Health assessment for rolling bearing based on local characteristic-scale decomposition — Approximate entropy and manifold distance

Abstract

The normal operation of rolling bearing is of great significance to the functionality and efficiency of complex primary system. Thus, health assessment for rolling bearing plays an important role in prognostics and health management for mechanical components. Considering the non-stationary and nonlinear characteristics of vibration signals of mechanical components, it's difficult to extract accurate features. This paper investigates the possibility of a novel bearing health assessment method using local characteristic-scale decomposition (LCD), approximate entropy (ApEn) together with manifold distance. First, the original vibration signals are decomposed into several intrinsic scale components (ISCs) by LCD. Second, the ApEns of each ISCs is computed, which is employed to compute the manifold distance with the ApEns of ISCs in normal condition. Finally the manifold distance is converted into confidence value (CV), resulting in a health degree of the rolling bearings. Validation data are collected to facilitate the evaluation of the proposed health assessment method. The results indicate that the proposed health assessment method for rolling bearing is of highly effectiveness.