Research And Application Of Seismic Random Noise Attenuation By Mproved Empirical Mode Decomposition Method

Noise suppression in seismic exploration is a key step in seismic data processing,which has been studied by many scholars.The Empirical Mode Decomposition(EMD)method proposed by Huang is very simple and intuitive.This method is an adaptive 'screening ' method.Compared with the 'windowing ' method such as wavelet transform,it is not affected by the basis function and the principle of uncertainty.Its complete adaptability will make the signal decompose according to the frequency from high to low,and then obtain the corresponding Intrinsic Mode Function(IMF)and a residual,and ensure that each Intrinsic Mode Function component decomposed has corresponding physical significance.In seismic exploration,the characteristics of noise are generally characterized by high frequency,so after Empirical Mode Decomposition,only the high frequency components are abandoned to complete the suppression of seismic noise.Therefore,many scholars have introduced Empirical Mode Decomposition into the random noise suppression of seismic exploration.Since the Empirical Mode Decomposition method has the properties of complete adaptability,completeness,approximate orthogonality,statistical significance of IMF components,and frequency modulation of IMF components,it ensures that each component——Intrinsic Mode Function obtained after EMD decomposition has corresponding physical significance.However,the Empirical Mode Decomposition method still has unresolved problems such as mode mixing and endpoint effect,which directly affects the decomposition results and IMF significance.Modal aliasing is generally caused by intermittent signal embedding and abnormal event interference,which is also very common for seismic exploration data.In this paper,Ensemble Empirical Mode Decomposition(EEMD)and Complementary Ensemble Empirical Mode Decomposition(CEEMD)are studied.Both methods use the characteristics of uniform distribution of white noise power spectral density to suppress the noise that would have modal aliasing phenomenon by introducing white noise.By studying the CEEMD method,it is found that this method can effectively embed the intermittent signal and separate the disturbance of abnormal events from the effective signal,thereby inhibiting the mode mixing phenomenon,which can be better applied to the field of seismic noise suppression.Through the theoretical seismic profile model calculation of EMD and CEEMD,it is found that the signal obtained by CEEMD decomposition is very clean,which verifies that the CEEMD method can be better applied to the attenuation of seismic noise and suppress the occurrence of modal aliasing.For the endpoint effect problem,this paper proposes a new extreme point parallel endpoint extension method to improve the original EMD,which solves the endpoint effect problem caused by 'overshoot ' and 'undershoot'.Through the research and analysis of the simulated signals synthesized in theory,and comparing the Empirical Mode Decomposition method with the improved Empirical Mode Decomposition method,it is found that the improved method in this paper has less decomposition times,higher decomposition effect and smaller error of decomposition results.Through the actual data test,it is found that the improved EMD prediction filtering method in f-x domain can better retain effective seismic data and suppress seismic noise.