on a cloud (or server). In this case, the placement support method is embodied in a system such as a cloud (or server) and includes only the identification step and the rearrangement step. In short, the placement support method is a method for supporting the determination of the placement of multiple components P1 in multiple supply units 3 with respect to the component mounting system 100. The component mounting system 100 generates a product P3 by mounting multiple components P1 supplied to the multiple supply units 3 on a target P2. The placement support method does not have a proposal step, but has a specification step and a rearrangement step. The specification step is a step of identifying a supply unit 3 that satisfies a specific condition as a target supply unit when a proposal for the placement of the multiple components for the multiple supply units is received and the proposed placement of the multiple components P1 is adopted. The specific condition is a condition regarding the time required for generating the product P3 in the component mounting system 100. The rearrangement step is a step of proposing an arrangement of the multiple components P1 to the target supply unit and a reference supply unit including at least one supply unit 3 other than the target supply unit among the multiple supply units 3.

Conversely, in embodiment 1, at least some of the functions of the placement assistance system 20 or the work system 200 that are distributed among multiple devices may be consolidated within a single housing. For example, some of the functions that are distributed between the component mounting system 100 and the placement assistance system 20 may be consolidated into the component mounting system 100.

In addition, the use of the component mounting system 100 is not limited to the manufacture of electronic devices in a factory. For example, when the component mounting system 100 is used to mount mechanical components on a glass plate, the component mounting system 100 performs the task of mounting a mechanical component, which is component P1, on a glass plate, which is an object P2.

In addition, the number of capture parts 41 provided in the mounting head 4 is not limited to the number described in embodiment 1. For example, the number of capture parts 41 may be 17 or more. Of course, the mounting head 4 may be provided with only one capture part 41. The mounting head 4 may be a rotary head.

In addition, at least one of the multiple mounting machines 10 may have only one supply unit 3, or three or more supply units 3. Similarly, at least one of the multiple mounting machines 10 may have only one mounting head 4, or three or more mounting heads 4.

In addition, in the first embodiment, the reference supply unit is composed of any supply unit 3 other than the target supply unit among the multiple supply units 3, but this is not limited to this example. For example, the supply unit 3 with the most leeway in terms of mounting time may be the reference supply unit. In other words, the component mounting system 100 sequentially mounts multiple components P1 arranged in multiple supply units 3, and the supply unit 3 that takes the shortest time to mount the component P1 among the multiple supply units 3 has the most leeway in terms of mounting time. In other words, the reference supply unit includes the supply unit 3 that takes the shortest time to generate the product P3 in the component mounting system 100 among the multiple supply units 3. According to this configuration, in the rearrangement, it is possible to shuffle multiple components P1 between the target supply unit that is a bottleneck in terms of mounting time and the reference supply unit that has the most leeway in terms of mounting time. As a result, the rearrangement makes it easier to improve the mounting time.

In addition, the specific condition for identifying the target supply unit may be a condition related to the mounting time, and is not limited to the mounting time being the longest among the multiple supply units 3 (the supply unit 3 being a bottleneck). For example, the specific condition may be that the time (mounting time) required to generate the product P3 in the component mounting system 100 is the Mth longest (M is an integer equal to or greater than 2) among the multiple supply units 3. Conversely, the specific condition may be that the time (mounting time) required to generate the product P3 in the component mounting system 100 is the shortest among the multiple supply units 3. In this way, depending on the specific condition, for example, it is possible to reduce the variation in the time required to generate the product P3 in the component mounting system 100 among the multiple supply units 3. In this case, for example, the concentration of the burden on a specific mounting machine 10 can be alleviated.

In addition, the target supply unit is not limited to one supply unit 3, but may be two or more supply units 3. Similarly, the reference supply unit is also not limited to one supply unit 3, but may be two or more supply units 3.

In addition, the supply unit 3 is not limited to a configuration in which a tape feeder is attached, and may be, for example, a tray on which multiple components P1 are placed. Furthermore, the component mounting system 100 as a whole may be a mixture of supply units 3 to which a tape feeder is attached and supply units 3 to which a tray is installed.

Furthermore, the placement assistance method may only propose the "placement" of parts P1. In this case, the placement assistance method divides multiple parts P1 into multiple parts, assigns one or more parts P1 after the division to one of multiple supply units 3, and even specifies the order (arrangement) of the parts within each supply unit 3. As a result, when multiple parts P1 are assigned to one supply unit 3, even the order of the parts P1 is specified.

In addition, the pre-learning of the trained model M1 may be performed in another system, not limited to the placement assistance system 20. In this case, the placement assistance system 20 uses the trained model M1 generated in the other system to perform a proposal for placement of the part P1 in the proposal unit 22 and the rearrangement unit 24.

In addition, it is not essential that the placement of the part P1 is proposed using the trained model M1 in both the proposal step and the rearrangement step. For example, of the proposal step and the rearrangement step, the placement of the part P1 may be proposed using the trained model M1 only in the proposal step, or the placement of the part P1 may be proposed using the trained model M1 only in the rearrangement step. Furthermore, the trained model M1 does not have to be used in either the proposal step or the rearrangement step. In other words, with regard to functions such as the allocation and rearrangement of the part P1 in a multi-stage manner, it is possible to realize functions similar to those of embodiment 1 without using the trained model M1.

In addition, in embodiment 1, it is not essential that the placement of component P1 is proposed using the trained model M1 in all of the primary proposal step, secondary proposal step, and tertiary proposal step. For example, the placement of component P1 may be proposed using the trained model M1 in only one or two of the primary proposal step, secondary proposal step, and tertiary proposal step.

Furthermore, the means for generating the trained model M1 is not limited to reinforcement learning, but may be, for example, "supervised learning," etc.

In addition, the placement support method according to the first embodiment is not limited to a case where all of the processes are executed by one or more processors, and at least some of the processes may be executed by a human being. For example, processes such as setting constraint conditions may be executed by a human being using a user interface or the like.

In addition, the means for calculating the implementation time used for generating the trained model M1 is not limited to the simulation in the simulation unit 26. For example, instead of the simulation, the implementation time may be calculated using a classifier generated by deep learning using a multi-layer neural network.

(Embodiment 2)
The placement support method according to this embodiment differs from the placement support method according to embodiment 1 in the processing part S200 for rearrangement (see FIG. 5). Hereinafter, the same components as those in embodiment 1 will be denoted by the same reference numerals and the description thereof will be omitted as appropriate.

That is, in the first embodiment, when rearranging part P1 in the rearrangement step, a supply unit 3 (target supply unit) that satisfies a specific condition is identified in the identification step, and part P1 that is arranged in the target supply unit is then targeted for rearrangement. Then, for the supply unit 3 (target supply unit) that satisfies the specific condition, if the same supply unit 3 still satisfies the specific condition as a result of rearrangement, it is determined that all the patterns for changing the arrangement of part P1 have been exhausted, and the processing part S200 for rearrangement is terminated.

In contrast, in this embodiment, even if the same supply unit 3 still satisfies the specific condition as a result of rearrangement of the supply unit 3 that satisfies the specific condition with respect to mounting time, the processing part S200 for rearrangement is continued. In other words, in the placement support method according to this embodiment, even if all the patterns for changing the placement of the component P1 with the supply unit 3 that satisfies the specific condition as the target supply unit are exhausted, the rearrangement step is repeated with another supply unit 3 as the target supply unit. Specifically, after all the patterns for changing the placement of the component P1 with the supply unit 3 that satisfies the specific condition as the target supply unit are exhausted, any of the multiple supply units 3 is selected in sequence as the target supply unit, regardless of whether it satisfies the specific condition.

As a result, according to the placement assistance method of this embodiment, it is possible to perform rearrangement of all of the multiple supply units 3, making it possible to find a more appropriate placement compared to embodiment 1.

As a variation of the second embodiment, the placement support method may not have a step of identifying a supply unit 3 that satisfies a specific condition as a target supply unit. In this case, the placement support method sequentially selects any supply unit 3 among the multiple supply units 3 as a target supply unit during relocation, regardless of whether the supply unit 3 satisfies the specific condition.

The configuration described in embodiment 2 (including modified examples) can be applied in appropriate combination with the configuration described in embodiment 1 (including modified examples).

(summary)
As described above, the placement support method according to the first aspect is a method for supporting the determination of placement of a plurality of components (P1) in a plurality of supply units (3) in a component mounting system (100), and includes an acquisition step and a proposal step. The component mounting system (100) is a system that generates a product (P3) by mounting a plurality of components (P1) supplied to the plurality of supply units (3) on a target object (P2). The acquisition step is a step of acquiring mounting data (D1) related to the plurality of components (P1) included in the product (P3). The proposal step is a step of proposing a placement of the plurality of components (P1) for the plurality of supply units (3) from the mounting data (D1) using the trained model (M1) so as to shorten the mounting time required for generating the product (P3) in the component mounting system (100). The trained model (M1) is a model generated in association with the characteristics of at least one of the plurality of supply units (3).

According to this aspect, in the suggestion step, rather than proposing a uniquely determined placement, it is possible to more easily find an appropriate placement by, for example, realizing in the learned model (M1) the skill of "deciding" the placement as done by an expert. In particular, since the learned model (M1) is a model generated in correspondence with the characteristics of at least one of the multiple supply units (3), the placement of the multiple components (P1) is proposed, for example, taking into account the performance or specifications of the mounting machine (10). Therefore, this placement support method makes it possible to propose an improved placement of the components (P1).

In the placement support method relating to the second aspect, in the first aspect, the trained model (M1) includes a model generated in correspondence with the characteristics of two or more supply units (3) among the multiple supply units (3).

According to this aspect, for example, even if multiple parts (P1) are allocated to two or more supply units (3) in the proposal step, it becomes easier to propose an appropriate arrangement using a trained model (M1) tailored to the two or more supply units (3) to which the parts are to be allocated.

In the placement support method relating to the third aspect, in the first or second aspect, the trained model (M1) includes a model generated in correspondence with one or more feature groups when the features of multiple supply units (3) are classified into one or more feature groups.

According to this aspect, for example, by grouping similar features into one feature group, even if there is a supply unit (3) with a variety of features, it becomes easier to suggest an appropriate arrangement using a trained model (M1) tailored to the supply unit (3).

In the placement support method relating to the fourth aspect, in any of the first to third aspects, the trained model (M1) is generated using reinforcement learning.

According to this aspect, it becomes easier to generate a trained model (M1) in cases where the correct answer is not clear and optimization over a huge number of combinations is required.

The placement support method of the fifth aspect, in any one of the first to fourth aspects, further includes an update step of updating the trained model (M1).

According to this aspect, during the operational phase of the placement support method, it is possible to generate an appropriate trained model (M1) in response to user requirements.

In the placement assistance method relating to the sixth aspect, in any of the first to fifth aspects, the trained model (M1) includes a model generated in correspondence with information regarding the product (P3) generated by the component mounting system (100).

According to this embodiment, a trained model (M1) is used that reflects not only the characteristics of the supply unit (3) but also the circumstances specific to the product (P3), making it easier to suggest an appropriate arrangement.

In the placement assistance method according to the seventh aspect, in any of the first to sixth aspects, the proposal step includes a primary proposal step and a secondary proposal step. The primary proposal step is a step of proposing a placement of a plurality of parts (P1) for a plurality of primary groups (G11, G12). The secondary proposal step is a step of proposing a placement of a plurality of parts (P1) for a plurality of secondary groups (G21 to G24) when each of the plurality of primary groups (G11, G12) is classified into a plurality of secondary groups (G21 to G24) after the primary proposal step. Each of the plurality of secondary groups (G21 to G24) includes at least one supply unit (3) of the plurality of supply units (3).

According to this aspect, the proposal of the placement of multiple parts (P1) is divided into a primary proposal step of placing the parts (P1) in multiple primary groups (G11, G12) and a secondary proposal step of placing the parts (P1) in multiple smaller secondary groups (G21 to G24). In short, the placement of multiple parts (P1) is realized not in a single stage process but in a hierarchically set multiple stage process, so that at each stage, the number of options for deriving an appropriate placement of multiple parts (P1) can be reduced. As a result, the process for proposing the placement of multiple parts (P1) at each stage can be simplified. Therefore, this placement support method makes it possible to propose an improved placement of parts (P1).

In the placement support method relating to the eighth aspect, in the seventh aspect, the trained model (M1) is different between the primary proposal step and the secondary proposal step.

According to this aspect, an arrangement of multiple parts (P1) can be proposed using an appropriate trained model (M1) in each of the primary proposal step and the secondary proposal step.

In the placement assistance method of the ninth aspect, in the seventh or eighth aspect, in at least one of the primary proposal step and the secondary proposal step, a placement of a plurality of parts (P1) is proposed using a trained model (M1).

According to this aspect, rather than proposing a uniquely determined arrangement, it is possible to more easily find an appropriate arrangement by realizing in the trained model (M1) the skill of "deciding" the arrangement, as an expert would do, for example.

In the wiring assistance method according to the tenth aspect, in any of the seventh to ninth aspects, a tertiary proposal step is further included. The tertiary proposal step is a step of proposing the placement of a plurality of components (P1) for a plurality of tertiary groups (G31 to G38) when each of the plurality of secondary groups (G21 to G24) is classified into a plurality of tertiary groups (G31 to G38) after the secondary proposal step. Each of the plurality of tertiary groups (G31 to G38) includes at least one supply unit (3) of the plurality of supply units (3).

According to this aspect, the proposal for the placement of multiple parts (P1) is realized in a primary proposal step and, in addition to the primary proposal step, a tertiary proposal step in which the parts (P1) are placed in a plurality of smaller tertiary groups (G31 to G38). As a result, the process for proposing the placement of multiple parts (P1) at each stage can be further simplified.

The placement assistance method of the eleventh aspect, in any one of the first to tenth aspects, further includes, after the proposal step, a rearrangement step of proposing the placement of a plurality of parts (P1) to a plurality of supply sections (3).

According to this aspect, even if the arrangement of the multiple parts (P1) proposed in the proposal step is not optimal, a new arrangement of the multiple parts (P1) can be proposed in the rearrangement step.

In the wiring assistance method according to the 12th aspect, in the 11th aspect, the rearrangement step proposes an arrangement of multiple components (P1) for at least the supply unit (3) that requires the longest time among the multiple supply units (3) to generate the product (P3) in the component mounting system (100).

According to this aspect, in the rearrangement step, a new arrangement of multiple components (P1) can be proposed for a component (P1) arranged in the supply section (3) that is a bottleneck in terms of the time required to generate the product (P3) in the component mounting system (100).

In the placement assistance method relating to the thirteenth aspect, in any of the first to twelfth aspects, when placing multiple parts (P1) to multiple supply units (3), a constraint is imposed that specifies whether or not a specific part (P1) can be placed on a specific supply unit (3).

According to this aspect, for a specific part (P1) that is specified in the constraints as being unable to be placed in a specific supply section (3), the specific supply section (3) can be excluded from the allocation targets, which makes it easier to simplify placement proposals.

In the placement support method of the 14th aspect, in the 13th aspect, the determination of whether or not a specific part (P1) can be placed for a specific supply unit (3) based on constraint conditions is made separately from the proposal of the placement of multiple parts (P1) for multiple supply units (3) using the trained model (M1).

According to this aspect, the number of options for deriving an appropriate placement of multiple parts (P1) can be reduced.

The method for generating a trained model relating to the 15th aspect generates a trained model to be used in any of the placement support methods 1 to 14 by associating it with the characteristics of at least one supply unit (3) out of multiple supply units (3).

According to this aspect, it is possible to propose an improved arrangement of part (P1).

The program relating to the 16th aspect is a program for causing one or more processors to execute the placement support method relating to any one of the first to 14th aspects or the method for generating a trained model relating to the 15th aspect.

According to this aspect, it is possible to propose an improved arrangement of part (P1).

The placement assistance system (20) according to the seventeenth aspect is a system for assisting in the determination of placement of a plurality of components (P1) in a plurality of supply units (3) for a component mounting system (100), and includes an acquisition unit (21) and a proposal unit (22). The component mounting system (100) is a system for generating a product (P3) by mounting a plurality of components (P1) supplied to the plurality of supply units (3) on a target object (P2). The acquisition unit (21) acquires mounting data (D1) related to the plurality of components (P1) included in the product (P3). The proposal unit (22) uses a trained model (M1) to propose a placement of the plurality of components (P1) for the plurality of supply units (3) from the mounting data (D1) so as to shorten the mounting time required for generating the product (P3) in the component mounting system (100). The trained model (M1) is a model associated with the characteristics of at least one of the plurality of supply units (3).

According to this aspect, it is possible to propose an improved arrangement of part (P1).

In the placement assistance system (20) according to the 18th aspect, in the 17th aspect, the proposal unit (22) includes a primary proposal unit (221) and a secondary proposal unit (222). The primary proposal unit (221) proposes placement of a plurality of parts (P1) for a plurality of primary groups (G11, G12). The secondary proposal unit (222) proposes placement of a plurality of parts (P1) for a plurality of secondary groups (G21 to G24) when each of the plurality of primary groups (G11, G12) is classified into a plurality of secondary groups (G21 to G24). Each of the plurality of secondary groups (G21 to G24) includes at least one supply unit (3) of the plurality of supply units (3).

According to this aspect, it is possible to propose an improved arrangement of part (P1).

In the placement assistance system (20) according to the 19th aspect, in the 17th aspect, the suggestion unit (22) includes a secondary suggestion unit (222). The secondary suggestion unit (222) proposes placement of a plurality of parts (P1) for a plurality of secondary groups (G21 to G24) when each of the plurality of primary groups (G11, G12) is classified into a plurality of secondary groups (G21 to G24). Each of the plurality of secondary groups (G21 to G24) includes at least one supply unit (3) among the plurality of supply units (3). The secondary suggestion unit (222) proposes placement of a plurality of parts (P1) for a plurality of secondary groups (G21 to G24) upon receiving a proposal for placement of a plurality of parts (P1) for the plurality of primary groups (G11, G12).

According to this aspect, it is possible to propose an improved arrangement of part (P1).

The work system (200) according to the twentieth aspect includes a placement assistance system (20) according to any one of the seventeenth to nineteenth aspects, and a component mounting system (100). The component mounting system (100) is used to place a plurality of components (P1) proposed by the placement assistance system (20) and generates a product (P3).

According to this aspect, it is possible to propose an improved arrangement of part (P1).

Not limited to the above aspects, various aspects (including modified examples) of the placement support method according to embodiment 1 and embodiment 2 can be embodied in a placement support system (20), an operation system (200), a program, and a non-temporary recording medium having the program recorded thereon.

The configurations relating to aspects 2 to 14 are not essential to the placement assistance method and may be omitted as appropriate.

The configurations relating to items 17 and 18 are not essential components of the placement support system (20) and may be omitted as appropriate.

3 Supply unit 20 Placement support system 21 Acquisition unit 22 Proposal unit 100 Component mounting system 200 Work system 221 Primary proposal unit 222 Secondary proposal unit D1 Mounting data M1 Trained models G11, G12 Primary groups G21 to G24 Secondary groups G31 to G38 Tertiary groups P1 Component P2 Object P3 Product

Claims (20)

1. A placement assistance method for a component mounting system that generates a product by mounting a plurality of components supplied to a plurality of supply units on a target object, the method comprising the steps of:
acquiring mounting data relating to the plurality of components included in the product;
and a proposing step of proposing an arrangement of the plurality of components for the plurality of supply units from the mounting data, using a trained model generated in association with a feature of at least one of the plurality of supply units, so as to shorten a mounting time required for generating the product in the component mounting system.
Placement assistance methods. The trained model includes a model generated in association with features of two or more supply units among the plurality of supply units.
The placement assistance method according to claim 1 . The trained model includes a model generated in association with one or more feature groups when the features of the plurality of supply units are classified into one or more feature groups,
The placement assistance method according to claim 1 or 2. The trained model is generated using reinforcement learning.
The method for assisting placement according to any one of claims 1 to 3. Further comprising an updating step of updating the trained model.
The placement assistance method according to any one of claims 1 to 4. The trained model includes a model generated in association with information about the product generated by the component mounting system.
The placement assistance method according to any one of claims 1 to 5. The proposing step includes:
a primary proposal step of proposing an arrangement of the plurality of components for a plurality of primary groups;
a secondary proposal step of proposing an arrangement of the plurality of components for a plurality of secondary groups when each of the plurality of primary groups is classified into a plurality of secondary groups including at least one supply unit among the plurality of supply units after the primary proposal step,
The placement assistance method according to any one of claims 1 to 6. The trained model is different between the first proposal step and the second proposal step.
The placement assistance method according to claim 7. In at least one of the primary proposal step and the secondary proposal step, a layout of the plurality of parts is proposed using the trained model.
The placement assistance method according to claim 7 or 8. a tertiary proposal step of proposing an arrangement of the plurality of components for a plurality of tertiary groups when each of the plurality of secondary groups is classified into a plurality of tertiary groups including at least one supply unit among the plurality of supply units after the secondary proposal step,
The arrangement assistance method according to any one of claims 7 to 9. The method further includes, after the proposing step, a rearrangement step of proposing an arrangement of the plurality of components with respect to the plurality of supply units.
The method for assisting placement according to any one of claims 1 to 10. In the rearrangement step, a placement of the plurality of components is proposed for at least a supply unit that requires the longest time to generate the product in the component mounting system among the plurality of supply units.
The method for assisting placement according to claim 11. When arranging the plurality of components to the plurality of supply units, a constraint is imposed that specifies whether or not a specific component can be arranged to a specific supply unit.
The method for assisting placement according to any one of claims 1 to 12. The determination of whether or not the specific part can be placed on the specific supply unit based on the constraint condition is performed separately from the proposal of the placement of the multiple parts on the multiple supply units using the trained model.
The method for assisting placement according to claim 13. A trained model used in the placement support method according to any one of claims 1 to 14 is generated in association with a feature of at least one of the plurality of supply units.
How to generate a trained model. A program for causing one or more processors to execute the placement support method described in any one of claims 1 to 14, or the method for generating a trained model described in claim 15. 1. A component mounting system that generates a product by mounting a plurality of components supplied to a plurality of supply units on an object, the component placement assistance system providing assistance in determining a placement of the plurality of components at the plurality of supply units, the system comprising:
an acquisition unit that acquires mounting data related to the plurality of components included in the product;
and a suggestion unit that uses a trained model associated with a feature of at least one of the supply units to propose an arrangement of the plurality of components for the plurality of supply units from the mounting data so as to shorten a mounting time required for generating the product in the component mounting system.
Placement support system. The suggestion unit,
a primary proposal unit that proposes an arrangement of the plurality of components for a plurality of primary groups;
a secondary suggestion unit that proposes an arrangement of the plurality of components for a plurality of secondary groups when each of the plurality of primary groups is classified into a plurality of secondary groups including at least one supply unit among the plurality of supply units,
20. The placement assistance system of claim 17. The suggestion unit,
a secondary suggestion unit that receives a proposal for an arrangement of the plurality of components for a plurality of primary groups, and proposes an arrangement of the plurality of components for the plurality of secondary groups when each of the plurality of primary groups is classified into a plurality of secondary groups including at least one supply unit among the plurality of supply units;
20. The placement assistance system of claim 17. The arrangement support system according to any one of claims 17 to 19,
the component mounting system being used in the arrangement of the plurality of components proposed by the arrangement assistance system to generate the product;
Working system.

Images