Research On The Construction And Application Of Knowledge Graph For Distribution Network Dispatching Fault Handling

With the development of China’s economy and society,the people and all walks of life have higher and higher requirements for the reliability of power supply in the distribution network.At the same time,the rapid development of refined digital distribution network has led to the continuous growth of knowledge in the field,and the difficulty of distribution network fault handling has been increasing,but at present,the knowledge of distribution network fault plan is still stored in unstructured text form,and the ability to check the overall analysis and fault handling decision of distribution network fault is low.In order to make the fault handling plan better assist dispatchers in fault handling,improve the efficiency and intelligence of distribution network fault handling,this paper studies the transformation of knowledge related to fault handling in distribution network into a structured form,constructs a knowledge graph for fault handling in distribution network dispatching,and assists the fault handling decision-making of distribution network based on the knowledge graph.The main work of this article is as follows:(1)This paper first sorts out the topology data of the distribution network and the text of the distribution network fault disposal plan,proposes a method of integrating the knowledge graph of the topology entity of distribution network equipment and the knowledge graph of the event of the fault disposal plan,and constructs a knowledge graph model for the fault handling of the distribution network.(2)Aiming at the problem of strong professionalism of knowledge in the electric power field and less labeled training data,this paper adopts the method of transfer learning to propose an EBERT-BiLSTM-CRF named entity recognition model based on the pre-trained BERT language model in the electric power field,which can reach 89.16%,92.24%and F1 score in the small-sample named entity recognition task,respectively.Moreover,the joint extraction pipeline model of entities and relationships designed based on the named entity recognition model has a joint extraction accuracy rate of 90.15%;(3)Aiming at the business needs of fault disposal in distribution network,this paper proposes a fault handling auxiliary decision-making strategy based on knowledge graph,which includes a sub-strategy of knowledge retrieval of the plan based on logical rules and an auxiliary decisionmaking sub-strategy based on topology analysis,which can realize the query of the existing disposal plan,and for the fault situation that cannot be matched by the historical plan,the power outage range analysis algorithm and the power supply path analysis algorithm are developed based on the breadth priority algorithm considering the equipment status.Finally,based on specific examples,the results of topology analysis and pre-plan knowledge graph query are used to generate a fault disposal and power supply guarantee scheme for distribution network,which proves the feasibility of this strategy.