Synchro spline-kernelled chirplet extracting transform: A useful tool for characterizing time-varying features under noisy environments and applications to bearing fault diagnosis

Abstract

Profiting from the good ability in time–frequency (TF) representation (TFR), synchroextracting transform (SET) has been extensively used for non-stationary signal processing. Vibration signals of bearing often contain strong noise, whose time-varying features are easily contaminated by noise information, affecting the TF readability of SET. Effectively identifying the time-varying features from noise are important for fault diagnosis. For this purpose, by introducing the frequency-rotating operator and frequency-shifting operator of spline-kernelled chirplet transform (SCT), this paper studies a new time–frequency analysis (TFA) method called synchro spline-kernelled chirplet extracting transform (SSCET). It is shown that the proposed method can effectively reveal the variations of time-varying features while retaining the energy concentration in noisy cases. Besides, the proposed method uses the studied binary TF image separation strategy to extract the time-varying features of multi-component signals. The comparative analysis results in simulations verified its effectiveness in energy concentration and anti-noise. The proposed method is finally successfully applied to the analysis of a bat echo signal as well as the fault diagnosis of a rolling bearing.