Rolling bearing fault diagnosis based on EMD-TEO and Mahalanobis distance

A intelligent rolling bearing fault diagnosis method is proposed on Empirical Mode Decomposition (EMD)-Teager Energy Operator (TEO) and Mahalanobis distance. EMD can adaptively decompose vibration signal into a series of Intrinsic Mode Functions (IMFs), that is, zero mean mono-component AM-FM signal. TEO can estimate the total mechanical energy required to generate signals, so it has good time resolution and self-adaptive ability to the transient of the signal, which shows the advantage to detect the signal impact characteristics. With regards to the impulse feature of the bearing fault vibration signals, TEO can be used to detect cyclical impulse characteristic caused by bearing failure, gain the instantaneous amplitude spectrum of each IMF component, then identify the characteristic frequency of the interesting and single IMF component in bearing faults by means of Teager energy spectrum. The amplitude of the Teager energy spectrum in inner race fault frequency, outer fault frequency and the ratio of the energy of the resonance frequency to the total energy were extracted as the feature vectors, which were used as training samples and test samples separately for fault diagnosis. Then the Mahalanobis distances between the real measure and different type overalls of fault sample are calculated to classify the real condition of rolling bearing. Experimental results was concluded that this method can accurately identify and diagnose different fault types of rolling bearing.