Optimization support device, optimization support method, and optimization support program for product life cycle management

Abstract

The present invention provides a technology that contributes to the optimization of product life cycle management using a component chart and a process chart.　This optimization support device for product life cycle management comprises at least one control unit. The control unit receives at least one piece of uncorrected data selected from a group consisting of at least one component chart and at least one process chart, receives a correction instruction for the uncorrected data, inputs the content of an input set that includes the uncorrected data and the correction instruction to a machine learning model, and outputs a calculation result obtained by the machine learning model having corrected the uncorrected data according to the correction instruction.

Description

Optimization support apparatus, optimization support method and optimization support program for product lifecycle management

[0001] The present invention relates to an optimization support apparatus, an optimization support method and an optimization support program for product lifecycle management.

[0002] In the manufacturing industry, systems for Product Lifecycle Management (PLM) that centrally manage technical information throughout the product lifecycle are known. For example, Patent Document 1 discloses a technology that creates a Bill of Materials (BOM) representing components included in a product and a Bill of Process (BOP) representing processes included in a routing, and performs integrated management by associating items included in the bill of materials with items included in the bill of processes.

[0003] International Publication No. 2018 / 088470

[0004] A technician of a company creates a bill of materials and a bill of processes in accordance with the creation standards established by the company, but the created bill of materials and bill of processes may not be optimal.

[0005] Therefore, an object of the present invention is to provide a technology that contributes to optimization of product lifecycle management using a bill of materials and a bill of processes.

[0006] The optimization support apparatus for product lifecycle management according to the present invention is an apparatus that supports optimization of product lifecycle management, and includes at least one control unit. The control unit receives at least one uncorrected data selected from the group consisting of at least one bill of materials and at least one bill of processes, receives a correction instruction for the uncorrected data, inputs the content of an input set including the uncorrected data and the correction instruction to a machine learning model, and outputs the calculation result obtained by the machine learning model correcting the uncorrected data in accordance with the correction instruction.

[0007] The product lifecycle management optimization support method according to the present invention is a method for supporting the optimization of product lifecycle management, comprising: receiving at least one pre-correction data selected from a group consisting of at least one bill of materials or at least one process chart; receiving a correction instruction for the pre-correction data; inputting the contents of an input set including the pre-correction data and the correction instruction into a machine learning model; and outputting a calculation result in which the machine learning model has corrected the pre-correction data according to the correction instruction.

[0008] The product lifecycle management optimization support program according to the present invention causes at least one processor to execute the optimization support method. The program may be stored in a computer-readable storage medium. The storage medium is a non-temporary, tangible medium.

[0009] According to the present invention, the output contains the modified parts list or process chart according to the user's modification instructions. By using the improved chart shown in these modifications, it is possible to contribute to the optimization of product lifecycle management.

[0010] Figure 1 is a schematic diagram of a product lifecycle management optimization support device according to an embodiment. Figure 2 is a diagram showing examples of design BOM, manufacturing BOM, and BOP stored in the database of Figure 1. Figure 3 is a block diagram of the server of Figure 1. Figure 4 is a block diagram of the information processing terminal of Figure 1. Figure 5 is a data flow diagram of the server processing of Figure 1. Figure 6 is a flowchart of the server processing of Figure 1. Figure 7 is a first example prompt screen of the information processing terminal of Figure 1. Figure 8 is a first example prompt screen of the information processing terminal of Figure 1. Figure 9 is a second example prompt screen of the information processing terminal of Figure 1. Figure 10 is a second example prompt screen of the information processing terminal of Figure 1. Figure 11 is a third example prompt screen of the information processing terminal of Figure 1. Figure 12 is a third example prompt screen of the information processing terminal of Figure 1. Figure 13 is a diagram illustrating an example of resolving BOM duplication using the machine learning model of Figure 1. Figure 14 is a diagram illustrating an example of performing overall optimization of multiple BOMs using the machine learning model in Figure 1. Figure 15 is a data flow diagram of a modified version of the server processing in Figure 1. Figure 16 is the interface screen of the display unit of the information processing terminal in Figure 1.

[0011] The embodiments will be described below with reference to the drawings.

[0012] Figure 1 is a schematic diagram of a product lifecycle management optimization support device 1 according to an embodiment. As shown in Figure 1, the optimization support device 10 is a system introduced to a company 1 belonging to the manufacturing industry to support the optimization of product lifecycle management. Company 1 has a head office 2 and multiple factories 3. The head office 2 and each factory 3 may be located in different countries or may be located within the same country. The optimization support device 10 includes a database DB, a server 11, a machine learning model 12, an information processing terminal 13, etc. Each database DB, server 11, machine learning model 12, and information processing terminal 13 are connected to a communication network N. The communication network N is, for example, the internet, but may also be an intranet, etc.

[0013] Each database DB corresponds to a factory 3 and stores the bill of materials and process schedule, as described later in Figure 2. Hereafter, the bill of materials will be referred to as BOM (Bill of Materials) and the process schedule as BOP (Bill of Process). The database DB also stores creation standards that define the standard methods for creating BOMs and BOPs. Server 11 can connect to each database DB via the communication network N. Server 11 can use the machine learning model 12 via the communication network N. Information processing terminal 13 can connect to Server 11 via the communication network N.

[0014] The machine learning model 12 may be located within the server 11, or it may be connected to the server 11 via a communication path different from the communication network N. The BOM, BOP, and creation standards may also be stored in other locations, as long as they are accessible from the server 11 and the information processing terminal 13.

[0015] Figure 2 is a diagram showing examples of design BOM, manufacturing BOM, and BOP stored in the database DB of Figure 1. In Figure 2, design BOM 41, manufacturing BOM, and BOP are shown as examples of BOMs and BOPs stored in the database DB. Design BOM 41 is created and used by engineers in the design department of Enterprise 1. Manufacturing BOM 42 and BOP 43 are created and used by engineers in the manufacturing department of Enterprise 1.

[0016] A Bill of Materials (BOM) is a list that shows the parts included in a product. A BOM may include attribute information such as the item number assigned to each part, part name, part description, size, properties, and performance, as well as the number of parts, unit of parts, relationship information between parts, information about the product configuration, variations, options, substitute information, version information, history information, and supplier information. BOMs can be created for different purposes, such as design, manufacturing, purchasing, quality control, after-sales service, logistics, and sales.

[0017] A BOP (Build-in-Place) is a table that represents the processes included in a construction sequence. A BOP may include process numbers assigned to each process, attribute information such as process descriptions, properties, and performance, work procedures, routing, work area information, relationship information between processes, information on equipment or materials associated with the process, variations, options, alternative processes, alternative work information, version information, history information, etc.

[0018] BOM 41, 42, and BOP 43 are written in structured text or natural language according to the creation standards established by Company 1. Specifically, the creation standards define how each item to be included in the BOM and BOP should be created, and what considerations should be made regarding the relationships between items and between the BOM and BOP. The creation standards are described below.

[0019] [BOM Creation Standards] "Definition of Parts": A part (item) is a physical unit. Identical items are registered as the same part in the BOM. Whether a part is the same as an existing part or a new part is determined based on Company 1's creation standards. The creation standards reflect Company 1's approach to requirements specifications for parts. For example, some companies specify different parts for different regions of milk production, while others manage them as the same part. When the requirements for milk are high, parts are managed in much smaller increments.

[0020] "Design of Part Numbers": The creation standards may include the number of digits in the part number, the meaning of each digit, the characters that can be used, rules, format, etc. The naming method for part numbers is determined by the company itself. The number of digits in the part number may be fixed or a maximum number of digits may be set and used accordingly.

[0021] "Meaningful Part Numbers": Part numbers that have a specific meaning are called meaningful part numbers. Typically, they are determined by combining information such as the manufacturing number, drawing number, product category, customer number, supplier number, serial number, version, and options, in a way that is beneficial to the company's operations. If existing rules are insufficient for numbering, measures may be taken such as assigning a category equivalent to "other," changing the rules for each digit, or increasing the number of digits. For example, if a company that manufactures metal products introduces a plastic material in a new product, it may add a category corresponding to the material to the product category, or change the rules to add a digit representing the material. With meaningful part numbers, there may be inconsistencies in numbering before and after revisions to the creation standard. In such cases, it is important to record which version of the creation standard each part number was designed based on.

[0022] "Meaningless Part Number": Meaningless part numbers are assigned using sequential numbers or random symbols, with only the number of digits and format defined as a minimum requirement. When using meaningless part numbers, no evaluation or modification of the part number is performed; if the number is unique, it is assigned and used immediately.

[0023] "Conversion to other purpose-specific BOMs": When converting a BOM for one purpose to a BOM for another purpose, the BOM is determined by considering the creation standards common to all departments or locations, as well as the creation standards of the source department or location and the creation standards of the destination department or location. For example, intermediate products or phantoms may be introduced in the manufacturing site. In that case, as shown in Figure 2, the intermediate product is newly added to the manufacturing BOM 42 as an item that was not in the design BOM 41.

[0024] "Design of Intermediate Products / Phantoms": In manufacturing, inventory management may involve items not included in the design BOM 41, such as assembled parts. In such cases, inventory management can be optimized by registering them as intermediate products in the manufacturing BOM 42. Items that are not actually managed as inventory units but are managed as a group to gain operational benefits may be registered as phantoms, for example, leftover parts from a temporarily generated parts kit. Whether to classify something as a phantom or an intermediate product is determined according to predetermined rules.

[0025] "Structural Design (Data Structure)": The BOM creation standards include rules regarding data structure, such as parent-child relationships between parts, part usage, and assembly units. Depending on the business objectives, various structures such as summary type, structure type, matrix type, and single-level type are used. A summary type BOM lists all the parts that make up a product. The names and attributes of each part in the BOM allow verification that there are enough parts to make the product. A structure type BOM defines the parent-child relationships between parts in a hierarchical manner. Rules such as the maximum number of levels and the granularity per level are determined for each company. A matrix type BOM represents the parts, options, etc. required for each configuration when there are multiple product configurations (such as color variations) on a matrix. The variables used for the matrix axes are selected according to rules determined for each company. A single-level type BOM defines the direct parent-child relationships between parts. Each company defines cases where parent-child relationships should not be set, and cases where it is recommended to include phantom or intermediate parts.

[0026] "Variations and Options": Variations and options refer to different specifications of the same product, or products with the same specifications but using different parts. Each company has its own terminology, such as variations, options, or alternatives. Whether to design a new part or a part as a variation or option of an existing part is decided according to each company's established rules and takes into account business considerations.

[0027] "Versions and Revisions": Versions are assigned each time the creation standard changes and are used to link data before and after the change. Version structuring and numbering are carried out according to rules determined by each company. Revisions are assigned for every finalized change, including those not used in business operations, and are used to link data before and after the change. Revision structuring and numbering are carried out according to rules determined by each company.

[0028] "Effective Date and Effective Quantity": When making changes to the Bill of Materials (BOM), the effective date and effective quantity are set based on a date or production lot, etc., in order to manage when the changes become effective.

[0029] [Standards for Creating a Business Plan (BOP)] "Definition of Process": Processes include various types such as material processing, parts assembly, surface treatment, painting, inspection, and packaging. The granularity of each process depends on the purpose of process control, required precision, etc., and is designed by referring to the company's rules.

[0030] "Definition of Routing Sequence": A combination of multiple processes is called a routing sequence. A routing sequence includes a series of processes necessary to produce a complete product, such as a product assembly line or an electronic circuit board mounting line. A combination of multiple routing sequences may be managed as a network or chain. The granularity of each routing sequence depends on the management purpose and required precision, and is designed by referring to the rules within the company.

[0031] "Designing process numbers": Designing process numbers is similar to designing item numbers.

[0032] "Conversion to other purpose-specific BOPs": Conversion to other purpose-specific BOPs is the same as conversion to other purpose-specific BOMs.

[0033] "Work Design": A process includes a series of specific tasks. The granularity of a single task depends on the management objectives, required precision, etc., and is designed in reference to the company's rules. Tasks may be further divided into smaller subtasks as needed.

[0034] "Designing Standard Capacity": The capacity of a process is calculated based on factors such as the process speed and utilization rate. In the stage where actual measurements are unavailable, the standard capacity is designed by referring to past knowledge and similar cases.

[0035] "Designing Standard Resources": Each process utilizes resources such as equipment, personnel, tools, and materials. Defining the resources required for a process as standard resources helps with resource allocation and other management purposes. Standard resources are determined by the characteristics and nature of the process, but also depend on the management objectives and required precision, and are designed by referring to the company's rules.

[0036] "Designing Phantom Processes": Phantom processes, while uncommon, can be designed for similar purposes to phantom BOMs. Registering processes as phantoms—processes not actually required in production but needed for on-site convenience—can offer management advantages. For example, processes that do not affect time, capacity, and resources, such as temporary fixing or treatment processes, can be managed as phantom processes. The criteria for designating a process as a phantom process are designed based on the company's rules.

[0037] "Structural Design": BOP is generally managed using the structure type described in the structural design section of BOM. When there are many options, variations, etc., representing them in a matrix type makes the relationship between processes and part assemblies easier to visualize.

[0038] "Variation and Option Design": Variations or options may be set for tasks included in a process, associated work areas, resources, etc. Variations in a process are determined according to rules established by each company, taking into consideration business needs.

[0039] Figure 3 is a block diagram of the server 11 shown in Figure 1. As shown in Figure 3, the server 11 comprises a control unit 21, a storage unit 22, and a communication unit 23. Specifically, the server 11 comprises a processor, system memory, storage memory, and a communication interface. The processor includes a CPU. The system memory includes volatile memory such as RAM. The storage memory includes non-volatile memory such as a hard disk and flash memory. The storage memory stores programs. The communication interface is an interface for connecting to a communication network N, and includes, for example, an Ethernet interface.

[0040] The control unit 21 is implemented by a processor that executes a program read from storage memory into system memory. The storage unit 22 is implemented by storage memory. The communication unit 23 is implemented by a communication interface. The functions disclosed in this embodiment can be executed using processing circuits, including general-purpose processors, dedicated processors, integrated circuits, ASICs, conventional circuits, and / or combinations thereof, configured to perform the functions.

[0041] Figure 4 is a block diagram of the information processing terminal 13 shown in Figure 1. As shown in Figure 4, the information processing terminal 13 comprises a control unit 31, a storage unit 32, a display unit 33, an input unit 34, and a communication unit 35. Specifically, the information processing terminal 13 comprises a processor, system memory, storage memory, a display, an input interface, and a communication interface. The processor includes a CPU. The system memory may include volatile memory such as RAM. The storage memory includes non-volatile memory such as a hard disk and flash memory. The storage memory stores programs. The display is capable of displaying a screen, such as a liquid crystal display. The input interface is capable of user input operation, such as a keyboard, mouse, or touch panel. The communication interface is an interface for connecting to a communication network N, and includes, for example, an Ethernet interface.

[0042] The control unit 31 is implemented by a processor that executes a program read from a storage memory to a system memory. The storage unit 32 is implemented by a storage memory. The display unit 33 is implemented by a display. The input unit 34 is implemented by an input interface. The communication unit 35 is implemented by a communication interface.

[0043] Figure 5 is a data flow diagram of the processing of the server 11 of FIG. 1. As shown in FIG. 5, the server 11 inputs, to the machine learning model 12, the content of an input set 48 including at least a BOM and / or BOP 40 selected from a database DB, and a correction instruction 46 including a correction policy for the BOM and / or BOP 40. Depending on the content of the correction instruction, the input set 48 further includes a creation standard 45 corresponding to the BOM and / or BOP selected from the database DB. The content of the input set may be information represented by the input set whose data format is changed from the input set, or may be the input set itself.

[0044] The machine learning model 12 is trained to be capable of predicting an output for an input including an instruction. For example, the machine learning model 12 includes generative AI. Generative AI generates new data not included in training data by learning the regularity and structure of the training data. The generative AI may be constructed using a large language model (LLM).

[0045] To the machine learning model 12, for example, the content of an input set 48 including the selected BOM and / or BOP 40, the creation standard 45 corresponding to the selected BOM and / or BOP 40, and the correction instruction 46 is input. The machine learning model 12 outputs a correction proposal for the BOM and / or BOP 40 as a result of calculation in accordance with the input correction instruction. The calculation result of the machine learning model 12 may be the content of the BOM and / or BOP that is corrected data and has a data format different from that of the original BOM and / or BOP, or may be the corrected BOM and / or BOP itself.

[0046] As described above, the corrected content of the BOM and / or BOP 40 is output in accordance with the correction instruction from the user, therefore, using the improved BOM and / or BOP that reflects the corrected content is useful for optimizing product life cycle management.

[0047] The machine learning model 12 may include interactive AI. In this case, even when the instruction input to the machine learning model 12 is incomplete, the interactive AI can output a question to the user and obtain an answer input from the user, thereby autonomously outputting an appropriate calculation result.

[0048] FIG. 6 is a flowchart of the process performed by the server 11 of FIG. 1. Hereinafter, the process will be described along the flow of FIG. 6 with appropriate reference to FIGS. 1 to 5. The process of the server 11 described below is executed by the control unit 21. The server 11 generates an input set for the machine learning model 12 by, for example, accessing a database DB in response to a command from the information processing terminal 13 operated by a user (step S1). The input set includes, for example, a creation standard 45, a BOM and / or BOP 40 created in accordance with the creation standard 45 (pre-correction data), and a correction instruction 46 input by the user from the information processing terminal 13.

[0049] The server 11 inputs the content of the input set to the machine learning model 12 (step S2). The server 11 outputs the calculation result obtained from the machine learning model 12 to the information processing terminal 13, and causes the information processing terminal 13 to display the calculation result (step S3). On the information processing terminal 13, a correction proposal for the BOM and / or BOP 40, which is pre-correction data, is displayed as the calculation result of the machine learning model 12.

[0050] The user inputs to the information processing terminal 13 whether to approve the correction proposal displayed on the information processing terminal 13. The server 11 determines whether the correction proposal displayed on the information processing terminal 13 has been approved by the user (step S4). If the server 11 determines that the correction proposal has not been approved, the process returns to step S1. If the server 11 determines that the correction proposal has been approved, the server 11 outputs corrected data that is a BOM and / or BOP 40 reflecting the proposed correction (step S5).

[0051] Server 11 may output the corrected data to the database DB where the pre-correction data, namely BOM and / or BOP 40, is stored, and overwrite the pre-correction data with the corrected data, or it may output the corrected data to the storage unit 22 of Server 11 or the storage unit 32 of the information processing terminal 13 and save it there temporarily.

[0052] Figure 7 shows the prompt screen 60A of the first example in the information processing terminal 13 shown in Figure 1. As shown in Figure 7, the server 11 displays the prompt screen 60A, which is the input screen for the machine learning model 12, on the display unit 33 of the information processing terminal 13 in response to a command from the information processing terminal 13 operated by the user. Figure 7 shows an example of resolving an error due to missing information in the BOM. The user operates the information processing terminal 13 and inputs an input set to the prompt screen 60A. The input set includes a correction instruction from the user written directly under "# Instruction", a creation standard written directly under "# Creation Standard", and the contents of the BOM, which is the pre-correction data written directly under "# BOM". Note that the creation standard and BOM may be input directly into the machine learning model 12 as attached files.

[0053] The correction instructions indicate that the correction policy is to modify the parts of the pre-correction BOM data that do not conform to the creation standards to conform to them. The creation standards require the inclusion of part numbers in the BOM, but the pre-correction BOM data is missing part numbers.

[0054] Figure 8 shows the prompt screen 60B of the first example in the information processing terminal 13 of Figure 1. As shown in Figure 8, when the input set shown in Figure 7 is input to the machine learning model 12, the server 11 displays the calculation result output from the machine learning model 12 on the prompt screen 60B. The machine learning model 12 determines that the missing part number does not conform to the creation standard and proposes a corrected BOM with the part number supplemented according to the creation standard by referring to other information in the BOM. As a result, even if information is missing that does not conform to the creation standard when an engineer at company 1 creates a BOM while referring to the creation standard, the information will be supplemented to conform to the creation standard. Therefore, adverse effects on product lifecycle management using the BOM are prevented.

[0055] Figure 9 shows the prompt screen 61A of the second example in the information processing terminal 13 of Figure 1. Figure 9 shows an example of resolving a typographical error in a BOM. The user operates the information processing terminal 13 and inputs an input set to the prompt screen 61A. The correction policy indicated by the correction instruction is to correct the parts of the BOM data that do not conform to the creation standard to conform to the creation standard. The creation standard defines how part numbers are written in the BOM, but the BOM data before correction contains typographical errors in the part numbers that do not conform to the creation standard.

[0056] Figure 10 shows the prompt screen 61B of the second example in the information processing terminal 13 of Figure 1. As shown in Figure 10, when the input set shown in Figure 9 is input to the machine learning model 12, the server 11 displays the calculation result of the machine learning model 12 on the prompt screen 61B. The machine learning model 12 determines that the part number error does not conform to the creation standard and, referring to other information in the BOM, proposes a corrected BOM with the part number error corrected in accordance with the creation standard. As a result, even if an error that does not conform to the creation standard occurs when an engineer at company 1 creates a BOM while referring to the creation standard, the information will be corrected to conform to the creation standard. Therefore, adverse effects on product lifecycle management using the BOM are prevented. In addition, if there is an error in the order in which each item in the BOM is arranged that does not conform to the creation standard, the pre-correction data may be modified so that the order in which each item in the BOM is arranged conforms to the creation standard.

[0057] Figure 11 shows the prompt screen 62A of the third example in the information processing terminal 13 of Figure 1. Figure 11 shows an example in which a correction to resolve a typographical error in the BOM is made, along with the justification for the correction. The user operates the information processing terminal 13 and inputs an input set to the prompt screen 62A. The correction instructions include not only a correction policy to correct parts of the BOM data that do not conform to the creation standard to conform to the creation standard, but also an instruction requesting that the justification for correcting the BOM data be shown. In the BOM data, there is a typographical error in the part number.

[0058] Figure 12 shows the prompt screen 62B of the third example in the information processing terminal 13 of Figure 1. As shown in Figure 12, when the server 11 inputs the input set shown in Figure 11 to the machine learning model 12, it displays the calculation results of the machine learning model 12 on the prompt screen 62B. The machine learning model 12 determines that the part number error does not conform to the creation standard, and by referring to other information in the BOM, it proposes a corrected BOM with the part number error corrected in accordance with the creation standard, and also presents the basis for proposing the correction of the BOM. In this way, by presenting the basis for correcting the pre-correction data to the user, the user can accurately evaluate the validity of the correction, and product lifecycle management can be optimized with high accuracy.

[0059] Figure 13 is a diagram illustrating an example of resolving the duplication of BOMs 41A and 41B using the machine learning model 12 in Figure 1. As shown in Figure 13, for example, suppose that BOMs 41A and 41B exist in the database of one factory. Looking at BOM 41A alone, it is correctly created according to the creation standard. Looking at BOM 41B alone, it is also correctly created according to the creation standard. However, due to some ambiguity in the creation standard, there is a variation in the expression of the part description between BOM 41A and BOM 41B, and in reality, the contents of BOM 41A and BOM 41B are substantially the same. In such a case, the input set entered into the prompt screen mentioned above includes a correction instruction that indicates a correction policy to resolve the data duplication in the multiple BOMs 41A and 41B, which are the pre-correction data.

[0060] By doing so, the machine learning model 12 proposes discarding either BOM41A or BOM41B. By discarding one of the BOM41A and BOM41B, which are substantially redundant, unnecessary duplication is eliminated, the burden of data management is reduced, and the efficiency of product lifecycle management is improved. Note that the elimination of duplication may also be achieved by reducing the number of items in a single BOM or BOP that have multiple overlapping items.

[0061] Figure 14 is a diagram illustrating an example of performing overall optimization of multiple BOMs 41C and 41D using the machine learning model 12 in Figure 1. As shown in Figure 14, for example, suppose that BOM 41C exists in the database DB of the first factory and BOM 41D exists in the database DB of the second factory. Looking only at BOM 41C, it is correctly created according to the creation standard. Looking only at BOM 41D, it is also correctly created according to the creation standard. However, due to some ambiguity in the creation standard, there is inconsistency in the expression of the part description between BOM 41C and BOM 41D.

[0062] In such cases, the input set entered into the prompt screen mentioned above includes correction instructions indicating a correction policy to unify similar descriptions of the same item among multiple BOMs 41 and 42, which are the pre-correction data, into the same description. By doing so, the machine learning model 12 will propose a correction to the pre-correction data to unify similar descriptions of the same item among multiple BOMs 41 and 42 into the same description for overall optimization. For example, in Figure 14, corrections are proposed to match the expression "aluminum handle frame" in BOM 41D to the expression "aluminum frame of the handle" in BOM 41C, and to match the expression "silver" in BOM 41D to the expression "silver" in BOM 41C.

[0063] Thus, even when multiple bills of materials or process schedules are partially optimized, the pre-correction data can be modified to unify the expressions of similar descriptions within the same item for overall optimization, thereby facilitating overall management of the product lifecycle. In addition, if there are similar parts (substitutes) from different suppliers across multiple BOMs, the pre-correction data may be modified to unify them to parts from a single supplier as part of overall optimization. Similarly, if there are similar processes (substitute processes) from different BOPs, the pre-correction data may be modified to unify them to the same type of process as part of overall optimization.

[0064] Figure 15 is a data flow diagram of a modified example of the processing of server 11 in Figure 1. As shown in Figure 15, server 11 may perform preprocessing 51 to format an input set 48, which includes a BOM and / or BOP 40 selected from the database DB, its creation standard 45, and a modification instruction 46 indicating a modification policy for the BOM and / or BOP 40, into a form suitable for machine learning model 12 before inputting it to machine learning model 12.

[0065] For example, as preprocessing 51, the server 11 may convert at least one data format of the BOM and / or BOP 40, the creation standard 45, and the modification instruction 46 so that the creation standard 45 and the modification instruction 46 become text data and the BOM and / or BOP 40 become text data or CSV data. In this way, even if the input set is in a data format that does not fit the machine learning model 12, the machine learning model 12 can be made to process it appropriately.

[0066] If the calculation result output from the machine learning model 12 differs from the data format of BOM and / or BOP, the server 11 performs post-processing 52 to create corrected BOM and / or BOP data based on the calculation result. In this way, even if the calculation result output from the machine learning model 12 is in a data format different from BOM and / or BOP, the BOM and / or BOP can be obtained as corrected data.

[0067] Furthermore, even if the correction instruction does not include a correction policy, the correction policy may be pre-set in the pre-processing step 51. That is, when the server 11 receives a correction instruction, it may input the pre-set correction policy into the machine learning model 12. If the correction policy is pre-set in the workflow in this way, the user does not need to specify the correction policy each time a correction instruction is given, and the data size of the input set can be reduced. This is suitable for batch processing of periodic inspections, etc.

[0068] Figure 16 shows the interface screen 70 of the display unit 33 of the information processing terminal 13 shown in Figure 1. Note that this interface screen 70 is merely an example, and various other configurations can be adopted as long as they reduce the user's workload for issuing correction instructions. As shown in Figure 16, the interface screen 70 of the information processing terminal 13 accessing the server 11 includes a BOM / BOP selection unit 71, a creation standard selection unit 72, a correction instruction creation unit 73, a correction instruction selection unit 74, and a correction execution command unit 75. The creation standard selection unit 71, BOM / BOP selection unit 72, correction instruction creation unit 73, and correction execution command unit 75 are icons that are selected by clicking with a mouse or the like. The correction instruction selection unit 74 includes a checkbox 74a that is selected by clicking with a mouse or the like. The interface screen 70 is used for preprocessing before the machine learning model 12 is executed.

[0069] When the BOM / BOP selection unit 71 is selected, the server 11 displays a BOM / BOP list screen on the display unit 33 of the information processing terminal 13, which allows the user to select a BOM and / or BOP from the database DB. The user operates the information processing terminal 13 to select the desired BOM and / or BOP from the BOM / BOP list screen.

[0070] When the creation standard selection unit 72 is selected, the server 11 displays a creation standard list screen on the display unit 33 of the information processing terminal 13, which allows the user to select a corresponding creation standard from the database DB. The user operates the information processing terminal 13 to select a creation standard corresponding to the selected BOM and / or BOP from the creation standard list screen. If a creation standard is linked to a corresponding BOM or BOP, the server 11 may automatically select the creation standard linked to the selected BOM or BOP in response to the selection of the BOM or BOP.

[0071] When the modification instruction creation unit 73 is selected, the server 11 displays a modification instruction creation screen on the display unit 33 of the information processing terminal 13 for the user to create a new modification instruction. The user can operate the information processing terminal 13 to create modification instructions for the BOM and / or BOP on the modification instruction creation screen. The created modification instructions are stored in the storage unit 22 of the server 11 and displayed on the creation instruction selection unit 74.

[0072] The creation instruction selection unit 74 displays the title or content of the modification instruction and the corresponding checkbox 74a. The user operates the information processing terminal 13 to check the checkbox 74a corresponding to the modification instruction they wish to execute. Of the multiple modification instructions stored in the storage unit 22, the modification instructions with checked checkboxes 74a are associated with the modification execution command unit 75.

[0073] When the correction execution command unit 75 is selected, the server 11 creates an input set such that the selected BOM and / or BOP, the selected creation standard, and the correction instructions checked in checkbox 74a are included in the input set. If a creation standard is not required for the correction of the BOM and / or BOP, the server 11 creates an input set such that the selected BOM and / or BOP and the correction instructions checked in checkbox 74a are included in the input set. The server 11 inputs the contents of the created input set into the machine learning model 12. The processing after the machine learning model 12 outputs the calculation result is as described above.

[0074] In this way, by preparing a workflow using the interface screen 70, the workload on users for issuing correction instructions can be reduced, and variations in the quality of correction instructions from user to user can be prevented.

[0075] In the example shown in Figure 16, the selection of the correction execution command unit 75 is exemplified as a predetermined operation by the user, but this is not limited to that. For example, when a predetermined operation is performed on the interface screen of the information processing terminal 13 in which a page transition operation is input by the user, the server 11 may create an input set such that the correction instruction is included in the input set.

[0076] Alternatively, instead of using the correction execution command unit 75, correction instructions may be executed by entering text into a prompt screen as shown in Figures 7, 9, and 11, or correction instructions may be executed in response to operations such as screen page transitions.

[0077] Furthermore, even if the correction instructions do not include the correction policy, the machine learning model 12 may have the correction policy incorporated into it. That is, the machine learning model 12 may be configured to autonomously determine the correction policy by being pre-trained using training data that includes a large number of combinations of pre-correction data and post-correction data. In this case, the user does not have to specify the correction policy each time a correction instruction is given, and the data size of the input set can be reduced. This is suitable for batch processing of periodic inspections, etc.

[0078] [Note] The embodiments described above are specific examples of the following disclosures.

[0079] (Note 1) A device for supporting the optimization of product lifecycle management, comprising at least one control unit, the control unit receiving at least one pre-correction data selected from a group consisting of at least one bill of materials or at least one process chart, receiving correction instructions indicating a correction policy for the pre-correction data, inputting the contents of an input set including the pre-correction data and the correction instructions into a machine learning model, and outputting a calculation result in which the machine learning model has corrected the pre-correction data according to the correction instructions.

[0080] (Note 2) The product lifecycle management optimization support device as described in Note 1, wherein the control unit receives input of a creation standard that defines a standard method for creating the bill of materials and the process chart, the input set further includes the creation standard, and the modification policy includes modifying the parts of the pre-modification data that do not conform to the creation standard to conform to the creation standard.

[0081] (Note 3) The aforementioned correction policy includes eliminating data duplication in the pre-correction data, and is an optimization support device for product lifecycle management as described in Note 1.

[0082] (Note 4) The pre-correction data includes multiple bills of materials or multiple process schedules, and the correction policy includes unifying similar descriptions of the same item among the multiple bills of materials or similar descriptions of the same item among the multiple process schedules, as described in Note 1, for the product lifecycle management optimization support device.

[0083] (Note 5) The product lifecycle management optimization support device described in Note 1, wherein the correction instruction includes an instruction requesting that the basis for correcting the pre-correction data be shown, the calculation result includes the basis, and the control unit outputs the basis.

[0084] (Note 6) The control unit prepares the input set in a form suitable for the machine learning model before inputting it to the machine learning model, thereby creating the content, and is an optimization support device for product lifecycle management as described in Note 1.

[0085] (Note 7) The control unit creates the bill of materials or process table as modified data based on the calculation results output from the machine learning model, and is an optimization support device for product lifecycle management as described in Note 1.

[0086] (Note 8) The product lifecycle management optimization support device according to Note 1, further comprising a storage unit for storing the correction instructions, wherein the control unit creates an input set such that the correction instructions stored in the storage unit are included in the input set when a predetermined operation is input by the user.

[0087] (Note 9) The product lifecycle management optimization support device described in Note 1, wherein the control unit displays an interface screen on the display unit that includes a modification execution command unit configured to be associated with the modification instructions stored in the storage unit, and the input of the predetermined operation includes the user selecting the modification execution command unit.

[0088] (Note 10) The product lifecycle management optimization support device described in Note 1, wherein receiving the correction instruction includes inputting the correction instruction on the prompt screen.

[0089] (Note 11) The machine learning model is a product lifecycle management optimization support device as described in Note 1, including generative AI.

[0090] (Note 12) The machine learning model is an optimization support device for product lifecycle management as described in Note 1, including an interactive machine learning model.

[0091] (Note 13) A method for supporting the optimization of product lifecycle management, comprising: receiving at least one pre-correction data selected from a group consisting of at least one bill of materials or at least one process chart; receiving instructions for correcting the pre-correction data; inputting the contents of an input set including the pre-correction data and the instructions into a machine learning model; and outputting a calculation result in which the machine learning model has corrected the pre-correction data according to the instructions.

[0092] (Appendix 14) An optimization support program for product lifecycle management that causes at least one processor to execute the optimization support method described in Appendix 13.

[0093] 10 Product Lifecycle Management Optimization Support Device 11 Server 12 Machine Learning Model 21 Control Unit 22 Memory Unit 40 BOM (Bill of Materials), BOP (Bill of Process) 45 Creation Standard 46 Modification Instruction 48 Input Set 70 Interface Screen 75 Modification Execution Command Unit

Claims

1. A device for supporting the optimization of product lifecycle management, comprising at least one control unit, wherein the control unit receives at least one pre-correction data selected from a group consisting of at least one bill of materials or at least one process chart, receives a correction instruction for the pre-correction data, inputs the contents of an input set including the pre-correction data and the correction instruction into a machine learning model, and outputs a calculation result in which the machine learning model has corrected the pre-correction data according to the correction instruction.

2. The product lifecycle management optimization support device according to claim 1, wherein the control unit receives input of a creation standard defining a standard method for creating the bill of materials and the process chart, the input set further includes the creation standard, and the modification instruction includes modifying portions of the pre-modification data that do not conform to the creation standard to conform to the creation standard.

3. The product lifecycle management optimization support device according to claim 1, wherein the correction instruction includes eliminating data duplication in the pre-correction data.

4. The product lifecycle management optimization support device according to claim 1, wherein the pre-correction data includes a plurality of bills of materials or a plurality of process schedules, and the correction instruction includes unifying similar descriptions of the same item among the plurality of bills of materials or similar descriptions of the same item among the plurality of process schedules.

5. The product lifecycle management optimization support device according to claim 1, wherein the correction instruction includes an instruction requesting that the basis for correcting the pre-correction data be indicated, the calculation result includes the basis, and the control unit outputs the basis.

6. The product lifecycle management optimization support device according to claim 1, wherein the control unit formats the input set into a form suitable for the machine learning model before inputting it to the machine learning model to create the content.

7. The product lifecycle management optimization support device according to claim 1, wherein the control unit creates the parts list or the process list as modified data based on the calculation result.

8. The product lifecycle management optimization support device according to claim 1, further comprising a storage unit for storing the correction instructions, wherein the control unit creates an input set such that the correction instructions stored in the storage unit are included in the input set when a predetermined operation is input by the user.

9. The product lifecycle management optimization support device according to claim 8, wherein the control unit causes the display unit to display an interface screen including a modification execution command unit configured to be associated with the modification instructions stored in the storage unit, and the input of the predetermined operation includes selecting the modification execution command unit.

10. The product lifecycle management optimization support device according to claim 1, wherein receiving the correction instruction includes inputting the correction instruction on a prompt screen.

11. The machine learning model includes generative AI, the product lifecycle management optimization support device according to claim 1.

12. The product lifecycle management optimization support device according to claim 1, wherein the machine learning model includes an interactive machine learning model.

13. A method for supporting the optimization of product lifecycle management, comprising: receiving at least one pre-correction data selected from a group consisting of at least one bill of materials and at least one process chart; receiving instructions for correcting the pre-correction data; inputting the contents of an input set including the pre-correction data and the instructions into a machine learning model; and outputting a calculation result in which the machine learning model corrects the pre-correction data according to the instructions.

14. An optimization support program for product lifecycle management, which causes at least one processor to execute the optimization support method described in claim 13.

Images