Abstract:Aiming at the nonlinear and nonstable characteristics of the fault signal of rotating machinery, a fault identification method based on wavelet packet sample entropy and GA-BP network is presented. Firstly, using the wavelet packet to decompose the fault signal, then the reconstructed signals in the first four sub larger sample entropy energy generation were calculated as the eigenvector. Then the feature vector was input into the generic algorithmback propagation (GA-BP) network model to identify the type of the fault, and at the same time comparing with the traditional BP network. The experimental results show that: The status of rotor bench is different, the wavelet packet sample entropy is also different, and this method is more effective for rotor fault discrimination, and the fault recognition rate is obviously improved.