Research On Application Of Knowledge Graph For Aircraft Maintenance

With the rapid development of digitalization in aviation maintenance industry,data management and decision-making in aircraft maintenance field have attracted much attention.Aircraft maintenance decision-making system shouldn’t only realize intelligent decisionmaking,but also put forward higher requirements for the collection,analysis and management of industry data.Previously,the knowledge graph research in the field of aircraft maintenance lacked data analysis,and ignored the few-shot problems of data missing,data sparse.Therefore,this paper focuses on the application of knowledge graph in the field of aircraft maintenance,proposes a knowledge completion model with few-shot based on fusion path,constructs knowledge graph of aircraft maintenance based on few-shot knowledge,effectively manages massive and complex domain data,and solves problems such as low effective information utilization rate and poor graph integrity caused by few-shot data.On the basis of ensuring full use of event information,knowledge reasoning algorithm is designed to realize maintenance decision-making of aircraft,improve the fault diagnosis coverage space,and improve the accuracy of maintenance decision.The main work and contributions of this paper are as follows:(1)In order to solve the problems of long-tail distribution and sparse data in domain knowledge,an adaptive attention network knowledge completion model with few-shot(PFAAN)based on fusion path was proposed to make full and efficient use of data information.The aircraft maintenance knowledge graph based on few-shot knowledge was constructed to improve the integrity of the graph.The experimental results show that the aircraft maintenance knowledge graph constructed after knowledge completion based on PFAAN can meet the requirements of the field and provide data support for aviation maintenance decisions.(2)Aiming at the problem that event information and correlation in domain knowledge graph are largely ignored,and it is difficult to effectively represent knowledge and answers,the aircraft maintenance event graph is further constructed to improve the utilization rate of domain knowledge and the availability of the graph;Aiming at the intelligent requirements of aircraft maintenance decision-making,a case-based reasoning algorithm(CBR-EEG)based on event enhanced graph is proposed,which can retrieve similar cases from the graph without training and expert experience and adaptively learn reasoning strategies.It solves the problems that machine learning consumes a lot of computing resources and reinforcement learning requires manual parameter adjustment.It can save computing resources and improve the accuracy of knowledge reasoning effectively.(3)The maintenance decision-making system of aircraft is developed to improve the efficiency and accuracy of maintenance decision.In this paper,based on the realization of aircraft maintenance knowledge graph construction and knowledge reasoning,based on the graph data,combined with the actual application scenarios,the system requirements were analyzed,and the aircraft maintenance decision-making system based on 737 NG aircraft was designed and implemented.The practice shows that the aircraft maintenance decisionmaking system can provide data support and decision aid in the daily maintenance process,which has high application value and practical significance.