Information support device, information support method, and information support program

Abstract

The ability for operators without advanced knowledge and understanding to easily schedule production plans. [Solution] The information support device comprises a first generation unit that generates prompts based on input information including first input information relating to the product production process and second input information relating to the product production site, and a second generation unit that inputs the generated prompts to a generation AI (Artificial Intelligence) and generates formalization information for representing the optimization problem of the product production plan as a mathematical model.

Description

Technical Field

[0001] The present disclosure relates to an information support device, an information support method, and an information support program.

Background Art

[0002] In plants and factories such as power generation, steel, chemical, and petroleum, for example, on-site personnel schedule production plans based on their knowledge. Various support technologies for scheduling production plans have been proposed (see, for example, Patent Document 1).

Prior Art Documents

Patent Documents

[0003]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0004] Scheduling the production plan of a plant or factory requires, for example, advanced knowledge and understanding of production facilities and the like, which is difficult for on-site personnel.

[0005] An embodiment of the present disclosure has been made in view of the above circumstances, and an object thereof is to provide an information support device, an information support method, and an information support program that can easily schedule a production plan even for an operator without advanced knowledge and understanding.

Means for Solving the Problems

[0006] An information support device according to an embodiment of the present disclosure is This is an information support device for optimizing product production plans. The information support device is... based on input information including first input information regarding a production process of a product and second input information regarding a production site of the product to specify the information to be treated as the objective function, constraints, and constants in the optimization problem of product production planning.The system comprises a first generation unit that generates prompts, and a second generation unit that inputs the generated prompts into a generation AI (Artificial Intelligence) and generates formalization information for representing the optimization problem of product production planning as a mathematical model. The first generation unit describes at least one of the following from the first input information—product information, product production process information, machine information used in the production process, and worker work information introduced into the production process—as prompts, treating them as constants in the optimization problem. It also describes at least one of the following from the second input information—production site operation rules, production site constraints, production site desired conditions, and delivery date conditions—as prompts, treating them as constraints or objective functions in the optimization problem. [Effects of the Invention]

[0007] According to one embodiment of the present disclosure, an information support device, an information support method, and an information support program are provided that enable even operators without advanced knowledge and understanding to easily schedule production plans. [Brief explanation of the drawing]

[0008] [Figure 1] This is a block diagram of an information support device according to one embodiment of the present disclosure. [Figure 2] A functional block diagram showing the functional configuration of an information support device according to one embodiment of this disclosure. [Figure 3] In one embodiment of this disclosure, this is a table showing the time taken for each process of each product produced in a factory. [Figure 4] This is a table showing the machines used in each process of each product produced in the factory, according to one embodiment of the present disclosure. [Figure 5] This figure shows a functional block of an information support device connected to a database in one embodiment of the present disclosure. [Figure 6] This figure shows an example of a prompt generated by the prompt generation unit of an information support device in one embodiment of the present disclosure. [Figure 7] This figure shows an example of an output result (explanation regarding the optimization problem) output from the output unit of an information support device in one embodiment of the present disclosure. [Figure 8] This figure shows an example of an output result (explanation regarding the optimization problem) output from the output unit of an information support device in one embodiment of the present disclosure. [Figure 9]A diagram showing an example of an output result (variable definition) output from an output unit of an information support device in one embodiment of the present disclosure. [Figure 10] A diagram showing an example of an output result (additional constraint) output from an output unit of an information support device in one embodiment of the present disclosure. [Figure 11] A diagram showing an example of an output result (formulation information) output from an output unit of an information support device in one embodiment of the present disclosure. [Figure 12] A diagram showing an example of an output result (modeling language) output from an output unit of an information support device in one embodiment of the present disclosure. [Figure 13] A diagram showing an example of an output result (modeling language) output from an output unit of an information support device in one embodiment of the present disclosure. [Figure 14] A flowchart showing a process executed by an information support device according to one embodiment of the present disclosure. [Figure 15] A diagram showing an array input by a data input unit of an information support device in a modified example of the present disclosure. [Figure 16] A diagram showing an array input by a data input unit of an information support device in a modified example of the present disclosure. [Figure 17] A diagram showing an example of an output result (explanation regarding an optimization problem) output from an output unit of an information support device in a modified example of the present disclosure. [Figure 18] A diagram showing an example of a prompt for a correction instruction input by an operator in a modified example of the present disclosure. [Figure 19] A diagram showing an example of an output result (formulation information after correction) output from an output unit of an information support device in a modified example of the present disclosure. [Figure 20] A diagram showing an example of an output result (modeling language) output from an output unit of an information support device in a modified example of the present disclosure. [Figure 21] A diagram showing an example of an output result (modeling language) output from an output unit of an information support device in a modified example of the present disclosure.

MODE FOR CARRYING OUT THE INVENTION

[0009] The following description relates to an information support apparatus, an information support method, and an information support program according to an embodiment of the present disclosure. For common or corresponding elements, the same or similar reference numerals are given, and overlapping descriptions are appropriately simplified or omitted as needed. In each figure, for convenience of explanation, the configuration is shown by appropriately enlarging, reducing, or omitting it. In order to improve the visibility of the drawings, elements in the figures may be shown by lines other than solid lines (such as dashed-dotted lines, broken lines, etc.) as needed.

[0010] Any reference to elements using terms such as "first", "second", etc. used in the present disclosure does not generally limit the quantity or order of those elements. These terms are used for convenience to distinguish between two or more elements. Therefore, references to the first and second elements do not mean, for example, that only two elements are adopted, that the first element must precede the second element, etc.

[0011] As used in the present disclosure, the term "processor" includes a single processor or a group of multiple processors. These processors may include a single-core processor, a multi-core processor, multiple processors within a single device, or multiple processors communicating with each other wired or wirelessly. Such processors may be distributed locally or remotely, and may operate collaboratively or in a distributed manner across a network of devices, the Internet, or the cloud, and can collectively execute tasks belonging to the "processor" described in the present disclosure. Similarly, the term "non-transitory computer-readable storage medium" includes a single storage medium or a group of multiple storage media. These storage media are distributed locally or remotely and can collectively store and provide access to instructions, data, or other information in a cooperative or distributed manner.

[0012] Figure 1 is a block diagram of an information support device 1 according to one embodiment of the present disclosure. As shown in Figure 1, the information support device 1 comprises a processor 10, memory 11, storage 12, communication device 13, input device 14, and output device 15. The information support device 1 is, for example, a PC (Personal Computer), a tablet terminal, or a smartphone.

[0013] The terms "device," "circuit," "unit," and "server" are interchangeable. The hardware configuration of the information support device 1 may include one or more of the devices shown in the figure, or it may be configured to omit some of the devices.

[0014] The processor 10 controls the entire information support device 1, which is a computer, by running an operating system, for example. The processor 10 may consist of a CPU (Central Processing Unit) that includes interfaces with peripheral devices, control devices, arithmetic units, registers, etc.

[0015] The processor 10 reads programs (program code), software modules, data, etc., from at least one of the storage 12 and the communication device 13 into the memory 11, and performs various processes according to these. The program used in this embodiment is one that causes the computer to execute at least a part of the operations described.

[0016] The memory 11 is a computer-readable recording medium and may consist of at least one of the following: ROM (Read Only Memory), EPROM (Erasable Programmable ROM), EEPROM (Electrically EPROM), RAM (Random Access Memory), or other suitable storage media.

[0017] Memory 11 may also be called a register, cache, main memory, etc. Memory 11 can store executable programs (program code), software modules, etc., for carrying out a method according to one embodiment of the present disclosure.

[0018] Storage 12 is a computer-readable recording medium. Storage 12 may consist of at least one of the following: a magneto-optical disk, a digital multipurpose disk, a removable disk, a hard disk drive, a smart card, a flash memory device (e.g., a card, stick, key drive), a magnetic stripe, a database, a server, or other suitable storage medium.

[0019] The communication device 13 is hardware (transceiver / receiver device) for communicating between computers via at least one of a wired network and a wireless network. The communication device 13 is also referred to as, for example, a network device, network controller, network card, or communication module. The communication device 13 may include a SIM card.

[0020] The communication device 13 connects, for example, the information support device 1 and the database 210 (see Figure 5) to enable mutual communication.

[0021] The input device 14 is an input device that accepts input from an external source (e.g., a keyboard, mouse, speech recognition software, etc.). The output device 15 is an output device that outputs to an external source (e.g., a display, speaker, etc.). The input device 14 and the output device 15 may be configured as an integrated unit (e.g., a touch panel).

[0022] Each device, such as the processor 10 and memory 11, is connected by a bus 16 for communicating information. The bus 16 may consist of a single bus or different buses may be used for communication between devices.

[0023] An information support device 1 according to one embodiment of this disclosure mathematically models and formulates the optimization problem of production planning for scheduling production plans in a factory that produces products. Production planning scheduling is the process of deciding which products to process, when, and with which equipment (machinery) in order to efficiently produce multiple products in a factory or plant. In this process, various objectives and constraints are taken into consideration, not just the order in which to process the products.

[0024] For example, when responding to fluctuations in product demand while equalizing the load on equipment and personnel, the goal is to minimize daily production volume variations in monthly and weekly plans, i.e., to level out production numbers. For example, when meeting deadlines or improving equipment operating efficiency, the goal is to shorten the time it takes to complete production of all products (makespan), i.e., to minimize the makespan. In these cases, it is necessary to satisfy various constraints such as different processes for each product, the inability to use equipment simultaneously, deadlines, and inventory.

[0025] For example, planners, such as on-site managers, need to solve combinatorial optimization problems (e.g., job shop scheduling problems, knapsack problems) that optimize the objective function while satisfying constraints. However, solving such optimization problems is difficult unless the planner possesses advanced knowledge and understanding of the factory (production equipment, etc.) and specialized knowledge of optimization. Typically, there are multiple products, processes, and types of equipment involved, resulting in a vast number of possible schedule combinations. Therefore, even experienced planners find it difficult to find the optimal solution manually.

[0026] In this embodiment, as described below, in a factory producing products A to C, given the processing time and machinery used for each process, the information support device 1 is used to formulate a job shop scheduling problem aimed at minimizing the makespan of all products. Even planners (hereinafter referred to as "operators") without advanced knowledge and understanding can easily schedule production plans.

[0027] Figure 2 is a block diagram showing the processes and machines M1 to M4 used for products A to C produced in the factory in this embodiment. Figure 3 is a table showing the time (processing time) required for each process of products A to C. Figure 4 is a table showing the machines used for each process of products A to C. The process performed by machine M1 is a cutting process. The process performed by machine M2 is a drilling process. The process performed by machine M3 is a polishing process. The process performed by machine M4 is an inspection process.

[0028] Products A through C are all produced in three processes S1 through S3. Product A is produced in the following order: cutting process by machine M1 (process S1), drilling process by machine M2 (process S2), and inspection process by machine M4 (process S3). Producing product A takes 4 hours in process S1, 3 hours in process S2, and 2 hours in process S3. Product B is produced in the following order: drilling process by machine M2 (process S1), polishing process by machine M3 (process S2), and inspection process by machine M4 (process S3). Producing product B takes 5 hours in process S1, 4 hours in process S2, and 3 hours in process S3. Product C is produced in the following order: cutting process by machine M1 (process S1), polishing process by machine M3 (process S2), and inspection process by machine M4 (process S3). Producing product C takes 6 hours in process S1, 5 hours in process S2, and 4 hours in process S3.

[0029] Figure 5 shows the functional blocks of the information support device 1 connected to the database 210. The database 210 may be provided by the information support device 1 or by an external device. The functional blocks shown in Figure 5 can be implemented by any combination of hardware and software. Furthermore, the means of implementing each functional block are not particularly limited. That is, each functional block may be implemented by a single physically connected device, or by two or more physically separated devices connected by wired or wireless connections.

[0030] The information support device 1 comprises a data input unit 110, a domain knowledge input unit 120, a prompt generation unit 130, a formula generation unit 140, an output unit 150, an input unit 160, and an interface unit 170.

[0031] The data input unit 110 is responsible for providing various values ​​necessary for production planning and scheduling. Specifically, the data input unit 110 inputs data D1 related to the product's production process. Data D1 is an example of first input information related to the product's production process. Data D1 can be input to the data input unit 110, for example, manually, or collected and input using measuring instruments such as sensors. Data D1 may also be acquired and input to the data input unit 110 in cooperation with an external system. Data D1 serves as basic data for generating and formulating prompts.

[0032] In addition, the database 210 stores data collected in conjunction with measuring instruments and external systems, for example, as data D1. The database 210 also stores newly created data D1 in response to a request from the data input unit 110. The database 210 can also store data D1 in other formats or intermediate calculation results. The data input unit 110 inputs the data D1 stored in the database 210 in a timely manner.

[0033] Data D1 includes, for example, at least one of the following: product information, production process information, and machine information. Product information may include, for example, product name, product type, and required quantity. Process information may include, for example, the order of processes and processing time for a product. For example, the process information for product A is "first, a cutting process for 4 hours, then, a drilling process for 3 hours, and finally, an inspection process for 2 hours." Machine information may include, for example, the type of machine, capacity, and operating hours. For example, the machine information for machine M1 is "a cutting machine for cutting metal, operating hours 8:00~17:00." Manual work by workers (such as screw tightening and assembly) may be introduced into the production process. In this case, data D1 may include, for example, information on the work performed by workers introduced into the production process (such as screw tightening and assembly). That is, data D1 may include, for example, at least one of product information, process information, machine information, and work information.

[0034] The operator can input data such as "product name," "process," and "processing time" as data D1 on a web screen, for example. The operator can also input various types of data D1 in bulk by uploading a file, for example, in CSV (Comma-Separated Values) format. The data input unit 110 may also collect real-time data from an MES (Manufacturing Execution System) or IoT sensors and input it automatically as data D1.

[0035] The domain knowledge input unit 120 inputs data D2 related to the product's production site. Data D2 is an example of second input information related to the product's production site. Data D2 includes, for example, at least one of the following: production site operation rules, production site constraints, production site desired conditions, and delivery date conditions. Data D1 and Data D2 may partially overlap in content.

[0036] The domain knowledge input unit 120 has the function of inputting site-specific rules and desired conditions necessary for formulating the optimization problem in natural language. This allows operators, for example, to input operational rules and desired conditions they are aware of on-site and have them reflected in the mathematical model, even without specialized knowledge of mathematical optimization.

[0037] The domain knowledge input unit 120 accepts natural language inputs such as "Machine M3 cannot perform multiple tasks simultaneously" and "Product B must be drilled before polishing" as operational rules and constraints. The domain knowledge input unit 120 also accepts natural language inputs such as "I want to finish as quickly as possible," "I want to ensure uniform machine operation," "I will always meet deadlines," and "I want to level out production volume daily" as desired conditions. Temporary constraints may also be entered, such as "Some machines are out of order and unavailable" as a temporary constraint.

[0038] For example, "Machine M3 cannot perform multiple tasks simultaneously" and "Product B must be drilled before polishing" are domain knowledge specific to the factory and equipment, and are natural language that corresponds to the constraints of the actual work site. In a mathematical model, the former can be translated into machine capacity constraints, and the latter into process sequence constraints.

[0039] For example, "I want to finish as quickly as possible" is natural language that corresponds to an objective function for minimizing the make-span. For example, "I want to make the machine operation uniform" and "I want to level out the production volume day by day" are natural language that corresponds to an objective function for leveling the load.

[0040] The domain knowledge input unit 120 can accept natural language, such as spoken language, as input. Even natural language, such as spoken language, can be converted into a mathematical model in a later functional block. For example, even tacit operational rules that are difficult to articulate concisely and clearly can be easily reflected in a mathematical model by the operator using the information support device 1.

[0041] The database 210 can store data D2, which may include, for example, rules for natural language usage. The domain knowledge input unit 120 can input data D2 stored in the database 210 in a timely manner.

[0042] The prompt generation unit 130 generates prompt P1 by combining data D1 (product information, process information, machine information, etc.) input by the data input unit 110 and data D2 (operational rules, constraints, desired conditions, etc.) input by the domain knowledge input unit 120. Prompt P1 is a prompt for formulating an optimization problem and is an instruction to tell the generating AI (Artificial Intelligence) what kind of optimization problem it wants to formulate.

[0043] The prompt generation unit 130 is an example of a first generation unit that generates a prompt P1 based on input information including data D1 and data D2. That is, the prompt generation unit 130 generates a prompt P1 that includes multiple elements for formulating the optimization problem based on the input information. These multiple elements include, for example, an overview of the optimization problem, specification of the objective function, specification of constraints, definition of variables, and specification of the output format.

[0044] The prompt generation unit 130 generates prompt P1, for example, using a template. In this embodiment, prompt P1 is written in Japanese. Prompt P1 may also be written in another language, such as English. For example, by creating input support rules or guidelines in advance for generating prompt P1 suitable for the generation AI, the accuracy and quality can be ensured regardless of who gives instructions to the generation AI through prompt P1.

[0045] The prompt generation unit 130 can, for example, automatically select an appropriate template from among multiple templates, or it can select a template specified by the operator. The prompt generation unit 130 embeds the data D1 input by the data input unit 110 and the data D2 input by the domain knowledge input unit 120 into the template. For example, the prompt generation unit 130 embeds information such as the number of products, the number of processes, processing time, and machine assignment, objectives such as "minimize makespan", constraints such as "machine M3 cannot perform multiple tasks simultaneously", and output format (natural language, modeling language (AMPL, Python, etc.)) into the template. The prompt generation unit 130 may also adjust the description of prompt P1 as appropriate so that it becomes an appropriate instruction for the generated AI.

[0046] The operator can, for example, perform an input operation to provide additional information (an example of first additional information) to the prompt generation unit 130. This additional information may include, for example, instructions to add or modify data D1 such as product information, process information, or machine information, or instructions to add or modify data D2 such as operational rules, constraints, or desired conditions.

[0047] The operator may create the entire prompt P1 manually, for example, without using a template.

[0048] The prompt generation unit 130 can generate prompt P1 using, for example, a trained model. In this case, it is desirable to use, for example, a trained model that has been fine-tuned for scheduling factory production plans. The operator can create prompt P1 by interacting with the trained model, either through the prompt generation unit 130 or without the prompt generation unit 130. In addition, the operator can create prompt P1 by performing input operations and providing the trained model with additional information (an example of first additional information), such as instructions, questions, correction requests, and confirmations. The prompt generation unit 130 can include the first additional information from the operator in prompt P1 (reflect it in prompt P1). The first additional information entered by the operator itself (questions, confirmations, etc.) can also be treated as prompt P1.

[0049] The prompt generation unit 130 can, for example, generate multiple prompts P1 using a trained model and propose them to the operator. The operator can then select the optimal prompt P1 from among the multiple options.

[0050] Figure 6 shows an example of a prompt P1 generated by the prompt generation unit 130. In the example in Figure 6, a prompt P1 is generated to formulate an optimization problem for scheduling factory production plans. According to this embodiment, even if, for example, the operator is unfamiliar with the technical terms, the prompt generation unit 130 automatically generates a prompt P1 suitable for the generated AI. As a result, the accuracy of the subsequent formulation process is improved.

[0051] The formula generation unit 140 is an example of a second generation unit. The formula generation unit 140, which is an example of a second generation unit, inputs the prompt P1 generated by the prompt generation unit 130 to the generation AI and generates formula information F1 for representing the optimization problem of the product production plan as a mathematical model. The generation AI is, for example, an LLM (Large Language Model). The formula information F1 includes, for example, an explanatory text (explanation of the optimization problem), an objective function, constraints, variable definitions, and code in a modeling language.

[0052] The output unit 150 outputs the formulation information F1 generated by the formulation generation unit 140 to a display device and presents it to the operator. The formulation information F1 is presented in a step-by-step format, for example, in the order of explanatory text, objective function, variable definitions, constraints, and output format. In this case, for example, explanatory text, mathematical formulas, and code examples can be presented at each step. The operator can check each element (objective function, constraints, variable definitions, etc.) in order, making it easy to give instructions for modifications or additions. For example, they can give instructions for partial modifications, such as adding a variable in step 3.

[0053] Figure 7 shows the output results from the output unit 150. In the example in Figure 7, as an explanation of the optimization problem, for example, it is shown that job shop scheduling is performed to minimize the make span of a product.

[0054] Figure 8 shows the output results from the output unit 150. In the example in Figure 8, the objective is shown to be minimizing the makespan. Constraints such as process sequence and machine non-overlap are also shown.

[0055] Figure 9 shows the output results from the output unit 150. In the example in Figure 9, the variable definition for Makespan is shown.

[0056] Figure 10 shows the output results from the output unit 150. In the example in Figure 10, the constraints added by the generation AI are shown.

[0057] Figure 11 shows the output results from the output unit 150. It shows a summary of the formulation information F1 presented at each step. In this embodiment, a job shop scheduling is formulated to optimize the start time of each process and the processing order of each machine, with the aim of minimizing the make span.

[0058] As shown in Figure 11, the objective function is a function that aims to minimize the makespan, and Minimize C maxThis is shown. As constraints for this optimization problem, for example, "the process for each product satisfies the sequential constraint," "processes do not overlap on the same machine," "the start time is non-negative," and "the make span is greater than or equal to the final process completion time for each product" are derived, and the equation exemplified in Figure 11 (process sequence, non-overlapping machines, make span, non-negative constraint) is defined. The formula generation unit 140 works in cooperation with the generation AI, and sometimes based on input from the operator, to define variables and adjust additional constraints based on the defined variables for the defined optimization problem. This results in the make span C exemplified in Figure 11. max (Total completion time), start time of each process for each product s i,k This provides an example of setting the constant M. The constant M is a constant applied in the Big M optimization method.

[0059] The operator can update the formalization information F1 while interacting with the generating AI via the formalization generation unit 140. Furthermore, the operator can update the formalization information F1 by performing input operations and providing instructions, questions, correction requests, confirmations, etc., to the generating AI via the formalization generation unit 140 as additional information (an example of second additional information). In other words, the formalization generation unit 140, which is an example of a second generation unit, updates the formalization information F1 in accordance with the second additional information provided by the operator.

[0060] The formula generation unit 140, which is an example of a second generation unit, can also generate alternative versions of the formulation information F1 in response to additional information such as instructions, questions, modification requests, and confirmations from the operator. Examples of alternative versions include proposals that slightly modify the objective, or those that incorporate changes or relaxations of constraints.

[0061] According to this embodiment, even if the operator is unfamiliar with technical terms, the prompt generation unit 130 generates a prompt P1 suitable for the generated AI. This improves the accuracy of the formulation process. Furthermore, even if the operator is unfamiliar with technical terms, the accuracy of the formulation information F1 can be easily improved through interactive operation with the formulation generation unit 140.

[0062] The generating AI accumulates history. Therefore, the formalized information F1 can, for example, take history into consideration. For example, the formalized information generation unit 140 may work in cooperation with the generating AI to refer to the dialogue history with the operator input from the input unit 160 and generate or improve the formalized information F1 that takes the dialogue history into consideration.

[0063] Generally, generative AI is designed to generate conversations with a reasonable degree of diversity. Taking advantage of this characteristic of generative AI, the formula generation unit 140 may, for example, provide the generative AI with the same prompt P1 multiple times to generate multiple patterns of formulation information F1. The formula generation unit 140 can integrate and evaluate these multiple patterns of formulation information F1, for example, in cooperation with the generative AI or based on operator instructions. The formula generation unit 140 may, for example, acquire the formulation information F1 with the highest evaluation as the optimal solution.

[0064] The output unit 150 may present the formulation information F1 all at once, rather than in a step-by-step format.

[0065] As described above, the output format of the formalization information F1 can be specified by prompt P1. The formalization information F1 can be translated not only into natural language but also into modeling language and programming language. Figures 12 and 13 show the output results from the output unit 150. Figure 12 shows the content formalized in natural language converted into a modeling language. Figure 13 shows the content formalized in natural language converted into a different modeling language than that in Figure 12.

[0066] The input unit 160 inputs input from the operator via a character input device such as a keyboard, mouse, touch panel, or scanner, or a speech recognition device, to the formula generation unit 140. The input unit 160 can, for example, input a prompt P1 created by the operator to the formula generation unit 140. The input unit 160 can also input additional information, such as a question or correction request from the operator, as prompt P1 to the formula generation unit 140.

[0067] The interface unit 170 outputs, for example, the formulation information F1 (in the form of natural language, modeling language, programming language, etc.) generated by the formulation generation unit 140 to an external solver or simulation system. The formulation information F1 generated by the information support device 1 can be utilized by an external solver or simulation system.

[0068] Figure 14 is a flowchart showing the processing performed by an information support device 1 according to one embodiment of the present disclosure. In the information support device 1, data D1 (product information, process information, machine information, etc.) and data D2 (operational rules, constraints, desired conditions, etc.) are input by the data input unit 110 and the domain knowledge input unit 120 (step S101). The prompt generation unit 130 generates a prompt P1 based on data D1 and data D2 (step S102). The prompt P1 is modified or newly added as appropriate, for example, in response to additional information from the operator. The formulation generation unit 140 inputs the prompt P1 generated by the prompt generation unit 130 into the generation AI and generates formulation information F1 for representing the optimization problem of the product production plan as a mathematical model (step S103). The formulation generation unit 140 updates the formulation information F1 as appropriate through dialogue with the operator. The generated formulation information F1 is output (step S104). For example, the output unit 150 outputs the formulation information F1 generated by the formulation generation unit 140 to a display device for display. For example, the interface unit 170 outputs the formulation information F1 generated by the formulation generation unit 140 to an external solver or simulation system.

[0069] The above is a description of exemplary embodiments of the present disclosure. Embodiments of the present disclosure are not limited to those described above, and various modifications are possible within the scope of the technical idea of ​​the present disclosure. For example, embodiments of the present application include combinations of embodiments explicitly shown in the specification or obvious embodiments as appropriate.

[0070] Modifications of this disclosure will be explained using Figures 15 to 21. In these modifications, there are 10 types of products (products A to J) produced in the factory. Monthly production quantities and delivery dates are determined for each of products A to J. The purpose of these modifications is to level out the daily production quantities.

[0071] Figures 15 and 16 show the arrays entered by the data input unit 110 in tabular format. The data input unit 110 inputs the array shown in Figure 15, which consists of the required production quantity, production start date, and delivery date (final production date) for each product. The data input unit 110 also inputs the array shown in Figure 16, which consists of the daily and product-specific production limits.

[0072] The domain knowledge input unit 120 inputs, for example, that there are 10 types of products, and that the monthly required production quantity for each product, the production start date for each product, and the production end date (which is the delivery date) are all determined.

[0073] The prompt generation unit 130 generates prompt P1 based on the data input from the data input unit 110 and the domain knowledge input unit 120.

[0074] The formula generation unit 140 generates formulation information F1 based on the prompt P1 generated by the prompt generation unit 130. Figure 17 shows an example of the formulation information F1 (explanation of the optimization problem) generated in this modified example. In the formulation information F1 exemplified in Figure 17, the objective is defined as "minimizing the total delay amount, which is the sum of the shortages at the time of delivery for each product, and maximizing on-time delivery." In other words, the objective is defined as scheduling a production plan that does not exceed delivery dates as much as possible. However, the operator's objective is to level out the production quantity each day. The formulation information F1 does not reflect the operator's objective.

[0075] The operator then instructs the desired modification. Figure 18 shows an example of a prompt P1 instructing the desired modification. The input unit 160 inputs the modification instruction prompt P1. The formula generation unit 140 modifies the formulation information F1 according to the instructions of the input prompt P1.

[0076] Figure 19 shows an example of the revised formulation information F1 (description of the optimization problem). In the revised formulation information F1, as shown in Figure 19, the objective is to equalize the production quantity each day. Note that specific examples of the formulated objective function and constraints are omitted here.

[0077] In this modified example, the output format of the formulation information F1 can also be specified by prompt P1. Figures 20 and 21 show the output results from the output unit 150. Figure 20 shows the content formulated in natural language in this modified example converted to a modeling language. Figure 21 shows the content formulated in natural language in this modified example converted to a different modeling language than that in Figure 20. [Explanation of symbols]

[0078] 1: Information support equipment 10: Processor 11: Memory 12: Storage 13: Communication device 14: Input device 15: Output device 110: Data Entry Section 120: Domain Knowledge Input Section 130: Prompt generation unit 140: Formula generator 150: Output section 160: Input section 170: Interface section 210: Database

Claims

1. An information support device for optimizing the production plan of a product, A first generation unit generates prompts for specifying information to be treated as the objective function, constraints, and constants in the optimization problem of the production plan of the product, based on input information including first input information regarding the production process of the product and second input information regarding the production site of the product. The system includes a second generation unit that inputs the generated prompt to a generation AI (Artificial Intelligence) and generates formalization information for representing the optimization problem of the production plan of the product as a mathematical model, The first generation unit is, At least one of the information included in the first input information—information about the product, information about the production process of the product, information about the machinery used in the production process, and information about the work of the workers introduced into the production process—is described in the prompt as information treated as a constant in the optimization problem. At least one of the following, included in the second input information, is described in the prompt as a constraint or objective function in the optimization problem: the operational rules of the production site, the constraints of the production site, the desired conditions of the production site, and the delivery date conditions. Information support equipment.

2. The aforementioned formulation information includes at least one of the following: a description of the optimization problem, an objective function, constraints, variable definitions, and code in a modeling language. The information support device according to claim 1.

3. The first generation unit includes the first additional information provided by the operator in the prompt. The information support device according to claim 1.

4. The first additional information includes at least one of the operator's instructions, questions, requests for correction, and confirmations. The information support device according to claim 3.

5. The second generation unit updates the formulation information in accordance with the second additional information provided by the operator. The information support device according to claim 1.

6. The second generation unit generates alternative formulations of the formulation information according to the second additional information. The information support device according to claim 5.

7. The second additional information includes at least one of the operator's instructions, questions, requests for corrections, and confirmations. The information support device according to claim 5.

8. The system includes an output unit that outputs the formulation information generated by the second generation unit, The information support device according to claim 1.

9. The first generation unit generates the prompt using a template or a trained model. The information support device according to claim 1.

10. A computer for optimizing the production plan of a product, Based on input information including first input information regarding the product's production process and second input information regarding the product's production site, prompts are generated to specify information to be treated as the objective function, constraints, and constants in the optimization problem of the product's production plan. The generated prompt is input to the generating AI, which generates formalized information to represent the optimization problem of the production plan for the said product as a mathematical model. In generating the aforementioned prompt, At least one of the information included in the first input information—information about the product, information about the production process of the product, information about the machinery used in the production process, and information about the work of the workers introduced into the production process—is described in the prompt as information treated as a constant in the optimization problem. At least one of the following, included in the second input information, is described in the prompt as a constraint or objective function in the optimization problem: the operational rules of the production site, the constraints of the production site, the desired conditions of the production site, and the delivery date conditions. Information support method.

11. A computer for optimizing the production plan of a product, Based on input information including first input information regarding the product's production process and second input information regarding the product's production site, prompts are generated to specify information to be treated as the objective function, constraints, and constants in the optimization problem of the product's production plan. The generated prompt is input to the generating AI, which generates formalized information to represent the optimization problem of the production plan for the said product as a mathematical model. In generating the aforementioned prompt, At least one of the information included in the first input information—information about the product, information about the production process of the product, information about the machinery used in the production process, and information about the work of the workers introduced into the production process—is described in the prompt as information treated as a constant in the optimization problem. At least one of the following, included in the second input information, is described in the prompt as a constraint or objective function in the optimization problem: the operational rules of the production site, the constraints of the production site, the desired conditions of the production site, and the delivery date conditions. Information support program.

Images