Bearing Fault diagnosis based on improved VMD and AR

Abstract

Rolling bearing fault diagnosis plays a very important role in the troubleshooting of automobile chassis transmission. In this paper, an improved variational mode decomposition (VMD) method combined with AR spectrum analysis is proposed for fault diagnosis of rolling bearings. The bearing vibration signal is decomposed into several VIMF components by VMD method. It is suggested that the instantaneous mean value is used as the evaluation index to determine the number of VIMF components. The selected components are then analyzed by AR spectrum to produce the VMD-AR spectrum of the vibration signal and the cumulative energy of VMD-AR spectrum. The bearing fault is diagnosed by comparing and analyzing such two kinds of maps corresponding to the normal and faulty conditions. Practical experimental results prove the effectiveness of the proposed diagnostic method.