Research On Weak Seismic Signal Enhancement Method Based On Stochastic Resonance

With the continuous improvement of seismic acquisition technology,the exploration depth is getting deeper and deeper,and the seismic information contained in seismic data is becoming more and more abundant.In deep seismic exploration,due to the limitation of seismic data acquisition conditions and the influence of complex geological conditions,weak signals will be generated in the original seismic data,that is,there are serious random noise interference effective signals in the profile,which will bring difficulties to the processing and interpretation of seismic data.Therefore,it is necessary to study the enhancement methods of deep seismic weak signals,so as to improve the quality of deep seismic data,which is also the basic work for accurate seismic identification of exploration targets.In this paper,the formation mechanism of seismic weak signal is discussed,and the enhancement method of seismic weak signal is studied.The seismic weak signal enhancement method either suppresses noise or enhances signal,or both.The main research work is as follows:Firstly,the formation mechanism of seismic weak signal and the common methods of weak signal enhancement are discussed,and the advantages and disadvantages of several conventional seismic weak signal enhancement methods are compared and analyzed.It is studied in detail that the widely used methods of seismic weak signal enhancement under the background of high random noise include Fourier transform,wavelet transform,curved wave transform,dictionary learning and so on.Wavelet transform is a method based on multi-scale analysis theory,which can show local features well,so it can have very good feature expression ability in both time domain and frequency domain,while curved wave transform has more scales.therefore,it has the characteristics of anisotropy and direction,which can well deal with the edge characteristics of seismic data.In the case of sufficient data,the dictionary learning method can achieve better signal-to-noise separation by building a dictionary and making use of the difference between seismic signal and noise sparsity.According to the idea that stochastic resonance theorizes noise into effective signal,this paper intends to carry out simulation research on the method of seismic weak signal enhancement under the background of high random noise based on the theory of stochastic resonance.According to the idea of stochastic resonance theory: changing noise into useful signal under nonlinear action,in order to improve its signal-to-noise ratio,broaden the idea of noise enhancement of seismic weak signal,and find a new method to enhance seismic weak signal.On the basis of the theory of seismic weak signal enhancement method studied by predecessors,combined with the basic theory of stochastic resonance,the basic theory of seismic weak signal enhancement method is carried out,and the stochastic resonance model of bistable system is established.By introducing the optimal matching theory and signal array theory,the seismic weak signal enhancement method based on optimal matched array-stochastic resonance method(OMA-SRM)is proposed.Then the denoising simulation experiments of seismic data are carried out by using wavelet transform,curved wave transform,dictionary learning,stochastic resonance and OMA-SRM,respectively,and the results are compared to analyze the noise suppression and enhancement effect of weak seismic signals under these methods.Through the comparative tests of two-dimensional sigmod theoretical model,twodimensional actual data,three-dimensional theoretical model and three-dimensional post-stack actual data,the results show that the noise suppression and enhancement result of weak seismic signal using OMA-SRM method is the best in the case of high signal-to-noise ratio,and has outstanding advantages in signal enhancement ability.