Applied Research On Ontology Modeling And Reasoning Of Intelligent Mechanical Fault Diagnosis System

In recent years, the study of intelligent mechanical fault diagnosis has become a research focus in the field of mechanical fault diagnosis. The intelligent level and diagnostic accuracy of system depend on the quantity and quality of knowledge in knowledge base, knowledge organization and classification, and further knowledge sharing and reasoning as well. Ontology can describe knowledge model on semantic and knowledge level, and it also can provide conceptualization description on conceptions and relationship in a certain field, thereby making a foundation for knowledge sharing. Additionally, ontology supports distinction of knowledge information achieving level representation of domain knowledge. Class axioms and constraint axioms in ontology can be used to knowledge reasoning, and the existing ontology reasoner can provide a direct reasoning service for ontology.This thesis mainly study on how to build a knowledge base of intelligent mechanical fault diagnosis system by using ontology, including ontology reasoning to ensure the correctness of knowledge of knowledge base and achieve fault diagnosis. Overall, this thesis makes the following contributions:(1) Aiming at the demand of mechanical fault diagnosis, based on the existing ontology modeling method, this thesis proposes an ontology modeling method to apply to the field of mechanical fault diagnosis. Furthermore, a longitudinal modeling is proposed by analyzing mechanical fault diagnosis knowledge. According to the fault type, fault symptom, fault reasons and fault disposals, fault diagnosis knowledge is classified. By defining property to express the complex relationship among fault type, fault symptom, fault reasons and disposals, mechanical fault diagnosis knowledge becomes knowledge base with semantic relation.(2) According to the ontology modeling method and longitudinal modeling proposed in our thesis, an ontology of AC motor fault diagnosis is constructed by using Protégé4.0, and the feasibility of our method is verified.(3) A method of implementing fault diagnosis is proposed. Specifically, using constructor of OWL to define complex classes which is further declared as equivalent classes, the implementation fault diagnosis is achieved by reasoning equivalent classes.(4) Aiming at the logic error detected by ontology reasoner, the corresponding solutions are proposed. By ontology reasoning, fault diagnosises including starting fault, overheating fault and brush fault are achieved and the accuracy and rationality of the result are comparatively high.The result of our experiment shows the feasibility of the ontology modeling and reasoning method proposed in our thesis, which offers a new way for ontology modeling and reasoning research in the field of intelligent mechanical fault diagnosis.