The Study On Diagnosis Expert System Of Engine

The engine is the power source of the car, it have complex, high failure rate. The diagnosis of engine is difficult. So the diagnosis expert system of the engine is developed that is significant. The expert system is composed by the neural network, diagnosis tree and the traditional expert system. It combines their advantages. Neural network have learning function that can solve the problem that the traditional expert system get knowledge difficultly, and it also has the interpretation. The diagnosis tree has strong logic and reasoning abilities. The three together will be more accurate, more efficient in the fault diagnosis.The system uses Windows XP as the platform of development, MATLAB as the design software of neural network. It developed by the object-oriented programming language JAVA and database language MY SQL. System consists of 5 parts such as Man-machine interface, knowledge base, inference engine, the interpreter and the agency to get knowledge.System has achieved the fault diagnosis, knowledge base management, Auxiliary Maintenance and so on. Diagnosis is completed by the fault diagnosis model of neural network _ Diagnosis tree. The model includes inference engine 1 and inference 2. Inference Engine 1 is the neural network diagnostic model which is trained by the characteristic values of the Great Wall engine under different failure. Engine 2 is composed by the diagnosis tree. Diagnosis tree is the method of graphic interpretation, combining test set, Criteria set and the best search strategy. Its search direction is decided by the Symbolic instructions which are stored in the Y field and the N field, according to the instructions to determine the next search path. Knowledge Base has two parts such as including knowledge base of the neural network module and knowledge base of diagnostic tree module. Neural network module gets knowledge by self-learning, and the diagnosis tree module gets knowledge by hand. Interpreter play a secondary service functions in the diagnostic process, it can make an explanation to the reasoning process and make the suggestion to repair.When we is diagnosing fault in the system, first, we use the model of the neural network to diagnose failure of engine. We get the fault system that the failure belongs, and use the diagnosis tree to determine the cause of the failure. In this paper, we study the failure symptoms that ignition is not normal, using the diagnostic results to verify the reliability of the system.