Research On The Correlation Analysis Of Aviation Safety Incidents Based On Knowledge Graph

The civil aviation industry is continuous development,so the number of aircraft and the throughput of airports grow rapidly,the aviation safety incidents are increasing and diversifying.Continuously improving the safety level of civil aviation has become the primary task for the state and related enterprises to develop the civil aviation industry.It is an important way to make full use of the information of aviation safety incidents and to analyze the correlation of aviation safety incidents.In previous studies,it was mainly used to analyze a single event and extract the entity information and causal relationship in the event,while the correlation analysis and research on multiple events were relatively insufficient.On the basis of deeply analyzing the characteristics of the aviation safety incidents,this paper puts forward an improved algorithm of FP-growth by introducing hierarchical frequent items and repair items.The association rules we have mined include not only various types of events such as engine failure,heavy landing and flight safety but also attributes at different conceptual levels such as event cause,event result,event type and operation stage.The results of our research indicate that the occurrence of events is closely related to human factors,environmental factors,mechanical factors and other factors.Method based on data mining association rules mined not enough comprehensive and has no interpretability rules,therefore the knowledge graph is introduced on this basis.The knowledge graph of aviation safety incidents involves a wealth of event information,including concepts of aviation security domain and multiple attribute instances such as personnel attribute,aircraft attribute,engine attribute,environment attribute and other attribute.Combining with the characteristics of aviation safety incidents knowledge graph,this paper proposes a NRLvLR model for knowledge graph correlation analysis.The knowledge graph is used to represent learning and define scoring function to realize the mining of interpretable rules of the knowledge graph,and the rules including predicates and entities.The experimental results show that the model is of higher quality and faster speed than other models,and it can dig out valuable information of correlation hidden in the knowledge map,generate targeted safety suggestions and provide semantic decision service for civil aviation safety.