Artificial neural networks (ANN) constitute a powerful class of nonlinear function approximate. ANN has been widely used in pattern recognition, prediction and classification. The application of wavelets in the fields of engineering has grown rapidly in the past few years. To improve the limitation of conventional applying traditional fault diagnosis method on gas turbine, a novel diagnosis approach integrating the wavelet transform with neural network is proposed. It can overcome the problems caused by local mimima of optimization. The model uses wavelets as the activation functions in neural networks. The wavelet basis function is assigned for each neuron of hidden layer and each weight is determined by learning. With respect to six kinds of typical and common fault based on thermo dynamic parameter of gas turbine, use these data as sample to train wavelet neural network, and then according the output of network to determine the type of fault. The experimental result show that the proposed gas turbine fault diagnostic model based on wavelet neural networks can diagnose the fault of gas turbine effectively. The method can be generalized to other fault diagnosis.