Multi-lifting synchrosqueezing transform for nonstationary signal analysis of rotating machinery

Abstract

Blurry time–frequency representation (TFR) of most time–frequency analysis methods influenced by inaccurate instantaneous frequency estimation hinders the meaningful rotating machinery fault detection. To address the drawback, a novel time–frequency analysis method called multi-lifting synchrosqueezing transform (MLST) is proposed, which could enhance the energy concentration of TFR for fast-varying instantaneous frequency and strong noise signals of rotating machinery. Hereinto, the critical points are improved for nonstationary signal analysis: (1) construct a multisqueeze second-order lifting operator to accurately estimate the fast-varying instantaneous frequency; (2) propose a correction operator to correct the deviation in the process of assigning time–frequency energy; (3) put forward a time–frequency fault measure cluster (TFMC) to optimize and evaluate the result of MLST. Through such multi-lifting operations, the final TFR with a high readability could provide a promising support for rotating machinery fault diagnosis. The repeatable simulations and engineering applications are used to verify the effectiveness of the method.