Research On Adaptive New Method Of High-resolution Processing And Its Application To Seismic Data

In order to better display the details of the information contained in the signal,not only the mapping relationship between signals and time,but also the relationship between signals and frequency domain components is revealed.Previous studies have found that the high-frequency components in the signal are closely related to the resolution of the details.The collected signals are often affected by the surrounding environment,and some of the high frequency components are missing.Two new methods are proposed after multi-scale decomposition of the signal.One is the CEEMD-correlation coefficient threshold amplitude compensation method.The other is the extrapolation multi-resolution singular value decomposition method.The specific research contents are as follows:(1)This paper studies three classical seismic data processing methods,such as spectral whitening,inverse Q filtering and deconvolution,and selects two methods to compare the effects with the new method.(2)The Complementary Ensemble Empirical Mode Decomposition(CEEMD)method can multi-scale characterize the signal with single-component.The Pearson correlation coefficient can be used to express the correlation between two variables by a specific numerical value.According to the original signal and single component numerical characteristics,the Pearson correlation coefficient is improved.Combined with subtraction and high frequency amplitude compensation method,the CEEMD-correlation coefficient threshold high frequency amplitude compensation method is proposed.The method first obtains a spectrum of the signal to determine a threshold set G.Then researcher use CEMMD to decompose the signal to obtain the corresponding single component,and use the correlation coefficient to calculate the tightness of each single component and the original signal.The correlation coefficient of the adjacent single component is made subtraction,and the position of the threshold,the sudden change of the original signal real information and the false information,is found.Finally,a single component containing the real information of the signal is reconstructed to obtain a new signal,and the threshold set G is used to perform high frequency amplitude compensation processing to improve the signal resolution.The threshold set G is determined according to the spectrum of the signal itself.So that the high frequency amplitude compensation is more effective.The correlation coefficient threshold method can reduce the influence of false information such as noise on high-frequency amplitude compensation.(3)The Multi-Resolution Singular Value Decomposition method can display the details of the original signal at multiple scales.And the essence of singular value decomposition is the two processes of rotation and scaling.Inspired by high-frequency amplitude compensation method and deconvolution,combined with function fitting,maximum variance mode and other mathematical methods,this paper proposes the Extrapolated Multi-Resolution Singular Value Decomposition(EMRSVD)method.The method first uses the MRSVD method to decompose the signal to obtain a series of detailed singular values.Then we use the exponential function to fit the characteristics of the original detail singular value in the least square sense,and use the function to iteration extrapolate the high frequency singular value to construct the corresponding detail signal.Finally,under the control condition of the maximum variance mode,the constructed detail signal is added to the original signal to obtain the result of high resolution processing.EMRSVD adaptively extrapolates its high-frequency part based on the characteristics of the signal itself,not only retains the low-frequency information of the original signal,but also compensates the high-frequency information,and the resolution of the whole signal is improved.The introduction of the exponential function reduces the fitting error.At the same time,the introduction of the modified variance mode avoids adding too much noise.The signal-to-noise ratio before and after signal processing is almost the same.Two new methods of adaptive high-resolution processing are applied to the seismic data processing respectively.The processing results show that both methods can better improve the resolution of the signal and clearly display the detailed information of the signal.Moreover,the processing effect of the EMRSVD method is better than that of the conventional spectral whitening and deconvolution methods.Without increasing the seismic noise,the information of the entire section becomes clear,and the position of the thin layer can be directly observed.The two methods in this paper do not require researchers to rely on experience to set parameters.They can lay a good foundation for signal processing and interpretation.