Design Of Substation Equipment Fault Question Answering System Based On Knowledge Graph

With the intelligent and informatization of the power system,State Grid Corporation of China have accumulated a large number of power data,among which substation equipment faults and fault maintenance records have important reference significance for substation operation and maintenance and technical management personnel to cope with emergencies,and the reasonable use of these fault data is of great significance to ensure the safety and stability of the power grid.In order to facilitate the unified management and accurate query of the historical data of substation equipment faults,this paper and Jiangxi Electric Power Research Institute jointly construct a knowledge question and answer system for substation equipment faults based on knowledge graph.At present,the research of knowledge graph in the field of electric power mainly focuses on construction and visualization of knowledge graphs,there are relatively few the Knowledge Question Answering Technology field of knowledge graphs,especially the research on knowledge questions and answers in the field of substation equipment faults.In this paper,the technology of knowledge graph,graph database,natural language processing,knowledge question answering and other knowledge is combined with the Electric power field by deep learning method,and the knowledge graph of substation equipment fault vertical domain based on Neo4 j is constructed by top-down method using the substation equipment fault troubleshooting report of Jiangxi Electric Power Research Institute in recent years as the data source.On this basis,a substation equipment fault question answering system based on knowledge graph is designed and implemented.The main contributions of this paper are as follows:(1)Build a knowledge graph in the field of substation equipment faults.Through the substation equipment fault maintenance report of Jiangxi Province in the past four years,the substation equipment fault dataset with substation equipment as the core was collected and sorted,and the substation equipment fault dataset was imported into the Neo4 j graph database by data modeling,knowledge extraction,knowledge storage and other methods,and the knowledge graph in the field of substation equipment fault was constructed,which realized the unified management and maintenance of data information in the field of substation equipment fault.(2)Build a substation equipment fault naming entity recognition model based on Bi-LSTM-CRF.The substation fault brief introduction collected and sorted out was selected as the substation equipment fault naming entity identification corpus,and the self-built substation equipment fault data is serially labeled according to each substation equipment fault entity category by BIO method,and the substation equipment fault naming entity identification dataset is established.The superiority of the Bi-LSTM-CRF model is verified by several comparative experiments,and the experimental results show that the F1 of the model is 0.97 on the self-built dataset,which can well complete the task of naming entity recognition of the substation equipment fault question answering system.(3)Build a substation equipment fault question intent recognition model based on Bert-Text CNN.Firstly,according to the characteristics of substation equipment fault data and knowledge graph in the field of substation equipment faults,13 kinds of substation equipment fault question intentions are designed,and then the substation equipment fault question Corpus set is created by adapting the public medical question and answer dataset and Self designed,and the substation equipment fault intention identification dataset is constructed according to the corresponding substation equipment fault intention annotated one by one.The experimental results show that the F1 of the model is 0.96 on the self-built dataset,which can meet the intention recognition task of the substation equipment fault question answering system.(4)On the basis of the above content,a knowledge question answering system in the field of substation equipment faults based on knowledge graph is designed and implemented.The system is composed of four modules: knowledge graph construction,semantic analysis,knowledge query and interactive interface.Taking the substation equipment fault knowledge graph as the knowledge base,the substation equipment fault question named entity recognition model and the intent recognition model constitute a semantic analysis module to parse the user’s natural language input,and establish a knowledge query module by designing semantic slots and Cypher query templates to retrieve substation equipment fault knowledge and return results,through The personal We Chat interface itchat module uses Python to call We Chat and builds a We Chat chat window as an interactive interface.It provides users with intelligent substation equipment fault knowledge Q&A services,and realizes the management,maintenance and query functions of substation equipment fault knowledge.