Matching Squeezing Transform With Application to Fault Diagnosis

Time-frequency (TF) analysis (TFA) is a vital tool for processing nonstationary signals. Synchro-squeezing transform (SST) and synchro-extracting transform (SET) are two important TFA tools, but they are defective in terms of sharpening performance or reconstruction performance. Inspired by the SET, this article theoretically derives the matching instantaneous frequency (IF) estimator and proposes a new SST called the matching squeezing transform (MST). This MST obtains a more concentrated TF representation by squeezing the short-time Fourier transform (STFT) coefficients onto the IF trajectory while keeping its reconstruction capability. Synthetic signals and bearing fault signals are utilized to analyze the MST’s performance in TF concentration and signal reconstruction. Compared with state-of-the-art TFA methods, the proposed MST has better potential to address strongly nonstationary signals.