Research On Fault Diagnosis Of Interlocking System Based On Knowledge Graph

Railway interlocking system is one of the key systems to ensure the safe operation of trains.The system structure is complex and the working environment is complex.Because of the complex fault causes,when the interlocking system fails,if it cannot be dealt with in time,it will endanger driving safety and cause serious consequences.Knowledge graph emerges from the research upsurge of semantic network and is a hot developing trend in the field of artificial intelligence in recent years.The combination of knowledge graph and fault diagnosis of interlocking system is beneficial to improve the efficiency of fault processing and shorten the period of fault diagnosis.At present,most fault knowledge data are recorded in the form of unstructured natural language,which contains abundant information.However,due to the limitation of these data storage formats,much effective information cannot be well applied.To solve the above problems,this thesis uses knowledge extraction and other technical means to extract fault knowledge from unstructured text data,so as to construct fault knowledge map of interlocking system,dig out the internal relationship between fault knowledge,and realize fault diagnosis of interlocking system based on knowledge map.And use the Django framework to design and implement a knowledge query application system based on knowledge graph,and explore the application of knowledge graph in interlocking fault diagnosis.The main research contents of this thesis include:(1)Collect unstructured text knowledge data related to interlocking system fault processing,and make a data set of interlocking system fault knowledge for entity relationship extraction by using the collected data.(2)Construction of fault knowledge graph of interlocking system,including construction of ontology framework of knowledge graph based on seven-step method and design of pattern layer of knowledge graph.Entity extraction of fault knowledge of interlocking system based on BILSTM-CRF.On the basis of entity extraction results,relational extraction and model of interlocking system fault knowledge based on CNN.Fault knowledge fusion of interlocking system based on Birch clustering algorithm.(3)Based on the constructed knowledge graph,the graph and its nodes are analyzed and evaluated.A statistical calculation method of fault cause of interlocking system based on knowledge graph data is designed and implemented to realize fault diagnosis,and the results of fault diagnosis are visualized by graph database.(4)Based on the constructed interlocking fault knowledge graph,the fault knowledge query system based on the knowledge graph is designed and implemented.The whole process includes system demand analysis,scheme design,implementation method and testing.The functions of fault diagnosis,knowledge question and answer,knowledge graph visualization,and knowledge graph update are realized.