CN-121984015-A - Scheduling decision determining method and device for power grid, storage medium and electronic equipment

CN121984015ACN 121984015 ACN121984015 ACN 121984015ACN-121984015-A

Abstract

The application discloses a scheduling decision determining method and device of a power grid, a storage medium and electronic equipment. The method comprises the steps of constructing a target knowledge graph based on text data of a target power grid, obtaining feature vectors corresponding to a plurality of entities respectively based on the target knowledge graph, integrating the feature vectors to obtain a plurality of optimized feature vectors, determining historical fault events of the target power grid, operation rules of fault equipment entities and a topological structure of a target area taking the fault equipment entities as centers, determining initial scheduling decisions of the target power grid based on the feature vectors, the optimized feature vectors, the historical fault events, the operation rules and the topological structure corresponding to the non-equipment entities respectively, verifying the initial scheduling decisions, and determining the initial scheduling decisions as target scheduling decisions of the target power grid under the condition that verification is passed. The application solves the technical problem of low accuracy of the scheduling decision determination result of the power grid in the related technology.

Inventors

WANG CHUNLING
ZHENG SHUWEI
WU YAN
CHENG HAIXING
SHA LICHENG
WANG HAIYUN
SUN HELIN
CHEN QIAN
FU LEI
YAO YIDI
LI YINGHAO
YANG LIPING

Assignees

国网北京市电力公司

Dates

Publication Date: 20260505
Application Date: 20260130

Claims (10)

1. A scheduling decision determination method for an electrical network, comprising: Constructing a target knowledge graph based on text data of a target power grid, wherein the target knowledge graph comprises an entity set and a relation set, the entity set comprises a plurality of entities in the target power grid, the plurality of entities comprise a plurality of equipment entities and a plurality of non-equipment entities, the plurality of equipment entities comprise fault equipment entities and non-fault equipment entities, and the relation set comprises relations among the plurality of entities; Based on the target knowledge graph, obtaining feature vectors respectively corresponding to the entities; Integrating a plurality of feature vectors to obtain a plurality of optimized feature vectors, wherein the plurality of feature vectors are in one-to-one correspondence with the plurality of equipment entities; Determining historical fault events of the target power grid, operation rules of the fault equipment entity and a topological structure of a target area taking the fault equipment entity as a center; determining an initial scheduling decision of the target power grid based on the feature vectors respectively corresponding to the plurality of non-equipment entities, the plurality of optimized feature vectors, the historical fault event, the operation rule and the topological structure; And verifying the initial scheduling decision, and determining the initial scheduling decision as a target scheduling decision of the target power grid under the condition that verification is passed.
2. The method according to claim 1, wherein the obtaining feature vectors respectively corresponding to the plurality of entities based on the target knowledge-graph includes: encoding the plurality of entities included in the target knowledge graph respectively to obtain semantic feature sub-vectors respectively corresponding to the plurality of equipment entities and semantic feature sub-vectors respectively corresponding to the plurality of non-equipment entities; Determining topological feature sub-vectors respectively corresponding to the plurality of equipment entities and measurement feature sub-vectors respectively corresponding to the plurality of equipment entities, wherein the topological feature sub-vectors are obtained by adopting a perception graph attention network model, and the measurement feature sub-vectors are used for representing the running state of the corresponding equipment entities in the target power grid; Determining fusion feature vectors corresponding to the plurality of equipment entities based on semantic feature sub-vectors corresponding to the plurality of equipment entities respectively, topology feature sub-vectors corresponding to the plurality of equipment entities respectively, and measurement feature sub-vectors corresponding to the plurality of equipment entities respectively; and determining the feature vectors respectively corresponding to the plurality of entities based on the semantic feature sub-vectors respectively corresponding to the plurality of non-equipment entities and the fusion feature vectors respectively corresponding to the plurality of equipment entities.
3. The method of claim 2, wherein the determining topological feature sub-vectors to which the plurality of device entities respectively correspond comprises: For any one of the plurality of equipment entities, determining attention weights between the any one equipment entity and a neighbor equipment entity of the any one equipment entity based on the electrical parameter data of the any one equipment entity, wherein the attention weights are used for quantifying the importance of electrical interaction between the any one equipment entity and the neighbor equipment entity; determining a topological feature sub-vector of the any equipment entity based on the electrical parameter data and the attention weight; And determining the topological characteristic sub-vector corresponding to each of the plurality of equipment entities by adopting a mode of determining the topological characteristic sub-vector of any equipment entity.
4. The method of claim 1, wherein the integrating the plurality of feature vectors to obtain a plurality of optimized feature vectors comprises: For any one of the plurality of equipment entities, performing similarity analysis based on the feature vector of the any one equipment entity and the feature vectors of the rest equipment entities to obtain a first similarity result of the any one equipment entity, wherein the rest equipment entities are equipment entities except the any one equipment entity in the plurality of equipment entities, and the first similarity result is used for indicating the similarity degree between the feature vector of the any one equipment entity and the feature vector of the rest equipment entity; Determining a second similarity result of the any equipment entity based on the neighbor equipment entity of the any equipment entity and the neighbor equipment entities of the rest equipment entities, wherein the second similarity result is used for indicating the similarity degree between the topological structure of the any equipment entity and the topological structure of the rest equipment entity; determining a target similarity result of the any equipment entity based on the first similarity result and the second similarity result, wherein the target similarity result is used for indicating whether the any equipment entity and the rest equipment entities are equivalent equipment entities or not; determining an equivalent equipment entity of any equipment entity based on a target similarity result of the any equipment entity; determining equivalent equipment entities corresponding to the plurality of equipment entities respectively by adopting a mode of determining the equivalent equipment entity of any equipment entity; dividing the plurality of equipment entities into a plurality of equipment entity sets based on equivalent equipment entities respectively corresponding to the plurality of equipment entities, wherein the equipment entities in the equipment entity sets are equivalent equipment entities; and integrating the feature vectors respectively corresponding to the equipment entities based on the equipment entity sets to obtain the optimized feature vectors, wherein the optimized feature vectors are in one-to-one correspondence with the equipment entity sets.
5. The method of claim 4, wherein the performing similarity analysis based on the feature vector of the any device entity and the feature vectors of the remaining device entities to obtain the first similarity result of the any device entity comprises: carrying out similarity analysis on semantic feature sub-vectors included in the feature vectors of any equipment entity and semantic feature sub-vectors included in the feature vectors of the rest equipment entities to obtain first similarity between any equipment entity and the rest equipment entities; carrying out similarity analysis on the topological feature sub-vector included in the feature vector of any equipment entity and the topological feature sub-vector included in the feature vector of the rest equipment entity to obtain second similarity between any equipment entity and the rest equipment entity; performing similarity analysis on the measurement feature sub-vector included in the feature vector of any equipment entity and the measurement feature sub-vector included in the feature vector of the rest equipment entity to obtain a third similarity between any equipment entity and the rest equipment entity; and obtaining a first similarity result of any equipment entity based on the first similarity, the second similarity and the third similarity.
6. The method of claim 1, wherein the determining historical fault events for the target grid, the operational rules for the faulty equipment entity, and the topology of the target area centered on the faulty equipment entity, comprises: Based on the optimized feature vector of the fault equipment entity, screening the historical fault event from a historical fault database of the target power grid in a similarity analysis mode; Screening out the operation rule from the operation rule base of the target power grid based on the fault condition description of the fault equipment entity; The topology is determined based on a target radius of the target region.
7. The method according to any one of claims 1 to 6, wherein said validating the initial scheduling decision, and in case of a validation passing, determining the initial scheduling decision as a target scheduling decision for the target grid, comprises: determining voltage data, current data and steady-state analysis results of the target power grid after the initial scheduling decision is executed; determining a first verification result based on the voltage data and a preset maximum voltage deviation threshold of the target power grid; Determining a second verification result based on the current data and a preset maximum current value of the target power grid; determining a third validation result based on the steady state analysis result; And under the condition that the first verification result, the second verification result and the third verification result all indicate verification pass, determining that the initial scheduling decision passes, and determining the initial scheduling decision as a target scheduling decision of the target power grid.
8. A scheduling decision determining apparatus for an electrical network, comprising: The system comprises a target knowledge graph construction module, a target knowledge graph construction module and a relation analysis module, wherein the target knowledge graph construction module is used for constructing a target knowledge graph based on text data of a target power grid, the target knowledge graph comprises an entity set and a relation set, the entity set comprises a plurality of entities in the target power grid, the plurality of entities comprise a plurality of equipment entities and a plurality of non-equipment entities, the plurality of equipment entities comprise fault equipment entities and non-fault equipment entities, and the relation set comprises relations among the plurality of entities; the feature vector determining module is used for obtaining feature vectors corresponding to the entities respectively based on the target knowledge graph; the integration module is used for integrating the plurality of feature vectors to obtain a plurality of optimized feature vectors, wherein the plurality of feature vectors are in one-to-one correspondence with the plurality of equipment entities; The first determining module is used for determining historical fault events of the target power grid, operation rules of the fault equipment entity and a topological structure of a target area taking the fault equipment entity as a center; the second determining module is used for determining an initial scheduling decision of the target power grid based on the feature vectors respectively corresponding to the non-equipment entities, the optimized feature vectors, the historical fault event, the operation rule and the topological structure; the verification module is used for verifying the initial scheduling decision, and determining the initial scheduling decision as the target scheduling decision of the target power grid under the condition that verification is passed.
9. A non-volatile storage medium, characterized in that it stores a plurality of instructions adapted to be loaded by a processor and to perform the scheduling decision determining method of the electrical network according to any one of claims 1 to 7.
10. An electronic device comprising one or more processors and a memory for storing one or more programs, wherein the one or more programs, when executed by the one or more processors, cause the one or more processors to implement the scheduling decision determination method of the power grid of any of claims 1-7.

Description

Scheduling decision determining method and device for power grid, storage medium and electronic equipment Technical Field The application relates to the field of power systems, in particular to a scheduling decision determining method and device of a power grid, a storage medium and electronic equipment. Background Along with the acceleration of the construction of a novel power system, the field of power grid dispatching faces the dual challenges of explosive growth of multi-source heterogeneous data and rapid increase of cross-system collaborative dispatching demands, and data such as equipment ledgers, SCADA (Supervisory Control and Data Acquisition, data acquisition and monitoring control systems) and the like form serious information islands through real-time measurement, weather early warning and historical dispatching instructions, so that the accuracy and timeliness of dispatching decisions are restricted. In the related art, a scheduling decision of the power grid is determined by adopting an artificial intelligence and deep learning based method, and the model may fail when new problems in the power grid are processed due to the quality and representativeness of training data affecting the performance of the model. Therefore, the technical problem of low accuracy of the scheduling decision determination result of the power grid exists in the related technology. In view of the above problems, no effective solution has been proposed at present. Disclosure of Invention The embodiment of the application provides a scheduling decision determining method, a scheduling decision determining device, a storage medium and electronic equipment for a power grid, which at least solve the technical problem that the accuracy of a scheduling decision determining result of the power grid is low in the related technology. According to one aspect of the embodiment of the application, a scheduling decision determining method of a power grid is provided, which comprises the steps of constructing a target knowledge graph based on text data of the target power grid, wherein the target knowledge graph comprises an entity set and a relation set, the entity set comprises a plurality of entities in the target power grid, the plurality of entities comprise a plurality of equipment entities and a plurality of non-equipment entities, the plurality of equipment entities comprise fault equipment entities and non-fault equipment entities, the relation set comprises relations among the plurality of entities, the characteristic vectors corresponding to the plurality of entities are obtained based on the target knowledge graph, integrating the plurality of characteristic vectors to obtain a plurality of optimized characteristic vectors, the plurality of characteristic vectors correspond to the plurality of equipment entities one by one, determining a historical fault event of the target power grid, an operation rule of the fault equipment entity and a topological structure of a target area taking the fault equipment entity as a center, determining an initial scheduling decision of the target power grid based on the characteristic vectors corresponding to the plurality of the non-equipment entities, the historical fault event, the operation rule and the topological structure, verifying the initial scheduling decision, and determining the initial scheduling decision as the target scheduling decision of the target power grid under the condition that verification passes. According to another aspect of the embodiment of the application, a scheduling decision determining device of a power grid is provided, which comprises a target knowledge graph construction module, an integration module and a first determining module, wherein the target knowledge graph construction module is used for constructing a target knowledge graph based on text data of the target power grid, the target knowledge graph comprises an entity set and a relation set, the entity set comprises a plurality of entities in the target power grid, the plurality of entities comprise a plurality of equipment entities and a plurality of non-equipment entities, the plurality of equipment entities comprise fault equipment entities and non-fault equipment entities, the relation set comprises a relation between the plurality of entities, the feature vector determination module is used for obtaining feature vectors corresponding to the plurality of entities respectively based on the target knowledge graph, the integration module is used for integrating the feature vectors to obtain a plurality of optimized feature vectors, the feature vectors are in one-to-one correspondence with the plurality of equipment entities, the first determining module is used for determining historical fault events of the target power grid, operation rules of the fault equipment entities and a topological structure of a target area centered on the basis of the feature vectors corresponding to the plurality o