Weak Feature Extraction of Rolling Bearing Fault Based on Generalized Variational Mode Decomposition

Abstract

Aiming at the deficiency of variational mode decomposition (VMD) in on-demand extraction of weak feature components, a generalized VMD (GVMD) is proposed to extract the weak features of rolling bearing faults. GVMD has excellent multi-scale and fixed frequency decomposition performance in the frequency domain. The spectrum decomposition positions and frequency domain decomposition scales of the algorithm can be flexibly dominated by prior center frequencies and scale parameters to realize on-demand decomposition. The simulation and experimental results show that, compared with VMD, GVMD can accurately extract weak feature components of bearing faults as desired by taking full advantage of bearing fault frequency information and bandwidth information, and the algorithm is robust to noise.

Keywords

• Variational mode decomposition;Rolling bearing fault;Weak signal extraction;On-demand decomposition