An enhanced variational mode decomposition based on correntropy and a periodicity-assisted log-cycligram for bearing fault diagnosis

Abstract

Variational mode decomposition (VMD) outperforms other decomposition methods in extracting fault-related modes. However, the major issue in the practical application of VMD is how to determine some key parameters appropriately, especially the mode number. To overcome this problem, an enhanced VMD is presented in this paper. A powerful tool, the rate of change of correntropy, is applied to estimate the mode number precisely. Then, the target mode is selected by a periodicity-assisted log-cycligram rather than conventional kurtosis. The advantages of the enhanced VMD are exposed by applying it to the experimental data collected from two different bearing test benches. Compared with the conventional fast kurtogram and empirical mode decomposition method, the enhanced VMD can extract the target fault mode from the collected vibration signal with strong background noise.