Based On The Conjugate Gradient Method Improved Fuzzy Neural Network Fault Diagnosis Method

With the rapid development of modern city, civil aviation aircraft flight demand is big ger.Civil aviation aircraft maintenance is based on. To improve the efficiency of maintenance, the artificial intelligence method applied to civil aviation aircraft fault diagnosis, at present the usual part are the neural network, expert system and fault tree applications. the B737-800 for model of aircraft flight control system between each subsystem and the internal components of the subs ystem between into chain relations were analyzed according to the flight control system fault types of complicated characteristics, a total of a gradient method based on improved fuzzy neural network method, The research and development of the flight control system fault diagnosis expert system, Through the example simulation results prove the method can improve the overall level of fault diagnosis, it has the very high practical value. Through comparing different methods, it draw the fuzzy mathematics and neural network combining the method of fault diagnosis for aircraft of the advantages and disadvantages, and analysis B737-800 type of flight control system function and structure, The fault characteristics are studied and analyzed. Through the fault phenomena and failure cause summarized to find the inner link, so it can generate fuzzy neural network model. With autopilot subsystem failure as an example, it verify the improved fuzzy neural network fault diagnos is method of feasibility, At last, the Visual C++ combined with SQL Server 2005 database software development flight control system fault diagnos is expert system, With autopilot fault as an expert system it run interface simulation, the system for the user to fault diagnosis.