Abstract: A pattern recognition method for composite fault diagnosis based on wavelet neural networks is presented. Considering the multiple faults in complex nonlinear system such as a fighter, a multiple-layer wavelet neural network is constructed. Residual signal is processed by discrete binary wavelet transformation at the input layer, and the detail coefficients are obtained under multi-resolution as fault character vectors, finally these character vectors are sent to the neural network classifier to complete fault pattern recognition. Simulation results reveal that the presented method is effective for composite fault diagnosis.