Abstract: A method is presented for fault diagnosis of an airplane steering surface system.The nonlinear fitting ability of self-organizing neural network is used to expand the information measured by related sensors,and output information of the related sensors is synthesized with D-S(Dempster-Shafer) proof algorithm.Based on the synthesized information,the information fusion diagnosis strategy can define the fault and identify the fault signals.The mathematical model of the airplane steering surface system fault diagnosis is constructed,and computer simulations are made.Simulation results show that this diagnosis structure can identify and warn the normal failures of the steering surface.It can significantly reduce the uncertainty of fault diagnosis and improve the recognition rate on fault modes.