Rolling Bearing Fault Diagnosis Using Improved Lifting Scheme
Vibration signal carries dynamic information of rotating machinery and the useful information often is corrupted by noise. Effective signal de-noising and feature extraction methods are necessary to analyze these signals. In this paper, an improved lifting scheme is proposed for such vibration signal analysis. The auto-correlation factor of scale decomposition vibration signal is used as to optimize the prediction operator and update operator at each sample point, which can adapt to the vibration signal characteristic. Improved lifting scheme decomposition and reconstruction procedures are designed. Experimental results confirm the advantage of the proposed method over redundant wavelet transform for rolling bearing fault diagnosis, and the typical fault features in time domain are desirably extracted by improved lifting scheme.