Power grid standard intelligent verification system based on semantic model

Abstract

The invention discloses a power grid standard intelligent verification system based on a semantic model. The power grid standard intelligent verification system comprises a data acquisition module, a preprocessing module, a semantic analysis module, a knowledge base module, a verification module and a user interface module. The system obtains power grid standard related data from structured and unstructured data sources through the data acquisition module, and data cleaning and format conversion are performed through the preprocessing module. An improved deep learning model in the semantic analysis module is used for analyzing document semantics, a multi-level and multi-dimensional knowledge graph is constructed, and high-dimensional vector representation is generated through embedding. And storing the dynamically updated knowledge graph in the knowledge base module. And according to the semantic analysis result and the knowledge base, performing intelligent verification on the power grid standard in the verification module. And the user interface module provides an interactive interface and displays a verification result and a correction suggestion. According to the system, efficient and accurate verification of the power grid standard document is realized, dynamic updating is supported, and the document management automation level is remarkably improved.

Description

Technical Field

[0001] This invention relates to the field of document information verification technology, and in particular to a smart verification system for power grid standards based on a semantic model. Background Technology

[0002] With the development of the power industry and the gradual intelligentization of power systems, the management and verification of power-related documents have become particularly important. The safe and stable operation of power systems relies on numerous complex engineering specifications and standards. In power engineering projects, construction documents, standard documents, and equipment operation and maintenance documents must adhere to specific technical requirements and format specifications. However, due to the wide variety, complexity, and high level of specialization of power documents, traditional manual verification methods are inefficient and prone to errors, failing to meet the growing demands of engineering management. Therefore, semantic model-based intelligent verification systems, designed to improve the automation and accuracy of document verification, have emerged. Currently, some related technologies have attempted to automate the verification of various documents in the power industry. For example, some existing methods utilize natural language processing technology to semantically understand power documents and then verify their content and format; other methods utilize deep learning models to optimize the verification of power construction project documents or bidding documents. These technical solutions have made some progress in improving the efficiency and accuracy of document verification, but several shortcomings remain. To further enhance the practicality and reliability of document verification technology, exploring more accurate and efficient verification methods has become a technological trend.

[0003] In traditional power document management and verification processes, manual review plays a dominant role. Document writers and reviewers meticulously check construction documents, operation and maintenance documents, and related technical data item by item according to power engineering specifications and standards. This method is not only time-consuming and labor-intensive but also prone to omissions or misjudgments due to human error, especially when dealing with large volumes of technical documents or complex technical clauses. Furthermore, manual review is highly subjective; different reviewers may arrive at different verification results, making it difficult to guarantee objectivity and consistency. To address the inefficiency of manual verification, some research has proposed rule-based document verification systems. These systems automatically verify the content and format of documents using pre-defined verification rules. For example, the system can check whether a document contains all necessary clauses and whether it conforms to established format requirements. While rule-based verification systems improve verification efficiency to some extent, their capabilities are limited when dealing with complex semantic understanding and logical verification. This is because traditional rule matching methods have weak semantic understanding capabilities and struggle to handle the complex semantic relationships implicit in natural language text. Natural Language Processing (NLP) technology has developed rapidly in recent years and is increasingly being applied to the automated verification of power documents. By utilizing techniques such as word segmentation, syntax analysis, and semantic understanding, Natural Language Processing (NLP) can extract specific structured information from text, thereby enabling document content verification. For example, Named Entity Recognition (NER) technology can be used to identify proper nouns, equipment names, and place names in power documents, helping the system understand the document content. While NLP technology has shown great potential in document verification, it still has limitations in handling domain-specific knowledge. The power industry is characterized by numerous technical terms and complex content, requiring the integration of domain ontology and knowledge graph technologies to further enhance semantic understanding capabilities. Patent CN202211021256.1 describes a method for verifying power construction documents, focusing on determining document type through a classification model and performing different verifications based on document type. This method improves verification speed and accuracy through logical and standardized verification of document content, but its understanding and processing of deep semantics are relatively limited, failing to achieve comprehensive analysis of complex semantic relationships. Patent CN202210611493.7 proposes a document verification method based on template rules, which uses document layout verification rules and content verification rules to perform real-time document verification. This method creates the target document using template content and performs verification during the document editing process, exhibiting good real-time performance. However, this method has limited support for non-template-based content or technical documents with complex formats, and struggles to handle the diverse document structures in the power industry. Patent CN201911421189.0 discloses an intelligent verification method for enterprise bidding documents, employing a plug-in design and utilizing an index database and text and image processing tools to implement verification, providing marking prompts after verification.Although this method reduces document error rates and improves document writing efficiency, its design is primarily geared towards bidding documents and does not take into account the specific standards, specifications, and complex document structure requirements unique to the power industry.

[0004] While existing technologies have achieved initial success in automating document verification, they still have the following shortcomings: Insufficient semantic understanding: Many methods only perform surface-level semantic analysis when processing document content, failing to accurately parse the complex technical concepts and semantic relationships involved, especially when dealing with power engineering terminology and multi-dimensional constraints. Lack of dynamic update capabilities: Power industry standards and specifications are frequently updated, and existing systems lack the flexibility to handle new standards and knowledge updates, making it impossible to adjust and optimize verification rules in a timely manner. Difficulty in integrating multi-source data: Power industry documents involve various data types, including structured data (such as equipment parameters), semi-structured data (such as maintenance records), and unstructured data (such as text content).

[0005] Therefore, there is a need for a smart verification system for power grid standards based on semantic models. Summary of the Invention

[0006] To address the shortcomings of existing technologies, this invention proposes a semantic model-based intelligent verification system for power grid standards, thereby solving the problems mentioned in the background. This semantic model-based intelligent verification system for power grid standards combines deep learning, knowledge graphs, and multi-source data processing technologies to achieve intelligent verification of power grid standard documents. The system includes the following modules: P1: Data acquisition module, used to acquire power grid standard-related data from multiple data sources. By establishing a unified data interface, it acquires power grid standard data from both structured and unstructured data sources. P2: Preprocessing module, which cleans, converts, and performs preliminary analysis on the collected data; P3: Semantic analysis module, which uses an improved deep learning model to perform semantic parsing on preprocessed data and extract concepts, attributes and relationships in power grid standards. It includes a graph construction submodule and a semantic embedding submodule. The semantic embedding submodule uses an improved multi-hypergraph neural network to embed nodes and edges of the knowledge graph to obtain a high-dimensional vector representation. P4: Knowledge base module, which stores a multi-level, multi-dimensional dynamic knowledge graph in the power grid field; P5: Verification module, which performs intelligent verification of power grid standards based on semantic analysis results and the knowledge base; the intelligent verification uses a similarity calculation submodule for judgment, which adopts a novel similarity calculation method based on multiple hypergraph embedding and weighted path distance, as shown in the following formula: in, Cosine similarity of node vectors; For nodes and Weighted shortest path distance in a multihypergraph; and They are nodes and The weight vector; , , To adjust the parameters, it must satisfy... ; This represents the similarity value. P6: User interface module, providing an interactive interface for users to view and process verification results.

[0007] As a preferred embodiment of the present invention, the structured data source in module P1 includes: Power Grid Standards Database: Structured data including version information, clause numbers, content summaries, and revision records of power grid standards obtained from the official standard databases released by the State Grid Corporation of China and China Southern Power Grid Company. Equipment Management System: Obtain data on the model, parameters, operating status, and maintenance records of power grid equipment from the power grid company's internal equipment management system; Monitoring and Dispatch System: Obtains real-time operating data of power load, voltage, and current from the power grid real-time monitoring and dispatch system; Geographic Information System: Acquires spatial data on the geographical location of power grid facilities, the route of power lines, and the distribution of substations.

[0008] As a preferred embodiment of the present invention, the preprocessing module in module P2 cleans, converts, and performs preliminary analysis on the power grid standard-related data acquired by the data acquisition module, specifically including: Data cleaning: missing value handling, detecting and filling missing values ​​in the data; for numerical data, the mean or median can be used for filling; duplicate data removal, detecting and removing duplicate records in the dataset to ensure the uniqueness and integrity of the data; error correction, correcting spelling errors, formatting errors and logical errors in the data based on power grid standards and domain knowledge. Format conversion: Data standardization, converting data from different sources into a unified encoding and format; Unit conversion, performing unified conversion of different units of measurement used in different data sources; Structure conversion, converting unstructured data into structured format; Preliminary analysis: Word segmentation processing: Chinese word segmentation is performed on the text data, and the accuracy of word segmentation is improved by using professional dictionaries and custom thesaurus in the power grid field; Named entity recognition: proper nouns, technical terms, equipment names, place names, and personal names in power grid standards are identified to improve the understanding of power grid professional content.

[0009] As a preferred embodiment of the present invention, the graph construction submodule in the semantic analysis module P3 is specifically constructed using a multi-grid knowledge supergraph construction method. The graph construction submodule is constructed using the following steps: S1: Domain Ontology Construction: Based on the ontology and professional knowledge of the power grid domain, define concepts, attributes, and relationships to form an ontology set. ,in, For a set of concepts, For a collection of attributes, For a set of relations; the ontology of the power grid domain mentioned in step S1 includes: Power generation equipment itself: Defines the concept, attributes, and relationships of power generation equipment, including thermal power generating units, hydropower generating units, nuclear power generating equipment, wind power generating equipment, and photovoltaic power generating equipment; its attributes include rated power, power generation efficiency, fuel type, and emission indicators; its relationships include "connected to" the power grid and "controlled by" the dispatch center; The main body of the power transmission system: defines the concepts and attributes of power transmission equipment such as transmission lines, substations, transformers, circuit breakers, and surge arresters; describes the topology, voltage levels, and line parameters of the power transmission network; relationships include "connected to" other equipment and "protected by" protective devices; The power distribution system itself includes power distribution transformers, power distribution lines, switchgear, and user terminals; it defines the operating mode, load characteristics, and power supply area of ​​the power distribution system; the relationships include "power supply to" users and "monitoring by" sensors; Power Grid Dispatch and Control Ontology: Describes the power grid dispatch center, automated control system, dispatch instructions, and load forecasting model; attributes include dispatch strategy, control parameters, and forecast accuracy; relationships include "issuing" dispatch instructions and "executing" control strategies.

[0010] S2: Multi-level Knowledge Graph Construction: Utilizing ontology sets, concepts, attributes, and instances in power grid standards are organized into a multi-level, multi-dimensional knowledge graph. The graph structure is a multi-hypergraph, represented as follows: ,in, It is a collection of nodes, including concept nodes, attribute nodes, and instance nodes; The knowledge graph is a set of hyperedges, each connecting multiple nodes, representing complex multi-dimensional relationships and constraints. The knowledge graph described in step S2 is modeled using a multi-hypergraph structure, and the calculation formula for the multi-hypergraph structure is as follows: ,in, A set of nodes, including concept nodes. Instance nodes and attribute nodes ,Right now ; Let be a set of superedges, each superedge It is a binary tuple ,in, The set of nodes connected by the hyperedge, containing two or more nodes; For the type label of the superedge, This indicates the type of relation or semantic information expressed by the hyperedge.

[0011] S3: Weight Allocation: During graph construction, weights are introduced for nodes and hyperedges. The weight is used to quantify its importance in the power grid standard, and the weight calculation formula is as follows: in, For nodes or edges The weights; For nodes or edges Frequency of appearance in power grid standards; As a weighting factor; The total number of nodes or edges; As a preferred embodiment of the present invention, the knowledge base module in module P4 stores a multi-level, multi-dimensional dynamic knowledge graph in the field of power grids, and the structure of the knowledge graph specifically includes: Multi-layered structure: The concept layer contains core concept nodes in the power grid field, including power generation equipment, transmission systems, distribution systems, dispatch control, equipment maintenance, electricity markets, and policies and regulations. These concepts originate from the ontology and professional knowledge of the power grid field, forming a concept set. The relationship layer defines the edges between concepts, including "connected to," "controlled," "monitored," "followed," "influenced," "participated in," and "dependent on," forming a relationship set. The edges in the relationship layer represent the logical and functional relationships between concepts, supporting multiple relationship types and attributes. The instance layer contains specific instance nodes, including specific models of equipment, specific transmission lines, actual standards and specifications documents, and operational data records, forming an instance set. The instance layer is connected to the concept layer through "belongs to" or "instantiated" relationships. Multi-dimensional attributes: Attribute nodes attach attributes to concept and instance nodes, including technical parameters, status information, geographical location information, and timestamps, forming an attribute set; Attribute relationships: Attributes are connected to concept or instance nodes through attribute relationships, which describe the characteristics and attribute values ​​of the nodes and support complex attribute types and data formats.

[0012] As a preferred embodiment of the present invention, the power grid standard intelligent verification system supports dynamic updating of the knowledge graph, which specifically includes the following steps: New Data Acquisition: After the data acquisition module acquires newly added power grid standards, equipment information, operational data, and professional knowledge, the preprocessing module cleans, converts, and performs preliminary analysis to generate incremental datasets. ; Incremental knowledge graph update: New node identification and semantic analysis module process the incremental dataset to extract new concepts, instances, and attributes, forming a new set of nodes. New hyperedge generation: Based on the new relations and constraints in the incremental data, a new set of hyperedges is generated. The knowledge graph is updated by incorporating newly added nodes and hyperedges into the existing knowledge graph. The updated node set and hyperedge set are as follows: in, For the set of old nodes, For the old superedge set, For the new set of nodes, For the new superedge set; Node and hyperedge weight updates: Weight calculation, for newly added nodes and hyperedges, calculate their weights. The calculation formula is: in, For nodes or superedges Frequency of occurrence in incremental data; For nodes or superedges Frequency of occurrence in incremental data; For nodes or superedges Importance rating; For nodes or superedges Importance rating; This represents the total number of new nodes and superedges in the incremental data.

[0013] As a preferred embodiment of the present invention, the P5 verification module calculates the similarity... and preset threshold Intelligent verification of power grid standards: when At that time, it was determined that the input power grid data contained erroneous data; when At that time, the input power grid data is determined to be correct.

[0014] Compared with the prior art, the beneficial effects of the present invention are as follows: This invention comprises six modules: data acquisition, preprocessing, semantic analysis, knowledge base, verification, and user interface. The system acquires power grid standard-related data from both structured and unstructured data sources through the data acquisition module, followed by data cleaning and format conversion in the preprocessing module. The semantic analysis module utilizes an improved deep learning model to parse document semantics, constructing a multi-level, multi-dimensional knowledge graph, and generating high-dimensional vector representations through embedding. The dynamically updated knowledge graph is stored in the knowledge base module. Based on the semantic analysis results and the knowledge base, the verification module performs intelligent verification of power grid standards. The user interface module provides an interactive interface displaying verification results and correction suggestions. This system achieves efficient and accurate verification of power grid standard documents, supports dynamic updates, and significantly improves the automation level of document management. Attached Figure Description

[0015] To more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the accompanying drawings used in the description of the specific embodiments or the prior art will be briefly introduced below. In all the drawings, similar elements or parts are generally identified by similar reference numerals. In the drawings, the elements or parts are not necessarily drawn to scale.

[0016] Figure 1 This is a system schematic diagram of the present invention. Detailed Implementation

[0017] 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.

[0018] It should be understood that, when used in this specification and the appended claims, the terms "comprising" and "including" indicate the presence of the described features, integrals, steps, operations, elements and / or components, but do not exclude the presence or addition of one or more other features, integrals, steps, operations, elements, components and / or collections thereof.

[0019] It should also be understood that the terminology used in this specification is for the purpose of describing particular embodiments only and is not intended to limit the invention. As used in this specification and the appended claims, the singular forms “a,” “an,” and “the” are intended to include the plural forms unless the context clearly indicates otherwise.

[0020] It should also be further understood that the term "and / or" as used in this specification and the appended claims refers to any combination of one or more of the associated listed items and all possible combinations, and includes such combinations.

[0021] Example 1: As Figure 1 As shown, the intelligent verification system for power grid standards based on a semantic model mainly consists of the following modules: a data acquisition module, a preprocessing module, a semantic analysis module (including a graph construction submodule and a semantic embedding submodule), a knowledge base module, a verification module, and a user interface module. The system's operation flow is as follows: First, the data acquisition module acquires relevant data on power grid standards; then, the preprocessing module cleans and converts the data; next, the semantic analysis module parses the data, constructs a multi-level knowledge graph, and generates high-dimensional vector embeddings; finally, the verification module performs intelligent verification based on the knowledge graph and deep semantic analysis results, while the user interface module provides users with interactive verification result display and processing functions.

[0022] The data acquisition module is used to obtain power grid standard-related data from various structured and unstructured data sources. Specifically, it includes: Power Grid Standards Database: The data acquisition module obtains version information, clause numbers, content summaries, and revision records of power grid standards from the official standard databases published by the State Grid Corporation of China and China Southern Power Grid Company via a dedicated API interface. These data sources are typically stored in the form of SQL databases, and the system executes SELECT queries through the data interface to retrieve the required data.

[0023] Equipment Management System: This system retrieves equipment model, parameters, operating status, and maintenance records from the power grid company's equipment management system. By integrating the OPC (OLE for Process Control) protocol interface, the data acquisition module can access equipment operating data in real time and perform batch imports.

[0024] Monitoring and Dispatch System: This module connects to the power grid monitoring and dispatch system, acquiring real-time operational data such as power load, voltage, and current through real-time data streams (e.g., SCADA systems). To this end, the system utilizes a data-push-based message bus architecture (e.g., Kafka or MQTT) to achieve efficient acquisition of large-scale real-time data.

[0025] Geographic Information System (GIS): The data acquisition module calls the GIS Web service API to obtain the geographical location, route, and substation distribution data of power grid facilities in JSON format, ensuring the integrity and accuracy of spatial data.

[0026] The preprocessing module is responsible for cleaning, format conversion, and preliminary analysis of the data acquired from the data acquisition module to ensure the accuracy of subsequent semantic analysis. The specific steps are as follows: Data Cleaning: Missing Value Handling: The system employs different strategies to handle missing values ​​in the collected data. For numerical data, the mean or median is used for imputation; for categorical data, pattern imputation or context-based semantic analysis-based missing value completion methods are used. Deduplication: The system utilizes hash fingerprinting technology to detect duplicates in large-scale data and employs parallel processing technology to accelerate the deduplication process, ensuring data uniqueness. Error Correction: The system performs error correction based on predefined power grid standards and a domain knowledge base, including spell checking, format standardization, and logical consistency checks. For example, the system automatically identifies the equivalence between "kilowatt" and "kW" and performs format conversion.

[0027] Format Conversion: Data Standardization: The system converts data from different sources into standardized encoding and formats. Power parameters are uniformly converted to international standard units (e.g., converting "megawatt" to "MW"). Structured Conversion: For unstructured text data, the system uses natural language processing technology to convert it into structured data. Specifically, it utilizes word segmentation and syntactic analysis to extract key attributes and store them in JSON or relational database formats. Preliminary Analysis: Word Segmentation: The system uses a deep learning-based Chinese word segmentation tool, combined with a professional dictionary for the power grid field and a custom thesaurus, to improve segmentation accuracy. For example, "transmission line protection device" is recognized as a complete term, rather than a separate word. Named Entity Recognition (NER): Using a pre-trained BERT model combined with a power industry terminology database, the system identifies equipment names, technical terms, place names, and personal names in documents, providing support for subsequent semantic analysis.

[0028] The semantic analysis module performs in-depth analysis of the preprocessed data, extracting concepts, attributes, and relationships from power grid standards, and constructing a dynamic knowledge graph. This module includes a graph construction submodule and a semantic embedding submodule.

[0029] Graph Construction Submodule (P3-1): Domain Ontology Construction: The system integrates professional knowledge from the power engineering field to construct a complete domain ontology, including power generation equipment, transmission systems, distribution systems, and dispatch control systems. Each ontology contains detailed definitions of concepts, attributes, and relationships. For example, the concept of "power generation equipment" includes "thermal power generating units," "hydropower generating units," etc., attributes include "rated power," "fuel type," etc., and relationships include "connected to."

[0030] Multi-level knowledge graph construction: The system organizes the above ontology information into a multi-level, multi-dimensional knowledge graph. By using a multi-hypergraph structure, the system can represent complex multi-dimensional relationships. For example, the relationship between "transmission line" and "surge arrester" can be represented as a weighted hyperedge, describing its protective function.

[0031] Weighting: Each node and hyperedge in the knowledge graph is assigned a weight, which is determined by the frequency and importance score of the node or edge. For example, "substation" is a key node and has a high weight.

[0032] Semantic Embedding Submodule (P3-2): The system uses a multi-hypergraph neural network to embed nodes and edges into the knowledge graph. Specifically, an improved GNN model is used to vectorize node features into high dimensions, generating semantic vector embeddings to enhance semantic representation capabilities. Each node vector represents the multi-dimensional semantic features of a concept or instance and forms semantic associations with other nodes in the knowledge base.

[0033] The knowledge base module stores a dynamic knowledge graph in the power grid field, supporting multi-level and multi-dimensional knowledge representation. The design of the knowledge base includes the following details: Conceptual layer: Contains core concept nodes in the power grid field, such as "power generation equipment" and "power transmission system". Each node is connected to related attributes and instances.

[0034] Relationship layer: Defines various relationships between concepts, including "connected to", "monitoring", "protecting", etc., and supports weighted relationships.

[0035] Instance layer: Stores specific device instances, standard documents, and real-time operating data, such as the operating parameters of a certain type of transformer.

[0036] Dynamic updates: When the data acquisition module acquires new data, the system will automatically update the knowledge graph, identify new nodes and hyperedges, and recalculate the weights.

[0037] The verification module performs intelligent verification tasks based on semantic analysis results and a knowledge base. The verification method is based on similarity. and preset threshold Intelligent verification of power grid standards: when At that time, it was determined that the input power grid data contained erroneous data; when At that time, the input power grid data is determined to be correct.

[0038] The user interface module provides an interactive interface, allowing users to easily view and process verification results. Specific functions include: Verification Result Display: The system displays the verification results in a visual manner, highlighting the parts that do not meet the standards, and providing detailed reasons for the errors and suggestions for correction.

[0039] Search and Query: Users can search for standards and equipment information in the knowledge base using keywords and obtain verification reports for relevant documents.

[0040] Visual analysis: The system generates a visual display of the knowledge graph, allowing users to intuitively understand the relationships and verification logic between power grid standards.

[0041] Example 2: This example demonstrates how to use a semantic model-based intelligent verification system for power grid standards to perform intelligent verification of power grid standard documents. By simulating the actual operation process, and combining simulation data and tables to display the system's specific functions and verification results, the reliability and effectiveness of the system are verified.

[0042] The system's workflow includes the following steps: data acquisition; data preprocessing; semantic analysis and knowledge graph construction; intelligent document verification; verification result analysis and user feedback.

[0043] This embodiment simulates a power grid standard document, which includes the following key parts: Basic document information: standard number, publication date, and version number; Technical clauses: descriptions of power grid equipment parameters, operating conditions, installation requirements, etc.; Equipment list: specific equipment models and their parameters; Logical relationship description: connection, protection, and monitoring relationships between equipment. Simulation data is shown in Table 1. Table 1 Simulation data of power grid standard documents Users upload standard documents to the system's user interface. The system automatically invokes the data acquisition module to extract relevant data from multiple sources, including the standard database, equipment management system, and monitoring and scheduling system. The system successfully retrieves the basic information of the document and its associated equipment parameters from the database and stores the data in system memory for preprocessing. The data acquisition is shown in Table 2.

[0044] Table 2 Simulation data of power grid standard documents The system cleans and converts the collected data, handling missing values, duplicate data, spelling errors, etc., and standardizes the data. All data is standardized; for example, rated power is standardized to the international standard unit kVA. Spelling errors are corrected, and duplicate entries are deleted. The preprocessing results are shown in Table 3.

[0045] Table 3 Data Preprocessing Results The system utilizes an improved deep learning model to perform semantic parsing on standard documents, extracting concepts, attributes, and relationships, and constructing a multi-layered dynamic knowledge graph. The system successfully generated a knowledge graph containing nodes such as "transformer," "surge arrester," "busbar," and "sensor," and assigned weights to each node. The knowledge graph structure is stored in the form of a multi-hypergraph, and the node and edge information is shown in Table 4.

[0046] Table 4 Knowledge Graph Node and Edge Information Based on the knowledge graph and semantic analysis results, the system performs complete specification and logical verification on the standard documents. The system calculates the similarity between nodes and checks whether the technical clauses and equipment parameters meet logical and standard requirements. The system detected a potential error in the technical clauses: when describing the rated voltage of the surge arrester, the user mistakenly wrote "200kV" instead of "220kV," and the system issued a warning and provided correction suggestions. The system further verified the logical relationships, confirming that the protection relationship between the transformer and the surge arrester, as well as the monitoring relationship of the sensors, all comply with the specifications. The verification results are shown in Table 5.

[0047] Table 5 Document Validation Results The system displays the verification results visually on the user interface, allowing users to view the specific content, results, and correction suggestions for each verification item. Users correct the document content according to the prompts and resubmit for a second verification. For example, the user modified the rated voltage description of the surge arrester in the document based on the correction suggestions. After resubmitting, the system showed that all verification items met the standards and the document was compliant.

[0048] The above operational process demonstrates that this system possesses the following characteristics: Highly efficient data processing: The system can automatically collect multi-source data and perform standardized preprocessing, significantly improving data quality. Precise semantic analysis: Utilizing deep learning and knowledge graph technologies, the system accurately extracts concepts and relationships from documents, achieving deep semantic understanding. Comprehensive intelligent verification: The system employs multiple verification methods to ensure the standardization and consistency of document content and logic, reducing human error.

[0049] The above embodiments merely illustrate implementation methods of the present invention, and while the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the invention. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of the present invention, and these all fall within the protection scope of the present invention.

Claims

1. A semantic model based grid standard intelligent verification system, characterized in that, The application relates to an intelligent power grid standard checking system and method. The data acquisition module is used for acquiring power grid standard related data from various data sources, and the power grid standard data is acquired from structured and unstructured data sources through establishment of a unified data interface; The preprocessing module is used for cleaning, format conversion and preliminary analysis of the acquired data; The semantic analysis module is used for semantic analysis of the preprocessed data by using an improved deep learning model, and concepts, attributes and relationships in the power grid standard are extracted, the semantic analysis module comprises a graph construction submodule and a semantic embedding submodule, wherein the semantic embedding submodule is used for embedding representation of nodes and edges of the knowledge graph by using an improved multiple hypergraph neural network, so that high-dimensional vector representation is obtained; The knowledge base module is used for storing a multi-level and multi-dimensional dynamic knowledge graph in the power grid field; The checking module is used for intelligent checking of the power grid standard according to the semantic analysis result and the knowledge base; the intelligent checking is determined by using a similarity calculation submodule; The user interface module is used for providing an interactive interface for a user to check and process the checking result.

2. The system of claim 1, wherein, The structured data sources comprise the following: The power grid standard database is used for acquiring version information, clause numbers, content abstracts and revision records of the power grid standard from a standard database officially published by the State Grid Corporation of China and the Southern Power Grid Corporation; The equipment management system is used for acquiring model numbers, parameters, running states and maintenance record data of power grid equipment from an equipment management system in the power grid enterprise; The monitoring and dispatching system is used for acquiring real-time running data of power load, voltage and current from a real-time monitoring and dispatching system of the power grid; The geographic information system is used for acquiring spatial data of geographic positions, line directions and distribution of power grid facilities.

3. The system of claim 1, wherein, The preprocessing module is used for cleaning, format conversion and preliminary analysis of the power grid standard related data acquired by the data acquisition module, and the cleaning, format conversion and preliminary analysis specifically comprise the following steps: Data cleaning: missing value processing, detection and filling of missing values in the data, for numerical data, mean value or median value filling can be adopted; repeated data deletion, detection and deletion of repeated records in the data, so that the uniqueness and integrity of the data are ensured; error correction, correction of spelling errors, format errors and logical errors in the data according to the power grid standard and field knowledge; Format conversion: data standardization, conversion of data from different sources into unified coding and format; unit conversion, unified conversion of different measurement units used in different data sources; structured conversion, conversion of unstructured data into structured format; Preliminary analysis: word segmentation processing, Chinese word segmentation of text data, and use of a professional dictionary and a self-defined word library in the power grid field to improve the accuracy of word segmentation; named entity recognition, recognition of special names, technical terms, equipment names, place names and personal names in the power grid standard, and improvement of understanding of the professional content of the power grid.

4. The system of claim 1, wherein, The similarity calculation submodule adopts a novel similarity calculation method based on multiple hypergraph embedding and weighted path distance, and the formula is as follows: wherein, is the cosine similarity of the node vectors; is the node and weighted shortest path distance in a multi-hypergraph; and are the weight vectors of the nodes and respectively; , , is a tuning parameter satisfying ; is the similarity value.

5. The system of claim 1, wherein, The graph construction submodule is constructed by using the following steps: S1: Domain ontology construction: based on the ontology of the power grid domain and professional knowledge, define concepts, attributes and relationships to form an ontology set wherein, is a concept set, is an attribute set, is a relationship set; S2: Multi-level knowledge graph construction: Use ontology set to organize the concepts, attributes and instances in the power grid standard into a multi-level and multi-dimensional knowledge graph. The graph structure is a multi-supergraph, represented as wherein, is a node set, including concept nodes, attribute nodes and instance nodes; is a superedge set, each superedge connecting multiple nodes, representing complex multi-element relationships and constraint conditions; S3: Weight distribution: In the process of graph construction, the weights of nodes and hyper-edges are introduced , to quantify its importance in the grid standard, the weight calculation formula is: wherein, is the weight of a node or edge ; is the frequency of occurrence in the grid standard ; is the weight factor ; is the total number of nodes or edges.

6. The system of claim 5, wherein, The ontology in the power grid field in step S1 comprises the following: Power plant ontology: define the concept, property and relationship of power plant, including thermal power unit, hydroelectric power unit, nuclear power plant, wind power plant, photovoltaic power plant; its properties include rated power, power generation efficiency, fuel type, emission indicators; relationships include "connected to" the power grid, "controlled by" the dispatch center; Transmission system ontology: define the concept and properties of transmission lines, substations, transformers, circuit breakers, and arrester transmission equipment; describe the topology, voltage level, and line parameters of the transmission network; relationships include "connected to" other devices, "protected by" protection devices; Distribution system ontology: includes distribution transformers, distribution lines, switching devices, and user terminals; defines the operation mode, load characteristics, and power supply area of the distribution system; relationships include "power supply to" users, "monitored by" sensors; Grid dispatch and control ontology: describes the grid dispatch center, automation control system, dispatch instructions, and load forecasting model; attributes include dispatch strategy, control parameters, and prediction accuracy; relationships include "issue" dispatch instructions, "execute" control strategies.

7. The system of claim 5, wherein, The knowledge graph in step S2 is modeled by a multi-hypergraph structure, and the calculation formula of the multi-hypergraph structure is wherein, is a node set, including a concept node , an instance node and an attribute node , i.e. ; is a hyperedge set, and each hyperedge is a binary tuple , wherein, is a node set connected by the hyperedge, containing two or more nodes; is a type label of the hyperedge, indicating the relationship type or semantic information expressed by the hyperedge.

8. The system of claim 5, wherein, The knowledge base module stores a multi-level, multi-dimensional dynamic knowledge graph in the power grid field, and the structure of the knowledge graph specifically includes: Multi-level structure: the concept layer contains core concept nodes in the power grid field, including power plants, transmission systems, distribution systems, dispatching and control, equipment maintenance, power markets, and policies and regulations. These concepts are derived from the ontology and professional knowledge of the power grid field to form a concept set. The relationship layer defines the relationship edges between concepts, including "connected to", "controlled by", "monitored by", "followed by", "affected by", "participated in", and "dependent on", to form a relationship set. The edges of the relationship layer represent the logical and functional relationships between concepts, supporting multiple relationship types and attributes. The instance layer contains specific instance nodes, including specific types of equipment, specific transmission lines, actual standard and specification documents, and operation data records, forming an instance set. The instance layer is connected to the concept layer through "belongs to" or "instantiates" relationships; Multi-dimensional attributes: attribute nodes add attributes to concept and instance nodes, including technical parameters, state information, geographic location information, and timestamps, forming an attribute set. Attribute relationships: attributes and concept or instance nodes are connected through attribute relationships, describing the characteristics and attribute values of the nodes, supporting complex attribute types and data formats.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium includes a stored program, wherein when the program runs, the device where the computer-readable storage medium is located executes the semantic model-based power grid standard intelligent verification system of any one of claims 1-8.

10. A processor, comprising: The processor is configured to run a program, wherein the program runs to execute the semantic model-based power grid standard intelligent verification system of any one of claims 1-8.

Images