An adaptive lifting scheme and its application in rolling bearing fault diagnosis

Vibration signals of rolling bearings usually are corrupted by heavy noise and it is very important to extract fault features from such signals. In this paper, an adaptive lifting scheme is proposed for fault diagnosis of rolling bearings. The kurtosis indexes of scale decomposition signals are used as the optimization indicator to select the prediction operator and update operator, which can adapt to the dominant signal characteristics, and reveal the fault feature. Fourier transform is adopted to remove the overlapping signal frequency components at every scale decomposition signal. Experimental results confirm the advantage of the adaptive lifting scheme over lifting scheme for feature extraction, and the typical features of rolling bearing in time domain are successfully extracted by adaptive lifting scheme.