Weak fault signature extraction of rotating machinery using flexible analytic wavelet transform

Extracting and revealing the weak periodic fault vibration impulses is reasonable for damage detection of rotating machinery. However, the widely used dyadic WT suffers fixed frequency partition manner and low oscillating bases which would weaken its performance in weak fault detection. A new method based on flexible analytic wavelet transform (FAWT) is proposed in this article. Employing fractional and arbitrary scaling and translation factors, FAWT possesses attractive properties such as flexible time-frequency (TF) covering manner, better shift-invariance and tunable oscillatory nature of the bases, offering proper wavelet frame and bases shape to match the weak fault components. Moreover, FAWT is effective in revealing the amplitude modulation feature of the periodic fault impulses that occur in some damaged rotating components. The applications to a rolling bearing, a planetary gearbox of, and a flue gas turbine unit show that the proposed method is effective in extracting weak impulsive fault signature.