Proper and Smooth Orthogonal Decompositions for Detection of Inner Race Defects in Rolling Element Bearings with Variable Rotational Speeds

Abstract

Rolling element bearings represent one of the prominent sources of nonlinearity in rotating systems, which makes fault diagnostics one of the most challenging tasks. This paper presents the application of proper and smooth orthogonal decomposition methods as feature extraction techniques for rolling bearing diagnostics under variable operating conditions. Multiple operation configurations are investigated using proper and smooth orthogonal-based features in order to characterize bearings with inner race defects. Results demonstrate the effectiveness of the proposed techniques for optimal detection of inner race defects under variable rotational speeds.