Research On MCA Denoising Technique In Sparse Domain

The quality of seismic data is the basis of imaging and interpretation of the formation,but the presence of noise will have a serious impact on the subsequent processing and interpretation of seismic data,especially in the weaker part of the effective signal energy,the noise and the effective signal are seriously aliased,which It is very difficult to recover seismic data using conventional methods,especially the frequency bands of some noises and effective signal frequency bands are superimposed on each other,making it difficult to apply traditional noise suppression algorithms to seismic data.Therefore,it is necessary to develop an effective random noise suppression algorithm.At the same time,better retain effective information.With the change of exploration targets,hydraulic fracturing microseisms have gradually become a hot topic of concern to the majority of scholars.In hydraulic fracturing microseisms,they are usually divided into underground microseismic monitoring and ground microseismic monitoring according to the location of the detection points.Unlike downhole monitoring,there is usually more serious noise interference in the seismic data detected by surface microseismic.Ground-mounted geophones are more susceptible to disturbances such as wind,machine operation,and industrial AC power.Since micro-seismic signals are relatively weak relative to environmental noise,they are usually covered by environmental noise.Therefore,the problem of denoising ground microseismic also needs to find a way denoising method to get better suppression effect.In this paper,the effective signals in the seismic data are distinguished from the noises such as surface waves and random noise in morphology,and the morphological component analysis method in the blind source separation problem is used to de-noise the seismic data.The research content mainly includes the basic problems of blind source separation and two main methods-independent component analysis and morphological component analysis,indepth study of the principle of morphological component analysis,and explores conventional sparse expression algorithms,such as BP,OMP,Gradient Tracking and LARS.Through a series of numerical simulation experiments including one-dimensional signals and two-dimensional images,the feasibility of denoising by morphological component analysis was verified.According to the different morphology,the noise is divided into regular linear noise and irregular random noise,and the denoising experiment of morphological component analysis is carried out by taking the random noise in surface wave and simulated microseism as an example to verify the morphological component in conventional seismic data and microseismic Medium denoising effect.After actual data verification,the morphological component analysis seismic data noise reduction algorithm can effectively eliminate the linear interference in the seismic data,and better handle the situation of strong random interference similar to the microseismic record,which can be retained in the data.Multiple texture features highlight effective waves,improve signal-to-noise ratio and improve the quality of seismic data.