Automatic examination method and system for three-dimensional model of power grid

Abstract

The invention discloses an automatic review method and system for a three-dimensional model of a power grid, and belongs to the technical field of power grid engineering design and intelligent review, and the system comprises the steps: analyzing the three-dimensional model of the power grid into a JSON file containing model attribute information; based on the model attribute information in the JSON file, retrieving from the OWL ontology knowledge base through an RAG retrieval enhancement module to obtain related standard knowledge; based on the model attribute information and related standard knowledge in the JSON file, executing model attribute automatic verification through a review rule script, and obtaining a model attribute verification result; and based on the model attribute information, the related standard knowledge and the model attribute verification result in the JSON file, guiding a power grid bright large model reasoning unit to generate a review result through a cue word template. According to the method, the problems of low efficiency, non-uniform standard, high error rate and the like of existing manual review are solved, and automatic and intelligent review of the three-dimensional model of the power grid is realized.

Description

Technical Field

[0001] This invention relates to an automatic review method and system for three-dimensional power grid models, belonging to the field of power grid engineering design and intelligent review technology. Background Technology

[0002] In the construction of power grid transmission and transformation projects, 3D models (such as GIM models) have become the core data carrier for design, construction, and operation and maintenance, and their accuracy directly affects the quality and efficiency of the project. Currently, the review of power grid 3D models mainly relies on manual work. Reviewers need to check the geometric parameters, attribute information, topological relationships, and other contents of the model one by one against the relevant specifications of the State Grid transmission and transformation projects (such as 3D design data interaction specifications, modeling technical guidelines, etc.).

[0003] However, the existing review methods have many problems: on the one hand, manual review is inefficient, especially for complex 3D models of large-scale projects, the review cycle is long and it is difficult to meet the project schedule requirements; on the other hand, the review standards rely on personal experience, and different reviewers have different understandings of the specifications, resulting in poor consistency of review results and easy omissions and false positives; in addition, the State Grid's specifications are numerous and frequently updated, and it is difficult for manual reviewers to fully and in real time grasp the latest requirements, which poses a risk of incomplete implementation of the specifications.

[0004] While some automated review tools exist in existing technologies, they mostly rely on fixed rule engines, which have limited ability to understand complex semantic specifications. Furthermore, these review methods require large amounts of labeled data for model training, resulting in high data collection and labeling costs, and their generalization ability is limited by the range of training data. Therefore, there is an urgent need for an intelligent review method and system that requires no additional training and can flexibly adapt to rule updates. Summary of the Invention

[0005] The purpose of this invention is to overcome the shortcomings of the prior art and provide an automatic review method and system for three-dimensional power grid models, solving the problems of low efficiency, inconsistent standards, and easy errors in existing manual review, and realizing the automated and intelligent review of three-dimensional power grid models.

[0006] To achieve the above objectives, the present invention is implemented using the following technical solution: In a first aspect, the present invention provides an automatic review method for a three-dimensional model of a power grid, comprising: Parse the power grid GIM 3D model into a JSON file containing model attribute information; Input the JSON file into a pre-built 3D model review agent that includes an OWL ontology knowledge base, a RAG retrieval enhancement module, review rule scripts, prompt word templates, and a power grid-based large model inference unit, and perform the following steps: Based on the model attribute information in the JSON file, relevant specification knowledge is retrieved from the OWL ontology knowledge base through the RAG retrieval enhancement module; Based on the model attribute information and relevant specification knowledge in the JSON file, the model attribute is automatically validated by the review rule script, and the model attribute validation results are obtained. Based on the model attribute information, relevant specification knowledge, and model attribute verification results in the JSON file, the power grid's large-scale model inference unit is guided to generate review results through prompt word templates.

[0007] Furthermore, the step of parsing the power grid GIM 3D model into a JSON file containing model attribute information includes: extracting geometric parameters, material information, topological relationships, and core engineering attributes from the GIM model, and mapping them into a key-value pair structure of a JSON file according to a preset format.

[0008] Furthermore, based on the model attribute information in the JSON file, the RAG retrieval enhancement module retrieves relevant specification knowledge from the OWL ontology knowledge base, including: Vectorize the specification clauses and model attribute constraints in the OWL ontology knowledge base to build a retrieval vector library; A retrieval query is generated based on the model attribute information in the JSON file, and relevant specification knowledge is obtained from the retrieval vector library through a vector similarity matching algorithm.

[0009] Furthermore, based on the model attribute information and relevant specification knowledge in the JSON file, the automatic validation of model attributes is performed through a review rule script to obtain the model attribute validation results, including: The mandatory clauses, recommended clauses, and verification logic for the review of the three-dimensional power grid model are converted into executable Python scripts. Based on the model attribute information in the JSON file and the relevant specification knowledge, the model attribute format validation, range validation, and logical consistency validation are implemented, and the model attribute validation results are obtained.

[0010] Furthermore, the step of guiding the power grid's large-scale model inference unit to generate review results through prompt word templates includes: converting the model attribute information in the JSON file into a structured review task through prompt word templates, constraining the output format through prompt words, integrating relevant standard knowledge and model attribute verification results, and generating review results.

[0011] Secondly, this invention provides an automatic review system for a three-dimensional power grid model, employing a B / S architecture. The system includes: The browser-side module is used to receive the power grid GIM 3D model uploaded by the user, parse it to generate a JSON file, and upload the JSON file to the server. The server-side module is used to receive the JSON file, automatically review the JSON file through a 3D model review agent, generate review results, and return them to the browser. The 3D model review agent includes the OWL ontology knowledge base, the RAG retrieval enhancement module, the review rule script unit, and the Guangming large model reasoning module.

[0012] Furthermore, the browser-side module includes a model upload and parsing module, which supports batch upload of GIM format files. During the parsing process, a format verification mechanism is used to provide error messages for invalid model files, and the generated JSON file is uploaded to the server via an encryption protocol.

[0013] Furthermore, the server-side module also includes a knowledge storage unit for storing the OWL ontology knowledge base and the RAG retrieval vector library, supporting dynamic updates of specification clauses, and maintaining the consistency of relationships between entities through the ontology reasoning engine.

[0014] Furthermore, the server-side module also includes a result storage module, which records the model information, review time, review results and operators for each review, and supports the query and tracing of review history.

[0015] Furthermore, the server-side module interacts with the browser through the communication module, and transmits the JSON file and review results using an encrypted RESTful API protocol.

[0016] Thirdly, the present invention provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of any of the methods described above.

[0017] Fourthly, the present invention provides an electronic device, comprising: Memory, used to store computer programs / instructions; A processor for executing the computer program / instructions to implement the steps of any of the methods described above.

[0018] Fifthly, the present invention provides a computer program product, including a computer program / instructions that, when executed by a processor, implement the steps of any of the methods described above.

[0019] Compared with the prior art, the beneficial effects achieved by the present invention are as follows: This invention provides an automatic review method and system for 3D power grid models. By using automated parsing and intelligent agent review, it replaces traditional manual point-by-point verification, significantly shortening the review cycle, and is particularly suitable for batch review of large-scale power grid 3D models. Based on the OWL ontology knowledge base, RAG technology, and prompt-word optimized large-model reasoning logic, this invention ensures consistency between the review criteria and established specifications, reducing missed or false detections caused by human misunderstanding. This invention reduces reliance on professional reviewers, lowers manual input, and assists designers in quickly optimizing models through rectification suggestions, improving overall work efficiency. Utilizing RAG retrieval enhancement technology and rule script-assisted reasoning, this invention eliminates the need for specialized training of the large-scale power grid 3D model review scenario, avoiding the high costs of annotation data preparation, computing power consumption, and technical debugging in traditional model training, significantly reducing the economic investment in system construction and maintenance. Attached Figure Description

[0020] Figure 1 This is a flowchart of an automatic review method for a three-dimensional power grid model provided in an embodiment of the present invention; Figure 2 This is an architecture diagram of an automatic review system for a three-dimensional power grid model provided in an embodiment of the present invention; Figure 3 This is a schematic diagram of the composition of the three-dimensional model review agent provided in the embodiment of the present invention. Detailed Implementation

[0021] The present invention will be further described below with reference to the accompanying drawings. The following embodiments are only used to more clearly illustrate the technical solution of the present invention, and should not be used to limit the scope of protection of the present invention.

[0022] Example 1: This example introduces an automatic review method for a three-dimensional power grid model, including: Parse the power grid GIM 3D model into a JSON file containing model attribute information; Input the JSON file into a pre-built 3D model review agent that includes an OWL ontology knowledge base, a RAG retrieval enhancement module, review rule scripts, prompt word templates, and a power grid-based large model inference unit, and perform the following steps: Based on the model attribute information in the JSON file, relevant specification knowledge is retrieved from the OWL ontology knowledge base through the RAG retrieval enhancement module; Based on the model attribute information and relevant specification knowledge in the JSON file, the model attribute is automatically validated by the review rule script, and the model attribute validation results are obtained. Based on the model attribute information, relevant specification knowledge, and model attribute verification results in the JSON file, the power grid's large-scale model inference unit is guided to generate review results through prompt word templates.

[0023] The automatic review method for the three-dimensional model of the power grid provided in this embodiment involves the following steps in its application process: Power Grid GIM 3D Model Parsing: This function parses the input power grid GIM 3D model, extracts core attributes such as geometric parameters (e.g., dimensions, coordinates, topological connections), material information (e.g., material type, fire resistance rating), and engineering attributes (e.g., equipment model, installation location, design stage), and maps these attributes to a JSON file according to a preset key-value pair structure, thereby achieving structured storage and transmission of model data.

[0024] Construction of the 3D Model Review Agent: This agent is the core of the review method and consists of the following four parts: The OWL ontology knowledge base is constructed based on relevant standards for State Grid transmission and transformation projects, covering 3D design model data interaction specifications, modeling technical guidelines, and deliverables transfer requirements. It uses OWL ontology to define three core entity types: "model entities" (e.g., transformer models, tower models), "attribute constraints" (e.g., dimensional tolerance ranges, material compliance requirements), and "standard clauses" (e.g., mandatory clauses, recommended clauses). It also defines the relationships between entities (e.g., "model entity - must satisfy - attribute constraint," "attribute constraint - originates from - standard clause," etc.), forming a structured knowledge system.

[0025] RAG retrieval enhancement module construction: This involves vectorizing the specification clauses, attribute constraints, and other knowledge in the OWL ontology knowledge base (using the Sentence-BERT model to convert text into vectors) to build a retrieval vector library. During the review process, retrieval queries (such as "permissible range of transformer model height deviation") are generated based on model attribute information in JSON files. Relevant specification knowledge is quickly retrieved from the vector library using a vector similarity matching algorithm, providing accurate support for large model inference.

[0026] Review rule script construction: Key rules for reviewing 3D power grid models are identified, including mandatory verification rules (such as consistency between equipment models and engineering design requirements), recommended verification rules (such as completeness of material labeling), and logical verification rules (such as the rationality of topological connections). These rules are then converted into executable Python scripts to automatically verify model attributes, such as determining whether model dimensions are within the allowable deviation range of the specifications and whether material labeling conforms to the preset format.

[0027] Deployment of the Guangming Large Model Inference Unit: The Guangming large model of the power grid serves as the "brain" of the intelligent agent, responsible for natural language understanding, review task decomposition, logical reasoning, and dynamic decision-making. This large model does not require specialized training for power grid 3D model review scenarios. It optimizes the reasoning process through preset prompt templates, directly relying on the RAG retrieval enhancement module to call upon normative knowledge in the knowledge base, and combining the execution results of the review rule scripts to perform reasoning. This significantly reduces the costs associated with traditional model training, such as massive amounts of labeled data collection, high-performance computing equipment investment, and dedicated algorithm personnel for debugging. After receiving model attribute information from a JSON file, this unit accurately transforms structured data into specific review tasks using prompts containing parameters such as entity type and attribute name (e.g., "Based on {tower model} {grounding device attributes}, generate a review task requiring integrity verification"). It then calls the RAG retrieval enhancement module to obtain the relevant normative clauses for the task; triggers the corresponding review rule script for automated verification; and finally, through preset format prompts (e.g., "Integrate results according to 'problem description: [deviation content]; based on clause: [normative number]; rectification suggestion: [adjustment plan]'), it generates standardized review conclusions through logical reasoning.

[0028] Automated review execution: The 3D model review agent loads a JSON file and guides the large model through prompts to complete the entire automated review process according to the workflow of "task transformation → knowledge retrieval → rule verification → conclusion generation". It outputs the final review results, including approved items, problematic items, and comprehensive score.

[0029] The following description, in conjunction with a preferred embodiment, illustrates the content involved in the above embodiments.

[0030] Step 1: Model Parsing. The user uploads a GIM 3D model file of a 220kV substation via the client. The client's model upload and parsing module calls the GIM parsing engine to extract the geometric parameters (e.g., transformer length × width × height is 5m × 3m × 2.5m), material information (e.g., switchgear material is Q235 steel), and engineering attributes (e.g., equipment model is SFZ11-25000 / 220) of entities such as transformers, switchgear, and cable trenches from the model, and maps this information into a JSON file.

[0031] Step 2: Agent Initialization. The server loads the 3D model review agent. The OWL ontology knowledge base has predefined the constraint relationship between the "transformer" entity and the 3D specification guideline "transformer external dimension deviation should be ≤±10mm"; the RAG retrieval vector library has stored the vectorized data of this guideline; the review rule script contains the logic for calculating the dimension deviation; the Guangming large model completes the inference environment initialization. At this point, there is no need to load any specific training parameters for the review of this substation model; the work can be carried out directly based on general inference capabilities combined with the knowledge base.

[0032] Step 3: Automatic Review. After receiving the JSON file, the agent generates precise instructions based on the prompt "convert transformer.height=2.5m into a review task containing the verification target (height deviation) and entity type (transformer)"; it calls the RAG module to retrieve relevant specification clauses; it triggers the rule script to calculate that the deviation between the model height and the design value (assuming the design value is 2.49m) is 0.01m (10mm), which meets the deviation requirement; finally, it generates a standardized review result based on the prompt "output conclusion in the format of 'problem description - based clause - rectification suggestion'", marking the transformer height verification item as "passed".

[0033] Step 4: Results Feedback. The server sends a report containing all entity review results to the client. Users can view the results through the review report preview module, and after confirming there are no issues, export the PDF report for archiving.

[0034] Example 2: This example provides an automatic review system for a 3D model of a power grid, using a B / S architecture. The system includes: The browser-side module is used to receive the power grid GIM 3D model uploaded by the user, parse it to generate a JSON file, and upload the JSON file to the server. The server-side module is used to receive the JSON file, automatically review the JSON file through a 3D model review agent, generate review results, and return them to the browser. The 3D model review agent includes the OWL ontology knowledge base, the RAG retrieval enhancement module, the review rule script unit, and the Guangming large model reasoning module.

[0035] The browser-side module includes a model upload and parsing module, which supports batch upload of GIM format files. During the parsing process, a format verification mechanism is used to provide error messages for invalid model files, and the generated JSON file is uploaded to the server via an encryption protocol.

[0036] The server-side module also includes a knowledge storage unit for storing the OWL ontology knowledge base and the RAG retrieval vector library, supporting dynamic updates of specification clauses, and maintaining the consistency of relationships between entities through the ontology reasoning engine.

[0037] The server-side module also includes a result storage module, which records the model information, review time, review results and operators for each review, and supports the query and tracing of review history.

[0038] The server-side module interacts with the browser through the communication module, and transmits the JSON file and review results using the encrypted RESTful API protocol.

[0039] The specific deployment and operation process of the automatic review system for the 3D power grid model provided in this embodiment is as follows: The system is accessed via a web page in a browser, requiring no client software installation. After designers complete the 3D model design of the substation, they log in to the system through a browser and select three GIM model files using the "Batch Upload" function on the "Model Upload" page. The system front-end automatically performs format verification, and a pop-up window indicates that one file cannot be parsed due to an outdated version (not GIM 2.0 or later). Designers then update the file and re-upload it. The two models that pass parsing in the browser generate JSON data and are uploaded to the server via encrypted HTTPS protocol.

[0040] After receiving the JSON data, the 3D model review agent initiates the review process: the OWL ontology knowledge base provides the latest modeling specifications, the RAG module accurately matches the review clauses corresponding to each entity, the rule script automatically performs attribute verifications such as dimensions and materials, and the integrated results from the large model database generate a detailed review report. Upon completion of the review, the results storage module records the model name, review time, and results, and simultaneously sends the report to the browser.

[0041] Designers reviewed the review report preview page in a browser and found that a switch cabinet model had a "missing material label" issue. Based on Article 5.2.3 of the "3D Design Data Interaction Specification," the suggested rectification was to "add the switch cabinet's material attribute as 'cold-rolled steel plate.'" The designers modified the model according to the suggestion and re-uploaded it for review until all issues were resolved.

[0042] This invention automates and intelligently reviews 3D power grid models by constructing an intelligent review method and system, effectively improving review efficiency and accuracy. At the same time, it significantly reduces system construction costs through a large-scale model application mode that eliminates the need for training, thus providing strong support for the quality of power grid engineering design.

[0043] Example 3: This example provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of any of the methods described in Example 1.

[0044] Example 4: This example provides an electronic device, including: Memory, used to store computer programs / instructions; A processor for executing the computer program / instructions to implement the steps of any of the methods described in Embodiment 1.

[0045] Example 5: This example provides a computer program product, including a computer program / instructions, which, when executed by a processor, implement the steps of the method described in any one of Examples 1.

[0046] The above description is only a preferred embodiment of the present invention. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the technical principles of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

[0047] Those skilled in the art will understand that embodiments of this disclosure can be provided as methods, systems, or computer program products. Therefore, this disclosure can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this disclosure can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0048] This disclosure is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this disclosure. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create a machine for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0049] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0050] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0051] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of this disclosure and not to limit its protection scope. Although this disclosure has been described in detail with reference to the above embodiments, those skilled in the art should understand that after reading this disclosure, they can still make various changes, modifications or equivalent substitutions to the specific implementation of the invention, but these changes, modifications or equivalent substitutions are all within the protection scope of the pending claims.

Claims

1. An automatic review method for a three-dimensional power grid model, characterized in that, include: Parse the power grid GIM 3D model into a JSON file containing model attribute information; Input the JSON file into a pre-built 3D model review agent that includes an OWL ontology knowledge base, a RAG retrieval enhancement module, review rule scripts, prompt word templates, and a power grid-based large model inference unit, and perform the following steps: Based on the model attribute information in the JSON file, relevant specification knowledge is retrieved from the OWL ontology knowledge base through the RAG retrieval enhancement module; Based on the model attribute information and relevant specification knowledge in the JSON file, the model attribute is automatically validated by the review rule script, and the model attribute validation results are obtained. Based on the model attribute information, relevant specification knowledge, and model attribute verification results in the JSON file, the power grid's large-scale model inference unit is guided to generate review results through prompt word templates.

2. The automatic review method for a three-dimensional power grid model according to claim 1, characterized in that, The process of parsing the power grid GIM 3D model into a JSON file containing model attribute information includes: extracting geometric parameters, material information, topological relationships, and core engineering attributes from the GIM model, and mapping them into a key-value pair structure of a JSON file according to a preset format.

3. The automatic review method for a three-dimensional power grid model according to claim 1, characterized in that, The model attribute information in the JSON file is used to retrieve relevant specification knowledge from the OWL ontology knowledge base through the RAG retrieval enhancement module, including: Vectorize the specification clauses and model attribute constraints in the OWL ontology knowledge base to build a retrieval vector library; A retrieval query is generated based on the model attribute information in the JSON file, and relevant specification knowledge is obtained from the retrieval vector library through a vector similarity matching algorithm.

4. The automatic review method for a three-dimensional power grid model according to claim 1, characterized in that, Based on the model attribute information and relevant specification knowledge in the JSON file, the model attribute is automatically validated through a review rule script to obtain the model attribute validation results, including: The mandatory clauses, recommended clauses, and verification logic for the review of the three-dimensional power grid model are converted into executable Python scripts. Based on the model attribute information in the JSON file and the relevant specification knowledge, the model attribute format validation, range validation, and logical consistency validation are implemented, and the model attribute validation results are obtained.

5. The automatic review method for a three-dimensional power grid model according to claim 1, characterized in that, The step of guiding the power grid's large-scale model inference unit to generate review results through prompt word templates includes: converting the model attribute information in the JSON file into a structured review task through prompt word templates, constraining the output format through prompt words, integrating relevant standard knowledge and model attribute verification results, and generating review results.

6. An automatic review system for a three-dimensional power grid model, employing a B / S architecture, characterized in that, The system includes: The browser-side module is used to receive the power grid GIM 3D model uploaded by the user, parse it to generate a JSON file, and upload the JSON file to the server. The server-side module is used to receive the JSON file, automatically review the JSON file through a 3D model review agent, generate review results, and return them to the browser. The 3D model review agent includes the OWL ontology knowledge base, the RAG retrieval enhancement module, the review rule script unit, and the Guangming large model reasoning module.

7. The automatic review system for the three-dimensional model of a power grid according to claim 6, characterized in that, The browser-side module includes a model upload and parsing module, which supports batch upload of GIM format files. During the parsing process, a format verification mechanism is used to provide error messages for invalid model files, and the generated JSON file is uploaded to the server via an encryption protocol.

8. The automatic review system for the three-dimensional model of a power grid according to claim 6, characterized in that, The server-side module also includes a knowledge storage unit for storing the OWL ontology knowledge base and the RAG retrieval vector library, supporting dynamic updates of specification clauses, and maintaining the consistency of relationships between entities through the ontology reasoning engine.

9. The automatic review system for the three-dimensional model of a power grid according to claim 6, characterized in that, The server-side module also includes a result storage module, which records the model information, review time, review results and operators for each review, and supports the query and tracing of review history.

10. The automatic review system for the three-dimensional model of a power grid according to claim 6, characterized in that, The server-side module interacts with the browser through the communication module, and transmits the JSON file and review results using an encrypted RESTful API protocol.

Images