Denoising Of Seismic Data Based On Convolution Neural Network

High quality seismic data is the basis of formation imaging and interpretation,but the existence of random noise seriously affects the processing and interpretation of subsequent seismic data,so it is difficult to judge the accurate location of oil and gas.Especially,the random noise in remote areas has the characteristics of non-stationary,high energy and serious aliasing of effective signal and random noise in frequency domain,which makes it very difficult for conventional denoising methods to recover seismic data.It is difficult for the traditional seismic data noise suppression algorithm to achieve the ideal effect.For this phenomenon,it is necessary to develop an efficient denoising algorithm to remove random noise and retain the complex edge information and rich texture information of the signal as far as possible in order to restore seismic data and improve the utilization rate of seismic data.In this paper,convolution neural network is applied to seismic data processing,and related algorithms are proposed to Denoise seismic data and enhance visual quality.The main contents of the study are as follows:1.Research on seismic data denoising based on convolution neural network.In view of the traditional denoising method,it is necessary to accurately model the signal and noise and optimize and adjust the manual input parameters,which makes it difficult to remove the noise from seismic data.It is found that the residual learning of DnCNN network can adaptively Denoise the different SNR data.Convolution neural network has the characteristics of end-to-end deep learning data,which can provide optimal sparse expression of local seismic in-phase axis.According to the characteristics of automatic feature extraction and blind denoising of DnCNN network,which combines batch standardization,residual learning and adaptive moment estimation,is used to optimize the network depth,training set and network parameters.An adaptive seismic data denoising algorithm based on depth convolution neural network is implemented.Through the comparative experimental analysis,it can be seen that the algorithm has a good denoising effect in seismic data,eliminates a large number of random noise,and preserves the texture features in the data2.Research on seismic data denoising based on dilated convolution network.In order to solve the problem of loss of texture details in edge part of seismic data denoising by DnCNN.It is found that dilated convolution can expand the receptive field without increasing parameters,and can make the network obtain more wave characteristics of seismic records,so as to retain more signal edge and texture information.According to the characteristics of dilated convolution,the DnCNN structure is improved.Combined with residual learning,batch standardization,adaptive moment estimation and other methods,the network framework is built,and the network expansion rate,training set and network parameters are optimized.A seismic data denoising algorithm based on dilated convolution network is implemented.Through comparative experimental analysis,it can be seen that the seismic data denoising algorithm proposed in this paper can effectively remove the noise similar to the signal frequency band in seismic records,retain more texture features in the data,and highlight the effective waves.Improve signal-to-noise ratio and improve the quality of seismic data...