摘要

In this paper, variational mode decomposition (VMD) method for wind turbine fault diagnosis was proposed. VMD is a non-recursive signal decomposition method, which can effectively avoid mode mixing existing in recursive decomposition model including empirical mode decomposition (EMD) or local mean decomposition (LMD) when it is used for actual vibration signal analysis. Mode mixing caused by intermittent and closed frequency components were analyzed respectively, and VMD shows a better performance than LMD. Finally, VMD was introduced into fault diagnosis of wind turbine transmission system and a decomposition characteristic comparison between VMD and LMD was done. The results show that the method based on VMD has a better performance than LMD not only in detecting unbalance fault but also monitoring incipient fault of bearing, due to its wavelet package-like filtering characteristic.