A new approach for performance degradation feature extraction based on generalized pattern spectrum entropy

Abstract

Performance degradation feature extraction is the basis of degradation condition recognition and remaining service life prediction. In this paper, morphological corrosion operator is introduced into mathematical morphological particle analysis (abbreviated as MMP), proposing a new analytical method named generalized mathematical morphological particle analysis (abbreviated as GMMP). On this basis, a new approach for degradation feature extraction based on generalized pattern spectrum entropy (abbreviated as GPSE) is proposed taking GMMP and information entropy as the theoretical foundation. In this approach, GPSE is calculated as degradation feature parameter in describing performance degradation degree of machinery equipment. Simulation analysis is processed, and the result shows that the value of GPSE will increase correspondingly along with the deepening of the degradation degree and the relevance between GPSE and degradation degree is stable. The effectiveness and practicality of the approach is tested through rolling bearing whole lifetime vibrating data. Rolling bearing fatigue life enhancement testing was carried out in Hangzhou Bearing Test & Research Center, getting the whole lifetime data which is able to cover each degradation condition from normal to invalidation. The approach is applied into analysis of rolling bearing data and the results verify its validity and feasibility.