Fault Diagnosis Method for Bearings Based on D-S Evidence Theory of Wavelet Packet Energy and Hierarchical Entropy

Aiming at the problems of low recognition rate and unstable fault signal of gearbox bearings, a bearing fault diagnosis method based on multisource data fusion of wavelet packet energy and hierarchical entropy was proposed. Firstly, the wavelet packet was used to decompose and reconstruct the vibration signals of bearing under normal condition, inner ring fault, outer ring fault and rolling element fault, and the wavelet packet energy and hierarchical entropy (sample entropy) of each frequency band were obtained as the fault eigenvector sets respectively. After the two kinds of eigenvectors were processed by the normalization method, BP neural network models were established respectively to realize the diagnosis of different fault modes of bearings. Finally, the D-S evidence theory was used to realize the information fusion of the two diagnostic results based on wavelet packet energy and hierarchical entropy. Through comparing the diagnostic results by wavelet packet energy, hierarchical entropy and information fusion, it is shown that the method based on the combination of neural network and D-S evidence theory is feasible and effective for the fault diagnosis of complex machinery.