Developmentof automobile fault diagnosis expert system based on fault tree — Neural network ensamble

摘要

The structure of cars is increasingly complex, the fault type, fault phenomena and fault causes of cars is more complicated, therefore ordinary users are stranded when fault happens. To solve the above questions, this paper outlines a kind of on-vehicle fault diagnosis expert system based on fault tree — neural network ensamble. The knowledge base of the expert system is divided into two parts, the fault tree analysis based knowledge base and neural network ensemble based knowledge base. For the faults easy to form production rules, the fault tree analysis is used to form expert rules. For those difficult to find specific expression between failure mode and fault reason, the neural network ensemble based method is adopt to form a diagnosis model, and it is tested by the simulation examples. Finally, the program language EVC++ under Windows CE is used to develop the fault diagnosis expert system for cars, which had better man-machine interacted interface.