JP2012242888A - Object rearrangement planning device, method, and program - Google Patents

Abstract

PROBLEM TO BE SOLVED: To automatically create an object rearrangement plan for rearranging a plurality of objects from positions on a layout before transfer to positions on a layout after transfer under constraints on object transfer, by using calculation means such as a personal computer.SOLUTION: An object rearrangement planning device extracts transfer from a departure position on a layout before transfer to a target position on a layout after transfer for each object as transfer element, and searches for constraints on order of a plurality of transfer elements. Then, the device creates a directed graph which regards the plurality of transfer elements as nodal points, and constraints on the order of the plurality of transfer elements as directed arrows connecting the nodal points. When there is a loop formed by the nodal points and the directed arrows on the directed graph, the device resolves the loop by decomposing the nodal points included in the loop. Further, the device determines partial order of the transfer elements under constraints on the order of the transfer elements, and converts the partial order of the transfer elements into total order which makes the shortest operation time required for rearranging objects, by using a genetic algorithm.

Description

  The present invention relates to a technique for planning an object moving procedure for rearranging a plurality of objects to a next destination.

  2. Description of the Related Art Conventionally, in a vehicle outfitting factory where operations such as rail car assembly and painting are performed, a work area is determined for each work, and it is necessary to move the vehicle to a work area corresponding to the work process. In the outfitting factory, a plurality of work sections are arranged in parallel with the rails on the laid rail to form a work section lane. A plurality of work area lanes are lined up in the direction crossing the rail to form one area. A plurality of areas are provided in a direction parallel to the rails, and a traverser is provided between the areas. The traverser moves the vehicle in a direction intersecting with the rail, and the vehicle can be moved to another work area lane via the traverser.

  In the outfitting factory as described above, the vehicle can move in the work area lane to which the work area where the vehicle exists, but cannot overtake other vehicles or move to the work area where other vehicles exist. . Furthermore, the vehicle can move across the work area lane, but in this case, it is necessary to pass through a traverser. Since there are such restrictions on the movement of the vehicle in the outfitting factory, deadlock (deadlock) may occur if the vehicle is moved unplanned. Therefore, a vehicle rearrangement plan for moving each vehicle from the arrangement of the vehicle before movement to the arrangement of the vehicle after movement is created without deadlock and taking as much time as possible. Each vehicle is moved sequentially according to the above. The vehicle rearrangement plan defines a rearrangement process such as a vehicle moving order and a movement route for rearranging the vehicle from the current layout to the next layout. At present, skilled workers create a vehicle relocation plan based on their experience. As described above, since the creation of the vehicle rearrangement plan is based on the experience of the operator, it may be difficult to secure personnel who can create the vehicle rearrangement plan. In addition, there is no way to evaluate the vehicle rearrangement plan, and the created vehicle rearrangement plan is optimal, that is, deadlock does not occur and the movement of the vehicle can be completed in the shortest time. It cannot be guaranteed.

  The above-mentioned problem is not limited to the rearrangement of vehicles in a railway vehicle outfitting factory, but also occurs when a plurality of objects are rearranged from a pre-movement layout to a post-movement layout under conditions where there are restrictions on the movement of the objects. Therefore, a system that automatically creates an object rearrangement plan that determines the movement order and movement path of objects in order to relocate a plurality of objects from the pre-movement layout to the post-movement layout, not limited to the railroad car outfitting factory Development is desired.

  By the way, as shown in Patent Documents 1 and 2, for example, methods for planning a procedure for rearrangement of objects under movement restrictions different from those of the above-mentioned outfitting factory have been conventionally proposed. In Patent Document 1, the storage items are rearranged in an automatic warehouse having a plurality of storage shelves arranged in the horizontal direction and the vertical direction, and a lifting carriage for performing replacement work of the storage items on these storage shelves. A technique for doing this is shown. In the method described in Patent Document 1, first, a stored item such that a stored item with a high frequency of loading / unloading requests is stored in a higher-ranked storage shelf position (that is, a storage shelf position closer to the carriage home position). Create a relocation plan. Next, the rearrangement plan is optimized so that the replacement is executed in order from the combination of the stored items having the large replacement effect (that is, the overall operation cost and the improvement rate of the access time).

  Further, in Patent Document 2, in a container yard having a plurality of transport machines on the same lane, in order to carry out the container smoothly and quickly, the interference between the transport machines is eliminated and the total work time for the rearrangement is as long as possible. A technique for creating a container rearrangement plan that can be shortened is shown. In the method described in Patent Document 2, a genetic algorithm is applied to a population having a gene expression representing a container processing order processed by each transport machine, crossover and mutation are performed, and a predetermined fitness function is used. The optimal plan is obtained while repeating the generation update with the kite.

JP-A 61-295902 JP-A-8-272863

  In Patent Document 1, the carriage can take out stored items from a plurality of storage shelves arranged in the horizontal direction and the vertical direction without restriction, and move the stored items to any empty storage shelf position. As described above, Patent Document 1 describes the creation of a relocation plan because the stored items can be replaced relatively freely. However, the method for determining the moving order of the stored items and the movement route is described in detail. Absent.

  In addition, the technique described in Patent Document 2 assumes a case in which an import container is rearranged from a buffer to a yard and an export container is rearranged from a yard to a buffer with a plurality of transport machines moving on the same lane. The container is planned to be delivered without the collision or overtaking of the plurality of transport machines. The technique described in Patent Document 2 can be applied to the simple movement as described above. However, when the movement becomes complicated, there is a high possibility that the evaluation function of the genetic algorithm will fail. There may be situations where a solution cannot be obtained.

  Therefore, the present invention provides an object rearrangement plan that does not cause a deadlock (i.e., an object relocation plan) in order to rearrange a plurality of objects from a pre-movement layout to a post-movement layout under conditions where the movement of the objects is limited. It is an object of the present invention to provide a technique that can automatically create a movement order and a movement route using a computing means such as a personal computer. In addition, an object of the present invention is to provide a technique for creating an optimal object rearrangement plan in which the work time required for object rearrangement is the shortest.

In the object rearrangement planning apparatus according to the present invention, there are a plurality of lanes in which one or more blocks in which objects can be arranged one by one are arranged, the object moves on the lane without overtaking other objects, In a stage allowed to move between the lanes,
Relocate the objects on the stage from the pre-movement layout where each of the objects on the stage is located in the block corresponding to the starting position to the layout after the movement where each of the objects on the stage is located in the block corresponding to the target position An object rearrangement planning device for creating an object rearrangement plan for
A moving element extraction unit that extracts movement from the starting position to the target position for each object as a moving element based on the pre-movement layout and the post-movement layout;
An order constraint search unit that searches for a constraint on the order between the plurality of moving elements based on the layout before moving, the layout after moving, and the configuration of the stage;
And a directed graph creating unit that creates a directed graph with the extracted moving elements as nodes and a directional arrow connecting the nodes as a restriction on the order between the moving elements. In the above, “layout” means the overall arrangement of objects on the stage.

Similarly, in the object rearrangement planning method according to the present invention, there are a plurality of lanes in which one or more blocks where objects can be arranged one by one exist, and the object moves on the lane without overtaking other objects. And in a stage where the object is allowed to move between the lanes,
Relocate the objects on the stage from the pre-movement layout where each of the objects on the stage is located in the block corresponding to the starting position to the layout after the movement where each of the objects on the stage is located in the block corresponding to the target position An object rearrangement planning method for performing
Extracting the movement from the starting position to the target position for each object as a moving element based on the pre-movement layout and the post-movement layout;
Obtaining a restriction on the order between the plurality of moving elements based on the layout before moving, the layout after moving, and the configuration of the stage;
And a step of creating a directed graph with the extracted moving elements as nodes and a restriction of the order between the moving elements as a directed arrow connecting the nodes. In the above, “layout” means the overall arrangement of objects on the stage.

Similarly, the object rearrangement planning program according to the present invention includes a plurality of lanes in which one or more blocks in which objects can be arranged one by one are arranged, and the object moves on the lane without overtaking other objects. And a block in which each of the objects on the stage corresponds to a target position from a pre-movement layout in which each of the objects on the stage corresponds to a starting position in a stage in which the object is allowed to move between the lanes. A program executed on a computer for creating an object rearrangement plan for rearranging an object on the stage to a post-movement layout arranged in
A process of extracting movement from the starting position to the target position for each object as a moving element based on the pre-movement layout and the post-movement layout;
A process for obtaining an order restriction between the plurality of moving elements based on the layout before moving, the layout after moving, and the configuration of the stage;
The computer is caused to execute a process of creating a directed graph in which the extracted plurality of moving elements are nodes, and the order restriction between the plurality of moving elements is a directed arrow connecting the nodes. In the above, “layout” means the overall arrangement of objects on the stage.

  According to the object rearrangement planning apparatus, the object rearrangement planning method, and the object rearrangement planning program, the created directed graph shows the movement of the objects by the nodes, and the moving order of the objects by the directed arrows. In this directed graph, it is possible to know whether or not a deadlock that occurs during the work of rearranging a plurality of objects occurs by the presence or absence of a loop composed of nodes and directed arrows. Therefore, in order to relocate multiple objects on the stage from the pre-movement layout to the post-movement layout under conditions where the movement of the objects is limited, an object relocation plan that does not cause a deadlock (ie, the object movement order) And a movement route) can be automatically created using a computing means such as a personal computer.

  In the object rearrangement planning apparatus, the order constraint search unit is configured such that the movement of one or more objects related to a block is a movement starting from the block, a movement passing through the block, and a movement arriving at the block. Thus, it is possible to search for an order restriction between the plurality of moving elements.

  By searching for the order restriction between the plurality of moving elements as described above, it is possible to obtain the order restriction between the plurality of moving elements by a simple calculation.

  The object rearrangement planning apparatus may include a loop extraction unit that extracts a loop formed by the nodes and the directed arrow in the directed graph by a depth-first search method.

  According to the object rearrangement planning apparatus, it is possible to extract all the loops that mean deadlock that occurs during the rearrangement work of a plurality of objects existing on the directed graph.

  In the object rearrangement planning device, when the loop is extracted by the loop extraction unit, one node included in the loop is temporarily saved from the starting position of the object corresponding to the node. The first node representing the moving element consisting of the movement to the retracted position to be moved and the second node representing the moving element consisting of the movement of the object from the retracted position to the target position are eliminated. It is preferable to provide a loop elimination unit. Here, when the plurality of loops are extracted by the loop extraction unit, and the extracted plurality of loops include a loop with two nodes, the loop cancellation unit has two nodes. It is preferable to replace the nodes included in a part of the path of the other moving element with priority to the first and second nodes among the nodes constituting the loop.

  According to the object rearrangement planning method or the object rearrangement planning apparatus, since the loop that existed on the directed graph is eliminated, a plurality of objects are prevented so as not to cause a deadlock between the operations of the rearrangement of a plurality of objects. It becomes possible to determine the order of the moving elements.

  The object rearrangement planning apparatus may include an order determination unit that determines an order of the plurality of moving elements based on a restriction on an order between the plurality of moving elements represented by the directed arrow.

  According to the object rearrangement planning apparatus, an object rearrangement plan (that is, a movement order and a movement path of each object) is created so as not to cause a deadlock during the work of rearranging a plurality of objects on the stage. be able to.

  In the object rearrangement planning apparatus, the order determination unit assigns a random number to each moving element, and determines a priority order based on the random number under the restriction of the order between the plurality of moving elements. Determine the order of the moving elements, calculate the work time required to rearrange the plurality of objects in the determined order of the plurality of moving elements, use the numerical sequence of random numbers given to each moving element as a gene, The time calculation process is used as an evaluation function, and it is preferable to search for a solution by repeating crossover so that the gene having the shortest working time remains using a genetic algorithm, and determine the order of the plurality of moving elements.

  According to the object rearrangement planning apparatus, the determined order of the plurality of moving elements is such that the work time required for rearranging the objects on the stage is the shortest. Therefore, it is possible to create an optimal object rearrangement plan in which deadlock does not occur during the work of rearranging the objects on the stage and the work time required for the rearrangement of objects is the shortest time. Furthermore, since a partial order set of a plurality of moving elements is determined in advance and a genetic algorithm is used to convert this partial order set into a full order set, the evaluation function of the genetic algorithm may fail. Therefore, an appropriate solution can be obtained in a short time.

  According to the present invention, in order to relocate a plurality of objects on the stage from a pre-movement layout to a post-movement layout under a condition in which the movement of the object is limited, an object rearrangement plan that does not cause a deadlock (ie, an object It is possible to automatically create a movement order and a movement route) using a computing means such as a personal computer.

It is a block diagram which shows schematic structure of the object rearrangement planning apparatus which concerns on embodiment of this invention. It is a figure which shows an example of the condition before and behind rearrangement of the stage in which the some object was arrange | positioned, (a) shows the layout before a movement, (b) has shown the layout after a movement. It is a figure explaining the order restriction | limiting of the movement regarding a block (2, e). It is a flowchart explaining the flow of an object rearrangement plan process. It is an example of the graph which uses the movement of an object as a node. It is an example of the graph to which the directed arrow which connects a node was added. It is an example of the graph by which the loop was eliminated by decomposition | disassembly of a node. It is a figure (1) explaining the method of determining the order of a node by the random number given as a gene. It is a figure (2) explaining the method of determining the order of a node by the given random number. It is a figure (3) explaining the method of determining the order of a node by the given random number. It is FIG. (4) explaining the method of determining the order of a node by the given random number. It is FIG. (5) explaining the method of determining the order of a node by the given random number. It is FIG. (6) explaining the method of determining the order of a node with the given random number. It is a figure (7) explaining the method of determining the order of a node with the given random number. It is a figure which shows the modification of the stage by which the several object is arrange | positioned.

  Hereinafter, embodiments for carrying out the present invention will be described in detail with reference to the drawings. In the following description, the same or corresponding elements are denoted by the same reference symbols throughout the drawings, and redundant description thereof is omitted.

  First, the object rearrangement planning apparatus according to the present embodiment will be described. FIG. 1 is a block diagram showing a schematic configuration of an object rearrangement planning apparatus according to an embodiment of the present invention. As shown in the figure, the object rearrangement planning apparatus 1 mainly comprises one or a plurality of computers 2, and each computer 2 is a CPU (central processing unit), an object rearrangement planning program executed by the CPU, and this program. Main storage device for storing rewritable data, secondary storage device for temporarily storing data when CPU executes program, interface for connecting CPU and external device, internal path for connecting these, etc. (Both not shown). The object rearrangement planning program is stored in various storage media such as a flexible disk, a CD-ROM, and a memory card, and is installed in the computer 2 from these storage media. The object rearrangement planning program is executed by the CPU of the computer 2, thereby realizing the function as the object rearrangement planning apparatus 1. The object rearrangement planning program includes at least a moving element extraction routine, an order constraint search routine, a directed graph creation routine, a loop extraction routine, a loop elimination routine, and an order determination routine. These function as a moving element extraction unit 61, an order constraint search unit 62, a directed graph creation unit 63, a loop extraction unit 64, a loop elimination unit 65, and an order determination unit 66, respectively, in the object rearrangement planning apparatus 1. The functions of these functional units of the object rearrangement planning apparatus 1 will be described in detail later.

  The computer 2 is connected to an input device 3 such as a mouse pointer or a keyboard via an interface. When the user performs an input operation on the input device 3, a signal indicating the operation content is input to the computer 2, and the computer 2 performs processing based on the input signal. The computer 2 is connected to an output device 4 such as a display or a printer via an interface. Various types of information that the computer 2 provides to the user in operation are appropriately displayed by characters and symbols on the output device 4. Further, the computer 2 is connected to the storage device 5 through an interface. The storage device 5 stores at least information related to the stage 10 on which the object is arranged, and information related to the pre-movement layout and the post-movement layout of a plurality of objects. Note that these pieces of information may be stored in a storage device included in the computer 2 instead of the storage device 5.

  Next, a stage on which an object is arranged and restrictions on the movement of the object on this stage will be described. 2A and 2B show an example of a stage on which a plurality of objects are arranged. FIG. 2A shows the layout before moving the object, and FIG. 2B shows the layout after moving the object. In the stage 10, one or more blocks 13 are arranged in one direction (left and right direction in FIG. 2) to form one lane L. A plurality of lanes L are arranged substantially parallel to each other in the direction intersecting the lanes (the vertical direction in FIG. 2) to form one area 12.

In the example shown in FIGS. 2A and 2B, the stage 10 is formed with two areas, a first area 12a and a second area 12b, arranged in the row direction. Three lanes L ₁₁ , L ₁₂ and L ₁₃ are formed in the first area 12 a, and three lanes L ₂₁ , L ₂₂ and L ₂₃ are formed in the second area 12 b. Yes. The lane L ₁₁ is provided with two blocks 13 whose coordinates are represented by (1, b) and (1, c), respectively. The lane L _12, coordinates (2, a) and (2, b) and (2, c) in three blocks 13, represented respectively provided. The lane L _13, coordinates (3, a) and (3, b) and (3, c) in three blocks 13, represented respectively provided. Similarly, a plurality of blocks 13 are provided in the lane L ₂₁ , the lane L ₂₂ , and the lane L ₂₃ . The stage 10 is provided with a total of 15 blocks 13, of which seven objects 20 from A to G are arranged in seven blocks 13, respectively. Hereinafter, the block 13 in which the objects A to G shown in FIG. 2A are arranged may be referred to as a “starting position”, and the block in which the objects A to G shown in FIG. 2B are arranged. 13 is sometimes referred to as a “target position”.

  A traverser 11 that is movable in a direction intersecting with the lane is provided between the first area 12a and the second area 12b of the stage 10. The traverser 11 can carry the objects 20 one by one in the direction crossing the lane. When the object 20 existing in a certain lane L moves to another lane L, that is, when the object 20 moves across the lane L, the traverser 11 is used.

  In the stage 10 configured as described above, the object 20 is subjected to the following movement restrictions <1> to <3>. <1> The stage 10 on which an object is arranged has a plurality of lanes L in which one or more blocks 13 on which one object 20 can be arranged are arranged, and the object 20 can move along the lane L. . <2> The object 20 cannot move to the block 13 where another object exists, or cannot move beyond the other object 20. <3> The object 20 can move to another lane L, but must always pass through the traverser 11. This is because each lane L has a block corresponding to an entrance (in the example shown in FIGS. 2A and 2B, a block adjacent to the traverser 11). 20 means that it cannot move to another lane L.

  The following movement order restrictions arise from the above movement restrictions. FIG. 3 is a diagram for explaining the movement order restriction regarding the block (2, e). Hereinafter, “block (coordinate)” indicates a block of the coordinate. For example, when the movement of the object 20 related to the block (2, e) is extracted from the arrangement before and after the movement shown in FIGS. 2A and 2B, the object E becomes the block (2, e) as shown in FIG. Moving from object to block (2, c), moving object A from block (1, b), passing through block (2, e) to block (2, f), object There are three movements, D moving from block (1, c) and arriving at block (2, e). From the standpoint of the block (2, e), the movement is not established unless it is the order of the starting movement, the passing movement, and the arriving movement, and this causes a movement order restriction. Therefore, the order restriction of the movement of the objects A, D, and E with respect to the block (2, e) is the order of the object E → the object A → the object D. Although the object movement order restriction related to the block (2, e) has been described above, the object movement order restriction similarly occurs in the other blocks.

  Subsequently, referring to FIG. 4, an object rearrangement plan for rearranging a plurality of objects from the pre-movement layout to the post-movement layout, that is, a processing procedure for determining the movement order and movement path of the objects is illustrated. This will be described using the examples shown in 2 (a) and 2 (b). FIG. 4 is a flowchart for explaining the flow of the object rearrangement planning process. The CPU of the computer 2 of the object rearrangement planning apparatus 1 executes the object rearrangement planning program and performs predetermined calculations in the following procedure while appropriately reading out the information stored in the storage device 5 to thereby perform object rearrangement. A plan can be created.

  The directed graph creation unit of the object rearrangement planning apparatus 1 first acquires information on the stage 10 on which the plurality of objects 20 are arranged, and information on the pre-movement layout and the post-movement layout of the plurality of objects 20 (step). S0). Here, the CPU of the computer 2 reads out information related to the stage 10 on which the plurality of objects 20 are arranged, and information related to the pre-movement layout and the post-movement layout of the plurality of objects 20 from the storage device 5. However, these pieces of information are not limited to those stored in the storage device 5 in advance. For example, the computer 2 prompts the user to input information related to the pre-movement layout and the post-movement layout of the plurality of objects 20 with the output device 4, and before the movement of the plurality of objects 20 input by the user via the input device 3. Information related to the layout and the post-movement layout can also be acquired.

  The information about the stage 10 on which the plurality of objects 20 acquired by the object rearrangement planning apparatus 1 and the information about the layout before and after the movement of the plurality of objects 20 are included in the configuration of the stage 10 and the information on the stage 10. It includes at least a restriction on movement of the object 20, a pre-movement layout of the plurality of objects 20 (coordinates of starting positions of the respective objects), and a layout after movement of the plurality of 20 objects (coordinates of the target positions of the respective objects).

Next, the moving element extraction unit 61 of the object rearrangement planning apparatus 1 extracts the movement from the starting position to the target position for each object 20 as a moving element (step S1). Note that the moving elements include those in which the starting position and the target position are the same and the object does not move as a result. The directed graph creation unit 63 of the object rearrangement planning device 1 creates a graph in which each extracted moving element is represented as a node (node) (step S2). When the total number of objects 20 is n, n moving elements are extracted, and n nodes N _{1 to} N _n are created on the graph.

As shown in FIG. 5, in the example shown in FIGS. 2A and 2B, a total of seven nodes N _{1 to} N ₇ are created on the graph for each of the objects A to G. The node N _{1 of the} object A is composed of a moving element from the starting position (1, b) to the target position (2, f), and the moving element is represented as “A: 1b → 2f”. Similarly, the node N ₂ of the object B is composed of a moving element from the starting position (2, d) to the target position (2, a), and the moving element is represented as “B: 2d → 2a”. The node N _{3 of the} object C is composed of a moving element from the starting position (2, b) to the target position (1, c), and the moving element is expressed as “C: 2b → 1c”. The node N _{4 of the} object D includes a moving element from the starting position (1, c) to the target position (2, e), and the moving element is represented as “D: 1c → 2e”. The node N _{5 of the} object E is composed of a moving element from the starting position (2, e) to the target position (2, c), and the moving element is expressed as “E: 2e → 2c”. The node N _{6 of the} object F is composed of a moving element from the starting position (1, d) to the target position (3, a), and the moving element is expressed as “F: 1d → 3a”. The node N _{7 of the} object G is composed of a moving element from the starting position (3, b) to the target position (3, b), and the moving element is expressed as “G: 3b → 3b”. It should be noted here that for an object that does not move before and after rearrangement (object G), a node is created as having a moving element from the starting position to the same target position as the starting position. .

Subsequently, the order constraint search unit 62 of the object rearrangement planning device 1 calculates a movement route from the departure position to the target position for all the objects 20 (step S3), and displays the movement route on the graph based on the calculated movement route. A search is made for the order restriction of movement between the created nodes N _{1 to} N _n (step S 4). The order restriction of movement between nodes is, in other words, the order restriction of movement between moving elements.

The order restriction of movement between nodes is determined based on the order of the movement of an object related to a block in the order of starting movement, passing movement, and arriving movement. For example, as shown in FIG. 3, the order of movement of the block (2, e) is as follows: the movement of the object E that leaves this block, the movement of the object A that passes through this block, and the movement of the object D that arrives at this block. The order of movement. From this, it is possible to determine the order of movement of the nodes, with the node N ₅ first, the node N ₁ after, the node N ₁ first, and the node N ₄ next.

Based on the order restriction of movement between nodes determined as described above, the directed graph creation unit 63 of the object rearrangement planning apparatus 1 has edges (edges) connecting the nodes N _{1 to} N _n created on the graph. It creates with a directional arrow (step S5). This directed arrow represents the order restriction of movement between nodes, and the root of the directed arrow is first and the direction of the arrow is later. In this way, a directed graph is created in which each node representing a moving element is connected by a directed arrow representing the order constraint of the moving element.

As shown in FIG. 6, an arrow from the node N ₅ to the node N ₁ and an arrow from the node N ₁ to the node N ₄ are created due to the movement order restriction regarding the block (2, e). Similarly, an arrow heading from the node N ₇ to the node N ₆ and an arrow heading from the node N ₆ to the node N ₇ are created due to the order restriction of the movement related to the block (3, b). Due to the movement order restriction on the block (2, c), an arrow from the node N ₃ to the node N ₅ and an arrow from the node N ₂ to the node N ₅ are created. Due to the movement order restriction on the block (1, c), an arrow from the node N ₄ to the node N ₁ and an arrow from the node N ₁ to the node N ₃ are created. An arrow heading from the node N ₃ to the node N ₂ is created due to the movement order restriction on the block (2, b). Due to the movement order constraint on block (2, d), an arrow from node N ₂ to node N ₁ , an arrow from node N ₂ to node N ₅ , and an arrow from node N ₂ to node N ₄ are created. The

  In the directed graph created as described above, there may be a closed loop formed by nodes and directed arrows. The loop on the directed graph is a moving element order constraint loop, which means deadlock. In the stage 10, the object 20 cannot pass the other object 20 or move to the same block 13 as the other object 20 at the same time, so that deadlock may occur. This deadlock is a loop of movement order restriction. Appears in the directed graph. Therefore, it is possible to know whether or not a deadlock that occurs during the relocation operation of a plurality of objects occurs depending on the presence or absence of a loop that exists in the directed graph created by the directed graph creation unit of the object rearrangement planning apparatus 1. .

  The loop extraction unit 64 of the object rearrangement planning apparatus 1 extracts a closed loop formed by a directed arrow in the directed graph created as described above (step S6). A depth-first search of graph theory is used to extract all the loops in the directed graph. Note that when the rearrangement of a plurality of objects is relatively simple, there is a case where no loop exists on the directed graph, that is, no deadlock occurs. In such a case, the loop is not extracted.

The directed graph shown in FIG. 6 includes six loops (a loop of nodes N ₁ and N _4, a loop of nodes N ₁ , N ₃ and N _2, a loop of nodes N ₁ , N ₃ and N ₅ , and nodes N ₁ and N ₃ , N ₂ , N ₅ , nodes N ₂ , N ₄ , N ₁ , N ₃ , and nodes N ₆ , N ₇ ), and all these loops are extracted.

  In the state where the loop exists on the graph, as described above, the deadlock occurs, and thus the movement is not established. Therefore, when the loop extraction unit 65 of the object rearrangement planning apparatus 1 extracts a loop, the loop resolving unit 65 decomposes the nodes for all the extracted loops and eliminates all the loops (step S7). .

  In the above, “decomposition of a node” means that one of the nodes constituting the loop is selected, and that one node is replaced with a node representing two movements that pass through a temporarily defined save position. To do. In other words, “decomposition of a node” means that one moving element is replaced with a moving element from the departure position to the retracted position and a moving element from the retracted position to the departure position.

For example, as shown in FIG. 6, there are loops of nodes N ₆ and N ₇ on the graph. When the node is decomposed for this loop, as shown in FIG. 7, one of the nodes N ₆ and N ₇ constituting the loop (here, node N ₇ ) is selected and this node is selected. N ₇ is decomposed into a node N ₇₁ representing movement from the departure position (3, b) to the save position W3 and a node N ₇₂ representing movement from the save position W3 to the target position (3, b). Note that the save position W3 must be different from the departure position and the target position. As described above, the node N ₇ is decomposed into the node N ₇₁ and the node N ₇₂ , thereby eliminating the loop of the nodes N ₆ and N ₇ .

  In performing the decomposition of the nodes, the number of nodes that can be decomposed is not limited to one among the nodes constituting the loop. However, when a plurality of nodes can be decomposed, a rule is set for selecting nodes to be decomposed. The first rule is that in a loop of length 2 consisting of two nodes and two arrows, the path of the moving element is selected with priority given to the nodes included in part of the path of other moving elements. It is. According to this rule, when a node having the same starting position and target position is included in the loop, the node is always decomposed. The second rule is to preferentially select nodes for which more loops are eliminated by decomposition. An increase in the number of nodes (that is, moving elements) due to the decomposition of the nodes leads to an increase in time until the rearrangement is completed. Therefore, in order to reduce the number of nodes on the graph as much as possible, nodes that eliminate more loops must be preferentially selected.

In the graph shown in FIG. 6, nodes N ₃ , N ₄ , and N ₇ to be decomposed according to the above rules are selected, and all loops can be eliminated by decomposing these nodes. There are two loops of length 2 on the graph shown in FIG. In accordance with the first rule, in the loop of nodes N ₁ and N ₄ , the path of the moving element of the object D at the node N ₄ is included in a part of the path of the moving element of the object A. Node N ₄ is selected. Further, according to the first rule, in the loop of the nodes N ₆ and N ₇ , the node N ₇ has the same starting position and target position, so the node N ₇ is selected as the node to be decomposed. The loops other than the length 2 remaining after the length 2 loop is decomposed include the nodes N ₁ , N ₃ , N ₂ , the nodes N ₁ , N ₃ , N ₅ and the nodes N ₁ , N _{There are 3} , N ₂ , and N ₅ loops. Here, according to the second rule, the node N ₁ or the node N ₃ is preferentially selected as a node to be decomposed. In the example shown in FIG. 7, the node N ₃ is selected, the node N ₃₁ is the node N ₃ representing the moving element from the starting position (2, b) to the retracted position W1, the target position (1 a retracted position W1, It is broken down into nodes N ₃₂ representing moving elements up to c). As a result of eliminating all the loops on the graph by the decomposition of the nodes, the number of nodes on the graph is increased from 7 to 10.

  When all the loops on the graph are resolved by repeating the decomposition of the nodes as described above, a partially ordered set of nodes, that is, a partially ordered set of moving elements is completed. In a partial order set of moving elements, after moving a certain moving element, it is not possible to uniquely select a moving element that represents the next movement to be performed. Therefore, the object rearrangement planning apparatus 1 searches for the order of moving elements that completes the rearrangement of objects in the shortest time based on the partial order set of moving elements completed as described above. That is, the partial order set of moving elements is converted into a full order set. There are several methods for converting a partially ordered set of moving elements into a fully ordered set. Here, a genetic algorithm (GA) is used.

Specifically, the order determining unit 66 of the object rearrangement planning apparatus 1 gives a random number to each of the nodes N _{1 to} N _n representing the moving element (step S8), and satisfies the condition that the order restriction on the movement of the nodes is observed. Below, the total order of the nodes N _{1 to} N _n is determined (step S9). If the total order of the nodes is determined, the order determining unit 66 of the object rearrangement planning apparatus 1 assigns a vacant block to the temporarily set saving position (step S10). If an actual block is assigned to the retreat position, the movement order and movement path of all the objects can be determined, and the work time required for the rearrangement of the objects can be calculated. The order determining unit 66 of the object rearrangement planning apparatus 1 obtains the work time required for the object rearrangement in the entire order of the determined nodes (step S11). Then, the order determining unit 66 of the object rearrangement planning apparatus 1 applies the genetic algorithm GA using the numerical sequence of random numbers given to the nodes as genes and using the process of obtaining the work time required for the rearrangement of the objects as an evaluation function. Crossover is performed so that the gene with the shortest working time remains, and the process of steps S8 to S11 is repeated to search for a solution (step S12). Finally, the order determining unit 66 of the object rearrangement planning apparatus 1 determines the entire order of the moving elements corresponding to the genes for which the work time required for the object rearrangement is the shortest as the optimal object rearrangement plan (step). S13). In general, the genetic algorithm GA may not function well if the possibility that the evaluation function fails is too high. On the other hand, in the flow of the object rearrangement planning process described above, the role of the genetic algorithm GA is narrowed down only to the degree of freedom resolution by obtaining a partial order set before applying the genetic algorithm GA. . As a result, the possibility of generating an incorrect movement order is largely eliminated, and as a result, it is possible to easily obtain the entire order of moving elements that minimizes the work time required for rearrangement of objects.

Here, the flow in which the object rearrangement planning apparatus 1 converts the partial order set of moving elements (nodes N _{1 to} N ₇₂ ) shown in FIG. 7 into a full order set using the genetic algorithm GA will be described. The object rearrangement planning apparatus 1 first gives random numbers [0, 1] to all the nodes N _{1 to} N ₇₂ on the graph shown in FIG. 7 as shown in the following Table 1 (step S8).

Next, the object rearrangement planning apparatus 1 uses a given numerical sequence of random numbers as a gene, and selects a node to which a maximum (or minimum) gene element is given under the condition that the order restriction of movement between nodes is observed. By selecting, the total order of the nodes N _{1 to} N ₇₂ is determined (step S9). As shown in FIG. 8, first, nodes N ₃₁ , N ₄₁ , and N _{71 that} are sources of directed arrows are extracted under the condition that the order restriction of movement between nodes is observed, and these nodes N ₃₁ , The genetic elements (0.4, 0.5, 0.0) given to N ₄₁ and N ₇₁ are compared. Among these, the node N ₄₁ to which the maximum gene element is given is selected, and the order of the node N ₄₁ is the first.

Next, as shown in FIG. 9, nodes N ₃₁ and N ₇₁ are extracted under the condition that the order restriction of movement between nodes whose order is not yet determined is satisfied, and given to these nodes N ₃₁ and N ₇₁ . Genetic elements (0.4, 0.0) are compared. Among these, the node N ₃₁ to which the maximum gene element is given is selected, and the order of the node N ₃₁ is the second.

Subsequently, as shown in FIG. 10, nodes N ₂ and N ₇₁ are extracted under the condition that the order restriction of movement between nodes whose order is undetermined is satisfied, and given to these nodes N ₂ and N ₇₁ . Genetic elements (0.1, 0.0) are compared. Among these, the node N ₂ to which the maximum gene element is given is selected, and the order of the node N ₂ is the third.

Subsequently, as shown in FIG. 11, nodes N ₅ and N ₇₁ are extracted under the condition that the order restriction of movement between nodes whose order is undetermined is kept, and given to these nodes N ₅ and N ₇₁ . Genetic elements (0.8, 0.0) are compared. Among these, the node N ₅ to which the maximum gene element is given is selected, and the order of the node N ₅ is the fourth.

Subsequently, as shown in FIG. 12, nodes N ₁ and N ₇₁ are extracted under the condition that the order restriction of movement between nodes whose order is not yet determined is satisfied, and given to these nodes N ₁ and N ₇₁ . Genetic elements (0.2, 0.0) are compared. Among these, the node N ₁ to which the largest gene element is given is selected, and the order of the node N ₁ is the fifth.

Subsequently, as shown in FIG. 13, nodes N ₃₂ , N ₄₂ , and N ₇₁ are extracted under the condition that the order restriction of movement between nodes whose order is undetermined is maintained, and these nodes N ₃₂ , N ₄₂ , genetic elements given to N ₇₁ (0.6,0.3,0.0) are compared. Among these, the node N ₃₂ to which the maximum gene element is given is selected, and the order of the node N ₃₂ is the sixth. Then, the gene elements (0.3, 0.0) given to the nodes N ₄₂ and N ₇₁ are compared, and the order of the node N ₄₂ to which the largest gene element is given is the seventh.

Finally, as shown in FIG. 14, the order of the node N ₇₁ is the eighth and the order of the node N ₆ is the ninth, under the condition that the order restriction of the movement of the nodes whose order is undetermined is observed. The order of the node N ₇₂ is the tenth. As described above, as shown in the following Table 2, the total order of the nodes N _{1 to} N ₇₂ is determined.

  When the total order of the nodes is determined, the object rearrangement planning apparatus 1 searches for a vacant block during the period in which the object stays at the retreat position, and assigns this empty block to the retreat position (step S10).

  As shown in Table 2, the object C stays in the first retreat position W1 between the second order and the sixth order. The block (3, d) that is vacant during this stay period is assigned to the first retreat position W1. The object D stays between the first and seventh positions in the second retreat position W2. The block (3, c) that is vacant during this stay period is assigned to the second retreat position W2. The object G stays between the eighth and tenth positions in the third retreat position W3. The block (3, d) that is vacant during this stay period is assigned to the third retreat position W3. When there are a plurality of empty blocks that can be assigned to the save position, a single empty block is selected using the genetic algorithm GA so that the work time required for the rearrangement of the object is the shortest. In this embodiment, an empty block is assigned to each save position, but the save position is not necessarily limited to the block 13 on the stage 10. For example, when a plurality of traversers 11 are provided on the stage 10, one or more of the traversers 11 can be assigned to any of the retracted positions. Further, one or a plurality of retracting positions different from the block 13 may be provided on the stage 10.

  The object rearrangement planning apparatus 1 is required to move the work time required for the rearrangement of objects, in other words, to move all the objects from the starting position to the target position with respect to the movement order determined as described above. A long working time is calculated by simulation (step S11). A simulation model is built in advance in the computer 2 of the object rearrangement planning apparatus 1, and the object rearrangement planning apparatus 1 calculates a work time required for the rearrangement of the object using the simulation model. The calculated work time required for the rearrangement of the object includes the movement time of the object on the lane L, the time when the object is transported by the traverser 11, the time when the traverser 11 moves without placing the object, and the like. .

  Then, the object rearrangement planning apparatus 1 uses the work time required for the rearrangement of the object as the evaluation value of the gene, and the gene (that is, the gene with the shortest work time required for the rearrangement of the object by the genetic algorithm GA remains (that is, The process of step S8 to step S11 is repeated while crossing the numerical value array of random numbers given to each node (step S12). Note that uniform crossover may be used as the crossover (recombination) method of the genetic algorithm GA. For example, as shown in Table 3 below, the parent gene 1 and the parent gene 2 that are numerical sequence of random numbers given to each node are uniformly recombined, and the parent gene has a probability of 1/2 for each element of the gene. A child gene is generated by taking over elements 1 and 2. In Table 3, a child gene is generated by combining gene elements surrounded by a thick frame of parent gene 1 and parent gene 2. And the process of step S8-step S13 is performed using this child gene.

  Finally, the object rearrangement planning apparatus 1 sets the movement order corresponding to the gene with the smallest evaluation value, in other words, the gene with the shortest work time required for the rearrangement of the object as the solution of the genetic algorithm GA, An object rearrangement plan is determined (step S13). The object rearrangement plan determined in this way includes the movement order and movement path of the objects, does not cause a deadlock, and has the shortest work time required for the rearrangement of the objects.

  In the object rearrangement planning process described above, the movement of each object from the starting position to the target position is a moving element, and each node representing the moving element is connected by a directed arrow that represents the order restriction of the moving element. By representing the directed graph, the constraint relationship of the moving order of the moving elements is arranged. Then, by applying graph theory to the directed graph, a semi-ordered set of nodes representing moving elements is created. Furthermore, by applying the genetic algorithm GA to this partial order set of nodes, the partial order set of nodes is converted into a full order set, and the optimal one of the full order sets of nodes is determined as an object rearrangement plan. Is done. According to the flow of processing of this object rearrangement plan, the object rearrangement plan can be created without depending on the experience of a skilled worker. Moreover, the optimality of the created object rearrangement plan is evaluated by the genetic algorithm GA, and the created object rearrangement plan does not cause deadlock and can complete the rearrangement of objects in the shortest time. It becomes. As described above, according to the present invention, it is possible to automatically create an optimal object rearrangement plan using a computing means such as a personal computer, and to reduce labor and time required for creating the object rearrangement plan. Can do. Introducing an embodiment that has achieved a reduction in the time taken to create an object rearrangement plan, the time required to create a rearrangement plan for 80 objects 20 in a stage with 100 blocks 13 is: Although it took several hours for the operator, it took several minutes for the object rearrangement planning apparatus 1 according to the present invention.

  The preferred embodiment of the present invention has been described above. However, the present invention is not limited to the above-described embodiment, and various design changes can be made as long as they are described in the claims. is there.

  For example, the form of the stage 10 on which the object 20 is rearranged is not limited to the above-described embodiment. The number of blocks 13 included in each lane L and the number of lanes L included in each area 12 can be changed as appropriate. Further, the stage 10 may be a flat surface or a solid body. When the stage 10 is a plane, a plurality of lanes L are arranged in the horizontal direction to form one area 12, and the traverser 11 reciprocates in the horizontal direction. When the stage 10 is a solid, a plurality of lanes L are arranged in the vertical direction to form one area 12, and the traverser 11 reciprocates in the vertical direction. Further, the number of areas 12 provided on the stage 10 and the number of traversers 11 are not limited to the above-described embodiment. For example, as shown in FIG. 15, three or more areas 12 may be provided on the stage 10 and two or more traversers 11 may be provided. The arrangement of the traverser 11 is not limited to between the areas 12 and may be provided at the end of the lane that is not between the areas 12. The traverser 11 is not limited as long as it can pass the object 20 between the lanes L, and the traverser 11 may be freely movable on the stage 10.

  Further, for example, in the above-described embodiment, the genetic elements of the genetic algorithm GA are numerical values [0, 1] and the crossover (recombination) method is uniformly crossed over. The crossover method is not limited to this.

  Note that the object rearrangement planning method and apparatus according to the present invention can be used to create a vehicle rearrangement plan in a railway vehicle outfitting factory, but the scope of use of the present invention is not limited to this. It can be widely used to create a rearrangement plan for rearranging a plurality of objects from a pre-movement layout to a post-movement layout under conditions where the movement of the objects is limited.

  The present invention automatically creates a rearrangement plan for rearranging a plurality of objects from a pre-movement layout to a post-movement layout under a condition in which the movement of the objects is restricted, using a computing means such as a computer. Can be used.

L Lane 1 Object rearrangement planning device 2 Computer 3 Input device 4 Output device 5 Storage device 61 Moving element extraction unit 62 Order constraint search unit 63 Directed graph creation unit 64 Loop extraction unit 65 Loop cancellation unit 66 Order determination unit 10 Stage 11 Traverser 12 Area 13 Block 20 Object

Claims (9)

There are multiple lanes with one or more blocks where objects can be placed one by one, and the object is allowed to move on the lane without overtaking other objects, and the object is allowed to move between the lanes. On stage
Relocate the objects on the stage from the pre-movement layout where each of the objects on the stage is located in the block corresponding to the starting position to the layout after the movement where each of the objects on the stage is located in the block corresponding to the target position An object rearrangement planning device for creating an object rearrangement plan for
A moving element extraction unit that extracts movement from the starting position to the target position for each object as a moving element based on the pre-movement layout and the post-movement layout;
An order constraint search unit that searches for a constraint on the order between the plurality of moving elements based on the layout before moving, the layout after moving, and the configuration of the stage;
An object rearrangement planning apparatus, comprising: a directed graph creating unit that creates a directed graph in which a plurality of extracted moving elements are nodes, and a restriction of an order between the plurality of moving elements is a directed arrow connecting the nodes.   The order constraint search unit may move the plurality of movements such that movement of one or more objects related to a block is in the order of movement starting from the block, movement passing through the block, and movement arriving at the block. The object rearrangement planning apparatus according to claim 1, wherein a restriction on an order between elements is searched.   The object rearrangement planning apparatus according to claim 1, further comprising a loop extraction unit that extracts a loop formed by the nodes and the directed arrow by a depth-first search method in the directed graph.   When the loop is extracted by the loop extraction unit, one node included in the loop is moved from the starting position of the object corresponding to the node to a retreat position for temporarily retreating the object. A loop canceling unit for canceling the loop by replacing the first node representing a moving element with a second node representing a moving element composed of movement of the object from the retracted position to the target position; Item 4. The object rearrangement planning apparatus according to Item 3. When the plurality of loops are extracted by the loop extraction unit, and the extracted plurality of loops includes a loop having two nodes.
The loop canceling unit gives priority to a node in which a path of the moving element is included in a part of a path of another moving element among the nodes constituting the loop having two nodes. Replace with 2 nodes,
The object rearrangement planning apparatus according to claim 4.   6. The apparatus according to claim 1, further comprising an order determination unit that determines an order of the plurality of moving elements based on a restriction on an order between the plurality of moving elements represented by the directed arrow. Object relocation planning device. The order determination unit includes:
A random number is given to each moving element, and the order of the plurality of moving elements is determined by determining a priority order based on the random number under the restriction of the order between the plurality of moving elements,
Calculate the work time required to reposition the plurality of objects in the determined order of the plurality of moving elements,
A numerical sequence of random numbers given to each moving element is a gene, the calculation process of the work time is an evaluation function, and a crossover is repeated using a genetic algorithm so that the gene with the shortest work time remains. The object rearrangement planning apparatus according to claim 6, wherein the object rearrangement planning apparatus searches for and determines an order of the plurality of moving elements. There are multiple lanes with one or more blocks where objects can be placed one by one, and the object is allowed to move on the lane without overtaking other objects, and the object is allowed to move between the lanes. On stage
Relocate the objects on the stage from the pre-movement layout where each of the objects on the stage is located in the block corresponding to the starting position to the layout after the movement where each of the objects on the stage is located in the block corresponding to the target position An object rearrangement planning method for performing
Extracting the movement from the starting position to the target position for each object as a moving element based on the pre-movement layout and the post-movement layout;
Obtaining a restriction on the order between the plurality of moving elements based on the layout before moving, the layout after moving, and the configuration of the stage;
An object rearrangement planning method including a step of creating a directed graph using the extracted plurality of moving elements as nodes and using a directed arrow connecting the nodes as a restriction on the order between the plurality of moving elements. There are multiple lanes with one or more blocks where objects can be placed one by one, and the object is allowed to move on the lane without overtaking other objects, and the object is allowed to move between the lanes. In the stage, from the pre-movement layout arranged in the block where each object on the stage corresponds to the starting position, to the post-movement layout arranged in the block where each object on the stage corresponds to the target position, the object on the stage An object rearrangement planning program executed on a computer for creating an object rearrangement plan for rearranging
A process of extracting movement from the starting position to the target position for each object as a moving element based on the pre-movement layout and the post-movement layout;
A process for obtaining an order restriction between the plurality of moving elements based on the layout before moving, the layout after moving, and the configuration of the stage;
An object rearrangement planning program for causing the computer to execute a process of creating a directed graph using the extracted plurality of moving elements as nodes, and a directional arrow connecting the order of the plurality of moving elements between the nodes. .

Images