| As one of the key equipment to ensure traffic safety,railway signal system plays a very important role in railway transportation system.The intelligent development of signal system has also become one of the important symbols of railway modernization.Under the background of the rapid development of modern railways,it is urgent to improve the digital and comprehensive level of operation and management of railway signal infrastructure equipment,and provide support and guarantee for the efficient operation and management of signal operation and maintenance personnel.Therefore,the intelligent development of signal technology needs to address the following urgent problems: to systematically integrate the existing equipment maintenance knowledge resources,to better provide support for equipment knowledge management and maintenance,and to improve the efficiency of operation and maintenance.In order to realize effective reuse of the knowledge of railway signal infrastructure equipment and improve the efficiency of equipment operation and maintenance,this thesis analyzes the semantic knowledge of relevant professional knowledge and fault maintenance case knowledge in the field of railway signal,and proposes the auxiliary decision method of railway signal infrastructure maintenance based on knowledge graph.The main work includes:(1)Aiming at the problems such as many professional terms,long semantic characters and complex structure existing in the corpus text data of knowledge in the maintenance domain of railway signal infrastructure equipment,a knowledge extraction model based on BERT-Bi LSTM-CRF in the maintenance domain of signal infrastructure equipment is proposed,and the corpus text is annotated with the annotation method of BIO corpus to construct the relevant data set,which realizes knowledge extraction task in this domain.(2)The knowledge characteristics and fault maintenance records in the field of railway signal infrastructure equipment maintenance are analyzed.Based on the relevant needs of equipment maintenance in this field,the conceptual model design of knowledge in this field is completed,including the knowledge of professional terms and the knowledge of fault maintenance cases.And the triad of equipment maintenance knowledge is formed by combining the results of knowledge extraction,and the knowledge triplet is divided according to the conceptual model design.(3)Based on the knowledge conceptual model of railway signal basic equipment and the related technology of knowledge map,a map construction method that conforms to the professional knowledge in the field is designed.Knowledge storage adopts the graph database storage method based on Neo4 j.The triples of knowledge of professional terms and knowledge of fault maintenance cases are stored in the graph database respectively to complete the construction of knowledge map of equipment maintenance in this field.(4)Aiming at the main business needs in the field of equipment maintenance,combined with the above knowledge graph construction technology,a knowledge graph based equipment maintenance decision-making method is proposed.Design related maintenance strategies and verify the feasibility of maintenance strategies with examples.A web-based knowledge graph visual management system for railway signal infrastructure equipment maintenance is designed and developed to verify the effectiveness of the proposed method and the scientific nature of the model. |
