Weak Fault Feature Enhancement of Acoustic Data Based on Variational Mode Decomposition

Abstract

In the prognostic and health management of rotating machinery, the characteristic frequency of early weak fault is usually difficult to be extracted. To overcome this difficulty, this paper presents a weak fault feature enhancement method of acoustic data for rolling bearings based on variational mode decomposition (VMD). Firstly, the acoustic data is decomposed into some band-limited intrinsic mode functions (BLIMF) by the optimized VMD. Then an adaptive signal-to-noise ratio (ASNR) estimation method is proposed to determine the optimal BLIMF. Finally, the fault types of rolling bearings are identified through Hilbert envelope transform. Experimental results show that the presented method can effectively enhance the feature for early weak fault in rolling bearings with acoustic data.