Information processing system, information processing method, and program

Abstract

The system comprises a management unit for managing design drawings, an auxiliary input unit for inputting the design drawings and design metadata representing an algorithm for generating a path that takes into account the processing order or processing conditions of the components in the design drawings, an input control unit for receiving input information related to the design drawings, and an output control unit for outputting output information based on a path generated based on the design metadata from a trained model into which the design drawings, design metadata, and the input information have been input.

Description

Technical Field

[0001] The present disclosure relates to an information processing system, an information processing method, and a program.

Background Art

[0002] In recent years, generative AI has been used in software development (Patent Document 1). On the other hand, in instrumentation and control systems, etc., the design, management, and operation of control logic related to hardware such as circuits and signal processing are performed. Non-Patent Document 1 describes an attempt to use generative AI for circuit design.

Prior Art Documents

Patent Documents

[0003]

Patent Document 1

Non-Patent Documents

[0004]

Non-Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0005] However, as described in Non-Patent Document 1, it is difficult to design circuits using generative AI. Circuit diagrams are diagrams that show the connections between power sources, electronic components, and wiring at a glance, but reading them requires experience. For example, a circuit diagram can be deciphered by understanding the meaning of circuit symbols and basic rules such as current flowing from left to right, and then examining the electrical connections. On the other hand, generative AI is a model trained to accurately understand natural language and programs by widely collecting and learning from texts generated in natural language, such as books, explanatory texts, news articles, and conversation logs. Generative AI is a model trained to understand natural language and accurately understand text, images, and programs in response to queries. Although generative AI can understand natural language and generate text, images, and programs in response to queries, it struggles to understand electrical connections after grasping the meaning of circuit symbols and basic rules. Furthermore, when the circuit diagram is image data created by some kind of conversion / output processing, such as scanning from paper media, the OCR processing that generative AI generally performs on image data is inaccurate in deciphering the strings written in the circuit diagram. Therefore, there was a problem in that it was difficult to provide design support using generative AI.

[0006] This disclosure was made to solve the above-mentioned problems and aims to provide an information processing system, information processing method, and program that can provide design support using generative AI. [Means for solving the problem]

[0007] To solve the above problem, one aspect of the present disclosure is an information processing system comprising: a management unit for managing design drawings; an auxiliary input unit for inputting design drawings and design metadata representing an algorithm for generating a path that takes into account the processing order or processing conditions of components in the design drawings; an input control unit for receiving input information relating to the design drawings; and an output control unit for outputting output information based on a path generated based on the design metadata from the design drawings, the design metadata, and a trained model into which the input information has been input.

[0008] To solve the above problem, one aspect of the present disclosure is an information processing method comprising the steps of: managing a design drawing; inputting the design drawing and design metadata representing an algorithm for generating a path that takes into account the processing order or processing conditions of the components in the design drawing; receiving input information relating to the design drawing; and outputting output information based on a path generated based on the design metadata from a trained model into which the design drawing, the design metadata, and the input information have been input.

[0009] To solve the above problem, one aspect of this disclosure is a program for causing a computer to function as a management unit for managing design drawings, an auxiliary input unit that inputs design drawings and design metadata representing an algorithm for generating a path that takes into account the processing order or processing conditions of the components in the design drawings, an input control unit that receives input information relating to the design drawings, and an output control unit that outputs output information based on a path generated based on the design metadata from the design drawings, the design metadata, and a trained model into which the input information has been input. [Effects of the Invention]

[0010] According to this disclosure, design support can be provided using generative AI. [Brief explanation of the drawing]

[0011] [Figure 1]This figure shows an example of screen G1, which displays a circuit diagram as a design drawing in an embodiment. [Figure 2] This figure shows an example of a firmware (FW) for displaying the AI ​​function generation in an embodiment. [Figure 3] This figure shows an example of a function description screen HW1 applied to the circuit diagram in the embodiment. [Figure 4] This figure shows an example of design metadata D11 applied to the circuit diagram in the embodiment. [Figure 5] This figure shows an example of screen G2, which displays a ladder diagram as a design drawing in an embodiment. [Figure 6] This figure shows an example of a function description screen HW2 applied to the ladder diagram in the embodiment. [Figure 7] This is a diagram of the design metadata D21 applied to the ladder diagram in the embodiment. [Figure 8] This is a block diagram showing an example of the application of the information processing system 10 in the embodiment. [Modes for carrying out the invention]

[0012] Embodiments of this disclosure will be described below with reference to the drawings. However, the scope of this disclosure is not limited to the embodiments described below and can be modified at will within the scope of the technical concept of this disclosure. Furthermore, in the following drawings, the scale and number of components in each structure may differ from those in the actual structure in order to make the configurations easier to understand.

[0013] The information processing system 10 of this embodiment provides design support services. These design support services utilize generation AI to assist with tasks related to design drawings, such as function descriptions, operation descriptions, and the creation of design drawings. By providing such design support services, the value of engineering tools used in design can be enhanced.

[0014] A generative AI is a model trained to accurately interpret an image or a question text called a prompt input into the generative AI and generate and output an answer to the question text by incorporating a language model. A language model is a model trained to accurately interpret natural language by widely collecting texts generated in natural language, which is a corpus on the web, such as books, explanatory texts, news articles, conversation logs, etc. As the language model, for example, Anthropic (registered trademark) Claude (registered trademark), ChatGPT (registered trademark), XLNet, T5, RoBERTa, ALBERT, etc. can be used.

[0015] An engineering tool is software or an application for supporting development, design, manufacturing, operation, etc. in products and systems. In this embodiment, in particular, an engineering tool using graphical representation is targeted. Graphical representation is a method of visually representing information using figures, pictures, graphs, tables, etc. Here, it includes those represented in a design drawing, for example, those representing the connection relationship of electrical components using a circuit diagram (see FIG. 1), and those visually representing control logic using a ladder diagram (see FIG. 5).

[0016] Hereinafter, a design drawing with graphical representation, for example, a circuit diagram or a ladder diagram, will be described as an example, but it is not limited thereto. Any graphical representation can be applied to this embodiment. For example, it can be applied to a system that controls a control target or manages a management target using a pedigree chart, a wiring diagram, etc. More specifically, it can be applied to a system that monitors the state of a plant based on a pedigree chart or a trend graph in instrumentation control applied to plants such as power plants, chemical industries, steel industries, food manufacturing, etc. Also, in wiring and piping work, it can be applied to a system that manages the design, implementation, and maintenance, etc. of sensors, control devices, and wiring, etc. arranged at the site based on the site layout.

[0017] In addition, in this embodiment, it is sufficient that at least information regarding the graphical representation shown in an engineering tool or the like can be output using a generative AI, and it is not necessarily required to use an engineering tool. For example, an image obtained by scanning a handwritten design drawing on paper media, or an image printed on paper media, or an image captured as a screenshot of a two-dimensional or three-dimensional drawing generated using CAD (Computer Aided Design) can be used.

[0018] FIG. 1 is a diagram showing an example of a screen G1 showing a circuit diagram as a design drawing in an embodiment. In this figure, an example in which the screen G1 is displayed on an engineering tool is shown. On the screen G1, a design screen DW1 and a parameter list screen PW are shown. The design screen DW1 is a screen that displays information graphically represented such as a design drawing, and in this example of the figure, a circuit diagram is shown. The parameter list screen PW is a screen that shows a list of parameters set for components shown in the circuit diagram shown on the design screen DW1, for example, components to which an electrical signal is input or output, components represented by electrical circuit symbols, components to which a control signal is input or output, components represented by control logic symbols, and the like.

[0019] In this figure, on the design screen DW1, a circuit diagram for outputting a command value of power generation output in an instrumentation control system for monitoring and controlling a power generation plant is shown. In this circuit diagram, as components that are components of the circuit, symbols indicating an input port, a selector as an electrical component, and an output port, and connection lines connecting the symbols are shown. The connection lines are indicated by arrows, and the direction of the arrows indicates the input / output direction of the signal. Also, above the symbol, a character string indicating the symbol name is shown. Also, depending on the symbol, a signal name is shown below the symbol. For example, in this figure, three selectors with symbol names SEL000249, SEL000250, and SEL000253 are shown. Of these, the selector enclosed in code C, with symbol name SEL000250, receives a signal from the input port with symbol name YR_ECYC_0011 and indicated as "XXXXX Power Output Command Value". This selector also receives a signal from the input port with symbol name W_1GTG_PV and indicated as "XXXXX Power PV Value" via components indicated by symbol names ENG000246 and ENG000247. Furthermore, this selector receives a control signal from the input port with symbol name CB_CYC_ALM004 and indicated as "XXXXX Communication Anomaly". From this diagrammatic representation, it can be seen that in this circuit diagram, the selector with symbol name SEL000250 outputs either a signal from symbol name YR_ECYC_0011 or a signal from symbol name W_1GTG_PV in response to a control signal from symbol name CB_CYC_ALM004.

[0020] In this embodiment, a generation AI function is provided to perform tasks related to such graphical representation, such as creating and modifying design drawings and providing functional descriptions, using generation AI. Figure 2 shows an example of a generation AI function display screen FW in an embodiment. Figure 2 shows an example where screen G2, which displays a circuit diagram as a design drawing, is displayed in the engineering tool. Screen G2 shows the design screen DW1 and the generation AI function display screen FW. The Generation AI Function Display Screen FW is a screen for supporting tasks related to the circuit diagram shown on the design screen DW1 using generation AI. In this diagram, the Generation AI Function Display Screen FW shows tabs for displaying a list of generation AI functions. The list of generation AI functions shows operation buttons for using each of the following functions: design diagram generation (modification), function explanation, operating condition explanation, test design, debugging, problem, solution, and others. The generation (modification) of design drawings refers to the function of having the generation AI generate new design drawings such as circuit diagrams and ladder diagrams, or modify existing design drawings. The function description is a feature that allows the generating AI to explain a graphically represented function, such as the signal flow shown in a design drawing. The function that describes the operating conditions allows the generating AI to explain the operating conditions for components that perform actions according to input conditions, such as selectors, switches, and logic circuits. Test design is a function that allows a generating AI to design input patterns for testing the operation of a design drawing, and the expected output patterns when those input patterns are input into the design drawing. Test execution is a function that allows the generating AI to test the operation of the design blueprint. Debugging is a function that allows the generating AI to discover and correct defects in the design blueprint. The "problem" refers to the function of having the generating AI identify and explain problems in the design drawings. The solution feature allows the generating AI to suggest ideas for resolving problems in the design drawings.

[0021] In the following section, we will illustrate an example of the execution of the generation AI function by describing a case where button B in Figure 2 is touched, that is, an operation is performed to execute the "function description" as a generation AI function, and the generation AI is instructed to execute the "function description".

[0022] <Comparative Example (Part 1)> First, let's explain Comparative Example (1). In Comparative Example (1), we assume a scenario where the selector enclosed by the symbol C in Figure 1 is the subject of explanation, and a prompt with a question such as "Please explain the input / output processing of SEL000250." is sent to the generating AI along with a captured image of the circuit diagram, causing the generating AI to perform a "function explanation."

[0023] In this case, while the generating AI can accurately understand the meaning of the question as natural language, it will have difficulty correctly grasping the signal flow in the circuit diagram if it has not been sufficiently trained on how to read circuit diagrams. As a result, the generating AI will output an answer that is far from accurate, such as the one shown below, using only the words "SEL000250" and "input / output processing" in the question and the words shown in the circuit diagram as clues.

[0024] <Answer in Comparative Example (1)> "SEL000250 is a selector block located in the center of this control logic diagram. The inputs to SEL000250 are as follows: 1. Input from the top: Output of SEL000249 2. Input from the bottom: Output of SEL000250 The output is connected to the right side and is sent to another selector block called SEL000253. The role of the SEL000250 is to determine which of the two input signals, one from the top and one from the bottom, will be selected as the output. The selection criteria are not explicitly shown in the diagram, but typically such selectors choose the input based on some condition or priority. This block likely processes signals related to power output commands and appears to be one of the key control points in the overall system. The upper input path seems to contain the PV power value for 1CTC, while the lower input path contains the power output command value for ECY.

[0025] As shown above, in comparative example (1), the generated answer is not very accurate. This is because, although the generating AI can understand natural language, it has not learned how to interpret graphically represented information such as blueprints, or its learning is insufficient. As a countermeasure, in this embodiment, design metadata can be added to the prompts input to the generating AI. Design metadata is information that helps in understanding the object represented in the diagram. By adding such design metadata, the generating AI can be made to correctly understand how to read design drawings and the like.

[0026] <Design metadata (Part 1)> As design metadata, for example, wording that conveys what is represented in the design drawing can be used. Specifically, in the prompt shown in the comparative example, wording was added to indicate that the circuit diagram is a diagram that represents connection relationships. For example, a prompt containing a question such as, "Please explain the input / output processing of SEL000250 while paying attention to the connection relationships," is sent to the generating AI along with a captured image of the circuit diagram. This allows the generating AI to understand that the circuit diagram graphically represents the connection relationships. The generating AI can then generate an answer that takes these connection relationships into account, such as the one shown below.

[0027] <Example response when design metadata (part 1) is provided> "I will now explain the input / output processing of SEL000250, paying attention to the connection relationships: input: 1. Signal from YR_DCYC_0011 (XXXXX power output command value) 2. Signal from W_1GTC_PV (XXXXX power PV value). This is processed via ENC000246 and ENC000247. 3. Signal from CR_CYC_ALN004 (XXXXX communication error) output: The output of SEL000250 is connected to SEL000253. process: 1. SEL000250 functions as a selector. 2. Under normal circumstances, the power output command value from YR_DCYC_0011 is selected. 3. However, if there is a signal from CR_CYC_ALN004, the processed signal from W_1GTC_PV will be selected. The role of this selector is to ensure system reliability by using command values ​​under normal circumstances and actual measured values ​​(PV values) as a backup in case of communication failures. The output of SEL000250 is further sent via SEL000253 and finally to AO000022 (XXXXX power generation setting) and CR_RMH_A022 (XXXXX pre-reply signal). This configuration enables a mechanism that selects the appropriate signal according to the communication status and controls the generator's output.

[0028] Thus, by adding a prompt that conveys what is represented in the design drawing as design metadata—that is, that the circuit diagram is a diagram that shows connection relationships—it becomes possible to explain things more accurately compared to when no such metadata is provided.

[0029] <Design metadata (part 2)> Additionally, design metadata can include, for example, the names of the components that make up the diagram.

[0030] The above example answer for design metadata (part 1) contains several errors in the character recognition of symbol names. For example, the above answer shows the signal corresponding to YR_DCYC_0011 as the first input to SEL000250, but the correct signal is "YR_ECYC_0011". Also, the second input to SEL000250 shows the signal W_1GTC_PV via ENC000246 and ENC000247, but the correct signal is "W_1GTG_PV" via "ENG000246" and "ENG000247". Furthermore, the third input to SEL000250 shows the signal corresponding to CR_CYC_ALN004, but the correct signal is "CB_CYC_ALM004".

[0031] For example, if you use a design drawing that contains illegible characters and try to have a generating AI explain the function and operation of that drawing, it is likely that errors in character recognition will occur. The following are some examples of cases where the characters are illegible. In circuit diagrams drawn by hand on paper, symbol names may become illegible due to any of the following cases or a combination thereof. • Cases where the handwriting is poor. • Cases that have weathered due to ink smudging, etc. • Cases where the contrast between the background and text is low In circuit diagrams printed on paper, symbol names may become illegible due to any of the following cases or combinations thereof. • In cases where the printer resolution is low • Cases where the scanner resolution is low In image data of circuit diagrams output from engineering tools, etc., symbol names may become illegible in any of the following cases or combinations thereof. • Cases where the display resolution is low • Cases where the font size is small

[0032] To address the issue of symbol names becoming difficult to read, the component name is now included in the prompts within the design metadata. For example, following a question such as, "Please explain the input / output processing of SEL000250 while paying attention to the connection relationships. However, please refer to the following for the symbol name," a prompt containing information including the symbol name (see Figure 4) is sent to the generating AI along with a captured image of the circuit diagram. Engineering tools draw a circuit diagram on the design screen DW1 by referring to information for drawing the circuit diagram, such as the position coordinates, size, and name of symbols as components, information indicating the type of connection terminals if any, and information indicating connections between symbols or terminals. This information referenced during drawing can be used as design metadata. Alternatively, since the netlist generated from a circuit diagram created with a circuit diagram editor, etc., contains connection information such as symbol names, symbol pin numbers, and connection destinations, the netlist can be used as design metadata. Furthermore, in cases where engineering tools are not used, if a handwritten circuit diagram or a circuit diagram printed on paper is used as the circuit diagram to be recognized by the generating AI, the circuit diagram can be clearly described or printed as a separate document as design metadata. Alternatively, the electronic data of the circuit diagram used during printing may be used as design metadata. By using this design metadata, the generating AI can understand that a circuit diagram graphically represents connection relationships and can correctly identify the components, or symbols, that make up the circuit diagram. This allows it to generate answers using the correct symbol names, such as those shown below.

[0033] <Example response when design metadata (part 1) and (part 2) are provided> "I will now explain the input / output processing of SEL000250, paying attention to the connection relationships: input: 1. YR_ECYC_0011 (XXXXX power output command value) 2. W_1GTG_PV (XXXXX Power PV Value) - via ENG000246 and ENG000247 3.CB_CYC_ALM004 (XXXXX communication error) output: The output of SEL000250 is connected to SEL000253. process: 1. W_1GTG_PV is processed by passing through ENG000246 and ENG000247. These ENG blocks are thought to be performing some kind of engineering transformation. 2. CB_CYC_ALM004 is a flag indicating a communication error and may affect the selection of SEL000250. 3. Two values, YR_ECYC_0011 and W_1GTG_PV (after processing), are entered into SEL000250. 4. Depending on the state of CB_CYC_ALM004, SEL000250 will select either YR_ECYC_0011 or W_1GTG_PV (after processing). 5. The selected value will be output to SEL000253. 6.SEL000253 is ultimately connected to AO000022 (XXXXX power generation setting) and used to set the power generation amount. This process ensures that, depending on the communication status and other conditions, an appropriate power output command value or measured value is selected and reflected in the power generation setting for No. 1GTG.

[0034] In this way, by adding the component name to the prompt as design metadata, it becomes possible to provide a more accurate description compared to when it is not included.

[0035] Figure 3 shows an example of a function description screen HW1 applied to the circuit diagram in the embodiment. The description screen HW1 pops up when, for example, button B in Figure 2 is touched and the operation to execute the "function description" as a generated AI function is performed. In this diagram, the explanation screen HW1 shows the prompt input field QF, data D10 which is image data of the circuit diagram, along with data D11 as design metadata and comment C1. Data D11 contains information including the names of the components shown in the circuit diagram. In this diagram, Data D11 is shown as design information (C1_RMT01_o001_design_information.txt), and an example is shown where text containing the names of the component symbols shown in the circuit diagram displayed on the design screen DW1 is used. Comment C1 contains information that indicates the objects represented in the circuit diagram, i.e., the connection relationships. In this diagram, comment C1 states, "Please refer to the attached design information for the details in the attached design drawing. Please pay attention to the connection relationships when answering."

[0036] Figure 4 shows an example of data D11 as design metadata for a circuit diagram in an embodiment. In this figure, data D11 contains information including the names of the symbols that make up the circuit diagram, following the phrase, "Please refer to the following for symbol names." For example, information including the symbol name SEL000250 is shown as "export.symbol.23=5,@@YSEL000250,,float,0.0,,,,inout". By adding such information including symbol names to the prompt, the generating AI can correctly recognize the symbol names shown in the circuit diagram.

[0037] Figure 5 shows an example of screen G2 displaying a ladder diagram as a design drawing in an embodiment. This figure shows an example of screen G2 being displayed in the engineering tool. Screen G2 shows the design screen DW and the generation AI function display screen FW. The design screen DW2 is a screen that displays graphically represented information such as design drawings, and in this example, a ladder diagram is shown. The generation AI function display screen FW is a screen for supporting work related to the ladder diagram shown on the design screen DW2 using generation AI. The generation AI function display screen FW is the same as in Figure 2, so a detailed explanation is omitted here.

[0038] In this diagram, design screen DW2 shows a ladder diagram illustrating the control logic for outputting command values ​​for power generation output in an instrumentation and control system that monitors and controls a power plant. In this ladder diagram, the components that make up the circuit are shown as follows: power lines indicated by vertical lines on the left, output lines indicated by vertical lines on the right, contacts provided in the ladder that connect the power lines and output lines, symbols representing processing instructions and end instructions, and connecting lines that connect the symbols to each other. In addition, strings corresponding to the symbols of the contacts, such as "ZR100.0", the symbols indicating that the processing instruction is a transfer instruction, such as "EMOV", and the register names of the source and destination, such as "ZR1001", are shown. For example, this diagram shows that when contacts ZR100.0 and ZR100.1 are both in the ON state and connected, the EMOV instruction is executed, and the value stored in the source register ZR1001 is transferred to the destination register ZR1011. From this diagrammatic representation, it can be seen that in this ladder diagram, the EMOV instruction is not executed if at least one of contacts ZR100.0 and ZR100.1 is disconnected, and the EMOV instruction is executed only when both contact ZR100.0 and contact ZR100.1 are connected.

[0039] <Comparative Example (Part 2)> Here, let's consider a comparative example (2). Specifically, we will consider a scenario where, regarding the ladder diagram shown in Figure 5, a prompt containing a question such as "Please explain the operation of the program in the diagram," is sent to the generating AI along with a captured image of the circuit diagram, causing the generating AI to perform a "function explanation."

[0040] In this case, while the generating AI can accurately understand the meaning of the question as natural language, it will have difficulty correctly grasping the control logic of the program shown in the ladder diagram if it has not been sufficiently trained on how to read ladder diagrams. As a result, the generating AI will output an answer that is not very accurate, such as the answer shown below, using only the words "diagram" and "program" in the question and the words shown in the ladder diagram as clues.

[0041] <Answer in Comparative Example (Part 2)> "This diagram shows a type of PLC programming language called a ladder diagram. The program works as follows: 1. The program is divided into two main sections: -ZR100 series (starts from line 0) -ZR200 series (starts from line 13) 2. Each section has four input conditions (e.g., ZR100.0, ZR100.1, ZR100.2, ZR100.3). 3. When each input condition is met, the corresponding EMOV (move) command is executed. 4. The EMOV instruction copies the value of one variable to another variable. For example: -When ZR100.1 is ON, the value of ZR1001 is copied to ZR1011. -When ZR200.2 is ON, the values ​​from ZR2002 are copied to ZR2012. 5. All conditions and instructions are evaluated and executed in parallel. This means that if multiple conditions are met simultaneously, the corresponding multiple EMOV instructions will be executed at the same time. 6. The program terminates with the END command on line 26. This program implements simple control logic that moves data from one location to another when certain conditions are met. This is a type of program commonly used in industrial control systems and automation processes.

[0042] Thus, in the comparative example (2), although the ladder diagram was correctly recognized as a type of PLC programming language, the generated answer did not take into account that when the two contacts ZR100.0 and ZR100.1 are connected in series, the two contacts form an AND condition, and the EMOV instruction is executed only when the condition that both contacts are connected is met.

[0043] <Design Metadata (Part 3)> As a countermeasure, in this embodiment, the design metadata (part 3) includes wording that conveys points to pay attention to in the ladder diagram and the items that are expected as answers. Specifically, the prompt, along with a captured image of the ladder diagram, was sent to the generating AI with the question, "List all operable paths in the diagram, paying attention to the symbol names, and explain the operation of the ladder diagram based on the listed paths." This allowed the generating AI to understand the importance of paying attention to symbol names in the ladder diagram, listing the paths shown in the diagram, and explaining the operation of those paths. The generating AI can then generate an answer based on the paths generated using the design metadata, such as the answer shown below.

[0044] <Example response when design metadata (part 2) is provided> "List the possible paths in this ladder diagram and explain how they work: 1.ZR100.0→ZR100.1→EMOV(ZR1001,ZR1011) 2.ZR100.0→ZR100.2→EMOV(ZR1002,ZR1012) 3.ZR100.0→ZR100.3→EMOV(ZR1003,ZR1013) 4.ZR200.0→ZR200.1→EMOV(ZR2001,ZR2011) 5.ZR200.0→ZR200.2→EMOV(ZR2002,ZR2012) 6.ZR200.0→ZR200.3→EMOV(ZR2003,ZR2013) Explanation of ladder diagram operation: 1. This ladder diagram consists of two main parallel circuits (ZR100 series and ZR200 series). 2. Operation of the ZR100 series: -When ZR100.0 is ON, if any of ZR100.1, ZR100.2, or ZR100.3 is ON, the corresponding EMOV instruction will be executed. -For example, if ZR100.1 is ON, the data from ZR1001 will be copied to ZR1011. 3. Operation of the ZR200 series: -When ZR200.0 is ON, if any of ZR200.1, ZR200.2, or ZR200.3 is turned ON, the corresponding EMOV instruction will be executed. -For example, if ZR200.2 is ON, the data from ZR2002 will be copied to ZR2012. 4. Each EMOV block will only transfer data if specific conditions are met. 5. Once all processing is complete, the program reaches the final END block and terminates. This ladder diagram is designed to execute different data transfer operations based on multiple conditions. The ZR100 and ZR200 series operate independently, performing the appropriate data transfer according to their respective conditions.

[0045] Thus, by adding design metadata such as prompts that convey points to pay attention to in the diagram and what is expected as a response—for example, prompts that in a ladder diagram should focus on the symbol names, list the paths, and explain how the paths work—it becomes possible to provide a more accurate explanation compared to when no such prompts are provided.

[0046] <Design metadata (part 4)> Furthermore, as design metadata, for example, an algorithm for components, specifically an algorithm for generating a path that takes into account the processing order or processing conditions of components in the design drawing, may be used.

[0047] The example answer for the design metadata (part 3) above lists the paths, but its operation is explained as follows: "When ZR100.0 is ON, if any of ZR100.1, ZR100.2, or ZR100.3 is ON, the corresponding EMOV instruction will be executed. For example, if ZR100.1 is ON, the data in ZR1001 will be copied to ZR1011." This explanation does not describe all operations individually. Therefore, if ZR100.0 is ON, the explanation could be interpreted as meaning that if ZR100.2 is ON, the data in ZR1001 will be copied to ZR1011.

[0048] As a countermeasure, we added the component's algorithm to the prompt as supplementary information. For example, a prompt containing information representing the algorithm (see Figure 7), along with a captured image of the ladder diagram, is sent to the generating AI, following a question such as, "(1) Following the algorithm, list all possible paths in the attached ladder diagram, paying attention to the symbol names. (2) Describe the operation of the ladder diagram based on the listed paths. Note that the direction of movement is represented by up, down, left, and right on the ladder diagram." By using this kind of design metadata, the generating AI can correctly understand the algorithms of the components shown in the ladder diagram and explain their operation more accurately. This allows it to generate answers such as the following:

[0049] <Example of response when design metadata (part 3) and (part 4) are provided> "According to the algorithm, we enumerate all possible paths in the ladder diagram: 1.ZR100.0→ZR100.1→EMOV(ZR1001,ZR1011) 2.ZR100.0→ZR100.2→EMOV(ZR1002,ZR1012) 3.ZR100.0→ZR100.3→EMOV(ZR1003,ZR1013) 4.ZR200.0→ZR200.1→EMOV(ZR2001,ZR2011) 5.ZR200.0→ZR200.2→EMOV(ZR2002,ZR2012) 6.ZR200.0→ZR200.3→EMOV(ZR2003,ZR2013) Based on these paths, we will explain the operation of the ladder diagram: 1. This ladder diagram consists of six parallel circuits. Each circuit operates independently. 2. The three circuits in the upper half (ZR100 series) and the three circuits in the lower half (ZR200 series) have the same structure. 3. Each circuit consists of two series contacts and an EMOV instruction. 4. Operation of the ZR100 series: -If ZR100.0 is ON and ZR100.1 is ON, EMOV(ZR1001,ZR1011) will be executed. If -ZR100.0 is ON and ZR100.2 is ON, EMOV(ZR1002,ZR1012) will be executed. -If ZR100.0 is ON and ZR100.3 is ON, EMOV(ZR1003,ZR1013) will be executed. 5. Operation of the ZR200 series: -If ZR200.0 is ON and ZR200.1 is ON, EMOV(ZR2001,ZR2011) will be executed. If -ZR200.0 is ON and ZR200.2 is ON, EMOV(ZR2002,ZR2012) will be executed. -If ZR200.0 is ON and ZR200.3 is ON, EMOV(ZR2003,ZR2013) will be executed. 6. The EMOV instruction copies the data in the first argument to the address in the second argument. 7. Each circuit operates independently and is not affected by the state of other circuits. This means that multiple EMOV instructions can be executed simultaneously if the conditions are met. 8. The at the end of the ladder diagram indicates the end of the program. This ladder diagram is designed to control multiple data transfer operations in parallel based on specific conditions (ON / OFF status of contacts).

[0050] In this way, by adding the component's algorithm to the prompt as design metadata, it becomes possible to describe the operation more accurately compared to when it is not included.

[0051] Figure 6 shows an example of a function description screen HW2 applied to the ladder diagram in the embodiment. The description screen HW2 is displayed as a pop-up when, for example, button B in Figure 5 is touched and the operation to execute the "function description" as a generated AI function is performed. In this diagram, the explanation screen HW2 shows the prompt input field QF, data D20 which is image data of the ladder diagram, along with design metadata data D21 and comment C2. The diagram also shows the question Q1 entered into the prompt, and the answer A1 from the generating AI to that question Q1. Data D21 contains information including the algorithms of the components shown in the ladder diagram. In this diagram, Data D21 is shown as design information (C1_RMT01_o001_design_information.txt), and an example is shown where text containing the algorithms of the component symbols shown in the ladder diagram displayed on the design screen DW2 is used. Comment C2 contains information that explains the points to focus on in the ladder diagram and the expected behavior as a response. In this diagram, comment C2 states: "The attached design drawing is a ladder diagram. Please refer to the attached design information for the details of the design drawing. Please pay attention to the design drawing and the symbol names in the design information, following the algorithm in the design information. When explaining the ladder diagram, please list all possible paths. Please provide your answer based on the listed paths." Question Q1 states, "Please explain the operation of the ladder diagram." Answer A1 shows the same answer as the example answer when design metadata (part 3) and (part 4) are provided, as described above.

[0052] Figure 7 shows an example of data D21 as design metadata for a ladder diagram in an embodiment. In this figure, data D21 includes the following text as comments described as design metadata (part 4), and text describing the algorithm of the component. "#Algorithm (1) Start from the top of the power source (vertical line on the far left) and proceed downwards. (2) Proceed along the direction of travel until you reach a junction or ground (vertical line on the far right). Add any contacts or coils along the way to the path. (3) When you reach a junction, process according to the conditional branching below and return to (2). ..." By providing prompts with information that includes such algorithms, the generating AI can more accurately describe the behavior and correctly recognize the algorithms of the components shown in the ladder diagram.

[0053] In the example above, we explained the case where design metadata (1) and (2) are applied to the circuit diagram, and design metadata (3) and (4) are applied to the ladder diagram. However, the combination of design diagrams and design metadata can be arbitrary. For example, design metadata (2) may be applied only to the circuit diagram, or design metadata (3) and (4) may be applied, or either design metadata (3) or (4) may be applied only to the circuit diagram. In other words, in order to have the generating AI explain the function of a circuit diagram, information including symbol names may be provided as explanatory supplementary information without mentioning that the circuit diagram shows connection relationships. In order to have the AI ​​explain the operation of a circuit diagram, either or both of the following may be provided as design metadata: wording that conveys points of focus in the connection diagram, or an algorithm that shows the operation of symbols in the connection diagram. Furthermore, in order to have the generating AI explain the operation of the ladder diagram, it is possible to provide only the algorithm that shows the operation of the symbols in the ladder diagram as explanatory supplementary information, without mentioning any wording that conveys the points of focus in the ladder diagram. In order to have the AI ​​explain the function of the ladder diagram, it is possible to provide either or both of the following as design metadata: that the ladder diagram shows control logic, and / or the symbol names.

[0054] Furthermore, while the above examples illustrate the use of a generating AI to explain the function of a circuit diagram and the operation of a ladder diagram, the AI ​​is not limited to these examples.

[0055] For example, the generation AI may be made to generate a design drawing. In this case, the information processing system 10 generates a design drawing in response to user input and stores design metadata corresponding to the generated design drawing. The design drawing here is, for example, a connection part or ladder diagram, and is a design drawing in which components that input or output electrical signals or control signals are connected, and which contains strings representing the names of each component. The auxiliary information here is, for example, data D11, and may be any of the following, or a combination thereof: information including the names of components, data D10, information representing the connection relationships between components such as netlists, information indicating points of focus in the design drawing, or information representing an algorithm for generating a path that takes into account the processing order or processing conditions of the components in the design drawing. In this way, the information processing system 10 generates a design drawing in response to user input and stores the design metadata corresponding to the generated design drawing. This allows the generating AI to use the design metadata corresponding to the design drawing being explained when it is asked to explain the functions and operations of the components shown in the design drawing.

[0056] Furthermore, the information processing system 10 may generate design drawings using a generation AI. For example, when a prompt is input by the user, the information processing system 10 causes the generation AI to input the input prompt and related design metadata. Here, the prompt input by the user requests the generation of a design drawing. The information processing system 10 causes the generation AI to input, along with the prompt, any of the following design metadata related to the design drawing, such as information including the names or naming rules of components, information representing examples of connections between components, information indicating points of focus in the design drawing, or information representing an algorithm for generating a path that takes into account the processing order or processing conditions of the components in the design drawing, or a combination thereof. As a result, the generation AI can understand the functions and / or operations of the components in the design drawing and generate the design drawing requested by the user. Therefore, it is possible to reduce the situation in which an image that is merely a design drawing and does not take into account the operation or algorithm is generated.

[0057] Furthermore, when the information processing system 10 generates a design drawing, it may use the generation AI to generate information related to the generated design drawing, such as design description information representing functions or operating conditions, and design evaluation information representing errors or problems. For example, the information processing system 10 inputs the design drawing and corresponding design metadata to the generation AI, along with a prompt requesting the generation of design description information and / or design evaluation information. The generation AI uses the input prompt, the design drawing, and the corresponding design metadata to generate the design description information and / or design evaluation information. Because the generation AI can understand the functions and operations of the components shown in the design drawing using the design metadata, it can generate more appropriate design description information and / or design evaluation information compared to cases where design metadata is not provided. By generating and presenting such appropriate design description information and / or design evaluation information to the user, the user can easily confirm whether the design drawing has been generated as intended, and whether there are any areas that need improvement.

[0058] Furthermore, when the information processing system 10 generates a design drawing, it may use the generation AI to add, delete, or modify components that are constituent elements of the design drawing, based on the design explanation information and / or design evaluation information, in order to generate a design drawing that avoids errors or problems.

[0059] Furthermore, when the information processing system 10 generates a design drawing, it may use a generation AI to generate design test information related to testing the design described in the design drawing. For example, the information processing system 10 inputs a design drawing and corresponding design metadata to the generation AI, along with a prompt requesting the generation of design test information. The generation AI uses the input prompt, the design drawing, and the corresponding design metadata to generate the design test information. Because the generation AI can understand the functions and operations of the components shown in the design drawing using the design metadata, it can generate more appropriate design test information compared to cases where design metadata is not provided. By using such appropriate design test information, it is possible to accurately test whether the design described in the design drawing is as intended.

[0060] Furthermore, when the information processing system 10 generates a design drawing, it may also perform a test of the design drawing using a generation AI based on the design test information.

[0061] Figure 8 is a block diagram illustrating an example of the application of the information processing system 10 in an embodiment. In this figure, the information processing system 10 is communicated with the generation AI system 20 and the user terminal 30 via a communication network model such as the Internet. As shown in this figure, the information processing system 10 may be communicated with multiple user terminals 30.

[0062] The generative AI system 20 is a system equipped with a language model as a generative AI, and performs natural language processing based on prompts, blueprints, and design metadata notified from an external source. In this embodiment, the generative AI system 20 receives prompts, blueprints, and design metadata transmitted from the information processing system 10, and generates a response corresponding to the prompt by performing natural language processing based on the received prompts, blueprints, and design metadata. The generative AI system 20 transmits the generated response to the information processing system 10.

[0063] The generation AI system 20 includes, for example, an input unit 22, an output unit 24, a control unit 26, and a storage unit 28. The input unit 22 receives information notified from the outside, such as prompts, blueprints, and design metadata corresponding to the blueprints. The output unit 24 outputs information to the outside. For example, the output unit 24 outputs the answer generated by the generation AI to the outside. The control unit 26 includes an operation acquisition unit 260 and a control unit 261. The operation acquisition unit 260 acquires operations for the generation AI, in this case the prompts, blueprints, and design metadata corresponding to the blueprints input to the input unit 22. The control unit 261 controls the generation AI. The control unit 261 causes the generation AI to generate an answer corresponding to the prompt by having the generation AI input the prompts, blueprints, and design metadata corresponding to the blueprints acquired by the operation acquisition unit 260. The storage unit 28 includes a model storage unit 280 and an external information storage unit 281. The model storage unit 280 stores model information of the model corresponding to the generation AI. The external information storage unit 281 stores information that exists externally, such as the Web, and / or information for accessing externally existing information, such as URLs (Uniform Resource Locators).

[0064] The user terminal 30 is a terminal device operated by a user who performs work related to design drawings using the information processing system 10, and is, for example, a PC (personal computer), smartphone, or tablet terminal. The user terminal 30 accesses the information processing system 10, refers to the design drawings stored in the information processing system 10, and performs operations to request an explanation of the function or operation of a specific component in the referenced design drawing, or an explanation of the operation of the design drawing as a whole.

[0065] The user terminal 30 includes, for example, an input unit 32, an output unit 34, a control unit 36, and a storage unit 38. The input unit 32 receives information indicating operations on the user terminal 30. The output unit 34 displays images and outputs sounds such as voice and alarm sounds. The control unit 36 ​​includes an operation acquisition unit 360 and a control unit 361. The operation acquisition unit 360 acquires information indicating operations on the user terminal 30, in this case, touch operations entered into the input unit 32. The control unit 361 controls the user terminal 30. The control unit 361 enables tasks related to design drawings via the information processing system 10 by displaying information in accordance with the operations acquired by the operation acquisition unit 260. The storage unit 38 includes an application storage unit 380 and an authentication information storage unit 381. The application storage unit 380 stores application programs. These application programs are programs for using the design support services provided by the information processing system 10. The authentication information storage unit 381 stores authentication information to prove that the user is eligible to use the design support services.

[0066] The information processing system 10 includes an input unit 12, an output unit 14, a control unit 16, and a storage unit 18. The input unit 12 receives information notified from an external source, such as a login request to use a design support service, a request to view design drawings via the design support service, and a request for an explanation of the function of components in the design drawings. The output unit 14 outputs information to the outside. For example, the output unit 14 sends a response to a login request to the user terminal 30 to use the design support service notified by the user terminal 30. The output unit 14 also sends a response to a request for an explanation of the function of a component in a design drawing, notified by the user terminal 30, to the user terminal 30.

[0067] The control unit 16 includes a management unit 160, an input control unit 162, an auxiliary input unit 164, an output control unit 166, a design unit 170, and a design storage unit 172.

[0068] The management unit 160 manages the design drawings. The management unit 160 manages the design drawings, for example, by storing the design drawings in the storage unit 18. The input control unit 162 receives input information related to the design drawing. This input information is text information entered into prompts and includes information such as questions, requests, and instructions regarding the circuit diagram and its components.

[0069] The auxiliary input unit 164 receives the design drawing and design metadata. The design metadata represents an algorithm for generating a path that takes into account the processing order or processing conditions of the components in the design drawing. The auxiliary input unit 164 directly or indirectly inputs the design metadata corresponding to the input information received by the input control unit 162 to the generating AI. The auxiliary input unit 164 may also be configured to input design metadata indicating that attention should be paid to connection relationships when a design drawing is input. The auxiliary input unit 164 may also be configured to input design metadata representing the connection relationships between components when a design drawing is input. The auxiliary input unit 164 may also be configured to accept design metadata, which represents an algorithm for generating a path that takes into account the processing order or processing conditions of multiple connected components, when a design drawing is input.

[0070] The output control unit 166 outputs output information output from the trained model. The trained model here corresponds to the generated AI. The output control unit 166 outputs output information based on the path generated based on the design metadata, as shown in the example answer when design metadata (part 2) is added, from the trained model into which the design drawing, design metadata, and input information have been input. For example, the input control unit 162 obtains input information from the user via the user terminal 30. The auxiliary input unit 164 reads a design drawing and design metadata from the storage unit 18 according to the input information. The information processing system 10 transmits the read design drawing and design metadata, as well as the input information input to the input control unit 162, to the generation AI system 20. Based on the design drawing, design metadata, and input information notified by the information processing system 10, the generation AI system 20 causes the generation AI to generate a response corresponding to the input information, and transmits the generated response to the information processing system 10. The information processing system 10 transmits the response to the user terminal 30 by outputting output information via the output control unit 166.

[0071] The design unit 170 generates a design drawing based on user input. The design drawing is a circuit diagram or ladder diagram, in which components that receive or output electrical signals or control signals are connected, and the design drawing contains the string representing the name of each component. The design unit 170 also outputs design metadata. The design metadata includes the names of the components and represents the connection relationships between the components. The design storage unit 172 stores design drawings and design metadata.

[0072] For example, the management unit 160 manages the design drawings generated by the design unit 170 and the design metadata corresponding to those design drawings. The auxiliary input unit 164 inputs the design metadata corresponding to the design drawings generated by the design unit 170 when inputting those design drawings.

[0073] The storage unit 18 includes a design drawing storage unit 180 and a design metadata storage unit 182. The design drawing storage unit 180 stores information indicating a design drawing, for example, image information indicating a design drawing. The design metadata storage unit 182 stores design metadata. The design metadata storage unit 182 stores, for example, one or more pieces of design metadata corresponding to a design drawing, in association with that design drawing.

[0074] The memory unit 18 stores control programs and various data that control each part of the information processing system 10. For example, the memory unit 18 is composed of components such as DRAM (Dynamic Random Access Memory), EEPROM (Electrically Erasable Programmable Read Only Memory), flash ROM, HDD (Hard Disk Drive), and SSD (Solid State Drive).

[0075] As described above, the information processing system 10 according to the embodiment comprises a management unit 160, an auxiliary input unit 164, an input control unit 162, and an output control unit 166. The management unit 160 manages the design drawings. The auxiliary input unit 164 receives the design drawings and design metadata as input. The design metadata represents an algorithm for generating a path that takes into account the processing order or processing conditions of the components in the design drawing. The input control unit 162 receives input information related to the design drawings. The output control unit 166 outputs output information based on the path generated based on the design metadata from the trained model into which the design drawings, design metadata, and input information have been input. As a result, the information processing system 10 according to the embodiment can generate a path based on the design metadata corresponding to the design drawings using the generated AI as a trained model, and generate an answer that appropriately explains the operation or function, etc., based on the generated path. For this reason, design support can be provided using the generated AI.

[0076] While embodiments of this disclosure have been described above, this disclosure is not limited to the configurations of the embodiments described above, and can be combined as appropriate within a range that does not contradict each other.

[0077] The model generation device 101 and the equipment management system 1 described above have an internal computer system. The processing steps of the above-described process are stored in program form on a computer-readable recording medium, and the above-described process is performed when the computer reads and executes this program. Here, a computer-readable recording medium refers to a magnetic disk, magneto-optical disk, CD-ROM, DVD-ROM, semiconductor memory, etc. Alternatively, this computer program may be distributed to a computer via a communication line, and the computer that receives the distribution may execute the program. [Explanation of Symbols]

[0078] 1…Equipment management system, 10…Information processing system, 20…Generative AI system, 30…User terminal, 160…Management unit, 162…Input control unit, 164…Auxiliary input unit, 166…Output control unit, 170…Design unit, 172…Design storage unit

Claims

1. The management department that manages the design drawings, The aforementioned design drawing and an auxiliary input unit that inputs design metadata representing an algorithm for generating a path that takes into account the processing order or processing conditions of the components in the aforementioned design drawing, An input control unit that receives input information related to the aforementioned design drawings, An output control unit outputs output information based on a path generated based on the design metadata, from a trained model into which the design drawing, the design metadata, and the input information have been input. An information processing system equipped with the following features.

2. The auxiliary input unit inputs the design metadata indicating that attention should be paid to connection relationships when the design drawing is input. The information processing system according to claim 1.

3. The auxiliary input unit, when the design drawing is input, inputs the design metadata representing the connection relationships between the components. The information processing system according to claim 1.

4. When the design drawing is input, the auxiliary input unit receives the design metadata, which represents an algorithm for generating a path that takes into account the processing order or processing conditions of the multiple connected components. The output control unit outputs the output information based on the path generated based on the design metadata. The information processing system according to claim 1.

5. A design unit generates a design drawing in which components that input or output electrical signals or control signals are connected, and which contains a string representing the name of each of the components, based on input from the user, and outputs design metadata including the names of the components, which represents the connection relationships between the components. A design storage unit that stores the design drawings and the design metadata corresponding to the design drawings, Equipped with, The management unit manages the design drawings and the design metadata corresponding to the design drawings. The auxiliary input unit inputs the design metadata corresponding to the design drawing when the design drawing is input. The information processing system according to claim 1.

6. Steps for managing design drawings, The steps include inputting the aforementioned design drawing and design metadata representing an algorithm for generating a path that takes into account the processing order or processing conditions of the components in the aforementioned design drawing, A step of receiving input information related to the aforementioned design drawings, A step of outputting output information based on a path generated based on the design metadata, from a trained model into which the design drawing, the design metadata, and the input information have been input. Information processing methods.

7. Computers, Management department that manages the design drawings, An auxiliary input unit that inputs the aforementioned design drawing and design metadata representing an algorithm for generating a path that takes into account the processing order or processing conditions of the components in the aforementioned design drawing. An input control unit that receives input information related to the aforementioned design drawing, An output control unit outputs output information based on a path generated based on the design metadata, from a trained model into which the design drawing, the design metadata, and the input information have been input. A program designed to function as such.

Images