The FERgram: A rolling bearing compound fault diagnosis based on maximal overlap discrete wavelet packet transform and fault energy ratio

Abstract

Compound fault features of the rolling bearing are difficult to separate and extract. To address this problem, the present paper proposed a diagnosis algorithm, namely FERgram, on the base of maximal overlap discrete wavelet packet transform (MODWPT) and fault energy ratio (FER). First, a group of frequency band signals are gained after MODWPT processing the initial vibration signal. Second, FER is chosen as the evaluation index, and then the FER values of each frequency band signal are calculated and used to generate FERgram. The frequency band signal with the maximum FER value containing plentiful fault information is chosen for envelope analysis. Finally, the fault type is determined by contrasting the prominent frequency component of the envelope spectrum with the fault feature frequency. The feasibility and superiority of the FERgram method are verified by four signals and four comparison methods. The results show that the FERgram method can effectively extract and accurately diagnose the compound fault of rolling bearing.