Sparsity-assisted signal representation for rotating machinery fault diagnosis using the tunable Q-factor wavelet transform with overlapping group shrinkage

Abstract

Rotating machinery fault diagnosis is of great importance for preventing catastrophic accidents. Effective signal processing techniques are in urgent demands to extract the fault features contained in the collected vibration signals. In this paper, a new sparsity-assisted feature extraction method is proposed for rotating machinery fault diagnosis. It is implemented using the tunable Q-factor wavelet transform (TQWT) with overlapping group shrinkage (OGS). The TQWT, for which the Q-factor is easily adjustable, is adopted as an effective tool to sparsely decompose vibration signals. Meanwhile, the OGS, which based on the minimization of a convex cost function incorporating a mixed norm, is employed to eliminate the irrelevant noise. The purpose of the proposed method is to extract useful features from observed signals. The effectiveness of the proposed method is demonstrated by extracting fault features from an engineering application case.