Fault Diagnosis for Rolling Element Bearings with VMD Time-Frequency Analysis and SVM

Abstract

This paper proposes a novel fault diagnosis method for rolling element bearings based on the newly developed adaptive signal processing technique -- variational mode decomposition (VMD), combining with time-frequency feature extraction and support vector machine (SVM). For the given samples of fault signals, VMD is firstly employed to decompose the signals into collections of intrinsic mode functions (IMFs). To extract more characteristics of fault information, 20 features of each IMF are calculated from time domain and frequency domain respectively. Then fault feature vectors of all samples are established by assembling features of the IMFs belonging to the same signal. Finally, all fault feature vectors are utilized to train SVM classifier, with which the fault modes of rolling element bearings are identified. To verify the effectiveness of the proposed model, EMD is utilized for comparison during the signal decomposing stage. The experimental result shows that the proposed method has better diagnosing performance.