基于改进自适应变分模态分解的滚动轴承微弱故障诊断

It is difficult to extract early fault information of rolling bearings because the signal is mixed with abundant compounded background noise. An adaptive variational mode decomposition (AVMD) with the Teager energy operator method was proposed. Firstly, the minimum mean envelope entropy (MMEE) was used to search the optimal value of parameters. Subsequently, the weighted kurtosis (WK) was adopted to select the effective modal components for signal reconstruction. Finally, the reconstructed signal was analyzed by Teager energy spectrum to identify fault frequency. The analysis of vibration signals of bearings with weak fault shows that the proposed method improves the decomposition accuracy, and has stronger noise robustness and fault identification ability than ensemble empirical mode decomposition and local mean decomposition.