Rolling element bearing defects are one of the dominant causes of mechanical system failure. Vibration-based diagnostic techniques are commonly used for detecting faulty conditions in rolling element bearings. However, fault characteristics can be easily affected by the normal vibration of other system components and even submerged in heavy background noise. To solve this issue, a method combining variational mode decomposition (VMD) and adaptive noise cancellation (ANC) using a normalized least mean square (NLMS) algorithm is presented in this paper. The ANC is applied to attenuate the noise within the complex vibration signal as a pre-processing step. After that, the denoised signal is decomposed into several different components via VMD. Finally, fault features are extracted from the decomposed signal components by envelope spectrum analysis. The simulation results show that the proposed method can precisely identify fault features and is much better than directly using original VMD without filtering under heavy noise conditions. The effectiveness of the proposed method is also verified by analysis of a practical bearing application.
