JP6853134B2 - Production support system and production support method - Google Patents

Description

The present invention relates to a production support system, a production support method, and a production support program.

Patent Document 1 discloses a planning method for reducing work-in-process inventory as a technique related to production planning in the manufacturing industry. That is, Patent Document 1 states, "Since the conventional production planning method and its device do not take into account the information on the alignment in the subsequent process, there is a problem that work-in-process inventory is generated at the end of the process. Therefore, it is possible to simultaneously supply a plurality of substrates required by a post-process unit in a face-to-face manner immediately before production in a post-process with a minimum in-process inventory. ”The planning method is described.

Further, Patent Document 2 describes a method of formulating a production plan using an empirical rule database of workers. That is, in Patent Document 2, "the first production lot to be an allocation candidate is evaluated based on the empirical rule database constructed by the selection result of the past allocation candidate for each evaluation item of the production condition information of the production lot. The evaluation value and the production lot as an allocation candidate were obtained from the second evaluation value, which is the prediction result of the production result obtained by performing the production plan simulation at each production line in order from the one with the highest first evaluation value. , The production lot of the allocation candidate and the production lot of the allocation candidate are ranked and output based on the third evaluation value which is the evaluation result of the allocation of the allocation candidate production line, and among the allocation candidates output with the above ranking. A production plan creation method of allocating a specific production lot to the production line, learning based on the specific production lot, and updating the empirical rule database is disclosed.

Japanese Unexamined Patent Publication No. 2004-145436 Japanese Unexamined Patent Publication No. 2012-194712

However, in many manufacturing industries, empirical factors such as considering the restart cost of machines (equipment) are also required when making a production plan. That is, it takes a considerable cost and time to start the machine used in each process, but such an element is because the optimum starting condition (starting number and starting time) is known from the experience of the operator. Is. However, since it is difficult to describe such elements with clear rules, these elements cannot be considered by the method of Patent Document 1.

On the other hand, Patent Document 2 formulates a plan using a database of worker's rules of thumb, but when there are a plurality of processes in production, there are constraints in each process, so these constraints are set. All must be satisfied. However, since the planning method of Patent Document 2 does not consider the constraint conditions, it cannot be applied to a production plan composed of a plurality of steps.

The present invention has been made based on such a background, and an object of the present invention is to provide a production support system, a production support method, and a production support program for appropriately producing a product produced by a plurality of processes. ..

One of the present inventions for solving the above problems is a production support system provided with a processor and a memory and supporting the production of a product produced by carrying out a plurality of steps, and the product and the production of the product. The process knowledge information storage unit that stores the process knowledge information, which is the relationship between the implementation time of each process that is executed for the purpose, the order information acquisition unit that acquires the information of the product to be produced, and the above-mentioned acquired unit. Based on the product information and the process knowledge information, an implementation process information generation unit that generates implementation process information, which is information on the implementation time of each process to be executed for the production of the acquired product, and an implementation process information generation unit set for each process. The constraint condition acquisition unit that acquires the constraint condition information, which is the information of the conditions that have been set, and the execution time of each process in the production of the acquired product that satisfies the set conditions based on the acquired constraint condition information. The feasible process information generation unit that generates the feasible process information that is information about, the learning unit that updates the process knowledge information based on the generated feasible process information, and the generated feasible process information. It includes an output unit that outputs information on the production process of the acquired product.

Issues, configurations and effects other than those described above will be clarified by the description of the following embodiments.

According to the present invention, it is possible to appropriately produce a product produced by a plurality of steps.

FIG. 1 is a diagram showing an example of the configuration of the production support system according to the present embodiment. FIG. 2 is a diagram illustrating an example of a hardware configuration of each information processing device constituting the production support system 1. FIG. 3 is a diagram illustrating an example of the functions included in the final / first process ideal plan creation device 104. FIG. 4 is a diagram illustrating an example of the functions included in the ideal planning apparatus 105. FIG. 5 is a diagram illustrating an example of the functions included in the plan candidate planning device 106. FIG. 6 is a diagram illustrating an example of the functions included in the plan candidate selection device 110. FIG. 7 is a diagram illustrating an example of a function provided in the worker terminal 109. FIG. 8 is a diagram showing an example of the order information DB 101. FIG. 9 is a diagram showing an example of the constraint condition DB 102. FIG. 10 is a diagram showing an example of the ideal plan DB 107. FIG. 11 is a diagram showing an example of the plan candidate DB 108. FIG. 12 is a diagram showing an example of the execution plan DB 111. FIG. 13 is a flowchart illustrating the production support process. FIG. 14 is a diagram illustrating an example of an ideal plan generation support screen. FIG. 15 is a diagram illustrating an example of an execution plan generation support screen. FIG. 16 is a flowchart illustrating the details of the calculation start process information generation process. FIG. 17 is a flowchart illustrating the details of the ideal plan creation process. FIG. 18 is a flowchart illustrating the details of the plan candidate creation process. FIG. 19 is a flowchart illustrating the details of the learning process.

A mode for carrying out the present invention will be described below with reference to the drawings.
(System configuration)
First, the system configuration will be described.

FIG. 1 is a diagram showing an example of the configuration of the production support system according to the present embodiment. As shown in the figure, the production support system 1 is an information processing system that supports the production of products produced by carrying out a plurality of processes, and includes a worker terminal 109 and an automatic planning system 103. It is configured. In the following, the process that is first performed in the production of a product (the most upstream process) is referred to as the first process, and the process that is finally performed in the production of the product (the most downstream process) is referred to as the final process.

The automatic planning system 103 includes each information processing device (computer) of the final / first process ideal plan creation device 104, the ideal plan planning device 105, the plan candidate planning device 106, and the plan candidate selection device 110. Further, the automatic planning system 103 includes databases of process knowledge DB 112, ideal plan DB 107, plan candidate DB 108, and execution plan DB 111 (DB: database; the same applies hereinafter).

The final / first process ideal planning device 104 and the ideal planning device 105 provide information on the execution time of each process to be performed for the production of products (for example, the order in which order lots are introduced into the equipment of each process, etc.). Information. Hereinafter, such information is referred to as an ideal plan).

The ideal plan, for example, when an order for the production of multiple products (lots) is received, determines which lot should be introduced and operated in which order (time zone) in each equipment used in each process. , Product production plan.

Specifically, the final / first process ideal plan creation device 104 specifies the process (hereinafter referred to as the final / first process or the calculation start process) for which the execution time is calculated first, and the specified final / first process and the specified process. , Information on the implementation time of the final / first process (hereinafter referred to as the final / first process ideal plan, based on the product order information (hereinafter referred to as order information; details will be described later) obtained from the worker terminal 109). Alternatively, it is called calculation start process information. Details will be described later). Then, the ideal plan planning device 105 generates an ideal plan based on the final / first process ideal plan generated by the final / first process ideal plan creation device 104.

The plan candidate planning device 106 generates a plurality of information (hereinafter, referred to as a plan candidate; details will be described later) regarding the execution time of each process that satisfies the specified conditions (conditions specified by the constraint condition information described later). ..

For example, in the above example, the plan candidate is operated by introducing which lot into each equipment in which order (time zone) after satisfying the operational and technical restrictions of each equipment used in each process. It is a practical production plan of the product that determines whether to make it.

The plan candidate selection device 110 identifies a predetermined number (for example, one) of plan candidates from the generated plurality of plan candidates. In the following, the specified plan candidate is referred to as an execution plan.

The process knowledge DB 112 (hereinafter, also referred to as a process knowledge information storage unit) is a knowledge database for creating an ideal plan. Specifically, the product and the implementation time of each process implemented for the production of the product. It memorizes the process knowledge information that is the relationship with. The process knowledge information is updated by machine learning or the like, as will be described later.

The order information DB 101 stores the order information necessary for calculating the ideal plan. The order information includes, for example, an order for a product to be produced and a past order history of the product. In addition, the order information includes, for example, information such as the type, quantity, and delivery date of the product requested by the customer.

The constraint condition DB 102 stores the constraint condition information which is the information of the condition set for each of the steps. The constraint condition information includes, for example, information such as the manufacturing capacity of the product in each process, the specifications of the product to be produced, the difference between the specifications of the product in each process and the specifications of the products in the processes before and after the process. In addition, the constraint condition information includes, for example, information regarding the delivery date of the product or the quantity of the product.

The ideal plan DB 107 stores the ideal plan for each process. Further, the plan candidate DB 108 stores the plan candidate for each process. The execution plan DB 111 stores the execution plan.

In this embodiment, each DB is composed of an information processing device such as a general PC (PC: Personal Computer) and general database software. These provide search and update functions.

(Hardware configuration)
Next, the hardware configuration of each information processing device constituting the production support system 1 will be described.

FIG. 2 is a diagram illustrating an example of a hardware configuration of each information processing device constituting the production support system 1. As shown in the figure, each information processing device includes a CPU 401, a memory 402, an interface 403, a network interface 404, a keyboard 405, an output device 406, a mouse 407, and a storage device 408.

The CPU 401 is, for example, a central processing unit, and is a processor capable of executing a program recorded in the memory 402 or previously transferred from the storage device 408 to the memory 402. The program may be introduced by a removable storage medium, if necessary. In this case, the device for reading the storage medium is connected to the interface 403. As a device for reading such a storage medium, for example, an optical disk (CD, DVD, Blu-ray disc, etc.) or a flash memory can be used. Further, if necessary, the program may be introduced into each information processing device by the network interface 404 via a communication medium (for example, a communication line or a carrier wave on the communication line).

The memory 402 temporarily records programs and data. The interface 403 is a device for connecting a storage medium or the like. The network interface 404 is a device for communicating with another information processing device.

The keyboard 405 accepts input of information. The output device 406 is a monitor or the like, and outputs various information. The mouse 407 changes the position of the pointer output to the output device 406, and also receives an input to a button provided on the mouse 407 to specify a position on the screen and give a predetermined processing instruction to the CPU 401. Tell. The mouse 407 can be replaced by, for example, a touch panel. In this case, the function related to the pointer may be omitted.

The storage device 408 stores a program or data. The storage device 408 is composed of, for example, a magnetic disk, a non-volatile memory (for example, a hard disk, SSD ()) or the like. In this case, the program or data stored in the storage device 408 is retained even when the power of the storage device 408 is turned on after the power is turned off, for example. An operating system (OS: Operating System) may be installed in the storage device 408 in advance. By doing so, the program can be specified by using the file name. Here, the OS is the basic software of the computer, and a generally widely known OS can be used. In this embodiment, it is assumed that the OS is installed.

(function)
Next, the functions provided by each information processing device will be described.

FIG. 3 is a diagram illustrating an example of the functions included in the final / first process ideal plan creation device 104. As shown in the figure, the final / first process ideal plan creation device 104 includes an order information acquisition unit 141 and a final / first process ideal plan creation unit 142. The order information acquisition unit 141 acquires information (order information) of the product to be produced. The final / first process ideal plan creation unit 142 generates the final / first process ideal plan (calculation start process information).

FIG. 4 is a diagram illustrating an example of the functions included in the ideal planning apparatus 105. As shown in the figure, the ideal plan planning device 105 includes an ideal plan generation unit 205, a learning method selection unit 201, a learning input processing unit 202, a learning unit 203, and a learning output processing unit 204.

The ideal plan generation unit 205 (hereinafter, also referred to as an execution process information generation unit) is a product obtained by the order information acquisition unit 141 based on the product information acquired by the order information acquisition unit 141 and the process knowledge information. Generates implementation process information (ideal plan), which is information on the implementation time of each process to be implemented for production.

Specifically, for example, the execution process information generation unit generates the execution process information by generating information for specifying the execution time of the process before the predetermined process based on the information regarding the predetermined process. Generate. Hereinafter, a method for generating information related to such a process is referred to as a forward method.

Further, for example, the execution process information generation unit generates information for specifying the execution time of the process to be executed after the predetermined process based on the information regarding the predetermined process. Hereinafter, a method for generating information related to such a process is referred to as a backward method.

Further, the execution process information calculation unit acquires an index for generating the implementation process information, and generates the implementation process information according to the acquired index. The index is, for example, information on the quantity of products or the production completion time (for example, delivery date) of products.

The learning unit 203 updates (learns) the process knowledge information based on the implementable process information generated by the enable process information generation unit described later. The learning unit 203 includes a learning model constructed by using a machine learning technique such as a neural network.

The learning unit 203 is a learning device that outputs information about a product production process, which is similar to a pattern that is conventionally performed, when input of information about a product and its implementation process is received. In the present embodiment, the learning unit 203 includes two types of learning devices. The first learner (hereinafter referred to as a learner in the reverse direction) is a backward-type learner that inputs a post-process and outputs a pre-process. On the other hand, the second learner (hereinafter referred to as a forward learner) is a forward-type learner that inputs the pre-process and outputs the post-process.

Specifically, the learning unit updates the process knowledge information by generating information that specifies the execution time of the process before the predetermined process based on the information about the predetermined process, or the learning unit updates the process knowledge information. Based on the information regarding the predetermined process, the process knowledge information is updated by generating information for specifying the execution time of the process to be performed after the predetermined process.

The learning method selection unit 201 accepts selection of a learning device in the forward direction or a learning device in the reverse direction. That is, the learning method selection unit 201 is a processing unit that selects whether to learn the product production process retroactively from the final process or to learn the product production process from the first process to the subsequent process.

The learning input processing unit 202 inputs information to the learning unit 203. The learning output processing unit 204 outputs the information learned (updated) by the learning unit 203.

FIG. 5 is a diagram illustrating an example of the functions included in the plan candidate planning device 106. As shown in the figure, the plan candidate planning device 106 includes a search input processing unit 301, a constraint condition input processing unit 302, a search unit 303, and a plan candidate output unit 304.

The search input processing unit 301 is a processing unit that reads the execution plan from the execution plan DB 111.

The constraint condition input processing unit 302 (hereinafter, also referred to as a constraint condition acquisition unit) acquires the constraint condition information, which is the information of the conditions set for each of the steps, from the constraint condition DB 102. For example, the constraint condition acquisition unit acquires information including information on usage conditions of equipment used in the process as the constraint condition information.

The search unit 303 (hereinafter, also referred to as an enableable process information generation unit) is a product obtained by the order information acquisition unit 141, which satisfies the set conditions based on the constraint condition information acquired by the constraint condition acquisition unit. Generate feasible process information (plan candidates), which is information on the implementation time of each process in manufacturing. That is, the search unit 303 is a processing unit that plans a plan candidate based on the constraint conditions of each process. The search unit 303 is configured to include a planning model constructed by using a mathematical planning technique such as constraint programming. The planning model can search for a plurality of planning candidates that satisfy the constraint conditions.

For example, the enablement process information generation unit generates a plurality of enablement process information satisfying the usage conditions of the equipment as the enablement process information.

Further, the enablement process information generation unit generates a plurality of the enablement process information, and calculates and calculates the similarity between each of the generated enablement process information and the generated implementation process information. Based on the similarities, the enablement process information that is most similar to the implementation process information is specified.

The plan candidate output unit 304 outputs the feasible process information (plan candidate) generated by the search unit 303 to the plan candidate DB 108.

FIG. 6 is a diagram illustrating an example of the functions included in the plan candidate selection device 110. As shown in the figure, the plan candidate selection device 110 includes a plan candidate selection unit 111. The plan candidate selection unit 111 identifies a predetermined number (for example, one) of the plan candidates among the plurality of plan candidates created by the search unit 303.

FIG. 7 is a diagram illustrating an example of a function provided in the worker terminal 109. The worker terminal 109 is an information processing device (computer) operated by a user (for example, an administrator of the production support system 1 or a person involved in the production of a product; the same applies hereinafter). As shown in the figure, the worker terminal 109 includes a user input unit 131 and a result output unit 132. The user input unit 131 accepts input of various information such as order information.

The result output unit 132 (hereinafter, also referred to as an output unit) outputs the feasible process information generated by the feasible process information generation unit as information on the production process of the acquired product by the order information acquisition unit 141. Specifically, the output unit outputs the enablement process information specified by the enablement process information generation unit as information on the production process of the acquired product by the order information acquisition unit 141.

(Database)
Next, the database provided in the production support system 1 will be described.

FIG. 8 is a diagram showing an example of the order information DB 101. As shown in the figure, the order information DB 101 has at least one or more records including each item of order number 501, delivery date 502, quantity 503, and product type 504. Information for identifying an order from a customer or the like (hereinafter referred to as an order number) is stored in the order number 501. The delivery date 502 stores a delivery time (for example, a delivery deadline, hereinafter referred to as a delivery date) for the order of the order number 501. For example, the delivery date 502 stores the number of days until shipment. The quantity 503 stores the number of products (hereinafter referred to as the order quantity) specified by the order of the order number 501. The product type 504 stores information (hereinafter, referred to as a product type) that identifies the type of the product specified by the order of the order number 501. In addition to the above items, the order information DB 101 may include other items necessary for the production of the product.

FIG. 9 is a diagram showing an example of the constraint condition DB 102. As shown in the figure, the constraint condition DB 102 has at least one or more records including each item of the process number 601, the decision variable 602, and the constraint condition 603. A number for identifying each process (hereinafter referred to as a process number) is stored in the process number 601. The coefficient of determination variable 602 stores information on the elements that determine the conditions or constraints imposed on the process of process number 601 (hereinafter referred to as the coefficient of determination variable), for example, the delivery date of the product, the order quantity of the product, or the order quantity of the product. Information indicating the product type is stored. The constraint condition 603 stores a condition based on the elements of the coefficient of determination variable 602 or the content of the constraint (hereinafter referred to as a constraint condition). For example, a condition related to the deadline of each process and a condition related to the equipment used in each process (machine, etc.) Information on the start-up cost and start-up time of the equipment) is stored. The coefficient of determination variable 602 and the constraint condition 603 may include information on the weather that affects the production of the product.

FIG. 10 is a diagram showing an example of the ideal plan DB 107. As shown in the figure, the ideal plan DB 107 has at least one or more records including each item of process number 702, production sequence 701, order number 703, delivery date 704, quantity 705, and product type 706. Of these, process number 702, order number 703, delivery date 704, quantity 705, and product type 706 are the process number 601 of the constraint condition DB 102, the order number 501 of the order information DB 101, the delivery date 502 of the order information DB 101, and the order information DB 101, respectively. It corresponds to the quantity 503 and the product type 504 of the order information DB 101. The manufacturing sequence 701 stores information (hereinafter, referred to as a manufacturing sequence number) that specifies the manufacturing sequence of the products (manufacturing order of each product with respect to the order) performed in the process of the step number 702. The production sequence number is a value indicating the priority of production.

FIG. 11 is a diagram showing an example of the plan candidate DB 108. As shown in the figure, the plan candidate DB 108 includes at least one or more records including each item of process number 802, plan candidate number 801, manufacturing sequence 803, order number 804, delivery date 805, quantity 806, and product type 807. Has. Of these, process number 802, manufacturing order 803, order number 804, delivery date 805, quantity 806, and product type 807 are the process number 601 of the constraint condition DB 102, the manufacturing order 701 of the ideal plan DB 107, and the order number 703 of the ideal plan DB 107, respectively. Corresponds to the delivery date 704 of the ideal plan DB 107, the quantity 705 of the ideal plan DB 107, and the product type 706 of the ideal plan DB 107.

Information for identifying a plan candidate (hereinafter referred to as a plan candidate number) is stored in the plan candidate number 801.

FIG. 12 is a diagram showing an example of the execution plan DB 111. The execution plan DB 111 includes at least one item including process number 903, adopted plan candidate number 901, distance from ideal plan 902, manufacturing order 904, order number 905, delivery date 906, quantity 907, and product type 908. Has the above records. Of these, process number 903, manufacturing order 904, order number 905, delivery date 906, quantity 907, and product type 908 are the process number 601 of the ideal plan DB 107, the manufacturing order 701 of the ideal plan DB 107, and the order number 501 of the order information DB 101, respectively. , Delivery date 502 of order information DB 101, quantity 503 of order information DB 101, and product type 504 of order information DB 101.

The adopted plan candidate number 901 stores information (hereinafter, referred to as an adoption plan candidate number) that identifies the adopted plan candidate (that is, the execution plan). The distance 902 from the ideal plan stores information indicating the similarity between the adopted plan candidate and the ideal plan. In the present embodiment, information indicating the distance between the two is stored. The method of calculating this distance will be described later.

The functions of each information processing device described above are executed by the hardware of each information processing device or by the CPU 401 (processor) of each information processing device by reading a program stored in the memory 402 or the storage device 408. It is realized by doing.

(processing)
Next, the processing performed in the production support system 1 will be described.

FIG. 13 is a flowchart illustrating a process (hereinafter, referred to as a production support process) of generating an implementation plan of a process performed for the production of a product and learning the process. The production support process is started, for example, when a predetermined input is made to the worker terminal 109.

As shown in the figure, first, the final / first process ideal plan creation device 104 acquires the order information input from the worker terminal 109 by communicating with the order information DB 101 (S1201). For example, the final / first process ideal planning apparatus 104 acquires order information of orders placed in a predetermined period (for example, orders for one week with a close delivery date).

Then, the worker terminal 109 acquires the order information acquired in S1201 from the final / first process ideal plan creation device 104, and outputs the acquired order information (S1202). As a result, the user is made to confirm the contents of the order information.

The above communication is performed by general communication, RPC (Remote Procedure Call), or the like. In the following as well, it is assumed that communication between each information processing device and each DB is performed by such a method.

Next, the final / first process ideal plan creation device 104 generates the final / first process ideal plan based on the order information acquired in S1201 (S1203). Details of this process (hereinafter referred to as calculation start process information generation process) will be described later.

Then, the ideal plan planning device 105 generates an ideal plan for each process based on the final / first process ideal plan calculated in S1203 (S1204). Details of this process (hereinafter referred to as ideal plan creation process) will be described later (FIG. 14). Further, the ideal plan planning device 105 registers the ideal plan generated in S1204 in the ideal plan DB 107 (S1205).

Further, the plan candidate planning device 106 acquires the constraint condition information from the constraint condition DB 102 (S1206). Then, the plan candidate planning device 106 generates at least one or more plan candidates based on the constraint condition information acquired in S1206 (S1207). Details of this process (hereinafter referred to as plan candidate planning process) will be described later.

に基づいて行われる。ここで、ＫＮ_i、及びＬＮ_iはそれぞれ、計画候補及び理想計画における製造順序ｉの注文の納期であり、ＫＳ_i、及びＬＳ_iはそれぞれ、計画候補及び理想計画における製造順序ｉの注文の数量であり、ＫＨ_i、及びＬＨ_iはそれぞれ、計画候補及び理想計画における製造順序ｉの注文の品種である。また、ＫＨ_iとＬＨ_iが同じ品種の場合はＫＨ_iとＬＨ_iとの差が０であり、ＫＨ_iとＬＨ_iが異なる品種の場合はＫＨ_iとＬＨ_iとの差が１であるとする。また、計画候補と理想計画との距離Ｄは、計画候補や理想計画における製造順序ｉの注文の距離Ｄ_iの総和である。なお、計画候補と理想計画との距離の算出に際しては、上記の各パラメータの代わりに、各パラメータに対して所定の重み付けの値を乗算した値（重み付けされた各パラメータ）を用いてもよい。 Next, the plan candidate selection device 110 calculates the similarity between each of the plan candidates created in S1207 and the ideal plan (S1208). In this embodiment, the distance is calculated in order to calculate the similarity. The calculation of this distance is, for example,

It is done based on. Here, KN _i and LN _i are the delivery dates of the orders of the production order i in the _{plan candidate and the ideal plan, respectively, and KS i} and LS _i are the quantities of the orders of the production order i in the plan candidate and the ideal plan, respectively. KH _i and LH _i are the ordered varieties of the production sequence i in the plan candidate and the ideal plan, respectively. Further, when KH _i and LH _i are the same variety, _{the difference between KH i} and LH _i is 0, and when KH _i and LH _i are different varieties, the difference between _{KH i} and LH _{i is 1.} To do. Further, the distance D between the plan candidate and the ideal plan is the sum of _{the distances D i} of the orders in the manufacturing sequence i in the plan candidate and the ideal plan. In calculating the distance between the plan candidate and the ideal plan, instead of each of the above parameters, a value obtained by multiplying each parameter by a predetermined weighting value (each weighted parameter) may be used.

The worker terminal 109 accepts from the user the selection of the plan candidate to be adopted from the plan candidates created in S1207 (S1209). Instead of the user selecting the plan candidate, the plan candidate selection device 110 or the like automatically selects the plan candidate having the shortest distance from the ideal plan (the plan candidate having the highest similarity to the ideal plan). You may.

The plan candidate selection device 110 registers the plan candidate selected in S1209 as an execution plan in the execution plan DB 111 (S1210). Further, the plan candidate selection device 110 outputs information regarding the execution plan selected in S1209 (S1211).

Subsequently, the ideal planning apparatus 105 learns the execution plan of each process based on the execution plan selected in S1209 (S1212). Details of this process (hereinafter referred to as learning process) will be described later. This completes the production support process (S1213).
Here, an example of the screen displayed in the production support process will be described.

FIG. 14 is a diagram illustrating an example of a screen (hereinafter, referred to as an ideal plan generation support screen) displayed when calculating an ideal plan among the screens output to the worker terminal 109. The ideal plan generation support screen is displayed, for example, when the production support process is started.

As shown in the figure, the ideal plan generation support screen 1000 has a worker number input box 1001, an order information display box 1002, a final / first process ideal plan selection box 1003, a final / first process ideal plan selection button 1004, and an ideal plan. Each column includes a creation index selection box 1005, an ideal plan creation index selection button 1006, a final / first process ideal plan creation button 1007, a final / first process ideal plan display box 1008, and a save button 1009.

The worker number input box 1001 receives input of an individual number such as an employee number from the user. The order information acquired in S1201 is output to the order information display box 1002.

The final / first process ideal plan selection box 1003 and the final / first process ideal plan selection button 1004 accept input of whether to select the final process or the first process as the calculation start process. The ideal plan creation index selection box 1005 and the ideal plan creation index selection button 1006 accept the selection of the index when generating the ideal plan. For example, it is an index of whether to create an ideal plan so that the delivery date of the product is quick, or whether to make an ideal plan so that the quantity of the product to be produced is large.

The final / first process ideal plan creation button 1007 accepts the start of creating an ideal plan based on the selected calculation start process. The final / first process ideal plan display box 1008 outputs the contents of the created ideal plan. The save button 1009 accepts the save of the created ideal plan.

FIG. 15 is a diagram illustrating an example of a screen displayed when creating an execution plan (hereinafter, referred to as an execution plan generation support screen) among the screens displayed on the worker terminal 109. The execution plan generation support screen is displayed, for example, when the creation of the ideal plan is started.

As shown in the figure, the execution plan generation support screen 1100 has a plan status display box 1101, an ideal plan display box 1102, a plan candidate display box 1103, a plan candidate selection box 1104, a plan candidate selection button 1105, and an execution plan display box 1106. , And each column of the save button 1107.

Information indicating the process for which the ideal plan has been created, the process for which the ideal plan is currently being created, and the process for which the ideal plan has not yet been created is output to the plan status display box 1101. Information on the process for which the ideal plan is currently being created is output to the ideal plan display box 1102. Information on the plan candidates of the process currently being created is output to the plan candidate display box 1103. The plan candidate selection box 1104 and the plan candidate selection button 1105 display the plan candidates and accept the selection of the plan candidates. Information on the plan candidate (that is, the execution plan) selected by the plan candidate selection box 1104 is output to the execution plan display box 1106. The save button 1107 accepts the saving of the execution plan information selected by the execution plan display box 1106. When the saving is accepted, the execution plan is stored in the execution plan DB 111.

Next, the details of each process in the production support process will be described.
<Calculation start process information generation process S1203>

FIG. 16 is a flowchart illustrating the details of the calculation start process information generation process. As shown in the figure, first, the worker terminal 109 selects a reference (whether the first process is the calculation start process or the final process is the calculation start process) when calculating the calculation start process information. Is received by the final / first process ideal plan selection box 1003, and the final / first process ideal plan creation device 104 acquires the contents of the accepted selection.

Further, the final / first process ideal plan creation device 104 determines an index for creating the ideal plan (S1302). Specifically, the worker terminal 109 accepts a selection from the user for the ideal plan creation index selection box 1005, and the final / first process ideal plan creation device 104 acquires the contents of the accepted selection. In the present embodiment, the above-mentioned indicators are an index for creating an ideal plan with an early delivery date and an index for creating an ideal plan with a large amount of production. You may use it.

The final / first process ideal plan creation device 104 generates calculation start process information based on the calculation start process acquired in S1301 and the index acquired in S1302 (S1303). Specifically, for example, when the index is the delivery date and the calculation start process is the final process, the final / first process ideal plan creation device 104 is a product based on an information processing program having a general calculation function and a sort function. At least one calculation start process information is generated so that the delivery date of the above can be shortened. When the index is the production amount and the calculation start process is the first process, the final / first process ideal plan creation device 104 is based on an information processing program having a general calculation function and a sort function, and the production amount of the product. At least one or more calculation start process information is generated so that

Then, the final / first process ideal plan creation device 104 outputs the calculation start process information generated in S1304 (S1304). This completes the calculation start process information generation process (S1305).

<Ideal plan creation process S1204>
Next, the ideal planning process will be described in detail.

FIG. 17 is a flowchart illustrating the details of the ideal plan creation process. As shown in the figure, the ideal planning device 105 selects the learning device by the learning method selection unit 201 (S1401). That is, when the final process is selected as the calculation start process by the calculation start process information generation process, the ideal planning apparatus 105 selects the learner in the opposite direction. Further, the ideal planning apparatus 105 selects a learning device in the forward direction when the first process is selected as the calculation start process by the calculation start process information generation process.

The ideal plan planning device 105 inputs the ideal plan (S1402). For example, the ideal plan planning device 105 inputs the final process ideal plan when the final process is selected as the calculation start process by the calculation start process information generation process. Further, when the first process is selected as the calculation start process by the calculation start process information generation process, the ideal plan planning device 105 inputs the first process ideal plan.

The ideal plan planning device 105 creates an ideal plan for the next process based on the ideal plan input in S1402 (S1403). For example, when the final process is selected as the calculation start process by the calculation start process information generation process, the ideal plan planning device 105 creates an ideal plan of the process before the final process. Further, when the first process is selected as the calculation start process by the calculation start process information generation process, the ideal plan planning device 105 creates the ideal plan after the final process.

Then, the ideal plan planning device 105 outputs the ideal plan created by S1403 (S1404). The processes of S1402 and S1403 are repeated until an ideal plan is created for all the steps.

For example, a neural network learner and an error backpropagation method are used to create an ideal plan. For more information on neural network learners and backpropagation, see, for example, Car-following behavior with instantaneous driver-vehicle reaction delay: A neural-network-based methodology, Transportation Research Part C 36 (2013) 339-351, Zheng et al. ], The method described in the above can be used.

The input made to the learning input processing unit 202 is, for example, information on the delivery date, quantity, and type of each product, and the information output by the learning output processing unit 203 is, for example, the production order of each product (in the equipment of each process). The order in which the order lots are introduced). For the production order that is a discrete value, the probability that each production order is selected is calculated, and the order is determined in descending order of the calculated probability. Further, the number of layers of the neural network learner is, for example, three, and the number of neurons in each layer is, for example, three in the first layer, ten in the second layer, and one in the third layer. Further, for the activation function between layers, for example, a sigmoid function is used.

For these calculations, generally available machine learning software can also be used. Further, in the above, specific numerical values, item names, function names and the like are given as information related to input / output, but other numerical values, item names, function names and the like may be used.

The ideal plan planning device 105 outputs the ideal plan related to each process generated by S1402 and S1403 to the ideal plan DB 107 (S1404). This completes the ideal plan creation process (S1405).

<Plan candidate creation process S1207>
Next, the plan candidate creation process will be described in detail.

FIG. 18 is a flowchart illustrating the details of the plan candidate creation process. As shown in the figure, the plan candidate planning device 106 reads the execution plan for each process from the execution plan DB 111 by the search input processing unit 301 (S1501).

Specifically, when the ideal plan is created by the learner in the reverse direction in the ideal plan creation process, the plan candidate planning device 106 uses the search input processing unit 301 to execute the execution plan of each process in the reverse direction from the first process. get. On the other hand, when the ideal plan is created by the learner in the forward direction in the ideal plan creation process, the plan candidate planning device 106 acquires the execution plan of each process in the reverse direction from the final process by the search input processing unit 301.

Further, the plan candidate planning device 106 reads the constraint condition information of each process from the constraint condition DB 102 by the constraint condition input processing unit 302 (S1502).

The plan candidate planning device 106 creates (searches) a plan candidate for each process by the search unit 303 (S1503).

This search method is based on, for example, constraint programming. In this case, the information input by the constraint condition input processing unit 302 is, for example, the delivery date, quantity, or product type of the product order, the production order of the product, and the constraint condition in each process. The output information is, for example, the manufacturing order of products in each process that satisfies the constraint conditions in each process. The search algorithm employs, for example, a search method called backtracking. Backtracking is one of the general solution search algorithms. When finding a certain solution, a procedure that may find the solution is tried, and the final solution is obtained by the tried procedure. If so, save the found solution, while if the trialed procedure finds that no solution is found, then go back to the previous state and try another step. It is a method called. In the backtracking search method, a plurality of solutions can be searched. This search is, for example,

It is done by calculating the distance between solutions using. Then, the solution that keeps a certain distance is adopted as a plan candidate. For example, when "Solution 2" is searched after "Solution 1" is searched, if the distance between "Solution 2" and "Solution 1" is larger than the preset distance X, "Solution 2" is adopted as a plan candidate. ..

For the above calculations, generally available mathematical planning software may be used. Further, in the present embodiment, specific numerical values, item names, and constraint conditions are given as input / output information, but other numerical values, item names, constraint conditions, and the like may be used.

Next, the plan candidate planning device 106 outputs the plan candidates for each process by the plan candidate output unit 304 (S1504). This completes the plan candidate creation process (S1505).

<Learning process S1212>
Next, the learning process will be described in detail.

FIG. 19 is a flowchart illustrating the details of the learning process. As shown in the figure, the ideal planning apparatus 105 first selects a learning device by the learning method selection unit 201 (S1601).

Specifically, the ideal planning apparatus 105 selects a learner in the opposite direction when the calculation start process is the final process in the calculation start process information generation process. Further, the ideal planning apparatus 105 selects a learning device in the forward direction when the calculation start process is the first process in the calculation start process information generation process.

The ideal plan planning device 105 reads the execution plan from the execution plan DB 111 by the learning input processing unit 202, and inputs each process in the read execution plan to the learning device (S1602). Specifically, the ideal planning apparatus 105 inputs the calculation start process selected by the calculation start process information generation process to the learner.

Then, the ideal planning device 105 outputs the execution plan of the next process (forward process or reverse process) based on the process of the execution plan input to the learning device by the learning output processing unit 204. (S1603). This process is repeated sequentially.

The ideal plan planning device 105 updates the process knowledge DB 112 based on the execution plan (execution plan that specifies the execution time of all processes) generated by S1602 and S1063. That is, the ideal plan planning device 105 learns the execution plan by the learning unit 203.

For input / output in this learning, for example, when a neural network learner is used, the delivery date, quantity, and product type in the calculation start process are input in the input layer, and the manufacturing order of the products in each other process is output in the output layer. Will be done. Further, the learning method updates the parameters of the neural network by using, for example, the backpropagation method.

At the time of learning the first learning device, the user inputs in advance predetermined information (execution plan history) regarding each process and the execution time of each process performed in the past in the input layer and the output layer. Based on this information, the parameters of the neural network may be updated using the backpropagation method.

This completes the learning process (S1605).

As described above, according to the production plan support system 1 of the present embodiment, based on the process knowledge information, the implementation process information which is the information regarding the implementation time of each process of the ordered product is generated, and each process is also generated. The feasible process information, which is the information about the implementation time of each process that satisfies the constraint condition, is generated by the mathematical planning method, the process knowledge information is updated based on the feasible process information, and the feasible process information is used as the product production plan. Therefore, a user such as a worker can appropriately produce a product produced by carrying out a plurality of steps. That is, according to the production planning support system 1 of the present embodiment, the process knowledge information is subjected to machine learning that is updated based on the feasible process information. Therefore, for example, a person who is skilled in the production of the product has performed the product. Production can be carried out based on the production plan. As a result, the user can perform production considering factors that require skillful experience and knowledge regarding product production, such as the restart cost of equipment used in each process, so that the product can be produced efficiently and at low cost. Can be produced at. Further, even if there are constraint conditions for each process such as the manufacturing capacity and quantity of the product, the user can perform appropriate production satisfying these constraint conditions.

For example, according to the production planning support system 1 of the present embodiment, a plurality of feasible process information satisfying the usage conditions of the equipment is generated as the feasible process information, so that the user can use various kinds of product production. Products can be produced based on an appropriate production plan corresponding to the equipment conditions.

Further, according to the production planning support system 1 of the present embodiment, the similarity between each of the plurality of generated feasible process information and the implementation process information is calculated, and among the calculated similarities, the most implementation process. Since the enablement process information similar to the information is specified and the specified enablement process information is output as the production process information of the product, the user can produce the most appropriate product in the production plan of the product that can be implemented. Production can be carried out based on the production plan that can be carried out.

Further, according to the production planning support system 1 of the present embodiment, the implementation process information is generated according to the index when the implementation process information is generated, so that the user can produce the product according to a predetermined target. ..

For example, according to the production plan support system 1 of the present embodiment, information on the quantity of products or the production completion time of products is acquired as an index when generating the implementation process information, and the implementation process information is obtained according to the acquired index. Since it is generated, the user can surely produce the product based on the production plan for producing the product in a predetermined quantity and the production plan for completing the production of the product on the predetermined delivery date.

Further, according to the production plan support system 1 of the present embodiment, the process knowledge information is updated by generating information for specifying the implementation time of the previous process based on the information about a certain process, or the process knowledge information is updated. Since learning is performed by updating the process knowledge information by generating information that specifies the execution time of the process to be executed after that based on the information about, the user has the highest number of production at the start of production of the product. It will be possible to produce products that will reduce the final in-process inventory.

As described above, an example applied to support the production planning of a product as an embodiment of the present invention has been described. The present invention is not limited to the above-described embodiment, and includes various modifications. For example, the above-described embodiment has been described in detail in order to explain the present invention in an easy-to-understand manner, and is not necessarily limited to the one including all the described configurations. Further, it is possible to replace a part of the configuration of one embodiment with the configuration of another embodiment, and it is also possible to add the configuration of another embodiment to the configuration of one embodiment. Further, it is possible to add / delete / replace a part of the configuration of each embodiment with another configuration.

For example, the present invention is not limited to the business of manufacturing a product (manufacturing industry), but also various other types of business (for example, decision-making business performed through a plurality of processes) in which the business is performed by a plurality of processes. It can be applied to various businesses.

Further, each of the above configurations, functions, processing units, processing means and the like may be realized by hardware by designing a part or all of them by, for example, an integrated circuit. Further, each of the above configurations, functions, and the like may be realized by software by the processor interpreting and executing a program that realizes each function. Information such as programs, tables, and files that realize each function can be placed in a memory, a hard disk, a recording device such as an SSD (Solid State Drive), or a recording medium such as an IC card, an SD card, or a DVD.

In addition, the control lines and information lines indicate those that are considered necessary for explanation, and do not necessarily indicate all the control lines and information lines in the product. In practice, it can be considered that almost all configurations are interconnected.

1 Production support system, 109 Worker terminal, 132 Result output unit, 112 Process knowledge DB, 141 Order information acquisition unit, 203 Learning unit, 205 Ideal plan generation unit, 302 Constraint input processing unit, 303 Search unit

Claims (14)

A production support system that is equipped with a processor and memory and supports the production of products produced by performing multiple processes.
An ideal plan for creating an ideal plan, which is information on the implementation time of each process, in the order of the first process, based on the process knowledge information which is the relationship between the product and the implementation time of each process and a predetermined index. With the creation department
Based on the constraints or conditions imposed on each process, a plan to generate a plurality of plan candidates which are information on the execution time of each process in the order of the processes in the direction opposite to the order of the first process. Candidate planning department and
A plan candidate selection unit that compares the similarity between the ideal plan and each of the plurality of plan candidates and selects the plan candidate having the highest similarity with the ideal plan as an execution plan.
A learning unit equipped with two different types of learning devices, each of which uses its own machine learning technology.
A learning method selection unit for selecting one of the two different types of learning devices according to each step is provided.
The learning unit updates the process knowledge information according to the selected execution plan based on each machine learning technique of the selected type of learning device.
Production support system. The ideal plan creation unit creates an ideal plan for each process in the order of the first process for creating the ideal plan for the subsequent process based on the ideal plan for the previous process.
The plan candidate planning unit creates a plurality of plan candidates for each process in an order opposite to the first process order for creating the plan candidates for the previous process based on the plan candidates for the subsequent process.
The production support system according to claim 1, wherein the production support system is characterized in that. The ideal plan creation unit creates an ideal plan for each process in the order of the first process for creating the ideal plan for the previous process based on the ideal plan for the subsequent process.
The plan candidate planning unit creates a plurality of plan candidates for each process in an order opposite to the order of the first process, which creates the plan candidates for the subsequent process based on the plan candidates for the previous process.
The production support system according to claim 1, wherein the production support system is characterized in that. The production support system according to claim 1, wherein the plan candidate selection unit selects a plan candidate having the highest degree of similarity to the ideal plan as the execution plan based on input information from the outside. The production support system according to claim 1, wherein the ideal plan creation unit further creates the ideal plan based on order information which is information on the product to be produced. The production support system according to claim 1, further comprising an ideal plan DB for storing the ideal plan for each process and a plan candidate DB for storing the plan candidate for each process. The production support system according to claim 1, wherein the predetermined index is information on the production amount of the product or the production completion time of the product. It is a production support method that supports the production of products produced by carrying out multiple processes.
An information processing device equipped with a processor and memory
An ideal plan for creating an ideal plan, which is information on the implementation time of each process, in the order of the first process, based on the process knowledge information which is the relationship between the product and the implementation time of each process and a predetermined index. Creation process and
Based on the constraints or conditions imposed on each process, a plan to generate a plurality of plan candidates which are information on the execution time of each process in the order of the processes in the direction opposite to the order of the first process. Candidate planning process and
A plan candidate selection process that compares the similarity between the ideal plan and each of the plurality of plan candidates and selects the plan candidate having the highest similarity with the ideal plan as an execution plan.
A process of selecting one of two different types of learners, each of which uses its own machine learning technique, according to each of the steps.
A learning process that updates the process knowledge information according to the selected execution plan based on the machine learning technology of each of the selected types of learners.
A production support method characterized by executing. In the ideal plan creation process, the ideal plan for each process is created in the order of the first process for creating the ideal plan for the subsequent process based on the ideal plan for the previous process.
In the plan candidate planning process, a plurality of plan candidates for each process are created in an order opposite to the order of the first process for creating the plan candidates for the previous process based on the plan candidates for the subsequent process.
The production support method according to claim 8, wherein the production support method is characterized in that. In the ideal plan creation process, the ideal plan for each process is created in the order of the first process for creating the ideal plan for the previous process based on the ideal plan for the subsequent process.
In the plan candidate planning process, a plurality of plan candidates for each process are created in an order opposite to the order of the first process for creating the plan candidates for the subsequent process based on the plan candidates for the previous process.
The production support method according to claim 8, wherein the production support method is characterized in that. The production support method according to claim 8, wherein the plan candidate selection process selects a plan candidate having the highest degree of similarity to the ideal plan as the execution plan based on input information from the outside. The production support according to claim 8, wherein the information processing apparatus executes a final / first process ideal plan creation process for creating the ideal plan based on order information which is information on the product to be produced. Method. The production support according to claim 8, wherein the information processing apparatus includes an ideal plan DB that stores the ideal plan for each process and a plan candidate DB that stores the plan candidate for each process. Method. The production support method according to claim 8, wherein the predetermined index is information on the production amount of the product or the production completion time of the product.

Images