Multiple synchro-tuning chirplet transform

Time-frequency analysis (TFA) is an effective tool for characterizing non-stationary signals. Due to the limited time-frequency resolution capability, traditional TFA methods cannot achieve the optimal time-frequency representation (TFR). Based on the post-processing technique, a novel TFA method named multiple synchro-tuning chirplet transform (MSTCT) is proposed to obtain high-resolution and robust TFR for non-stationary signals. The MSTCT first estimates the spectral energy, instantaneous frequency (IF), and chirp rate (CR) of non-stationary signals with high time-frequency resolution, enabling precise energy localization on the time-frequency plane. Then, a multiple reassignment procedure is proposed to concentrate the blurry time-frequency energy and yield a better TFR. Moreover, benefiting from the multiple squeezing mechanism in the frequency axis, the MSTCT holds the potential to reconstruct the signal with high accuracy. Both the simulation results and the real-world experiment using the bat echolocation data demonstrate that the proposed MSTCT algorithm achieves superior performance in the time-frequency resolution, noise immunity, and reconstruction capability compared to the state-of-art linear TFA methods.