Method of bearing fault diagnosis based on time-frequency manifold analysis

Bearing damage is one of the main causes of mechanical malfunction , and its vibration signal has the characteristics of nonlinear, non-stationary and difficult to be extracted. In order to solve the problem of eigenvalues and eigenvectors, the concept of multi-scale sub-band sample entropy is proposed, which cannot accurately extract weak signals from complexity. First, the wavelet packet decomposition of multi-scale signals is obtained, and then, the scale of each signal is sub-band-decomposing. Finally, the sample entropy of each subband can be solved. This method can deeply mine the basic characteristic signals. In this paper, a set of normal fault, inner ring fault, spherical fault and outer ring fault signal are used as the original data to verify the effectiveness of the method. The experimental results show that the method can effectively extract bearing fault features.