Abstract:Analysis of the rolling bearing fault vibration signal of the nonlinear and non stationary characteristics, using the method of empirical mode decomposition (EMD) in the treatment of the advantages of this kind of signal, the time-frequency analysis method is studied for the fault signal of the rolling bearing. The rolling bearing vibration signal through the EMD method was decomposed into a number of stable sum of intrinsic mode functions (IMF) component. Selected the top eight IMF energy value as frequency domain features and combined with time domain features, sum of features of fault vibration signals were constructed as the BP neural network input. The BP neural network model of rolling bearing fault diagnosis was set up, by using the BP network self-learning mechanism for network training, the mapping relationship between input features and fault mode was gotten. Based on different categories of rolling bearing fault diagnosis experiment, proves the feasibility of this method is proved.