A new ball bearing fault diagnosis method based on EMD and SVM

Abstract

This paper presents a new method which combines empirical mode decomposition (EMD) and support vector machine (SVM) together for bearing fault diagnosis in low speed-high load rotary machine. EMD is a novel self-adaptive method which is based on partial characters of the signal. Vibration signal measured from a defective rolling bearing is decomposed into a number of intrinsic mode functions (IMFs), with each IMF corresponding to a specific range of frequency components contained within the vibration signal. Then calculate the energy entropy mean of each IMF and normalization motor speed(RPM) to construct feature vector to train SVM classifiers. The results of application in simulation signal and practical bearing fault signal both show its efficiency. © 2011 Springer-Verlag Berlin Heidelberg.