Enhancing Time-Frequency Concentration and Accuracy Using an Improved Time-Frequency Synchrosqueezing Transform

The synchrosqueezing transform (SST) improves the readability of time-frequency representation (TFR). The time-frequency multisqueezing transform (TFMST) is the most advanced SST method to date. It combines the advantages of frequency-reassigned SST (FSST) for harmonic-like components and time-reassigned SST (TSST) for impulsive-like components, thereby significantly enhancing the quality of TFR. In segmenting the original signal into harmonic-like and impulsive-like components, TFMST applies the standard of equal time-frequency concentration in both the frequency and time directions. However, achieving equal time-frequency concentration does not guarantee equal accuracy in these directions, which may lead to the misidentification of harmonic-like and impulsive-like components. Using TSST for harmonic-like components and FSST for impulsive-like components can result in time-frequency blurring. In addition, the strategy of first segmenting the signal into two components and then applying squeezing in TFMST can cause frequency oscillations at the segmentation boundaries. To achieve precise segmentation of impulsive-like and harmonic-like components and to produce a high-quality TFR, this article proposes an improved time-frequency synchrosqueezing transform (ITFST). ITFST features a more accurate segmentation standard and a more effective squeezing strategy. The former achieves precise segmentation of harmonic-like and impulsive-like components by ensuring equal reassignment accuracy in both the frequency and time directions. The latter first employs synchrosqueezing before segments the TFR to avoid processing truncated signals and effectively eliminates frequency oscillations at the segmentation boundaries. The effectiveness of ITFST is demonstrated through simulations and experiments, and its exceptional performance is thoroughly analyzed and discussed.