Research And Application On CEEMD Method

In seismic exploration,the complexity of underground medium leads to the received seismic data is often complex nonlinear and non-stationary signal,and spectrum decomposition technique is a powerful tool for analyzing this kind of signal.At present,the commonly used spectrum decomposition methods are the traditional methods developed based on the Fourier transform theory,such as short time Fourier transform(STFT),continuous wavelet transform(CWT),S transform and generalized S transform and so on.However,the conventional spectral decomposition methods have some problems,such as selecting decomposition base with difficulty,restricted by Heisenberg's uncertainty principle,so that the analysis effect of seismic signal by using the methods is not very satisfactory.And it is difficult to meet the needs of high-resolution seismic data processing and interpretation.Hilbert-Huang transform(HHT)is a kind of powerful tool to be suitable for analyzing nonlinear and non-stationary signal,including both empirical mode decomposition(EMD)and Hilbert transform,and the basic idea is that it first uses EMD to decompose the nonlinear and non-stationary signal into a series of sub-signals called intrinsic mode function(IMF)with different frequency components,then applies the Hilbert transform for each IMF component to obtain the signal instantaneous attributes with obvious physical meaning,which has been already applied widely in seismic data processing and interpretation.However,the noise or intermittent signal can make the EMD generate serious mode mixing effect,which leads to the distortion of the IMFs and the generation of aliasing.EMD and its improved algorithms including EEMD and CEEMD are studied in this paper,and the latest algorithm variation called CEEMD is applied to seismic data denoising and the gas-bearing detection of thin sandstone reservoir.First,this paper introduces the principle of several time-frequency analysis methods,including STFT,CWT,three parameters wavelet transform,S transform,synchrosqueezing wavelet transform(SWT)and HHT known as the powerful tool for nonlinear and non-stationary signal analysis,and uses synthetic signal test to analyze and discuss the influence factors of time-frequency resolution performances of the methods.Then,this paper expounds the basic principle,screening iterative process,properties aswell as the existing problems of the EMD.To deal with the mode mixing problem of the EMD,a improved algorithm called ensemble empirical mode decomposition(EEMD)is introduced in this paper.It is a kind of noise-assisted analysis method,which can improve the mode mixing problem to some extent,but at the same time,it introduces some new problems,such as poor completeness,higher computational cost.Next,this paper studies the principle and implementation process of the complete ensemble empirical mode decomposition(CEEMD)known as the latest variant of the EMD algorithm,and uses synthetic signal test to verify the superiority of CEEMD relative to the EMD and EEMD method.CEEMD can not only overcome the mode mixing effect,but also provide an accurate reconstruction of original signal and relatively high computing efficiency than EEMD.Finally,in view of the superiority of CEEMD,this paper presents a wavelet threshold denoising method based on CEEMD and a high-resolution time-frequency analysis method called CEEMD/HT,then applies the two methods in the seismic data denoising and thin gas-bearing sandstone reservoir detection,respectively.In seismic denoising,the simulated signal test and actual seismic data analysis results show that wavelet threshold denoising based on CEEMD not only can suppress random noise effectively,but also maintain effective signal against losses.Thus it is proved to be an effective and relatively preserved-amplitude denoising method.In the detection of the thin gas-bearing sandstone reservoir,the reservoir location is more accurately depicted,and low-frequency abnormal phenomenon is more obvious on the CEEMD/HT frequency-division section.Meanwhile the analysis results of normalized formation absorption profile are consistent with instantaneous spectrum analysis and can indicate intuitively the existence of gas.On the other hand,the high-frequency and low-frequency absorption attenuation gradients extracted by CEEMD/HT method combining with wavelet spectrum fitting can clearly and accurately show the location and describe the space distribution of the gas-bearing sandstone thin reservoir.I believe CEEMD/HT time-frequency analysis method can be used as a powerful tool for thin layer fine interpretation.