Fault diagnosis for rolling element bearing using EMD-DFDA

Abstract

A new fault diagnosis method for rolling element bearing is proposed based on empirical mode decomposition (EMD) and fisher discriminant analysis (FDA). First, non-stationary vibration signals are processed by applying EMD technique, and stationary IMF components are obtained. Then, fault feature vectors with the moving time-lagged windows are composed using the absolute values of IMF components of healthy and detection bearings in order to consider the dynamic behavior. Finally, a DFDA model is construed and a linear discriminant matrix is obtained by which IMF components are projected into the low discriminant space. The diagnosis performance of the proposed method is tested using a dataset from bearing data center of Case Western Reserve University.