Power name specification detection method based on large model

Abstract

The invention relates to the field of data processing, and relates to an electric power name specification detection method based on a large model, which comprises the following steps: constructing an electric power equipment name specification knowledge base; obtaining a graph topology data set; obtaining a to-be-detected equipment name and a site name to which the to-be-detected equipment name belongs, and performing normative detection on the to-be-detected equipment name by utilizing the large model in combination with the power equipment naming specification knowledge base; selecting graph topological data corresponding to a site to which the to-be-detected equipment belongs from the graph topological data set, analyzing each component of the equipment name by using a large model, and further detecting whether the logical relationship between the components is consistent with the graph topological structure of the corresponding graph topological data or not, thereby obtaining a logic consistency detection result; and fusing the normative detection result and the logic consistency detection result to generate a final comprehensive detection result. By adopting the method provided by the invention, the automation degree, comprehensiveness and accuracy of power equipment name specification detection can be effectively improved.

Description

Technical Field

[0001] This invention relates to the field of data processing. More specifically, this invention relates to a method for detecting standardized electricity names based on a large model. Background Technology

[0002] In power systems, the standardization of equipment naming is directly related to the safe, stable, and efficient operation of the power grid. The power industry has formulated a number of standards that strictly regulate the format, hierarchical structure, and semantic content of equipment names.

[0003] Currently, the detection of equipment naming standardization in power systems mainly relies on two traditional methods: manual review and rule-based automated matching.

[0004] Manual review relies heavily on the experience and judgment of maintenance personnel. Faced with an ever-expanding power grid and a rapidly growing number of monitoring points, manual verification is not only inefficient and costly, but also susceptible to subjective factors, leading to risks of missed checks and misjudgments. Historical issues such as inconsistent naming and data entry errors further exacerbate the difficulty and unreliability of manual review.

[0005] Existing rule-based automated matching systems mostly employ static rules or regular expressions for format matching. While these methods can identify obvious format errors (such as illegal characters, missing fields, and hierarchical errors), they are essentially superficial syntactic checks, lacking a deep understanding of naming semantics. For example, when a device name is "110kVXxx line," the system cannot automatically verify whether the voltage level "110kV" matches the actual voltage level of "Xxx line" in the substation. Similarly, "10kV Zhangzhuang line" might be mistakenly written as "Zhangzhuang 10kV transmission line," which, while semantically similar, does not conform to the standard naming format. Such semantically non-standard expressions are difficult for rule-based systems to effectively identify. Furthermore, rule bases require frequent maintenance as standards are updated, have weak generalization capabilities, and struggle to adapt to complex and ever-changing real-world naming scenarios.

[0006] It is worth noting that the main electrical wiring diagram of a substation is the authoritative basis for reflecting the physical connections and electrical relationships between major electrical equipment such as transformers, circuit breakers, and lines within the substation. Cross-validating equipment naming information with this diagram is the fundamental way to ensure the accuracy and consistency of naming. However, existing methods generally fail to effectively integrate the topological structure and semantic knowledge contained in the wiring diagram, resulting in normative checks being limited to the isolated text level and unable to achieve linkage verification with the actual physical state of the power grid.

[0007] In summary, existing methods for standardizing the naming of power equipment suffer from technical problems such as low efficiency and low accuracy. Summary of the Invention

[0008] To address the technical problems of low efficiency and low accuracy in existing methods for standardizing the naming of power equipment, this invention provides solutions in the following aspects.

[0009] This invention provides a method for standardizing power name detection based on a large model, comprising: A knowledge base for power equipment naming standards will be constructed and used as the basis for large-scale model testing of equipment name standardization. The knowledge base covers equipment information standards, naming standards, and historical naming cases of power equipment. The electrical wiring diagrams of each site in the target area are converted into structured graph topology data using a large visual language model, thereby obtaining a graph topology dataset. The system obtains the name of the device to be tested and the name of the site to which it belongs, and uses a large model combined with a knowledge base of power equipment naming standards to perform a standardization test on the name of the device to be tested, thereby obtaining the standardization test results. Select the graph topology data corresponding to the site to which the device to be detected belongs from the graph topology dataset, use the large model to parse each component of the device name, and then detect whether the logical relationship between each component is consistent with the graph topology structure of the corresponding graph topology data, so as to obtain the logical consistency detection result. The results of the standardization test and the logical consistency test are combined to generate the final comprehensive test result.

[0010] Preferably, constructing a knowledge base for power naming standards includes: Collect standard documents related to the naming of power equipment and historical naming cases of power equipment. The standard documents include national and industry standards and specifications as well as internal naming rules documents of enterprises; the historical naming cases of power equipment include correct naming examples that comply with the specifications and incorrect naming examples that do not comply with the specifications. Preprocessing of the collected documents includes: identifying and filtering out content in the document content that is irrelevant to the naming conventions through text analysis, and extracting the naming rules from the filtered document content; The extracted naming rules and historical naming cases of power equipment are stored in a knowledge base; Extract new naming rules from documents related to the naming of new power equipment, and store the new naming rules in the knowledge base.

[0011] Preferably, content unrelated to naming conventions includes document background information and the purpose of its creation.

[0012] Preferably, before storing the extracted naming rules and historical naming cases of power equipment into the knowledge base, the method further includes: The extracted naming rules are divided into two categories using a large model: general rules and equipment-specific rules. Historical naming cases of power equipment are divided into positive and negative cases. The general rules are the basic naming requirements applicable to all power equipment, including the naming format framework, separator usage specifications, and alphanumeric encoding rules. The equipment-specific rules are the exclusive naming requirements for different types of special equipment, including main transformers, circuit breakers, lines, and busbars.

[0013] Preferably, before storing the new naming rules in the knowledge base, the method further includes: performing conflict detection: comparing whether there is a contradiction between the requirements of the new rules and the existing rules under the same applicable conditions; when a rule conflict is detected, generating a conflict report and pausing the knowledge base update, and waiting for manual confirmation before completing the knowledge base update.

[0014] Preferably, the method further includes: experts correcting the generated final comprehensive test results and writing the corrected comprehensive test results into the power equipment naming standard knowledge base.

[0015] Preferably, the standardization detection of the equipment name to be tested using a large model combined with a power equipment naming standardization knowledge base includes: writing system prompt words suitable for equipment name standardization detection. The system prompt words are used to drive the large model to detect the standardization of the equipment name according to the subsequently specified general rules, specific equipment rules, positive cases and negative cases, and output the specific reasons and the rules violated when the name is not standard. Keywords are extracted from the device name, and relevant general rules and device-specific rules are retrieved from the knowledge base using these keywords. Vector similarity retrieval is used to retrieve relevant positive and negative cases from the case knowledge base, and the top 5 retrieved cases are selected as reference cases for the large model to check the standardization of device names. Based on the device name, the specially written prompt words, and the retrieved rules and reference cases, input content that meets the input requirements of the large model is constructed, and the input content is input into the large model to obtain the detection results; The detection results output by the large model are analyzed, and a normative verification report is generated, which includes the violated rules and suggestions for improvement.

[0016] Preferably, converting the electrical wiring diagrams of each site into structured graph topology data using a large visual language model includes: Collect the original power wiring diagrams of existing sites in the current target area, and preprocess and enhance them to obtain improved power wiring diagrams; The improved electrical wiring diagram is input into the visual language model to obtain the corresponding topological relationships; the visual language model is used to output the corresponding topological relationships based on the electrical wiring diagram. The standardized topology data model is used to convert the topological relationships corresponding to the power wiring diagram into structured topology data and save it to the database. The standardized topology data model is stored in a graph data structure, where nodes represent devices and edges represent connections. The attribute set of a node includes device type, voltage level, identifier, and location information.

[0017] Preferably, it further includes: in response to an update to the power wiring diagram of a certain station, deleting the original topology data of that station from the graph topology dataset, and obtaining the structured topology data corresponding to the new power wiring diagram of that station, and updating it to the graph topology dataset.

[0018] Preferably, detecting whether the logical relationships between the components are consistent with the graph topology structure of the corresponding graph topology data includes: The device name is preprocessed to extract the primary connection relationships between its components; Based on the corresponding graph topology, obtain the actual connection relationships between the components of the device name; Compare the first connection relationship with the actual connection relationship to see if they are consistent, and generate a logical consistency verification report. The logical consistency verification report includes the reasons for the inconsistency.

[0019] The beneficial effects of this invention are as follows: By combining a knowledge base of power equipment naming standards with large-scale model technology, the method of this invention achieves dual detection from two dimensions: textual standardization and consistency of topological logic in drawings. Compared with traditional detection methods that rely solely on manual verification or simple rule matching, this invention utilizes a visual language large-scale model to transform unstructured electrical wiring diagrams into structured topological data. Combined with the semantic understanding capabilities of the large-scale model, it can not only detect errors in naming format but also deeply identify logical errors where equipment names do not match their actual physical connections in the electrical wiring diagram. This significantly improves the automation, comprehensiveness, and accuracy of power equipment name standardization detection. Attached Figure Description

[0020] Figure 1 This is a schematic flowchart illustrating a power name specification detection method based on a large model according to an embodiment of the present invention; Figure 2 This is a flowchart illustrating the construction of an electrical naming convention knowledge base according to an embodiment of the present invention. Detailed Implementation

[0021] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. 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.

[0022] The specific embodiments of the present invention will now be described in detail with reference to the accompanying drawings.

[0023] Example of a method for standardizing electricity names based on a large model: like Figure 1 As shown, the power name standardization detection method based on a large model of the present invention includes: S101. Construct a knowledge base for power equipment naming standards. Specifically, construct a knowledge base for power equipment naming standards, which will serve as the basis for large-scale model testing of equipment name standardization. The content of the power equipment naming standard knowledge base will cover equipment information standards, naming standards, and historical naming cases of power equipment. S102. Obtain the graph topology dataset, specifically by using a large visual language model to convert the electrical wiring diagrams of each station in the target area into structured graph topology data, thereby obtaining the graph topology dataset. The structured graph topology data serves as the basis for logical consistency checks on device names in subsequent steps.

[0024] In this embodiment, the visual language large model can be the Qwen3-VL large model or other suitable visual language large models.

[0025] S103. Obtain the standardization test results, specifically: obtain the name of the equipment to be tested and the name of the station to which it belongs, and use the large model combined with the power equipment naming standard knowledge base to perform standardization test on the name of the equipment to be tested, thereby obtaining the standardization test results.

[0026] S104. Obtain the logical consistency detection result, specifically: select the graph topology data corresponding to the site to which the device to be detected belongs from the graph topology dataset, use the large model to parse each component of the device name, and then detect whether the logical relationship between each component is consistent with the graph topology structure of the corresponding graph topology data, thereby obtaining the logical consistency detection result.

[0027] S105. Integrate the results of the standardization test and the logical consistency test to generate the final comprehensive test result.

[0028] The large model and the visual language large model mentioned in this embodiment can be two different large models (one of which is the visual language large model and the other can be a regular large model), or the same visual language large model can be used.

[0029] By verifying the logical relationship between each part of the device name and the actual topology, we can ensure that the name is accurate in the context of the power system.

[0030] This embodiment combines a knowledge base of electrical equipment naming conventions with large-scale model technology to achieve dual detection from two dimensions: textual standardization and consistency of topological logic in drawings. Compared to traditional detection methods that rely solely on manual verification or simple rule matching, this invention utilizes a visual language large-scale model to transform unstructured electrical wiring diagrams into structured topological data. Combined with the semantic understanding capabilities of the large-scale model, it can not only detect errors in naming format but also identify logical errors where equipment names do not match their actual physical connections in the electrical wiring diagram. This significantly improves the automation, comprehensiveness, and accuracy of electrical equipment naming convention detection.

[0031] In one embodiment, such as Figure 2 As shown, the construction of the power naming standard knowledge base includes: S201. Collect original data, specifically including: collecting standard documents related to the naming of power equipment and historical naming cases of power equipment. The standard documents include national and industry standards and specifications as well as internal naming rules documents of enterprises; historical naming cases of power equipment include correct naming examples that conform to the specifications and incorrect naming examples that do not conform to the specifications.

[0032] S202. Extracting knowledge content, specifically including: preprocessing the collected documents, including: identifying and filtering out content in the document content that is not related to the naming convention through text analysis, and extracting the naming rules from the filtered document content.

[0033] In this embodiment, content unrelated to naming conventions includes non-substantive content such as document background information and compilation purpose.

[0034] This embodiment explicitly removes irregular text such as document background descriptions and compilation purposes. This helps reduce interfering information in the input context of large models, allowing the model to focus more on learning and applying core naming rules, thereby improving retrieval and matching accuracy with limited computing resources.

[0035] S203. Perform knowledge storage, specifically including: storing the extracted naming rules and historical naming cases of power equipment into a knowledge base.

[0036] S204. Perform dynamic updates, specifically including: extracting new naming rules from documents related to the naming of new power equipment, and storing the new naming rules in the knowledge base.

[0037] Through the above construction process, a well-structured, comprehensive, and dynamically updatable knowledge base for power equipment naming standards is formed, providing a basis for the standardization testing of power equipment names.

[0038] By collecting, cleaning, and extracting rules from standard documents and historical naming cases from multiple sources, a high-quality dedicated knowledge base was constructed. Compared to directly using the original documents, this invention pre-filters noisy data unrelated to naming conventions and distinguishes between positive and negative cases, providing a more accurate and pure knowledge background for subsequent large-scale models, effectively improving the inference performance and retrieval efficiency of large-scale models in specific vertical domains.

[0039] In one embodiment, before storing the extracted naming rules and historical naming cases of power equipment into the knowledge base, the method further includes: The extracted naming rules are divided into two categories using a large model: general rules and equipment-specific rules. Historical naming cases of power equipment are divided into positive and negative cases. The general rules are the basic naming requirements applicable to all power equipment, including the naming format framework, separator usage specifications, and alphanumeric encoding rules. The equipment-specific rules are the exclusive naming requirements for different types of special equipment, including main transformers, circuit breakers, lines, and busbars.

[0040] In this embodiment, a large model is used to achieve refined classification and structured storage of knowledge base content. By dividing rules into general and device-specific rules, and cases into positive and negative instances, more targeted searches can be performed based on specific device types during subsequent detection, avoiding misjudgments of special devices by general rules and improving the scenario adaptability and recall accuracy of the detection system.

[0041] In one embodiment, before storing the new naming rule in the knowledge base, the method further includes: performing conflict detection: comparing whether there is a contradiction between the requirements of the new rule and the existing rule under the same applicable conditions; when a rule conflict is detected, generating a conflict report and pausing the knowledge base update, and waiting for manual confirmation before completing the knowledge base update.

[0042] This embodiment introduces a rule conflict detection mechanism to ensure the consistency and authority of the knowledge base. During the rule update iteration process, it can automatically identify logical contradictions between old and new rules and pause the update until manual confirmation is received. This avoids the large model from producing "illusions" or giving ambiguous detection results due to internal logical conflicts in the knowledge base, thus ensuring the stability of the system.

[0043] In one embodiment, the method further includes: experts revising the generated final comprehensive test results and writing the revised comprehensive test results into the power equipment naming standard knowledge base.

[0044] By writing the comprehensive test results revised by experts into the knowledge base, a closed-loop learning mechanism is formed.

[0045] This embodiment establishes a "human-machine loop" optimization mechanism. By writing the expert-corrected results back to the knowledge base, the system can continuously accumulate experience in handling difficult and marginal cases, realizing dynamic updates of the knowledge base and continuous evolution of the model. As usage time increases, the system's detection accuracy will continue to improve.

[0046] In one embodiment, the standardization check of the equipment name to be tested using a large model combined with a power equipment naming standard knowledge base includes: S301. Write dedicated prompt words, including: write system prompt words applicable to equipment name standardization detection. The system prompt words are used to drive the large model to detect the standardization of equipment names according to the subsequently specified general rules, specific equipment rules, positive cases and negative cases, and output the specific reasons and the rules violated when the names are not standard. S302. Retrieve the knowledge base, including: extracting keywords from the device name, using the keywords to retrieve relevant general rules and specific device rules from the knowledge base; using vector similarity retrieval to retrieve relevant positive and negative cases from the case knowledge base, and selecting the top 5 retrieved cases as reference cases for the large model to check the standardization of device names; In this embodiment, a pre-built knowledge retrieval enhancement generation mechanism can be used to retrieve knowledge from the knowledge base. The knowledge retrieval enhancement generation mechanism includes a keyword retrieval module and a semantic understanding module. The keyword retrieval module is used to perform fast matching based on keywords in the naming elements, mainly targeting specific device rules. The semantic understanding module retrieves similar cases from the cases based on the semantics of the device name.

[0047] S303. Constructing input content, including: constructing input content that meets the input requirements of the large model based on the device name, written special prompt words, and retrieved rules and reference cases, and inputting the input content into the large model to obtain the detection results; S304. Analyze the detection results, including: analyzing the detection results output by the large model, outputting a standardization verification report, including the violated rules and improvement suggestions.

[0048] The method described in this embodiment can achieve comprehensive and accurate detection of the standardization of power equipment names, not only determining whether they are standard, but also accurately locating problems and providing improvement suggestions.

[0049] By combining system prompt word engineering with Retrieval Augmentation (RAG) technology, efficient normative detection is achieved. Using positive and negative reference cases retrieved through vector similarity as context, the system guides the large model to accurately understand and apply complex rules. Simultaneously, the system can output specific reasons for violations and improvement suggestions, solving the problem of lack of interpretability in traditional "black box" models and facilitating rapid correction by users.

[0050] In one embodiment, converting the electrical wiring diagrams of each site into structured graph topology data using a large visual language model includes: S401. Collect the original power wiring diagrams of existing sites in the current target area, and preprocess and enhance them to obtain improved power wiring diagrams. The original electrical wiring diagram can be in PDF format or other formats.

[0051] S402. Input the improved electrical wiring diagram into the visual language large model to obtain the topological relationship corresponding to the electrical wiring diagram; the visual language large model is used to output the corresponding topological relationship based on the electrical wiring diagram. S403. Use a standardized topology data model to convert the topology relationships corresponding to the power wiring diagram into structured topology data and save it to the database. The standardized topology data model is stored using a graph data structure, where nodes represent devices and edges represent connections. The attribute set of a node includes device type, voltage level, identifier, and location information.

[0052] The method described in this embodiment can convert unstructured PDF drawings into structured, computable topological data, providing a data foundation for subsequent logical consistency testing.

[0053] In this embodiment, a large visual language model is used to solve the problem that traditional OCR technology struggles to understand the topology of complex electrical wiring diagrams. Through image enhancement, topology extraction, and standardized storage, unstructured drawings are transformed into a computer-readable graph database format, providing a solid data foundation for subsequent logical consistency verification and realizing the digitization and semanticization of drawing information.

[0054] In one embodiment, the method further includes: in response to an update to the power wiring diagram of a certain site, deleting the original topology data of that site from the graph topology dataset, obtaining the structured topology data corresponding to the new power wiring diagram of that site, and updating it to the graph topology dataset.

[0055] The method in this embodiment enables dynamic maintenance of graph topology data. When the site wiring diagram changes, the system can update the corresponding topology data in a timely manner, ensuring that logical consistency detection is always based on the latest power grid operating status. This avoids false detections caused by outdated drawing versions and guarantees the timeliness and reliability of the detection results.

[0056] In one embodiment, the method further includes: in response to the emergence of a new site in the target area, obtaining the structured topology data corresponding to the power wiring diagram of the new site, and updating it to the graph topology dataset.

[0057] In one embodiment, the training process of the large visual language model includes: S501. Collect multiple original electrical wiring diagrams in PDF format, and preprocess and enhance them to obtain improved electrical wiring diagrams corresponding to each original electrical wiring diagram.

[0058] In this embodiment, the preprocessing and enhancement of the original electrical wiring diagram in PDF format includes: (1) Convert the original electrical wiring diagram in PDF format into a high-resolution image using professional tools; (2) Image enhancement algorithms are used to optimize the image in order to improve the visual quality of the original drawing. The optimization process includes improving the clarity of lines and enhancing the contrast between text and background.

[0059] S502. Text annotation is performed on the improved electrical wiring diagram corresponding to the original electrical wiring diagram to describe its key electrical elements and the relationships between the key electrical elements, thereby obtaining the training set of the visual language large model; key electrical elements include circuit breakers, disconnect switches, transformers, busbars and lines; S503. Train the large visual language model using the training set.

[0060] In one embodiment, detecting whether the logical relationships between the components are consistent with the graph topology structure of the corresponding graph topology data includes: S601. Preprocess the device name and extract the first connection relationship between its components.

[0061] In this embodiment, a dedicated parser can be set up to extract the relationships between the components of the device name.

[0062] The device name is composed of electrical elements.

[0063] S602. Obtain the actual connection relationships between the components of the device name based on the corresponding graph topology; S603. Compare whether the first connection relationship is consistent with the actual connection relationship, and generate a logical consistency verification report. The logical consistency verification report includes the reasons for the inconsistency.

[0064] In this embodiment, a topology verifier can be used to verify whether the first connection relationship is consistent with the actual connection relationship.

[0065] For example: If the device name is "502 switch on the 10kV side of main transformer No. 2", the first connection relationship is that switch 502 is connected to the 10kV side of main transformer No. 2. However, the actual connection relationship obtained according to the topology diagram is that switch 502 is connected to the 10kV side of main transformer No. 1. The system finds that the two are inconsistent after comparison, and the generated logical consistency verification report contains the following content: inconsistent. The reason for the inconsistency is: the connection relationship corresponding to switch 502 is incorrect.

[0066] This embodiment details the logical consistency verification process. By comparing the implicit connection relationships parsed from the device name with the actual physical connection relationships in the graph database, deep-seated errors of "name-to-real" discrepancies can be accurately identified. This method, based on semantic and topological bidirectional verification, effectively fills the technical gap where traditional methods cannot verify the authenticity of location / connection information in names.

[0067] While this specification has shown and described numerous embodiments of the invention, it will be apparent to those skilled in the art that such embodiments are provided by way of example only. Many modifications, alterations, and alternatives will occur to those skilled in the art without departing from the spirit and essence of the invention. It should be understood that various alternatives to the embodiments of the invention described herein may be employed in the practice of this invention.

Claims

1. A method for standardizing power name detection based on a large model, characterized in that, include: A knowledge base for power equipment naming standards will be constructed and used as the basis for large-scale model testing of equipment name standardization. The knowledge base covers equipment information standards, naming standards, and historical naming cases of power equipment. The electrical wiring diagrams of each site in the target area are converted into structured graph topology data using a large visual language model, thereby obtaining a graph topology dataset. The system obtains the name of the device to be tested and the name of the site to which it belongs, and uses a large model combined with a knowledge base of power equipment naming standards to perform a standardization test on the name of the device to be tested, thereby obtaining the standardization test results. Select the graph topology data corresponding to the site to which the device to be detected belongs from the graph topology dataset, use the large model to parse each component of the device name, and then detect whether the logical relationship between each component is consistent with the graph topology structure of the corresponding graph topology data, so as to obtain the logical consistency detection result. The results of the standardization test and the logical consistency test are combined to generate the final comprehensive test result.

2. The method for detecting standardized power names based on a large model as described in claim 1, characterized in that, The construction of a knowledge base for power naming standards includes: Collect standard documents related to the naming of power equipment and historical naming cases of power equipment. The standard documents include national and industry standards and specifications as well as internal naming rules documents of enterprises; the historical naming cases of power equipment include correct naming examples that comply with the specifications and incorrect naming examples that do not comply with the specifications. Preprocessing of the collected documents includes: identifying and filtering out content in the document content that is irrelevant to the naming conventions through text analysis, and extracting the naming rules from the filtered document content; The extracted naming rules and historical naming cases of power equipment are stored in a knowledge base; Extract new naming rules from documents related to the naming of new power equipment, and store the new naming rules in the knowledge base.

3. The method for detecting standardized power names based on a large model as described in claim 2, characterized in that, Content unrelated to naming conventions includes document background information and the purpose of its creation.

4. The method for detecting standardized power names based on a large model as described in claim 2, characterized in that, Before storing the extracted naming rules and historical naming cases of power equipment into the knowledge base, the following steps are also included: The extracted naming rules are divided into two categories using a large model: general rules and equipment-specific rules. Historical naming cases of power equipment are divided into positive and negative cases. The general rules are the basic naming requirements applicable to all power equipment, including the naming format framework, separator usage specifications, and alphanumeric encoding rules. The equipment-specific rules are the exclusive naming requirements for different types of special equipment, including main transformers, circuit breakers, lines, and busbars.

5. The method for detecting standardized power names based on a large model as described in claim 2, characterized in that, Before storing the new naming rules in the knowledge base, the process also includes: performing conflict detection: comparing the requirements of the new rules and existing rules under the same applicable conditions to see if there is any contradiction; when a rule conflict is detected, a conflict report is generated and the knowledge base update is paused, and the knowledge base update is completed after manual confirmation.

6. The method for detecting standardized power names based on a large model as described in claim 1, characterized in that, Also includes: Experts revised the final comprehensive test results and wrote the revised comprehensive test results into the knowledge base of power equipment naming standards.

7. The method for detecting standardized power names based on a large model as described in claim 1, characterized in that, The process of using a large model in conjunction with a knowledge base of power equipment naming conventions to perform standardization checks on the names of the equipment to be tested includes: writing system prompts for equipment name standardization checks. These system prompts drive the large model to check the standardization of equipment names based on subsequently specified general rules, specific equipment rules, positive cases, and negative cases, and output the specific reasons and rules violated when the names are not standard. Keywords are extracted from the device name, and relevant general rules and device-specific rules are retrieved from the knowledge base using these keywords. Vector similarity retrieval is used to retrieve relevant positive and negative cases from the case knowledge base, and the top 5 retrieved cases are selected as reference cases for the large model to check the standardization of device names. Based on the device name, the specially written prompt words, and the retrieved rules and reference cases, input content that meets the input requirements of the large model is constructed, and the input content is input into the large model to obtain the detection results; The detection results output by the large model are analyzed, and a normative verification report is generated, which includes the violated rules and suggestions for improvement.

8. The method for detecting standardized power names based on a large model as described in claim 1, characterized in that, Using a large visual language model, the electrical wiring diagrams of each site are converted into structured graph topology data, including: Collect the original power wiring diagrams of existing sites in the current target area, and preprocess and enhance them to obtain improved power wiring diagrams; The improved electrical wiring diagram is input into the visual language model to obtain the corresponding topological relationships; the visual language model is used to output the corresponding topological relationships based on the electrical wiring diagram. The standardized topology data model is used to convert the topological relationships corresponding to the power wiring diagram into structured topology data and save it to the database. The standardized topology data model is stored in a graph data structure, where nodes represent devices and edges represent connections. The attribute set of a node includes device type, voltage level, identifier, and location information.

9. The method for detecting standardized power names based on a large model as described in claim 1, characterized in that, Also includes: In response to an update to the power wiring diagram of a certain site, the original topology data of that site is deleted from the graph topology dataset, and the structured topology data corresponding to the new power wiring diagram of that site is obtained and updated to the graph topology dataset.

10. The method for detecting standardized power names based on a large model as described in claim 1, characterized in that, Detecting whether the logical relationships between the components are consistent with the graph topology structure of the corresponding graph topology data includes: The device name is preprocessed to extract the primary connection relationships between its components; Based on the corresponding graph topology, obtain the actual connection relationships between the components of the device name; Compare the first connection relationship with the actual connection relationship to see if they are consistent, and generate a logical consistency verification report. The logical consistency verification report includes the reasons for the inconsistency.

Images