A Broken Rotor Bar Fault Diagnosis Approach Based on Singular Value Decomposition and Variational Mode Decomposition

Abstract

The low-amplitude side-band fault feature frequencies of broken rotor bar fault in induction motor closing to the fundamental frequency in the stator currents are easily overlapped by high-amplitude fundamental frequency. In order to solve this problem, a new fault diagnosis approach that includes singular value decomposition (SVD), variational mode decomposition (VMD) and Hilbert transform (Hilbert-VMD) is proposed. Above all, SVD efficiently filters out the fundamental component and the color noises from one phase stator current. Simultaneously, SVD also calculates the effective number of band-limited intrinsic mode functions (BLIMFs) of VMD. Next, VMD decomposes the filtered phase stator current into finite BLIMFs. Finally, Hilbert transform performs on each BLIMF and combines the Hilbert spectrum of every BLIMF. So far, the fault feature frequencies are observed easily from the processing consequences. The simulation consequences prove the excellence of the proposed approach for broken rotor bar fault diagnosis.