Gear fault diagnosis based on the improved wavelet neural network and simulation

In order to eliminate the noise in original signals about gear fault, wave filtering is put into use in this paper, and Wave Neural Network which is based on that is built. In the network training, it mainly applies Gradient Descent Method and Adaptive Learning Rate Adjustment Method to optimize every parameter. Furthermore the arithmetic of the gradient and learning rate is improved. Finally, the trained Wave Neural Network is used to diagnose gear fault. The simulation results show that the use of filtered information and Wavelet Neural Network can accurately identify the gear fault.