Neural Network-based Aircraft Fault Diagnosis Expert System

Based on the study of traditional expert system and the neural network in fault diagnosis of airplane, this paper formulated the total truss of airpane fault diagnosis expert system, inference method and realization technique.It mainly analyzed the function requirement of the ES, the realization technique of the primary function module, the design principle of the knowledge base and the inference method.The main goal of the research is to assist airplane crew, especially the maintenance man, to eliminate failures rapidly and accurately when they maintain the airplane, to enhance the work efficiency and gain more economic benefits and higher safety quotiety.According the deeply study of ANN, the paper gathered the PW4000 engine data ,combined with the typical fault data, then formulated the fault diagnosis and contrasted with the actual fault data. At last the paper showed the practicability of the system.The ANN module of this system in simulated example adopts a three-layer BP net structure, which has three inputs, twelve nodes in central layer and fourteen outputs confronted with the fault model. On algorithm,this paper used the gradient iteration Method, constantly correct the authority. After iteration 10 thousand times, the precision of study (the total error of this system) is E=0.001. According the study of the typical fault date of from engines, the formulated fault diagnosis can met the actual requirement.