Iterative matching synchrosqueezing transform and application to rotating machinery fault diagnosis under nonstationary conditions

Synchrosqueezing transform (SST) and its advanced version matching SST have been widely used for rotating machinery fault diagnosis. However, the short-time Fourier transform (STFT) based matching SST suffers from extra time-frequency blurriness when the analyzed signal is strongly time-varying and contaminated by noise. A novel method called iterative matching Synchrosqueezing transform is, therefore, proposed by iteratively applying reassignment operations to address the smearing problem of time-frequency representation (TFR). The capability of the proposed method in increasing the readability of TFR is mathematically analyzed. The signal reconstruction is also derived and realized by detecting instantaneous frequency (IF) ridges from the improved TFR. Analyzing results of simulated and experimental signals show that the proposed method could further sharpen the IF ridges and improve the TFR readability. The detailed IF features of vibrations can then be easily identified on the TFRs of the proposed method for rotating machinery fault diagnosis.