Research On Seismic Signal Processing Method Based On Spectral Decomposition And Time-frequency Filtering

As the major areas of seismic exploration in China gradually turn to complex regions including lithologic strata,conventional time-frequency analysis methods such as Fourier transform,short-time Fourier transform,and wavelet transform,which have low resolution due to their frequency attribute analysis methods,is unable to meet the requirements for the resolution and accuracy of seismic attribute analysis anymore.If the conventional time-frequency analysis method could be modified to obtain a highresolution seismic data section and the absorptive attenuation estimation of the seismic signal,the accuracy of the oil and gas identification and reservoir prediction of the seismic data will be improved effectively,providing effective support of the theory,method and technique for seismic exploration.This paper selects seismic signals as the research object and focuses on studying its time-frequency resolution as well as absorptive attenuation characteristics.Based on the analysis of simulation and actual signal experiments,a modified generalized Stransform method is employed to process the high-resolution filtering of seismic profiles.Moreover,the spectral decomposition method based on the Prony transform is analyzed to achieve the absorptive attenuation estimation of seismic signals.The main research contents are as follows:(1)Applied research on the existing spectral decomposition and time-frequency analysis methods in signal processing.Based on a systematic analysis of the principle of Fourier transform,short-time Fourier transform,Gabor transform,continuous wavelet transform and Wigner-Ville distribution method,the advantages and disadvantages of these conventional time-frequency analysis methods are compared and analyzed respectively in the time domain,the frequency domain and the timefrequency domain through simulation experiments.Combined with the experimental results,the existing problems of the above methods in dealing with non-stationary signals are analyzed.(2)Research on the high-resolution filtering method of seismic signal based on the generalized S-transform.The standard S-transform estimator suffers from some inherent disadvantages of poor energy concentration with higher frequencies.Therefore a modification to the existing S-transform is proposed to improve the performance of the distribution.Additional parameters are introduced to control the window's width more precisely which can greatly enhance the signal decomposition in the time-frequency domain.The performance of the modified generalized S-transform is evaluated by the theoretical simulation data as well as actual seismic data.The results show that this method can effectively improve the resolution of the seismic signal in the time and frequency domain,and make the target layer in the seismic profile more clear and coherent,which provides a good foundation for subsequent interpretation and analysis of seismic data.(3)Research on the absorptive attenuation estimation method of seismic signal based on Prony transform.Aiming at the limitation of the frequency attribute analysis of conventional time-frequency analysis method,the Prony transform method is adopted to directly obtain the absorptive coefficient of the seismic signal in the target layer,so as to improve the accuracy of the result.By selecting the sampling frequency,the time window and other parameters of the Prony transform and modifying its ordering method,the absorptive attenuation estimation of the seismic signal is improved,and therefore the absorptive attenuation anomaly of the seismic data in different frequency bands are obtained.