Fault Feature Extraction of Rolling Bearing based on Multi-scale Morphology and SVD

Abstract

Aiming at the problems of multi- scale morphological difference filter,a new fault feature extraction method of rolling bearing based on multi- scale morphological difference filter and singular difference spectrum is proposed,which could effectively filter out the noise and extract the fault details. With the help of the theory of feature energy ratio,two different kinds of multi- scale weighted method is proposed in the multi- scale morphological difference filter,and the results of multi- scale morphological difference filter in different biggest analysis scales are studied. Then,by using the singular difference spectrum,put the result of morphological filter which failed to effectively filter the Gaussian noise into SVD reconstruction,and the final feature extraction result is got. The simulation shows the method could effectively extract the pulse shock signal under the high noise background,and in the fault feature extraction of actual bearing fault signals is also.

Keywords

• Multi-scale;Morphological difference filter;Feature energy ratio;Singular value decomposition（SVD）;Difference spectrum