Research On Knowledge Extraction And Pushing Method Of Engine Fault Diagnosis Based On Demand Driven

With the development of technology,the complexity of engine structure is greatly increased,which leads to the greater difficulty of engine maintenance and repair process,and puts forward higher requirements for the knowledge level of maintenance engineers.The long-term accumulation of fault diagnosis knowledge data in enterprises has become an important reference for fault diagnosis and processing activities of maintenance engineers.However,with the continuous expansion of knowledge scale,it is very difficult for maintenance engineers to acquire the required knowledge in the face of fault diagnosis knowledge data which is multi-source heterogeneous,large number but lack of effective organization.Aiming at the above problems,this paper proposes a method of knowledge extraction and push of engine fault diagnosis based on demand driven,and studies the knowledge extraction algorithm,fault demand matching and push model,and key implementation technologies of engine fault diagnosis text data.The specific research contents are as follows:(1)This paper analyzes the common engine fault and diagnosis process and the text data characteristics of engine fault diagnosis knowledge.Taking the two types of engine failure to start and engine idling instability as examples,it studies the characteristics of common engine faults,further studies the engine fault diagnosis process,and analyzes the needs of fault diagnosis knowledge in different stages of the process.In addition,combined with the characteristics of common engine faults,the text data characteristics of engine fault diagnosis knowledge are studied.The results show that the engine fault types are complex,there is a strong coupling relationship between fault knowledge,and the knowledge demand of fault diagnosis changes with the process of fault diagnosis.(2)According to the characteristics of engine fault diagnosis knowledge text,a named entity model of fault diagnosis knowledge based on BERT-Bi LSTM-Att-CRF deep neural network and an entity relationship classification model based on SVM machine learning method are constructed.The engine plus fault diagnosis knowledge text is preprocessed,and the depth feature of the knowledge text is extracted,and the sequence annotation of the fault diagnosis knowledge entity is completed.On this basis,the knowledge entity relationship is classified,and then the knowledge extraction of the fault diagnosis knowledge text data is realized.(3)In view of the dynamic knowledge demand in the process of fault diagnosis,a knowledge push method based on demand driven is proposed.The model of knowledge push method driven by demand and the ontology of fault diagnosis knowledge are constructed,and the matching method of fault knowledge based on ontology description and the reasoning method of fault reason and fault processing method based on confidence degree connection are given.(4)The knowledge push service platform of automobile engine fault diagnosis is built,which realizes the text knowledge extraction of engine fault diagnosis,fault demand knowledge matching,fault cause and fault handler reasoning knowledge push service.The research results of this paper provide new ideas and methods for the realization of engine fault diagnosis knowledge push on demand,which has higher theoretical guidance and practical significance for improving the quality of engine fault maintenance,improving maintenance efficiency and reducing maintenance costs.