A method and system for fast retrieval of general power grid model based on vector

Abstract

The present invention discloses a method and system for rapid retrieval of a general power grid model based on vectors, the method comprising: extracting semantic information and vectorizing structured data of the general power grid model, extracting text information and vectorizing unstructured data, and also extracting feature vectors from pictures and videos; establishing a mapping relationship between model information and vectors, establishing a mapping relationship between unstructured data and vectors, and storing them in a vector database; when the system retrieves a device, the semantic information to be retrieved is converted into a vector, matching and retrieving each mapping table, and finding the corresponding general power grid model and its unstructured data based on the association established with the device as the center. The present invention takes the device as the benchmark, establishes vectors of association of multiple models, and performs storage and retrieval based on the vector database, which is more concise and effective.

Description

Technical Field

[0001] The invention belongs to the technical field of power grid universal model data management, and relates to a method and system for rapid retrieval of a vector-based power grid universal model. Background Art

[0002] The substation equipment model covers the information models of all equipment in the substation, including structured data such as the equipment general ledger model, power grid model, geographic area model, primary equipment model, secondary equipment model, and auxiliary equipment model, as well as unstructured data such as equipment maintenance records, fault reports, pictures, and videos. This information is distributed in different systems, and the power grid general model is a unified model system that integrates equipment models in multiple systems.

[0003] The unified model system requirements include:

[0004] The power grid model should include the power grid area model, substation model, voltage level model, interval model, and primary equipment topology model; the models are in the following relationship: the power grid area model includes the substation model, the substation model includes the voltage level model, and the voltage level model includes the interval model; the primary equipment topology model is modeled according to the physical connection relationship of the primary equipment of the plant and station, reflecting the topological connection relationship of the primary equipment. The substation geographical area model includes the substation area model, the panel cabinet model, the cable connection model, and the device equipment model; the substation area model includes models such as walls, switch yards, interval places, buildings, floors, rooms, roads, and cable trenches, and can define multi-level regional relationship models according to the actual geographical location relationship; relative to the upper area, all lower areas do not overlap. The primary equipment model should include equipment models such as transformers, circuit breakers, busbars, switches, lines, capacitors / reactors, mutual inductors, arc suppression coils, high-voltage reactors, lightning arresters, carts, and grounding transformers; the primary equipment model is included in the interval model in the power grid model, and its deployment location is included in the substation geographical area model; the primary model includes measurement, control, and online detection point models. The secondary equipment model should include equipment models such as measurement and control, protection, safety automatic device, metering, intelligent recorder, AC and DC system; its deployment location is included in the substation geographical area model, described as the area or cabinet attribute of the secondary equipment; there is a correlation between the primary equipment and the secondary equipment. The auxiliary equipment model should include equipment models such as online monitoring, power environment, security defense, fire protection, etc.; its deployment location is included in the substation geographical area model, described as the area or cabinet attribute of the secondary equipment; there is a correlation between the primary equipment and the auxiliary equipment. The equipment general ledger model should include two parts: general ledger and general technical parameters; the general ledger defines the basic equipment information such as equipment type, equipment name, manufacturer, production date, etc.; the general technical parameters include software and hardware version information, factory code, etc.; the secondary equipment and auxiliary equipment contain the general technical parameter model, and all equipment contain general ledger information. Unstructured data based on equipment, including maintenance records, fault reports, pictures, videos, etc., are stored in the system as files.

[0005] After the general power grid model integrates the power grid model, geographic area model, primary equipment, secondary equipment, auxiliary equipment, ledger model and unstructured data, it will become difficult for the application business system to retrieve the model in actual applications.

[0006] In addition, due to the existence of manual configuration in various systems, the same information may be presented or described differently in different application systems. Taking the four remote controls as an example, although the secondary system optimization stipulates the protection standard model, the versions of the existing equipment are different, and the models and descriptions are different. The configuration of on-site business data relies on manual configuration and association, which not only requires a large workload, but is also prone to configuration errors, affecting the correctness of subsequent event analysis and other functions. Text description information in different formats is machine-recognized while ensuring recognition accuracy. Conventional natural language recognition methods have a certain error rate and are difficult to meet high-reliability and high-accuracy application scenarios such as four-remote control matching and recording channel association analysis. Summary of the invention

[0007] In order to solve the deficiencies in the prior art, the present invention provides a method and system for rapid retrieval of a general power grid model based on vectors.

[0008] The present invention adopts the following technical solution.

[0009] A first aspect of the present invention provides a method for quickly retrieving a general power grid model based on a vector, comprising:

[0010] Step 1: extracting semantic information from structured data in the general power grid model, vectorizing the extracted semantic information and saving it in a model vector mapping table;

[0011] Step 2: vectorize the semantic information of the text of the unstructured data of the power grid system, extract feature vectors from the pictures or videos of the unstructured data, and store them in the corresponding file vector information table;

[0012] Step 3: When searching for data, vectorize the search semantic information, use vector matching to search the model vector mapping table and vector information table, and obtain the corresponding information.

[0013] Preferably, in step 1, semantic information is extracted from structured data in the general power grid model at different dimensions of the same device, the semantic information is converted into word vectors, the vectors are associated with the corresponding models and saved in the model vector mapping table to form a mapping based on the same device.

[0014] Preferably, the general power grid model includes a power grid model, a geographical area model and a ledger model.

[0015] Preferably, in step 1, the semantic information extracted from the power grid model includes the power grid area name, plant station name, voltage level name and equipment name;

[0016] The semantic information extracted from the geographic area model includes a substation geographic area, an in-station geographic area, a smaller level geographic area, and an equipment geographic area;

[0017] The semantic information extracted from the ledger model includes device name, Chinese description, device model, manufacturer, country of production, and production date.

[0018] Preferably, the model vector mapping table includes: power grid equipment ID, power grid equipment vector, geographic area equipment ID, geographic area equipment vector, equipment ledger ID, equipment ledger vector, wherein the power grid equipment ID, geographic area equipment ID, and equipment ledger ID are primary key IDs derived from the power grid model, geographic area model, and ledger model, respectively; the power grid equipment vector, geographic area equipment vector, and equipment ledger vector are semantic vectors associated with the power grid model, geographic area model, and ledger model, respectively.

[0019] Preferably, the unstructured data includes equipment maintenance records, equipment failure reports, and equipment pictures and videos.

[0020] Preferably, step 2 comprises: extracting semantic information from the equipment maintenance record, equipment fault report, equipment picture or video file description information text, converting the semantic information into word vectors, associating the vectors with the corresponding files and saving them in the equipment maintenance record, fault report, equipment video picture vector information table;

[0021] Extract feature vectors from the images or videos of the device image or video file, associate them with the corresponding files, and save them in a device video image vector information table.

[0022] Preferably, the fault report vector information table, the equipment maintenance record vector information table and the equipment video picture vector information table are in a one-to-many association relationship.

[0023] Preferably, in step 3, in the model vector mapping table, the power grid equipment vector, the geographic area equipment vector and the equipment inventory vector are retrieved by retrieving the semantic information word vector to obtain the record of the most matching vector, and the power grid equipment ID, geographic area equipment ID and equipment inventory ID in the record are used to retrieve the corresponding records from the power grid equipment model, the geographic area equipment model and the equipment inventory model to obtain various model specific data, and the vector field content in the record is used to retrieve the corresponding records with the highest vector matching degree from the fault report vector information table, the equipment maintenance record vector information table and the equipment video picture vector information table to obtain various file contents.

[0024] Preferably, step 3 also includes: performing a secondary search through the feature vectors of the pictures or videos in the acquired file content, retrieving pictures or videos whose matching degree exceeds a set value in the device video picture vector information table, realizing image search, and obtaining related fault reports and maintenance records based on the association relationship between the fault report vector information table, the equipment maintenance record vector information table and the equipment video picture vector information table, and integrating and uniformly displaying the various information obtained.

[0025] A second aspect of the present invention provides a system for rapid retrieval of a general power grid model based on a vector, comprising:

[0026] The structured data processing module is used to extract semantic information from the structured data in the general power grid model, vectorize the extracted semantic information and save it in the model vector mapping table;

[0027] The unstructured data processing module is used to vectorize the semantic information of the text of the unstructured data of the power grid system, extract the feature vector of the picture or video of the unstructured data, and store it in the corresponding file vector information table;

[0028] The retrieval module is used to vectorize the retrieval semantic information when searching for data, and use vector matching to retrieve the model vector mapping table and vector information table to obtain the corresponding information.

[0029] Compared with the prior art, the beneficial effects of the present invention include at least:

[0030] In view of the insufficient association, storage and retrieval capabilities of various models in the existing system after the fusion of the general power grid model, the present invention proposes a device-centric approach to establish the association and mapping of various models based on semantic vectors, and to store them based on a vector database to achieve fast retrieval.

[0031] The present invention is based on a universal model of various normalized structures in the power grid, extracts semantic information in the specifications, obtains semantic vectors generated by the respective descriptive information of the power grid model, geographic area model, and ledger model to which the equipment belongs, and transforms the original mapping method for establishing a strong consistent relationship between multiple models, that is, establishing a multi-model ID mapping method, into a method for establishing mapping with semantic vectors, abstracting the descriptive information in the model into a vector of mathematical expression, which is more concise and effective, can better express and understand the semantics of the data, makes the data easier to understand and use in application development and data analysis, and enables advanced applications based on multiple models to complete retrieval according to semantics.

[0032] The present invention vectorizes semantic information of text of unstructured data associated with the device, extracts feature vectors of pictures or videos, and establishes a mapping relationship between these vectors and unstructured data, so that unstructured data and structured data, unstructured data and unstructured data are semantically associated; the present invention stores the established vectors and model data in a vector database, and uses the retrieval function of the vector database to perform rapid model retrieval and semantic matching, which can provide support for advanced semantic-based applications across power grid systems.

[0033] The present invention is based on vectors and mappings, and performs semantic related queries in the system. For example, the power grid model information is input, and the information is converted into vectors, and then queried in the mapping table according to the vector matching degree. For the query results, the semantic vectors of other structured data are found through the mapping relationship, and then the vector matching retrieval of unstructured data is performed through the semantic vector. Finally, the associated pictures and video content can be retrieved through their feature vectors in the form of image search and feature vector similarity retrieval to retrieve more related reports, records and other unstructured data. It retrieves various related data in the power data through the matching retrieval of semantic vectors and feature vectors and the image search method, thereby realizing the rapid retrieval of the general power grid model. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] Figure 1 It is a flow chart of a method for rapid retrieval of a general power grid model based on vectors of the present invention;

[0035] Figure 2 It is a model entity vector transformation and relationship diagram in an embodiment of the present invention;

[0036] Figure 3 It is the model, vector mapping table and relationship diagram stored in the vector database in the embodiment of the present invention. DETAILED DESCRIPTION

[0037] In order to make the purpose, technical scheme and advantages of the present invention clearer, the technical scheme of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. The embodiments described in this application are only embodiments of a part of the present invention, rather than all embodiments. Based on the spirit of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work belong to the protection scope of the present invention.

[0038] like Figure 1-Figure 2 As shown, Embodiment 1 of the present invention provides a method for quickly retrieving a general power grid model based on a vector. In a preferred but non-limiting embodiment of the present invention, the method comprises the following steps:

[0039] Step 1: extracting semantic information from structured data in the general power grid model, vectorizing the extracted semantic information and saving it in a model vector mapping table;

[0040] This step extracts semantic information from the power model description information in the power system. The models include power grid model, regional model, primary equipment, secondary equipment, auxiliary equipment model, equipment ledger model, etc., and are associated with the equipment as the center.

[0041] Further preferably, semantic information is extracted from structured data in the general power grid model at different dimensions of the same device, the semantic information is converted into word vectors, the vectors are associated with the corresponding models and saved in the model vector mapping table to form a mapping based on the same device.

[0042] The general power grid model includes a power grid model, a geographical area model and a ledger model.

[0043] The semantic information of the equipment extracted from the data of the general power grid model includes: power grid model information, geographical area model information, and ledger model information; among them, the power grid model information includes the power grid area name, plant name, voltage level name, interval name, and equipment name; the geographical area model information includes the description information of the substation geographical area, the station geographical area, the smaller geographical area, and the equipment geographical area; the ledger model information includes the equipment name, Chinese description, equipment model, manufacturer, country of production, and production date.

[0044] Step 1 specifically includes:

[0045] 1) Perform vectorization association of the power grid model: extract the semantics of the equipment description information in the power grid model, such as "power grid area name, plant station name, voltage level name, equipment name", and convert it into a word vector, and associate the vector with the power grid model;

[0046] 2) Perform vectorization association of the geographic area model: extract the equipment geographic area description information "substation geographic area, station geographic area, small-level geographic area, equipment geographic area", perform word vector conversion, and associate the vector with the geographic area model;

[0047] 3) Perform vectorized association of the ledger model: Extract the semantic information of each attribute description of the general ledger information, such as: device name, Chinese description, device model, manufacturer, country of production, production date, etc., convert it into a word vector, and associate the vector with the ledger model;

[0048] 4) For the power grid model, geographic area model and ledger model, semantic vectors are extracted from different dimensions of the same device, and mapping is established based on the same device.

[0049] In the specific implementation, the semantic information of the device is extracted according to the relevant specifications of the power grid model. The following example takes the different model information of a device as an example.

[0050] (1) Extraction of semantic information from power grid models

[0051] The required format of the device description information in the power grid model is as follows:

[0052] "Grid area name. Plant name / voltage level name. Equipment name";

[0053] For example: "East China. Shanghai Road / 220kV.15B circuit breaker", the middle connectors are "." and " / ".

[0054] When extracting semantic information, the semantic information is extracted by using connectors as segments: {220kV 15B circuit breaker on Shanghai Road, East China}.

[0055] (2) Extraction of semantic information from geographic region models

[0056] The typical area division of a substation is usually created based on the principle of relatively independent physical area distribution. The typical area division is as follows:

[0057] The surrounding walls of the substation are divided into five areas, among which the wall with the gate is divided into two areas according to the walls on the left and right sides of the gate.

[0058] The outdoor area of ​​primary equipment can be divided into transformer outdoor area, 500kV outdoor area, 220kV outdoor area or 110kV outdoor area according to voltage level and equipment type.

[0059] Indoor primary switchgear, secondary equipment, station equipment, main control equipment, communication equipment, batteries, etc. should be divided into 500kV protection room, 220kV protection room, 35kV high-voltage room, station power distribution room, production complex or main control room according to relatively independent physical areas indoors.

[0060] The specific description information of the 15B circuit breaker is: "Shanghai Road 220kV line 15B interval place 15B circuit breaker", and the semantic information is extracted: {Shanghai Road 220kV line 15B interval place 15B circuit breaker}.

[0061] (3) Extraction of semantic information from ledger model

[0062] The equipment general ledger model should include two parts: general ledger and general technical parameters. The general ledger defines basic equipment information such as equipment type, equipment name, manufacturer, and production date. Each item of general ledger information is extracted as semantic information. For example, the ledger information of 15B circuit breaker is shown in Table 1:

[0063] Table 1 Account information

[0064] property information Device Type breaker Device Name 15B Chinese description 15B circuit breaker Device Model CBFM264 Manufacturer XXXX Country of production China Production Date October 31, 2022

[0065] Extract semantic information: {Circuit Breaker 15B 15B Circuit Breaker CBFM264 XXXX China October 31, 2022};

[0066] After extracting the semantic information, the semantic information extracted by each model is vectorized and saved. In the above steps, the semantic information has been extracted and segmented. It is only necessary to use a tool such as Word2Vec to complete the word vector conversion and generate a word vector. After the word vector is generated, it is stored in a vector database, specifically in a model vector mapping table, and the vector field is indexed. The model vector mapping table includes: power grid equipment ID, power grid equipment vector, geographic area equipment ID, geographic area equipment vector, equipment ledger ID, equipment ledger vector, wherein the power grid equipment ID, geographic area equipment ID, and equipment ledger ID are the primary key IDs derived from the power grid model, geographic area model, and ledger model, respectively; the power grid equipment vector, geographic area equipment vector, and equipment ledger vector are semantic vectors associated with the power grid model, geographic area model, and ledger model, respectively. The table structure is shown in Table 2:

[0067] Table 2 Model vector mapping table

[0068]

[0069] like Figure 3 As shown, the model vector mapping table and the power grid equipment table, geographic area equipment table, and equipment ledger table are all in a one-to-one relationship, and the model vector mapping table accesses the data of these model tables through foreign keys.

[0070] Step 2: vectorize the semantic information of the text of the unstructured data of the power grid system, extract the feature vector of the picture or video of the unstructured data, and store it in the corresponding file vector information table; that is, it includes vectorizing the semantic information of the text and extracting the feature vector of the image, wherein the feature vector extraction of the image is used for subsequent image search by feature vector matching.

[0071] Further preferably, the unstructured data includes equipment maintenance records, equipment failure reports, and equipment pictures and videos.

[0072] This step extracts semantic information from the text information of unstructured data, converts it into word vectors, associates it with the corresponding files, and saves it in the mapping table. Extract feature vectors from files such as pictures or videos and save them in the mapping table, including:

[0073] Extracting semantic information from the equipment maintenance record, equipment fault report, equipment picture or video file description information text, converting the semantic information into word vectors, associating the vectors with the corresponding files and saving them in the equipment maintenance record, fault report, equipment video picture vector information table;

[0074] Extract feature vectors from the images or videos of the device image or video file, associate them with the corresponding files, and save them in a device video image vector information table.

[0075] Through the above steps, vectors are extracted from unstructured data, such as equipment maintenance records, equipment fault report text records, equipment pictures and videos, etc., and then a mapping relationship table between files and vectors is established, including: fault vector information table, equipment maintenance record vector information table, equipment picture video vector information table, etc. In specific implementation, it is not limited to the above three types of unstructured data, and corresponding vector information tables are also established for other file types. The table structure is shown in Table 3-Table 5:

[0076] Table 3 Fault report vector information table

[0077] Field Name type illustrate Object ID Int64 Fault report file name String Fault Report Description String Fault Report Vector Float vector Vectors extracted from report content text

[0078] Table 4 Equipment maintenance record vector information table

[0079] Field Name type illustrate Object ID Int64 Maintenance record file name String Maintenance record description String Maintenance record vector Float vector Vector extracted from the text of the maintenance record

[0080] Table 5 Equipment picture video vector information table

[0081]

[0082]

[0083] The extraction of text information vector is the same as in step 1, using the same model and the same tools to maintain consistency.

[0084] The feature vectors extracted from images or videos are extracted using two commonly used algorithms, SIFT or SURF, and only one algorithm is used in a system.

[0085] like Figure 3 As shown, the fault report vector information table and the equipment maintenance record vector information table have a one-to-many relationship with the equipment video image vector information table, which means that the fault report and maintenance record will contain multiple images or videos.

[0086] Based on the above steps 1-2, a model vector mapping table, a maintenance record vector information table, a fault report vector information table, an equipment video image vector information table, and the association relationship between multiple tables are established. The system already includes structured data and unstructured data, and has vectorized the data, and has the conditions for semantic query.

[0087] Step 3: When searching for data, vectorize the search semantic information, use vector matching to search the model vector mapping table and vector information table, and obtain the corresponding information.

[0088] This step converts the semantic information to be retrieved into vectors. Based on the device-centric association, the model data (structured information) and unstructured data such as fault reports, maintenance records, pictures and videos are retrieved in a vector-based maximum matching manner. The entire retrieval process is divided into two steps. The first step is a semantic search. The search results include some pictures. Then, the feature vectors of these pictures are used to search for more unstructured data, including:

[0089] When searching for data in the system, input semantic information, for example: "Shanghai Road 15B Circuit Breaker".

[0090] First, the semantic information is converted into word vectors through the vector conversion tool.

[0091] Secondly, in the model vector mapping table, the power grid equipment vector, geographic area equipment vector and equipment inventory vector are retrieved using the search information word vector to obtain the record of the most matching vector.

[0092] Again, the corresponding records are retrieved from the power grid equipment model, geographic area equipment model and equipment inventory model using the power grid equipment ID, geographic area equipment ID and equipment inventory ID in the record to obtain specific data of various models.

[0093] Again, using the vector field content in the record, the corresponding record with the highest vector matching degree is retrieved from the fault report vector information table, the equipment maintenance record vector information table, and the equipment video image vector information table to obtain various file contents.

[0094] Again, a secondary search is performed through the feature vectors of the pictures or videos in the file content, and similar pictures or videos are retrieved in the device video picture vector information table to achieve image search, and other related reports, maintenance records and other information can be obtained based on the correspondence between fault reports, maintenance records and pictures.

[0095] Finally, the various information obtained through the query is integrated and displayed in a unified manner.

[0096] In summary, structured data and unstructured data are retrieved through semantic information.

[0097] Embodiment 2 of the present invention provides a system for rapid retrieval of a general power grid model based on a vector, comprising:

[0098] The structured data processing module is used to extract semantic information from the structured data in the general power grid model, vectorize the extracted semantic information and save it in the model vector mapping table;

[0099] The unstructured data processing module is used to vectorize the semantic information of the text of the unstructured data of the power grid system, extract the feature vector of the picture or video of the unstructured data, and store it in the corresponding file vector information table;

[0100] The retrieval module is used to vectorize the retrieval semantic information when searching for data, and use vector matching to retrieve the model vector mapping table and vector information table to obtain the corresponding information.

[0101] A terminal comprises a processor and a storage medium; the storage medium is used to store instructions; the processor is used to operate according to the instructions to execute the steps of the method.

[0102] A computer-readable storage medium stores a computer program, which implements the steps of the method when executed by a processor.

[0103] The beneficial effects of the present invention are as follows:

[0104] In view of the insufficient association, storage and retrieval capabilities of various models in the existing system after the fusion of the general power grid model, the present invention proposes a device-centric approach to establish the association and mapping of various models based on semantic vectors, and to store them based on a vector database to achieve fast retrieval.

[0105] The present invention is based on models of multiple standardized structures in the power grid, extracts semantic information in the specifications, obtains semantic vectors generated by the respective descriptive information of the power grid model, geographic area model, and ledger model to which the equipment belongs, and transforms the original mapping method for establishing a strong consistent relationship between multiple models, that is, establishing a multi-model ID mapping method, into a method for establishing mapping with semantic vectors, abstracting the descriptive information in the model into a vector of mathematical expression, which is more concise and effective, can better express and understand the semantics of the data, makes the data easier to understand and use in application development and data analysis, and enables advanced applications based on multiple models to complete retrieval according to semantics.

[0106] The present invention vectorizes semantic information of text of unstructured data associated with the device, extracts feature vectors of pictures or videos, and establishes a mapping relationship between these vectors and unstructured data, so that unstructured data and structured data, unstructured data and unstructured data are semantically associated; the present invention stores the established vectors and model data in a vector database, and uses the retrieval function of the vector database to perform rapid model retrieval and semantic matching, which can provide support for advanced semantic-based applications across power grid systems.

[0107] The present invention is based on vectors and mappings, and performs semantic related queries in the system. For example, the power grid model information is input, and the information is converted into vectors, and then queried in the mapping table according to the vector matching degree. For the query results, the semantic vectors of other structured data are found through the mapping relationship, and then the vector matching retrieval of unstructured data is performed through the semantic vector. Finally, the associated pictures and video content can be retrieved through their feature vectors in the form of image search and feature vector similarity retrieval to retrieve more related reports, records and other unstructured data. It retrieves various related data in the power data through the matching retrieval of semantic vectors and feature vectors and the image search method, thereby realizing the rapid retrieval of the general power grid model.

[0108] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the above embodiments, ordinary technicians in the relevant field should understand that the specific implementation methods of the present invention can still be modified or replaced by equivalents, and any modifications or equivalent replacements that do not depart from the spirit and scope of the present invention should be covered within the scope of protection of the claims of the present invention.

Claims

1. A method for rapid retrieval of a general power grid model based on vectors, characterized in that: The method comprises: Step 1: extract semantic information from structured data in the general power grid model in different dimensions of the same device, convert the semantic information into word vectors, associate the vectors with the corresponding models and save them in the model vector mapping table to form a mapping based on the same device; The model vector mapping table includes: power grid equipment ID, power grid equipment vector, geographic area equipment ID, geographic area equipment vector, equipment ledger ID, equipment ledger vector, wherein the power grid equipment ID, geographic area equipment ID, and equipment ledger ID are primary key IDs derived from the power grid model, geographic area model, and ledger model, respectively; the power grid equipment vector, geographic area equipment vector, and equipment ledger vector are semantic vectors associated with the power grid model, geographic area model, and ledger model, respectively; Step 2: vectorize the semantic information of the text of the unstructured data of the power grid system, extract feature vectors from the pictures or videos of the unstructured data, and store them in the corresponding file vector information table; Step 3: When searching for data, vectorize the search semantic information, use vector matching to search the model vector mapping table and vector information table, and obtain the corresponding information, including: In the model vector mapping table, the power grid equipment vector, the geographic area equipment vector and the equipment ledger vector are retrieved by retrieving the semantic information word vector to obtain the record with the best matching vector. The power grid equipment ID, the geographic area equipment ID and the equipment ledger ID in the record are used to retrieve the corresponding records from the power grid equipment model, the geographic area equipment model and the equipment ledger model to obtain the specific data of various models. The vector field content in the record is used to retrieve the corresponding record with the highest vector matching degree from the fault report vector information table, the equipment maintenance record vector information table and the equipment video picture vector information table to obtain various file contents. A secondary search is performed through the feature vectors of the pictures or videos in the obtained file content, and pictures or videos with a matching degree exceeding a set value are retrieved in the equipment video picture vector information table to realize image search, and based on the association relationship between the fault report vector information table, the equipment maintenance record vector information table and the equipment video picture vector information table, the associated fault reports and maintenance records are obtained, and the various information obtained are integrated and displayed in a unified manner.

2. The method for rapid retrieval of a general power grid model based on vector according to claim 1, characterized in that: The general power grid model includes a power grid model, a geographical area model and a ledger model.

3. The method for rapid retrieval of a general power grid model based on vector according to claim 2, characterized in that: In step 1, the semantic information extracted from the power grid model includes the power grid area name, plant station name, voltage level name and equipment name; The semantic information extracted from the geographic area model includes a substation geographic area, an in-station geographic area, a smaller level geographic area, and an equipment geographic area; The semantic information extracted from the ledger model includes device name, Chinese description, device model, manufacturer, country of production, and production date.

4. The method for rapid retrieval of a general power grid model based on vector according to claim 1, characterized in that: The unstructured data includes equipment maintenance records, equipment failure reports, and equipment pictures and videos.

5. The method for rapid retrieval of a general power grid model based on vector according to claim 4, characterized in that: Step 2 includes: extracting semantic information from the equipment maintenance record, equipment fault report, equipment picture or video file description information text, converting the semantic information into word vectors, associating the vectors with the corresponding files and saving them in the equipment maintenance record, fault report, equipment video picture vector information table; Extract feature vectors from the images or videos of the device image or video file, associate them with the corresponding files, and save them in a device video image vector information table.

6. The method for rapid retrieval of a general power grid model based on vectors according to claim 5, characterized in that: The fault report vector information table, the equipment maintenance record vector information table and the equipment video picture vector information table are in a one-to-many association relationship.

7. A system for rapid retrieval of a general power grid model based on vectors, using the method according to any one of claims 1 to 6, characterized in that: The system comprises: The structured data processing module is used to extract semantic information from the structured data in the general power grid model, vectorize the extracted semantic information and save it in the model vector mapping table; The unstructured data processing module is used to vectorize the semantic information of the text of the unstructured data of the power grid system, extract the feature vector of the picture or video of the unstructured data, and store it in the corresponding file vector information table; The retrieval module is used to vectorize the retrieval semantic information when searching for data, and use vector matching to retrieve the model vector mapping table and vector information table to obtain the corresponding information.

Images