The Improvements Of Time-frequency Analysis Based On SST And Its Application Research

As a forceful tool in the field of signal processing,Time-Frequency(TF)analysis helps us recognize and extract the characteristics of the signal in the meanwhile providing accesses to the subsequent processing.How to obtain TF results with high resolution for accurate signal description is the main target for time-frequency analysis.Synchrosqueezing transform(SST)is a novel TF analysis tool which can optimize the original representation and improve the resolution by means of "redistributing" the energy of TF plane.Resort to the complete mathematical theory and ability of reconstruction,it is also becoming a popular topic in the field of signal.Overcoming the drawbacks of SST itself and improving the accuracy can be considered as a critical issue in the application of SST for signal processing.After introducing the comprehensive theory of SST in this paper,we verify its performances of energy concentration and reconstruction,as well as the ability of suffering noise.For solving the smear occurred in SST,Second Order Synchrosqueezing Transform(SST2)can be regard as a revise for the instantaneous frequency estimation.We validate its applicability by detailed derivation and model testing.Inspired by the idea of utilizing high-order phase information to correct instantaneous frequencies,we deeply explore the principle of high-order SST and discuss its process of realization while the synthetic and real data are both involved for testing the results.Nevertheless,suffering from the inherent resolution of STFT,high-order SST still has defects for complex signal.A novel TF method called Adaptive Synchrosqueezing S Transform(ASST),as well as its second-order form(ASST2)using local operator,are firstly proposed in this paper and the corresponding improvement based on the correction of iterative SST(Multi ASST2)is also accomplished for reducing the error in ASST2.This new method is promoted from its own performance and the process of SST.Numerical simulations prove that can accurately locate the features of signals in the TF plane and provide representations with higher resolution.Finally,the aforementioned high-precision improvements based on SST are applied to the spectral decomposition and ground-roll suppression for evaluating their capability in seismic processing.