A multi-center collaboration capability governance method and system of an industrial intelligent agent construction system, an electronic device, and a medium

Abstract

The application provides a multi-center collaborative capability management method and system of an industrial agent construction system, an electronic device and a medium, relates to the technical field of industrial intelligent software platforms and agent systems, and comprises the following steps: establishing a multi-center collaborative capability system in a management backend; creating, accessing, configuring or arranging agent capabilities, knowledge capabilities, tool capabilities and model capabilities, and performing testing, publishing, registration and enabling through a unified management process; binding the published and enabled target capabilities with client menu function points; the client matches and calls corresponding capabilities according to function point calling requests, and records execution results, version information, running logs and audit information; the application can realize unified organization, unified publishing, unified binding and unified calling of multiple intelligent capabilities in the industrial agent system, supports continuous evolution of the background capabilities while maintaining the stability of the client interface, and improves the manageability, reusability and traceability of the system.

Description

Technical Field

[0001] This invention relates to the field of industrial intelligent software platforms and intelligent agent systems, and in particular to a multi-center collaborative capability governance method, system, electronic device and medium for an industrial intelligent agent construction system. Background Technology

[0002] With the development of large language models, tool invocation, and multi-agent orchestration techniques, agent systems have gradually expanded from general question answering to complex task automation scenarios. Existing research shows that multi-agent methods have advantages in complex task decomposition, collaborative reasoning, and automated workflow execution, but they still generally face problems such as unclear capability boundaries, unstable execution, and insufficient observation and governance in real production environments.

[0003] Meanwhile, research in the design and engineering fields has begun to explore combining large language models with computer-aided design processes to achieve design intent understanding, parameter generation, geometric modeling, engineering calculations, and iterative feedback. However, most existing solutions focus on automating single design tasks, generating CAD code, or assisting local processes, and are more of a point-based enhancement of a certain type of tool or workflow. They have not yet formed a general basic platform capability organization method suitable for industrial software environments.

[0004] For industrial intelligent agent systems, the real technical challenge lies not only in how to generate results from models, but also in how to organize intelligent agent capabilities, knowledge capabilities, engineering tool capabilities, and model capabilities into manageable, deployable, bindable, and auditable platform capability units, and stably map these capabilities to client-side functional entry points. Existing solutions typically suffer from the following shortcomings: 1. Multi-domain industrial knowledge integration and multi-type tool invocation are managed in a fragmented manner, lacking a unified organizational structure; 2. The lack of a unified testing, release, registration, and activation process before capability deployment makes it difficult to meet the stability and traceability requirements of industrial scenarios; 3. The lack of a clear correlation between the client-side functional interface entry point, which is mainly organized in a question-and-answer format, and the backend multimodal content generation capabilities, disrupting existing interaction habits of current industrial software and making the generation process difficult to control; 4. The lack of a decoupling mechanism for stable interfaces and variable capabilities makes it difficult to implement capability replacement, version rollback, and audit governance in a standardized manner.

[0005] Therefore, it is necessary to propose a multi-center collaborative capability governance method for industrial intelligent agent construction systems to solve the technical problem of the difficulty in unified governance and stable invocation of multiple intelligent capabilities in industrial intelligent agent systems. Summary of the Invention

[0006] The purpose of this invention is to provide a multi-center collaborative capability governance method, system, electronic device and medium for an industrial intelligent agent construction system, in order to solve the problems of capability dispersion, incomplete governance links, unstable functional entry points and excessive coupling between client and backend capabilities in existing intelligent agent platforms in industrial scenarios, and to realize the unified organization, unified release, unified binding and unified invocation of multiple capabilities in the industrial intelligent agent system.

[0007] To achieve the above objectives, this invention provides a method for governing the multi-center collaborative capabilities of an industrial intelligent agent construction system, comprising the following steps: Step S101: Establish a multi-center collaborative capability system in the management backend and classify and organize the capabilities of intelligent agents, knowledge capabilities, tool capabilities, and model capabilities. Step S102: The corresponding capabilities are created, accessed, configured, or orchestrated through the intelligent agent building platform, knowledge center, tool center, and large model center, respectively. Step S103: Perform a unified governance process on the capability through the capability release engine. The unified governance process includes testing, release, registration and activation. Step S104: The management backend establishes a binding relationship between the enabled target capabilities and the target function points in the client menu hierarchy; Step S105: The client receives the user's request to call the target function and matches the corresponding published capabilities according to the binding relationship; Step S106: Invoke the corresponding published capability to perform task processing and return the result to the client display area; Step S107: Record the version information, execution status, running logs, and audit information during the capability invocation process.

[0008] Preferably, the multi-center collaborative capability system includes: The Agent Center is used to manage agent registration information, responsibility domains, running status, and version information. The Knowledge Center is used to manage documents, drawings, data tables, knowledge graphs, and prompt word template resources; The Tools Center manages engineering calculation tools, simulation tools, analysis tools, optimization tools, and external system interfaces. The large model center is used to access and manage generative, inference, analytical, and extraction models. An agent building platform for visually creating, orchestrating, and debugging agent processes; The capability release engine is used to uniformly perform testing, release, registration, activation, and auditing operations.

[0009] Preferably, in step S102, capability creation, access, configuration, or orchestration includes one or more of the following methods: creating an intelligent agent process in the intelligent agent building platform, importing knowledge resources in the knowledge center, registering tools and their parameter templates in the tool center, and accessing the target model and configuring the model's purpose and status in the large model center.

[0010] Preferably, in step S103, the unified governance process also includes performing version recording, release auditing, activation status management and rollback control on the target capability; in step S104, the management backend verifies the release status and activation status of the target capability, and establishes a binding relationship only when the target capability has been released and activated.

[0011] Preferably, in step S104, the client menu hierarchy includes a first-level menu, a second-level function set, a third-level function point, and a fourth-level form, and the target function point and the target capability are bound together by a backend configuration method.

[0012] Preferably, the client is configured to only receive call requests, load published capabilities, and display processing results. The client's communication interface blocks the transmission channel for sending capability logic modification instructions to the management backend. Published capabilities can be replaced, upgraded, or rolled back without changing the layout of the client's function entry points.

[0013] A multi-center collaborative capability governance system for an industrial intelligent agent construction system includes a client and a management backend: The client is used to provide menu-level functions, receive user requests for target functions, and display the processing results. The management backend communicates with the client, including: The Agent Center is used to manage the registration information, domain of responsibility, running status, and version information of agents. The Knowledge Center is used to manage documents, drawings, data tables, knowledge graphs, and prompt word template resources; The Tools Center manages engineering calculation tools, simulation tools, analysis tools, optimization tools, and external system interfaces. The large model center is used to access and manage generative, inference, analytical, and extraction models. An agent building platform for visually creating, orchestrating, and debugging agent processes; The capability deployment engine is used to perform unified testing, deployment, registration, activation, and auditing operations on various capabilities. The menu binding configuration module is used to establish a backend binding relationship between the enabled target capabilities and the target function points in the client menu hierarchy; The management backend responds to the call request initiated by the client, matches the corresponding published capabilities according to the binding relationship, calls the corresponding capability to perform task processing and returns the result to the client, and records the version information, execution status, running logs and audit information during the call process.

[0014] Preferably, the intelligent agent construction platform includes an intelligent agent creation module, a node orchestration module, a parameter configuration module, a process debugging module, and a template reuse module; The intelligent agent center includes an intelligent agent registration module, a responsibility domain management module, a version management module, a runtime status management module, a dependency management module, and an execution history management module; The knowledge center includes a multimodal knowledge import module, a document extraction module, a knowledge graph module, a rule base module, a case library module, and a prompt word template module; The tool center includes a tool registration module, a tool version management module, a runtime environment management module, a parameter template module, a call permission module, and a tool call log module; The large model center includes a model access module, a model usage configuration module, a model start / stop module, a model replacement module, and a model call log module; The capability release engine includes a capability testing module, a release module, a registration module, an activation module, a rollback module, and a release audit module; The menu binding configuration module includes a menu hierarchy configuration module, a function point binding module, a function point status verification module, and a client synchronization module.

[0015] An electronic device includes a memory and a processor, the memory storing a computer program and the processor running the computer program to enable the electronic device to perform a multi-center collaborative capability governance method for an industrial intelligent agent construction system.

[0016] A computer-readable storage medium storing a computer program that, when executed by a processor, implements a multi-center collaborative capability governance method for an industrial intelligent agent construction system.

[0017] Therefore, this invention employs the aforementioned multi-center collaborative capability governance method, system, electronic device, and medium for an industrial intelligent agent construction system. First, a multi-center collaborative capability system is constructed in the management backend, consisting of an intelligent agent center, knowledge center, tool center, large model center, capability release engine, and intelligent agent construction platform. Second, a unified testing, release, registration, and activation governance process is implemented for multiple types of capabilities, and a backend binding relationship is established between released capabilities and client function points. Finally, the client matches and invokes the corresponding capability based on the function point call request, and records version, logs, and audit information. This invention improves the organizational efficiency and online governance capabilities of multiple capabilities in the industrial intelligent agent system, further enhancing capability reusability, version controllability, and client call stability, thereby supporting the continuous evolution of backend capabilities while maintaining interface stability in the industrial intelligent agent system.

[0018] The technical solution of the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. Attached Figure Description

[0019] Figure 1 This is a diagram illustrating the overall architecture of the multi-center collaborative intelligent agent platform according to an embodiment of the present invention. Figure 2 This is a flowchart of a multi-center collaborative capability governance method for an industrial intelligent agent construction system according to the present invention; Figure 3 This is a mapping diagram of the client menu function points and published capabilities in an embodiment of the present invention; Figure 4 This is a platform data flow diagram according to an embodiment of the present invention; Figure 5 This is a schematic diagram of multi-center collaborative invocation in an embodiment of the present invention for generating intelligent agents for mechanical products or mechatronic products. Detailed Implementation

[0020] The following detailed description of embodiments of the invention provided in the accompanying drawings is not intended to limit the scope of the claimed invention, but merely illustrates selected embodiments of the invention. All other embodiments obtained by those skilled in the art based on the embodiments of the invention without inventive effort are within the scope of protection of the invention.

[0021] Example like Figure 1As shown, the industrial intelligent agent basic platform consists of a client and a management backend. The management backend includes an intelligent agent center, a knowledge center, a tool center, a large model center, a capability publishing engine, and an intelligent agent building platform. Specifically, the intelligent agent center manages intelligent agent registration information, responsibility domains, operational status, and version information; the knowledge center manages knowledge resources such as documents, drawings, data tables, knowledge graphs, and prompt word templates; the tool center manages computational tools, simulation tools, analysis tools, optimization tools, and external system interfaces used in engineering; the large model center accesses and manages large-scale generative models, inference analysis models, and multimodal extraction models; the intelligent agent building platform provides a visual platform for creating, orchestrating, and debugging intelligent agent processes; and the capability publishing engine uniformly executes testing, publishing, registration, activation, and auditing operations.

[0022] like Figure 2 As shown, this invention provides a method for governing the multi-center collaborative capabilities of an industrial intelligent agent construction system, comprising the following steps: Step S101: Establish a multi-center collaborative capability system in the management backend, classifying and organizing agent capabilities, knowledge capabilities, tool capabilities, and model capabilities. The multi-center collaborative capability system includes: an agent center for managing agent registration information, responsibility domains, operational status, and version information; a knowledge center for managing documents, drawings, data tables, knowledge graphs, and prompt word template resources; a tool center for managing engineering calculation tools, simulation tools, analysis tools, optimization tools, and external system interfaces; a large model center for accessing and managing generation, inference, analysis, and extraction models; an agent building platform for visually creating, orchestrating, and debugging agent processes; and a capability release engine for uniformly executing testing, release, registration, activation, and auditing operations.

[0023] Step S102 involves creating, integrating, configuring, or orchestrating corresponding capabilities through the agent building platform, knowledge center, tool center, and large model center. Capability creation, integration, configuration, or orchestration includes one or more of the following methods: creating agent processes in the agent building platform, importing knowledge resources in the knowledge center, registering tools and their parameter templates in the tool center, and integrating target models and configuring their purpose and status in the large model center.

[0024] Step S103: Execute a unified governance process on the capability through the capability release engine. The unified governance process includes testing, release, registration, and activation. In an optional implementation, the unified governance process further includes version recording, release auditing, activation status management, and rollback control for the target capability; target capabilities that have not been released or activated must not be bound as executable function points on the client side.

[0025] Step S104 involves the management backend establishing a binding relationship between the enabled target capabilities and the target function points in the client's menu hierarchy. The client menu hierarchy includes a first-level menu, a second-level function set, a third-level function point, and a fourth-level form. The binding relationship between the target function point and the target capability is established through backend configuration. For example... Figure 3 As shown, a mapping relationship is established between client menu function points and published capabilities through backend configuration. Multiple second-level function sets are set under the first-level menu, and multiple third-level function points are set under each second-level function set. Each third-level function point is bound to a corresponding published capability. Target capabilities that have not been published or enabled cannot be bound as client-executable function points.

[0026] In step S105, the client receives the user's request to call the target function and matches the corresponding published capabilities according to the binding relationship.

[0027] Step S106: Invoke the corresponding published capability to perform task processing and return the result to the client display area.

[0028] Step S107: Record the version information, execution status, running logs, and audit information during the capability invocation process.

[0029] like Figure 2 As shown, in step S104, the client menu hierarchy includes a first-level menu, a second-level function set, a third-level function point, and a fourth-level form. The target function point and the target capability are bound together by a backend configuration method. Target capabilities that have not been published or enabled cannot be bound as client-executable function points.

[0030] As an optional implementation, the client does not have the permission to modify the backend capability logic, and is only used to receive call requests, load published capabilities and display processing results; published capabilities can be replaced, upgraded or rolled back without changing the layout of the client's function entry, thereby realizing a capability access mode with a stable interface and variable capabilities in the industrial infrastructure platform.

[0031] See Figure 4The data flow process of the industrial intelligent agent basic platform includes: client-side menu function triggering, four-level form input, project data, CAX (Computer Aided X) files or model data, and user interaction commands entering the platform control and routing layer; the control and routing layer performs routing control, process template management, context reading, call orchestration control, and log collection; subsequently, according to task needs, it schedules the intelligent agent center, knowledge center, tool center, and large model center, and interacts with relational databases, cache databases, graph databases, file storage, and publishes audit records; finally, it returns parameter results, design schemes, document files, decision interpretations, and process status information to the client result area.

[0032] In summary, this invention proposes a multi-center collaborative capability governance method for an industrial intelligent agent construction system. First, a multi-center collaborative capability system is established in the management backend. Then, unified testing, release, registration, and activation are performed on various capabilities, and binding relationships are established between these capabilities and client functionalities. Finally, the client invokes the released capabilities according to the binding relationships and records version, logs, and audit information.

[0033] The following uses the design task of a mechanical product or mechatronics product as an example to illustrate the multi-center collaborative process of an industrial intelligent agent construction system. Mechanical products or mechatronics products may include equipment support structures, installation and connection structures, tooling fixtures, transmission support components, box-type components, plate-and-frame structural parts, or other industrial products with geometric structures, design parameters, engineering rules, and manufacturing constraints. The following explanation uses the generation of an equipment support structure as an example, but this example does not constitute a limitation on the scope of protection of this invention.

[0034] like Figure 5 As shown, on the client side, the first-level menu can be configured as "Scheme Design" or "Scheme Generation," and the second-level function set can be configured as "Conceptual Scheme Generation," "Parameter Scheme Generation," and "Geometric Scheme Generation." Specifically, "Conceptual Scheme Generation" can be configured with third-level functions such as "Requirement Identification Agent," "Conceptual Scheme Matching Agent," and "Conceptual Scheme Reconstruction Agent"; "Parameter Scheme Generation" can be configured with third-level functions such as "Rule Parameter Generation Agent," "Hole Parameter Generation Agent," and "Special Constraint Generation Agent"; and "Geometric Scheme Generation" can be configured with third-level functions such as "Model Input Agent," "Special Structure Addition Agent," and "Generation Result Output Agent." The fourth-level form or input area can be configured with forms for uploading design documents, filling in requirement variables, selecting templates, confirming parameters, inputting constraints, inputting model files, and manual confirmation.

[0035] In the management backend, the Knowledge Center pre-imports or integrates product design documents, 3D model files, typical structural template libraries, design rule libraries, manufacturing constraint libraries, and historical solution case libraries. Design documents can include structured or unstructured files such as XML (Extensible Markup Language), JSON (JavaScript Object Notation), Word, and PDF (Portable Document Format). 3D model files can include formats such as STEP (Standard for the Exchange of Product Model Data), IGES (Initial Graphics Exchange Specification), and STP. The Knowledge Center performs document parsing, parameter extraction, requirement variable aggregation, and requirement triple representation on the above resources to form a structured requirement list that includes load conditions, installation interfaces, spatial constraints, opening requirements, material preferences, manufacturing constraints, and special structural requirements.

[0036] The intelligent agent construction platform creates intelligent agent creation schemes and generates intelligent agents, dividing this process into nodes such as requirement identification, concept scheme matching, concept scheme reconstruction, concept scheme output, parametric scheme generation, and geometric scheme generation. The requirement identification node calls the document parsing results from the knowledge center and the extraction model from the Large Language Model (LLM) center to output a requirement list and requirement triples. The concept scheme matching node calls the constraint satisfaction filtering tool, similarity matching tool, and multi-objective optimization tool from the engineering intelligent computing tool center. It first filters product structure templates based on constraints, then calculates template similarity based on requirement triples, and finally selects the target template from candidate templates using a multi-objective optimization strategy. The parametric scheme generation node generates key design parameters based on structural dimensions, hole positions, reinforcements, materials, loads, and manufacturing rules. The geometric scheme generation node calls the parametric template model and special structure addition tool to generate the structural scheme model of the target product.

[0037] The Intelligent Agent Center establishes registration records for the intelligent agents that generate solutions. These records include the agent's identifier, name, version number, domain of responsibility, supported input formats, output formats, dependent knowledge resources, dependent engineering computing tools, and dependent models. The Engineering Computing Tool Center registers constraint satisfaction filtering tools, similarity matching tools, multi-objective optimization tools, parameter rule generation tools, and parametric geometry generation tools, and records tool versions, parameter templates, operating environments, and invocation rules. The Large Model Center configures the purpose and activation status of target models for requirement extraction, semantic matching, solution interpretation, and anomaly alerts.

[0038] The capability release engine executes a unified governance process for the solution generation agent and its dependent knowledge, tools, and models. Specifically, the capability release engine reads a set of sample design documents and sample 3D models, and sequentially tests whether the document upload, design document parsing, requirement variable aggregation, requirement triple representation, template filtering, template similarity matching, multi-objective optimization selection, parameter adjustment, special constraint generation, and geometric model output meet preset rules. After the tests are passed, a release record, capability version number, dependency list, and audit record are generated, and the solution generation agent is marked as released and enabled.

[0039] After a client user clicks on a target function, the client only submits the current project context, CAX model context, four-level form parameters, and user interaction instructions, without directly modifying the backend agent logic. The platform control and routing layer loads the corresponding solution based on the function binding relationship, generates an agent version, and calls the Knowledge Center, Engineering Computing Tool Center, and Large Model Center to complete the task processing. It then returns the conceptual solution triplet, structural parameter table, parameter change table, reference rule table, structural solution model, deviation report, optimization selection reasons, and running status to the client's results area. Version information, input / output summaries, tool call logs, exception information, manual confirmation records, and audit information generated during execution are uniformly recorded for subsequent traceability, rollback, and version comparison.

[0040] To implement the above method, the present invention also provides an industrial intelligent agent construction system, the architecture of which includes a client and a management backend, such as... Figure 1 As shown. The client is used to receive requests for calling target functions, submit four-level forms or input parameters, load published capabilities, and display processing results. The management backend is used to host the intelligent agent building platform, intelligent agent center, knowledge center, engineering intelligent computing tool center, large model center, capability publishing engine, and menu binding configuration module.

[0041] The intelligent agent construction platform includes modules for intelligent agent creation, node orchestration, parameter configuration, process debugging, and template reuse. The intelligent agent center includes modules for intelligent agent registration, responsibility domain management, version management, runtime status management, dependency management, and execution history management. The knowledge center includes modules for multimodal knowledge import, document extraction, knowledge graph, rule base, case library, and prompt word templates. The engineering intelligent computing tool center includes modules for tool registration, tool version management, runtime environment management, parameter templates, access permissions, and tool call logs. The large model center includes modules for model access, model usage configuration, model start / stop, model replacement, and model call logs. The capability release engine includes modules for capability testing, release, registration, activation, rollback, and release auditing. The menu binding configuration module includes modules for menu hierarchy configuration, function point binding, function point status verification, and client synchronization.

[0042] Through the above module configuration, the industrial intelligent agent construction system can connect the intelligent agent creation process, capability governance process, function point binding process and client call process into a closed loop, enabling a stable collaborative relationship between multiple centers.

[0043] The present invention also provides an electronic device, including a memory and a processor, wherein the memory is used to store a computer program, and the processor runs the computer program to enable the electronic device to perform the above-described method.

[0044] The present invention also provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the above-described method.

[0045] Therefore, the present invention adopts the above-mentioned multi-center collaborative capability governance method, system, electronic device and medium of an industrial intelligent agent construction system, which has the following advantages: (1) It can organize intelligent agents, knowledge, engineering tools and model capabilities into unified platform capability units, thereby improving the organization of industrial intelligent capabilities. (2) It can improve the stability, controllability and traceability of capability deployment in industrial scenarios through a unified testing, release, registration and activation link. (3) It can realize an industrial basic platform access method with stable interface and variable capabilities through the background binding mechanism between client function points and released capabilities.

[0046] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit them. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the technical solutions of the present invention, and these modifications or equivalent substitutions cannot cause the modified technical solutions to deviate from the spirit and scope of the technical solutions of the present invention.

Claims

1. A method for governing the multi-center collaborative capabilities of an industrial intelligent agent construction system, characterized in that, The method is applied to an industrial intelligent agent infrastructure platform consisting of a client and a management backend. The management backend includes an intelligent agent center, a knowledge center, a tool center, a large model center, a capability publishing engine, and an intelligent agent building platform. The method includes the following steps: Step S101: Establish a multi-center collaborative capability system in the management backend and classify and organize the capabilities of intelligent agents, knowledge capabilities, tool capabilities, and model capabilities. Step S102: The corresponding capabilities are created, accessed, configured, or orchestrated through the intelligent agent building platform, knowledge center, tool center, and large model center, respectively. Step S103: Perform a unified governance process on the capability through the capability release engine. The unified governance process includes testing, release, registration and activation. Step S104: The management backend establishes a binding relationship between the enabled target capabilities and the target function points in the client menu hierarchy; Step S105: The client receives the user's request to call the target function and matches the corresponding published capabilities according to the binding relationship; Step S106: Invoke the corresponding published capability to perform task processing and return the result to the client display area; Step S107: Record the version information, execution status, running logs, and audit information during the capability invocation process.

2. The multi-center collaborative capability governance method for an industrial intelligent agent construction system according to claim 1, characterized in that: The multi-center collaborative capability system includes: The Agent Center is used to manage agent registration information, responsibility domains, running status, and version information. The Knowledge Center is used to manage documents, drawings, data tables, knowledge graphs, and prompt word template resources; The Tools Center manages engineering calculation tools, simulation tools, analysis tools, optimization tools, and external system interfaces. The large model center is used to access and manage generative, inference, analytical, and extraction models. An agent building platform for visually creating, orchestrating, and debugging agent processes; The capability release engine is used to uniformly perform testing, release, registration, activation, and auditing operations.

3. The multi-center collaborative capability governance method for an industrial intelligent agent construction system according to claim 1, characterized in that, In step S102, capability creation, access, configuration, or orchestration includes one or more of the following methods: creating an intelligent agent process in the intelligent agent building platform, importing knowledge resources in the knowledge center, registering tools and their parameter templates in the tool center, and accessing the target model and configuring the model's purpose and status in the large model center.

4. The multi-center collaborative capability governance method for an industrial intelligent agent construction system according to claim 1, characterized in that: In step S103, the unified governance process also includes performing version recording, release auditing, activation status management and rollback control on the target capability; in step S104, the management backend verifies the release status and activation status of the target capability, and establishes a binding relationship only when the target capability has been released and activated.

5. The multi-center collaborative capability governance method for an industrial intelligent agent construction system according to claim 1, characterized in that: In step S104, the client menu hierarchy includes a first-level menu, a second-level function set, a third-level function point, and a fourth-level form. The target function point and the target capability are bound together by a backend configuration method.

6. The multi-center collaborative capability governance method for an industrial intelligent agent construction system according to claim 1, characterized in that: The client is configured to only receive call requests, load published capabilities, and display processing results. The client's communication interface blocks the transmission channel for sending capability logic modification instructions to the management backend. Released capabilities support replacement, upgrades, or rollbacks without altering the layout of client-side function entry points.

7. A multi-center collaborative capability governance system for an industrial intelligent agent construction system, used to implement the multi-center collaborative capability governance method for an industrial intelligent agent construction system as described in any one of claims 1-6, characterized in that, Includes client and management backend: The client is used to provide menu-level functions, receive user requests for target functions, and display the processing results. The management backend communicates with the client, including: The Agent Center is used to manage the registration information, domain of responsibility, running status, and version information of agents. The Knowledge Center is used to manage documents, drawings, data tables, knowledge graphs, and prompt word template resources; The Tools Center manages engineering calculation tools, simulation tools, analysis tools, optimization tools, and external system interfaces. The large model center is used to access and manage generative, inference, analytical, and extraction models. An agent building platform for visually creating, orchestrating, and debugging agent processes; The capability deployment engine is used to perform unified testing, deployment, registration, activation, and auditing operations on various capabilities. The menu binding configuration module is used to establish a backend binding relationship between the enabled target capabilities and the target function points in the client menu hierarchy; The management backend responds to the call request initiated by the client, matches the corresponding published capabilities according to the binding relationship, calls the corresponding capability to perform task processing and returns the result to the client, and records the version information, execution status, running logs and audit information during the call process.

8. The multi-center collaborative capability governance system for an industrial intelligent agent construction system according to claim 7, characterized in that: The intelligent agent construction platform includes an intelligent agent creation module, a node orchestration module, a parameter configuration module, a process debugging module, and a template reuse module; The intelligent agent center includes an intelligent agent registration module, a responsibility domain management module, a version management module, a runtime status management module, a dependency management module, and an execution history management module; The knowledge center includes a multimodal knowledge import module, a document extraction module, a knowledge graph module, a rule base module, a case library module, and a prompt word template module; The tool center includes a tool registration module, a tool version management module, a runtime environment management module, a parameter template module, a call permission module, and a tool call log module; The large model center includes a model access module, a model usage configuration module, a model start / stop module, a model replacement module, and a model call log module; The capability release engine includes a capability testing module, a release module, a registration module, an activation module, a rollback module, and a release audit module; The menu binding configuration module includes a menu hierarchy configuration module, a function point binding module, a function point status verification module, and a client synchronization module.

9. An electronic device, characterized in that, The device includes a memory and a processor, the memory being used to store a computer program, and the processor running the computer program to cause the electronic device to perform the method according to any one of claims 1 to 6.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed by a processor, implements the method according to any one of claims 1 to 6.

Images