Weighted envelope spectrum based on reselection mechanism and its application in bearing fault diagnosis

Abstract

Abstract As a powerful algorithm, fast spectral correlation (Fast-SC) is widely used in bearing fault diagnosis. However, the interference of strong background noise and the existence of multiple fault-related frequency bands make it difficult to diagnose bearing faults using enhanced envelope spectrum and improved envelope spectrum on the basis of Fast-SC. To alleviate the problem, fast kurtogram is imitated to divide the spectral frequency bands; then, the threshold-based diagnostic feature (TDF) of each integration band is calculated to guide the weighting function of the corresponding level; finally, weighted envelope spectrum with the largest TDF is determined as the output; this process is called the reselection mechanism. Simulation and experimental results indicate that the design of weighting function highlights the contribution of the fault-related frequency bands, and the reselection mechanism further improves its performance.