LAUSR

Abstract In contrast to empirical mode decomposition (EMD), variational mode decomposition (VMD) has better robustness to noise and sampling. In the VMD, the mode number is a key parameter that VMD can be applied successfully. In order to enhance the reliability of the adaptive selection to this parameter, tentative variational mode decomposition (TVMD) is proposed in this study. On the other hand, when a real signal sampled in strong noise background is analyzed by TVMD, the mode mixing problem frequently happens, dynamic time warping (DTW) is thus adopted to reconstruct original signals well. The effectiveness of the proposed approach is verified by both simulation analysis and the vibration signals of bearings with an outer race, an inner race and a rolling element faults, respectively. The experimental results indicate that the proposed method can extract the bearing fault features and has more reliable ability for selecting the mode number. Compared with EMD, ensemble empirical mode decomposition (EEMD) and complete EEMD (CEEMD), the proposed method has superior performance in fault feature detection.