Research Of Automotive Fault Diagnosis System Based On CBR And Extenics

With the development of the automotive industry, automobile fault diagnosis system has become a popular research domain. Traditional automotive fault diagnosis systems are lack of systematic, comprehensive, and relying on dynamic signal processing. With the development of artificial intelligence technology, Automobile Fault Diagnosis Expert System is introduced in the field of vehicle maintenance based on neural networks, case-based reasoning and rule-based reasoning. As automotive systems are large and complex, and effective logs of vehicle maintenance are few, it is difficult to rely solely on mathematical models to solve automotive fault diagnosis problems. Further more, it is not easy to obtain rules of the comprehensive vehicle maintenance. Based on Case-based reasoning (CBR), the vehicle fault diagnosis system can use iterative methods to complete the comprehensive knowledge base. However, there are some problems in this system based on CBR. Firstly, the system is particularly sensitive for noise data, error data and redundant data. The efficiency and solving results of the system are affected in the retrieval system easily. Secondly, with the capability of the case base increasing, the retrieval efficiency maybe lower and the retrieval results are probably redundant. Thirdly, bottleneck problems exist in acquiring knowledge of case base.To solve above problems, automotive fault diagnosis system is researched based on CBR and Extenics in this paper. Matter-element theory has been applied to the process of setting up extension case model. First of all, extension models and tree data structure are used to build the knowledge expression, formalizing the description of vehicle fault cases. And then creating the index of the car fault diagnosis case is introduced. Besides, via the machine learning methods and users'feedback, the case base is maintained. Taking one piece of car fault diagnosis record as an example, the instance of case storage methods is shown. Secondly, Extension reasoning and retrieval strategies for case-based reasoning process are explained. Hence, the case base is extended in the logical layer. Finally, the objects, principles, design concepts, process and major functional modules of the system are described.Based on extension model, car fault cases are created, stored, maintained and searched readily, improving the efficiency and flexibility of this system. The main properties of cases are stored in the case base with the formalization. Based on the primary attribute analysis, the capacity of case base is compressed. Automotive Fault Diagnosis Case information is described by the natural language, and the information is not comprehensive. Based on CBR and the extension theory, extension retrieval strategy is designed to improve the efficiency of retrieving case and reducing the expertise requirement in automobile maintenance industry for user and R&D.