An operation evidence verification method for an industrial engineering software agent, a gating system, an agent, a non-transitory computer readable storage medium, an electronic device, and a computer program product

Abstract

The present application relates to the technical field of industrial engineering software intelligence, in particular to an operation evidence verification method for an industrial engineering software agent, a gating system, an agent, a non-transitory computer readable storage medium, an electronic device and a computer program product, wherein the method comprises: after the agent performs semantic recognition on a to-be-executed engineering automation task, establishing and initializing an operation evidence gating state through the gating system when the engineering automation task includes an operation evidence closed-loop requirement; after the agent sequentially performs compilation, operation and operation evidence reading verification operations on the engineering automation task, recording the operation evidence gating state after completing each operation; and before the agent outputs an execution condition reply, checking the operation evidence gating state through the gating system; when the preset condition is not met, constructing a continuation input to trigger the agent to complete the corresponding operation in the next execution round and then output the final execution condition reply.

Description

Technical Field

[0001] This invention relates to the field of intelligent industrial engineering software technology, and in particular to a method for verifying operational evidence of intelligent agents in industrial engineering software, a gating system, an intelligent agent, a non-transitory computer-readable storage medium, an electronic device, and a computer program product. Background Technology

[0002] With the development of intelligent industrial engineering software, intelligent agents are widely used for the autonomous execution of engineering automation tasks. Intelligent agents can receive task descriptions and autonomously invoke compilation and execution tools to complete the compilation, execution, and output of engineering objects. However, in practical applications, after completing compilation and execution, intelligent agents typically rely solely on their own memory or reasoning to generate execution status responses. This can lead to the agent outputting false responses that do not match the actual situation. For example, it might claim that the task was successfully executed even if compilation or execution failed, or it might fabricate execution results without reading execution evidence. If only the agent's generation process itself is relied upon, the responses generated by the agent may still include unverified execution conclusions. When these execution status responses are passed to downstream decision-making or recording stages, they can cause problems such as misleading decisions and distorted records.

[0003] Therefore, existing technologies have the problem that agents typically rely solely on their own memory or reasoning to generate execution status responses, which may result in false or erroneous responses generated by the agents, leading to misleading downstream decisions and distorted records.

[0004] It should be noted that the information disclosed in this background section is only for understanding the background technology of this application concept and is not considered to constitute prior art information. Summary of the Invention

[0005] The purpose of this invention is to provide a method for verifying operational evidence of intelligent agents in industrial engineering software, comprising: After the intelligent agent performs semantic recognition on the engineering automation task to be executed, the gating system determines whether the engineering automation task includes a closed-loop requirement for running evidence based on the semantic recognition result. If it includes a closed-loop requirement for running evidence, the running evidence gating state is established and initialized. After the intelligent agent sequentially performs compilation, execution, and execution evidence reading and verification operations on the engineering automation task, for each completed operation, the gating system records the execution evidence gating state based on the execution result of that operation; and Before the agent outputs the execution status of the engineering automation task, the gating system checks the gating status of the running evidence. If the status value does not indicate running failure but indicates missing evidence, a continuation input is constructed and sent to the agent to trigger the agent to complete the operation corresponding to the missing evidence in the next execution round before outputting the final execution status.

[0006] In another aspect, the present invention also provides a gating system for intelligent agents in industrial engineering software, including a computer device configured to include: The state creation module is configured to determine whether the engineering automation task to be executed includes a running evidence closed-loop requirement based on the semantic recognition result after the agent performs semantic recognition on the engineering automation task to be executed. If the running evidence closed-loop requirement is included, the running evidence gating state is established and initialized. The state maintenance module is configured to record the running evidence gating state based on the execution result of each operation after the intelligent agent sequentially performs compilation, running, and running evidence reading and verification operations on the engineering automation task. The final gating module is configured to check the running evidence gating status before the agent outputs the execution status response of the engineering automation task; if the status value does not indicate running failure but indicates missing evidence, a continuation input is constructed and sent to the agent to trigger the agent to complete the operation corresponding to the missing evidence in the next execution round before outputting the final execution status response.

[0007] Thirdly, the present invention also provides an industrial engineering software intelligent agent, comprising: The semantic recognition module is configured to perform semantic recognition on the engineering automation task to be executed in order to determine whether the engineering automation task includes a closed-loop requirement for operational evidence. The task execution module is configured to sequentially perform compilation, execution, and execution evidence reading and verification operations on the engineering automation task; The response generation module is configured to generate an execution status response for the engineering automation task, and upon receiving a continuation input, to suspend the generation of the current execution status response and discard the generated content, and in the next execution round, first complete the operation corresponding to the missing evidence, and then output the final execution status response.

[0008] Fourthly, the present invention also provides a non-transitory computer-readable storage medium storing instructions that, when executed by a processor, cause the processor to perform the aforementioned method for verifying operational evidence of intelligent agents in industrial engineering software.

[0009] Fifthly, the present invention also provides an electronic device, comprising: Non-transitory computer-readable storage medium; processor; The non-transitory computer-readable storage medium stores instructions that, when executed by a processor, cause the processor to perform the aforementioned method for verifying operational evidence of intelligent agents in industrial engineering software.

[0010] In a sixth aspect, the present invention also provides a computer program product, including instructions that, when executed by a processor, cause the processor to perform the aforementioned method for verifying operational evidence of an industrial engineering software intelligent agent.

[0011] The beneficial effects of this invention are that the operational evidence verification method for industrial engineering software agents establishes an operational evidence gating state when the engineering automation task includes a closed-loop requirement for operational evidence, and checks the operational evidence gating state before the agent generates an execution status response. This ensures that the agent can only output a final response when the operational evidence is completely closed-loop. This mechanism eliminates the possibility of the agent fabricating execution results from memory, ensuring that every execution status response is supported by genuine operational evidence. This not only reduces the risk of the agent outputting false responses and misleading downstream decisions, but also prevents unverified operational conclusions from directly entering the recording or decision-making process.

[0012] Other features and advantages of the invention will be set forth in the following description, and will be apparent in part from the description, or may be learned by practicing the invention. The objects and other advantages of the invention are realized and obtained through the structures particularly pointed out in the description and the drawings.

[0013] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, preferred embodiments are described below in detail with reference to the accompanying drawings. Attached Figure Description

[0014] To more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the specific embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0015] Figure 1 The diagram illustrates the steps of a method for verifying operational evidence of intelligent agents in industrial engineering software, as illustrated in some embodiments. Figure 2 A schematic diagram of a gating system for intelligent agents in industrial engineering software, as illustrated in some embodiments, is shown. Figure 3 A schematic diagram of the industrial engineering software intelligent agent involved in some embodiments is shown; Figure 4 Block diagrams of electronic devices involved in some embodiments are shown. Detailed Implementation

[0016] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0017] In practical applications, after compilation and execution, agents typically rely solely on their own memory or reasoning to generate responses regarding the execution status. This can lead to false responses that do not reflect reality, such as claiming task success despite compilation or execution failures, or fabricating results without verifying the execution process. Furthermore, relying solely on the agent's generation process itself may result in responses containing unverified conclusions. When these responses are passed to downstream decision-making or recording stages, they can cause misleading decisions and distorted records.

[0018] Therefore, at least one embodiment provides a method for verifying operational evidence for an industrial engineering software agent, comprising: after the agent performs semantic recognition on the engineering automation task to be executed, determining whether the engineering automation task includes operational evidence closed-loop requirements based on the semantic recognition results through a gating system; if operational evidence closed-loop requirements are included, establishing and initializing an operational evidence gating state; after the agent sequentially performs compilation, execution, and operational evidence reading and verification operations on the engineering automation task, recording the operational evidence gating state based on the execution result of each operation completed through the gating system; and checking the operational evidence gating state through the gating system before the agent outputs the execution status response of the engineering automation task; if the status value does not indicate execution failure but indicates missing evidence, constructing a continuation input and sending it to the agent to trigger the agent to first complete the operation corresponding to the missing evidence in the next execution round, and then outputting the final execution status response.

[0019] The operational evidence verification method for industrial engineering software agents in this embodiment establishes an operational evidence gating state when the engineering automation task includes a closed-loop operational evidence requirement. This gating state is checked before the agent generates an execution status response, ensuring that the agent can only output a final response when the operational evidence is completely closed-loop. This mechanism eliminates the possibility of the agent fabricating execution results from memory, guaranteeing that every execution status response is supported by genuine operational evidence. This not only reduces the risk of the agent outputting false responses and misleading downstream decisions but also prevents unverified operational conclusions from directly entering the recording or decision-making process.

[0020] The various non-limiting embodiments of the present disclosure will now be described in detail with reference to the accompanying drawings.

[0021] like Figure 1 As shown, some embodiments provide a method for verifying operational evidence of intelligent agents in industrial engineering software, including: Step S101: After the intelligent agent performs semantic recognition on the engineering automation task to be executed, the gating system determines whether the engineering automation task includes a closed-loop requirement for running evidence based on the semantic recognition result. If it includes a closed-loop requirement for running evidence, the running evidence gating state is established and initialized. Step S102: After the intelligent agent sequentially performs compilation, running and running evidence reading and verification operations on the engineering automation task, after each operation is completed, the running evidence gating status is recorded by the gating system according to the execution result of the operation. Step S103: Before the agent outputs the execution status response of the engineering automation task, the gating system checks the gating status of the running evidence. If the status value does not indicate running failure but indicates missing evidence, a continuation input is constructed and sent to the agent to trigger the agent to complete the operation corresponding to the missing evidence in the next execution round before outputting the final execution status response.

[0022] Specifically, the missing evidence includes: missing any one of the following: successful compilation, successful execution, or execution evidence reading verification; the operation corresponding to the missing evidence includes: compilation, execution, or execution evidence reading verification.

[0023] In some embodiments, the running evidence gating state includes: evidence requirement state, compilation success state, running success state, running failure state, evidence path binding state, and running evidence reading verification state; the engineering automation task includes: at least one running object; wherein, the method for recording the running evidence gating state according to the execution result of each operation includes: when the agent calls the compilation tool to compile the current running object in the engineering automation task and the compilation is successful, recording the compilation success state; after the agent calls the running tool to run the current running object, recording the running success state or running failure state according to the running result; when the running is successful and the running result contains a running evidence path, binding the running evidence path with the current running object, generating evidence path binding data and recording the evidence path binding state; after the agent calls the evidence reading tool to read the running evidence of the current running object, verifying whether the read running evidence path is consistent with the bound running evidence path and whether the read running evidence satisfies the non-empty and parsable conditions, and recording the running evidence reading verification state according to the verification result.

[0024] Specifically, parsable means that an intelligent agent can successfully convert the read operational evidence content into structured data that it can process internally, according to a predefined format or protocol.

[0025] Specifically, the evidence requirement status is used to mark whether evidence closure is needed. When it is determined that the engineering automation task includes the requirement for running evidence closure, the evidence requirement status is initialized to be marked as needed; the compilation success status is used to mark whether the compilation was successful; the running success status is used to mark whether the running was successful; the running failure status is used to mark whether the running failed; wherein, the running success status and the running failure status are mutually exclusive and cannot be true at the same time; the evidence path binding status is used to mark whether the running evidence path is bound to the corresponding running object; the running evidence reading and verification status is used to mark whether the running evidence has been read and verified.

[0026] In some embodiments, the method for determining whether the engineering automation task includes a runtime evidence closed-loop requirement based on the semantic recognition result includes: if the semantic recognition result includes compilation action semantics, runtime action semantics, and runtime evidence readback semantics, then the engineering automation task includes a runtime evidence closed-loop requirement; the runtime evidence closed-loop requirement includes: compilation requirement, runtime requirement, and runtime evidence readback requirement.

[0027] Specifically, the semantic recognition results also include: execution evidence path semantics and execution object identifiers. Among them, execution evidence path semantics refers to the language expression patterns that appear in the text description of the task and explicitly indicate the storage location of execution evidence or the target to be read.

[0028] In some embodiments, the continue run input includes: a list of missing evidence, a set of tools that are allowed to be invoked preferentially, the name of the run evidence path to be read, and a constraint flag that prohibits premature output of execution status responses.

[0029] Specifically, the missing evidence list tells the agent which evidence is currently missing from compilation success, execution success, or execution evidence reading and verification; the set of tools that can be prioritized tells the agent which tools (such as compilation tools, execution tools, and evidence reading tools) can be called in the next execution round; the expected execution evidence path name provides the agent with a clear evidence location identifier, ensuring that the agent can accurately find the required execution evidence in the next execution round; and the constraint flag that prohibits premature output of execution status responses forces the agent to complete the operation corresponding to the missing evidence first and cannot directly output a response.

[0030] In some embodiments, when the engineering automation task includes multiple running objects, an independent running evidence gating state is established for each running object, and the running evidence gating state of each running object is checked one by one before generating an execution status response. When recompilation or engineering task configuration modification is detected, the previously recorded compilation success state, running success state, and running evidence reading verification state of the affected running object are set to invalid states. When running object switching or evidence storage area switching is detected, the previously recorded running success state and running evidence reading verification state of the affected running object are set to invalid states. In the subsequent check before the execution status response, if the status value of the affected running object does not indicate running failure but indicates the lack of any evidence among compilation success, running success, or running evidence reading verification, the corresponding continuation input is constructed to the agent to trigger the agent to re-execute the corresponding compilation, running, and running evidence reading verification operations for the running object.

[0031] Specifically, this embodiment establishes an independent gating state for operational evidence for each operational object and checks the integrity of the evidence for each operational object one by one before generating an execution status response. This ensures that the evidence loop for each object is independent and complete in multi-operational object scenarios, and that evidence is not mixed due to object switching or configuration changes.

[0032] In some embodiments, the evidence path binding data includes: running object identifier, tool call round, evidence storage area name, evidence file size, and pending read action identifier.

[0033] Specifically, the running object identifier is used to identify which running object the evidence belongs to; the tool call round is used to identify which tool call generated the evidence (distinguishing multiple runs of the same object); the evidence storage area name is used to identify which evidence storage area the evidence is located in; the evidence file size is used to identify the size of the evidence file, which helps to verify the consistency of the evidence file; and the pending read action identifier is used to identify whether there is a pending read action.

[0034] The following describes in detail the method for verifying the operational evidence of industrial engineering software intelligent agents in this embodiment, using the operational evidence verification process for single-running objects and multiple-running objects respectively.

[0035] Case 1: This case uses the engineering automation task TaskTicket-RG-001 as an example. This task includes a run object RunCell-214, which is used to perform an industrial simulation calculation.

[0036] First, after receiving the engineering automation task TaskTicket-RG-001, the agent performs semantic recognition on the text description of the task. The agent recognizes the following semantic elements in the text description: the running object is identified as RunCell-214; the compilation action semantic is to call the compilation tool to compile RunCell-214; the running action semantic is to call the running tool to run RunCell-214; the running evidence path semantic is that the running result log is stored at evidence: / / run / RunCell-214 / session-042 / runtime.log; and the running evidence readback semantic is to request to read and verify the running result log.

[0037] Since the aforementioned semantics simultaneously include compilation requirements, runtime requirements, and runtime evidence readback requirements, the gating system determines that the engineering automation task TaskTicket-RG-001 includes runtime evidence closed-loop requirements. The runtime evidence gating state is then established and initialized, and the initialization results are shown in Table 1 below:

[0038] Then, the agent calls the compilation tool to compile the runtime object RunCell-214. The compilation tool returns a successful compilation result (BuildTicket-214-07), and the gating system records the successful compilation status as TRUE, indicating that the compilation was successful.

[0039] The agent invokes the runtime tool to run the running object RunCell-214. The runtime tool returns a successful execution result (RunTicket-214-08) and returns the runtime evidence path evidence: / / run / RunCell-214 / session-042 / runtime.log. The gating system records this runtime evidence path and records the successful execution status as TRUE, indicating successful execution.

[0040] Since the operation was successful, the gating system binds the runtime evidence path evidence: / / run / RunCell-214 / session-042 / runtime.log to the currently running object RunCell-214 and generates evidence path binding data. The evidence path binding data includes the data shown in Table 2 below:

[0041] At this point, the evidence reading and verification status is still FALSE, indicating that the evidence reading and verification has not yet been completed.

[0042] Next, the agent attempts to generate an execution status response. Before generating the response, the gating system checks the running evidence gating status and finds that the running failure status does not indicate failure, but the running evidence read verification status is FALSE, indicating a lack of evidence for running evidence read verification. Therefore, the agent is triggered to abandon response generation and construct a continuation input. In the continuation input, the missing evidence list includes: the running evidence read verification status; the set of tools allowed to be invoked first includes: the evidence reading tool ReadTool; the expected running evidence path name is: evidence: / / run / RunCell-214 / session-042 / runtime.log; and the constraint flag for prohibiting premature output of the execution status response is TRUE, indicating that premature output of the execution status response is prohibited.

[0043] After receiving the continuation input in the next execution round, the agent invokes the evidence reading tool to read the runtime evidence from the runtime evidence path evidence: / / run / RunCell-214 / session-042 / runtime.log. The gating system determines that the read runtime evidence path belongs to the evidence storage area run / RunCell-214 / session-042 / corresponding to the current running object, and that the runtime evidence path is exactly the same as the runtime evidence path evidence: / / run / RunCell-214 / session-042 / runtime.log bound to the current running object RunCell-214, thus the verification is successful. In addition, the evidence reading tool returns 36 lines of parsable text content, satisfying the non-empty and parsable conditions. Therefore, the gating system records the runtime evidence reading verification status as verified (TRUE) and updates the pending action flag (pendingRead=FALSE).

[0044] Then, the gating system checks the gating status of the running evidence again, confirming that the compilation success status is TRUE, the running success status is TRUE, the running failure status is FALSE (no running failure indicated), and the running evidence reading verification status is TRUE (verified). The evidence is complete, meaning that the current running object has completed a complete running evidence loop, triggering the agent to generate the final execution status response. At the same time, the gating system generates an auditable record, including the values ​​of all the above-mentioned running evidence gating statuses, as well as the evidence path binding data and the continuation decision result (continuation has been triggered and gating has been released after reading verification is completed).

[0045] Case 2: This case takes the engineering automation task TaskTicket-RG-516 as an example. This task includes two running objects: RunCell-516A and RunCell-516B.

[0046] Assume that the running object RunCell-516A has completed a full execution evidence loop in tool call round turn-18, with its compilation success status being TRUE, execution success status being TRUE, and execution evidence reading verification status being TRUE (verified). The bound execution evidence path is evidence: / / run / RunCell-516A / session-003 / runtime.log. The running object RunCell-516B has not yet started execution; its compilation success status is FALSE, execution success status is FALSE, and execution evidence reading verification status is FALSE.

[0047] The two running objects each have their own independent running evidence gating state and do not interfere with each other.

[0048] Scenario 1: Project task configuration modification trigger status failure After the user successfully ran RunCell-516A, they modified the project task configuration (such as changing compilation parameters). This modification triggered a recompilation (BuildTicket-516-19). The gating system detected the recompilation event and performed status invalidation processing on the RunCell-516A affected by the compilation: the compilation success status, running success status, and running evidence reading verification status of RunCell-516A were all set to FALSE. However, the running evidence gating status of RunCell-516B remained unaffected and unchanged.

[0049] Scenario 2: State switching triggered by running object fails. The agent switches the currently running object from RunCell-516A to RunCell-516B. The system detects the running object switching event and sets both the previously successful running status and the running evidence reading verification status of the running object to be switched, RunCell-516A, to FALSE.

[0050] In this scenario, the agent attempts to generate a response regarding the execution status of the engineering automation task TaskTicket-RG-516. Before generating the response, the gating system checks the gating status of the execution evidence for each running object. The check results are shown in Table 3.

[0051] Since both running objects lack the necessary evidence, the gating system constructs a continuation input, requiring the agent to proceed in the next execution round: For the running object RunCell-516A, the running tool and the evidence reading tool are invoked again to rerun and read the verification running evidence; for the running object RunCell-516B, the compilation tool is invoked first to compile, then the running tool is invoked to run, and finally the evidence reading tool is invoked to read the verification running evidence.

[0052] In some embodiments, when the run failure status indicates a run failure, a failure description consistent with the run failure diagnostic information returned by the run tool is generated as a response to the execution status.

[0053] Specifically, the execution failure status takes precedence over evidence missing checks. When the execution failure status indicates failure, regardless of the compilation success status, execution success status, or execution evidence reading and verification status, a failure description is directly generated as a response to the execution status. Continued input construction is not performed, avoiding resource waste and logical errors caused by blindly continuing evidence collection when the execution has clearly failed. At the same time, it provides users with accurate failure reasons and diagnostic information, facilitating subsequent troubleshooting and retries.

[0054] The following example illustrates the process of generating a failure description consistent with the failure diagnostic information returned by the running tool as a response to the execution status when the failure status indicates that the execution has failed.

[0055] This case study uses the engineering automation task TaskTicket-RG-617 as an example.

[0056] Suppose the gating system establishes a gating state for runtime evidence based on the semantic recognition results of the agent, and the agent successfully compiles the code using the compilation tool. Subsequently, it calls the runtime tool, which returns a runtime failure result (RuntimeFault-RG-617) along with runtime failure diagnostic information: "Memory overflow error, stack trace: [detailed stack information]".

[0057] The gating system records the running failure status as TRUE and the running success status as FALSE based on the running results.

[0058] The agent attempts to generate an execution status response. When the gating system checks the gating status of the execution evidence, it finds a failure status indicating execution failure. According to the rules, it does not construct a continuation input, but instead generates a failure statement consistent with the execution failure diagnostic information as the execution status response, as follows: "TaskTicket-RG-617 failed to execute: The runtime tool returned a memory overflow error (RuntimeFault-RG-617), stack trace: [detailed stack information]. It is recommended to check the runtime resource configuration and try again." In some embodiments, an auditable record is also generated when checking the gating status of running evidence; the auditable record includes: evidence requirement status, compilation success status, running success status, running failure status, evidence path binding status, running evidence reading and verification status, evidence path binding data, and continued running decision results.

[0059] Specifically, this embodiment generates auditable records when checking the gating status of operational evidence, making the entire evidence loop of each task execution traceable and auditable, thus meeting the stringent requirements of industrial engineering software for operational compliance and quality traceability.

[0060] The following case study illustrates in detail the process of generating auditable records.

[0061] Taking the scenario in Case 1 as an example, when generating the final response on the execution status, the system also generates auditable records, including the information shown in Table 4:

[0062] This auditable record can be used for subsequent tracing, debugging, and verification, ensuring that the entire operational evidence verification process is auditable and traceable.

[0063] like Figure 2 As shown, some embodiments also provide a gating system for industrial engineering software intelligent agents, including a computer device configured to include: The state creation module is configured to determine whether the engineering automation task to be executed includes a running evidence closed-loop requirement based on the semantic recognition result after the agent performs semantic recognition on the engineering automation task to be executed. If the running evidence closed-loop requirement is included, the running evidence gating state is established and initialized. The state maintenance module is configured to record the running evidence gating state based on the execution result of each operation after the intelligent agent sequentially performs compilation, running, and running evidence reading and verification operations on the engineering automation task. The final gating module is configured to check the running evidence gating status before the agent outputs the execution status response of the engineering automation task; if the status value does not indicate running failure but indicates missing evidence, a continuation input is constructed and sent to the agent to trigger the agent to complete the operation corresponding to the missing evidence in the next execution round before outputting the final execution status response.

[0064] The specific implementation functions of the state creation module, state maintenance module, and final gating module can be found in the aforementioned content on the operational evidence verification method for intelligent agents in industrial engineering software, and will not be repeated here.

[0065] like Figure 3 As shown, some embodiments also provide an industrial engineering software intelligent agent, including: The semantic recognition module is configured to perform semantic recognition on the engineering automation task to be executed in order to determine whether the engineering automation task includes a closed-loop requirement for operational evidence. The task execution module is configured to sequentially perform compilation, execution, and execution evidence reading and verification operations on the engineering automation task; The response generation module is configured to generate an execution status response for the engineering automation task, and upon receiving a continuation input, to suspend the generation of the current execution status response and discard the generated content, and in the next execution round, first complete the operation corresponding to the missing evidence, and then output the final execution status response.

[0066] The specific implementation functions of the semantic recognition module, task execution module, and response generation module can be found in the aforementioned content on the operational evidence verification method for intelligent agents in industrial engineering software, and will not be repeated here.

[0067] The electronic devices in the embodiments of this disclosure are described below from the perspective of hardware processing: The embodiments disclosed herein do not limit the specific implementation of the electronic device.

[0068] like Figure 4 As shown, some embodiments also provide an electronic device, including: a processor, a non-transitory computer-readable storage medium, a communication bus, and a communication interface; wherein the processor, the non-transitory computer-readable storage medium, and the communication interface communicate with each other through the communication bus; the non-transitory computer-readable storage medium stores instructions that, when executed by the processor, cause the processor to execute the aforementioned method for verifying the operational evidence of intelligent agents in industrial engineering software.

[0069] In some embodiments, computer devices, automobiles ECU Vehicle controllers, domain controllers, simulation equipment, or calibration equipment can also be considered as electronic devices.

[0070] Figure 3 The diagram shows a schematic structure of an electronic device, which is for illustration only and does not constitute a limitation on the electronic device. The electronic device may include fewer or more components than shown, or may combine components, or may use different component arrangements.

[0071] In some embodiments, the communication interface may include RS -232、 RS -485、 USB (include Type - C Physical interfaces such as Ethernet are used to connect external devices or bus adapters; they may also include wired network interfaces such as Ethernet. Wi - Fi Wireless network interfaces such as Bluetooth are used to establish communication connections between computer devices and other electronic devices.

[0072] In some embodiments, non-transitory computer-readable storage media include, but are not limited to: flash memory, hard disk, magnetic storage, magnetic disk, optical disk, and card-type storage (e.g., multimedia cards, secure digital cards). SD (e.g., memory). In some embodiments, the storage medium can serve as an internal storage unit of a computer device, such as a built-in hard disk; in other embodiments, it can serve as an external storage device, such as a plug-in hard disk or a smart memory card. SMC ), secure digital SD Cards, flash memory cards, etc. Furthermore, the storage medium can also include both internal storage units and external storage devices. This storage medium can be used to store application software and various types of data (such as computer program code) installed on the computer device, and can also be used to temporarily store data that has been output or will be output.

[0073] In some embodiments, the processor may be a central processing unit (CPU). CPU ), controller, microcontroller, microprocessor or other data processing chip, used to run program code in a storage medium and / or process data, such as executing a computer program.

[0074] In some embodiments, the communication bus may be an input / output bus, such as a peripheral component interconnect (PCI). PCI ) bus or extended industry standard structure ( EISA This bus can be divided into address bus, data bus, and control bus, etc.

[0075] Optionally, the computer device further includes a user interface. The user interface may include a display, an input unit (e.g., a keyboard), and a standard wired and / or wireless interface. Optionally, the display (or display module) may be... led Monitor, LCD monitor, touch screen LCD monitor, or OLED A monitor (or display module), also known as a display screen or display unit, is used to display information processed by a computer device and present a visual user interface.

[0076] When the processor executes the program, it implements the above. Figure 1 The steps in the embodiment of the method for verifying operational evidence of intelligent agents in industrial engineering software are shown. Alternatively, the processor executes the computer program to implement the functions of each module or unit in the above-described device embodiments.

[0077] Some embodiments also provide a non-transitory computer-readable storage medium storing instructions that, when executed by a processor, cause the processor to perform the operational evidence verification method for industrial engineering software intelligent agents described above.

[0078] Please refer to the detailed description of the operational evidence verification method for intelligent agents in industrial engineering software; it will not be repeated here.

[0079] Some embodiments also provide a computer program product including instructions that, when executed by a processor, cause the processor to perform the aforementioned method for verifying operational evidence for industrial engineering software agents.

[0080] In the several embodiments provided in this application, it should be understood that the disclosed apparatus and methods can also be implemented in other ways. The apparatus embodiments described above are merely illustrative; for example, the flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of apparatus, methods, and computer program products according to various embodiments of the present invention. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code, which includes one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions marked in the blocks may occur in a different order than those marked in the drawings. For example, two consecutive blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in a block diagram and / or flowchart, and combinations of blocks in block diagrams and / or flowcharts, can be implemented using a dedicated hardware-based system that performs the specified function or action, or using a combination of dedicated hardware and computer instructions.

[0081] In addition, the functional modules in the various embodiments of the present invention can be integrated together to form an independent part, or each module can exist independently, or two or more modules can be integrated to form an independent part.

[0082] If the aforementioned functions are implemented as software functional modules and sold or used as independent products, they can be stored in a non-transitory computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention.

[0083] Based on the above-described preferred embodiments of the present invention, and through the foregoing description, those skilled in the art can make various changes and modifications without departing from the inventive concept. The technical scope of this invention is not limited to the contents of the specification, but must be determined according to the scope of the claims.

Claims

1. A method for verifying operational evidence of intelligent agents in industrial engineering software, characterized in that, include: After the intelligent agent performs semantic recognition on the engineering automation task to be executed, the gating system determines whether the engineering automation task includes a closed-loop requirement for running evidence based on the semantic recognition result. If it includes a closed-loop requirement for running evidence, the running evidence gating state is established and initialized. After the intelligent agent sequentially performs compilation, running, and running evidence reading and verification operations on the engineering automation task, the running evidence gating status is recorded by the gating system based on the execution result of each operation after each operation is completed. as well as Before the agent outputs the execution status of the engineering automation task, the gating system checks the gating status of the running evidence. If the status value does not indicate running failure but indicates missing evidence, a continuation input is constructed and sent to the agent to trigger the agent to complete the operation corresponding to the missing evidence in the next execution round before outputting the final execution status.

2. The method according to claim 1, characterized in that, The operational evidence gating states include: evidence requirement state, compilation success state, operation success state, operation failure state, evidence path binding state, and operational evidence reading and verification state; The methods for recording the gating status of runtime evidence based on the execution results of each operation include: When the intelligent agent calls the compilation tool to compile the currently running object in the engineering automation task and the compilation is successful, the compilation success status is recorded; After the intelligent agent calls the running tool to run the current running object, it records the running success status or running failure status according to the running result; when the running is successful and the running result contains the running evidence path, it binds the running evidence path to the current running object, generates evidence path binding data and records the evidence path binding status; After the agent calls the evidence reading tool to read the running evidence of the currently running object, it verifies whether the reading running evidence path is consistent with the bound running evidence path and whether the reading running evidence satisfies the conditions of being non-empty and parsable, and records the running evidence reading verification status based on the verification results.

3. The method according to claim 1, characterized in that, Methods for determining whether the engineering automation task includes a closed-loop requirement for operational evidence based on semantic recognition results include: If the semantic recognition result includes compilation action semantics, running action semantics, and running evidence back-reading semantics, it is determined that the engineering automation task includes running evidence closed-loop requirements. The closed-loop requirements for runtime evidence include: compilation requirements, runtime requirements, and runtime evidence readback requirements.

4. The method according to claim 3, characterized in that, The missing evidence includes: missing any one of the following: successful compilation, successful execution, or execution evidence reading and verification. The operations corresponding to the missing evidence include: compiling, running, or running evidence reading and verification; and The inputs for the continued run include: a list of missing evidence, a set of tools that are allowed to be invoked first, the names of the expected execution evidence paths to be read, and constraint flags that prohibit premature output of execution status responses.

5. The method according to claim 2, characterized in that, When the execution failure status indicates that the execution has failed, a failure description consistent with the execution failure diagnostic information returned by the execution tool is generated as a response to the execution status.

6. The method according to claim 2, characterized in that, When the engineering automation task includes multiple running objects, an independent running evidence gating state is established for each running object, and the running evidence gating state of each running object is checked one by one before generating the execution status response. When recompilation or modification of project task configuration is detected, the previously recorded compilation success status, running success status, and running evidence reading verification status of the affected running object will be set to invalid status. When a change of running object or evidence storage area is detected, the previously recorded running success status and running evidence reading verification status of the affected running object will be set to invalid status. In the subsequent check before the execution status is responded to, if the status value of the affected running object does not indicate running failure but indicates that any evidence of successful compilation, successful running, or running evidence reading verification is missing, the corresponding continuation input is constructed to the agent to trigger the agent to re-execute the corresponding compilation, running, and running evidence reading verification operations on the running object.

7. The method according to claim 2, characterized in that, The evidence path binding data includes: running object identifier, tool call round, evidence storage area name, evidence file size, and pending read action identifier.

8. The method according to claim 2, characterized in that, Auditable records are also generated when checking the gating status of operational evidence; The auditable records include: evidence requirement status, compilation success status, execution success status, execution failure status, evidence path binding status, execution evidence reading and verification status, evidence path binding data, and continuation decision results.

9. The method according to claim 7, characterized in that, The semantic recognition result also includes: the semantics of the running evidence path and the identifier of the running object.

10. A gating system for intelligent agents in industrial engineering software, characterized in that, Includes a computer device, the computer device being configured to include: The state creation module is configured to determine whether the engineering automation task to be executed includes a running evidence closed-loop requirement based on the semantic recognition result after the agent performs semantic recognition on the engineering automation task to be executed. If the running evidence closed-loop requirement is included, the running evidence gating state is established and initialized. The state maintenance module is configured to record the running evidence gating state based on the execution result of each operation after the intelligent agent sequentially performs compilation, running, and running evidence reading and verification operations on the engineering automation task. The final gating module is configured to check the running evidence gating status before the agent outputs the execution status response of the engineering automation task; if the status value does not indicate running failure but indicates missing evidence, a continuation input is constructed and sent to the agent to trigger the agent to complete the operation corresponding to the missing evidence in the next execution round before outputting the final execution status response.

11. An intelligent agent for industrial engineering software, characterized in that, include: The semantic recognition module is configured to perform semantic recognition on the engineering automation task to be executed in order to determine whether the engineering automation task includes a closed-loop requirement for operational evidence. The task execution module is configured to sequentially perform compilation, execution, and execution evidence reading and verification operations on the engineering automation task; The response generation module is configured to generate an execution status response for the engineering automation task, and upon receiving a continuation input, to suspend the generation of the current execution status response and discard the generated content, and in the next execution round, first complete the operation corresponding to the missing evidence, and then output the final execution status response.

12. A non-transitory computer-readable storage medium, characterized in that, The system stores instructions that, when executed by a processor, cause the processor to perform the operational evidence verification method for industrial engineering software intelligent agents according to any one of claims 1-9.

13. An electronic device, characterized in that, include: Non-transitory computer-readable storage medium; processor; The non-transitory computer-readable storage medium stores instructions that, when executed by a processor, cause the processor to perform the operational evidence verification method for intelligent agents in industrial engineering software according to any one of claims 1-9.

14. A computer program product, characterized in that, Includes instructions that, when executed by a processor, cause the processor to perform the operational evidence verification method for industrial engineering software intelligent agents according to any one of claims 1-9.

Images