Study On Feature Reconstruction And Clustering Analysis Of Pre-stack Seismic Signals Based On Sparse Coding

Signal processing technology is widely used in today's production and life,and plays an extremely important role in the fields of communication engineering,medical and health care,military industry and seismic exploration.Signal processing technology plays an important role in seismic exploration and is applied to almost the whole process.Seismic phase analysis is an important application of seismic signal processing technology.Seismic facies analysis technology,based on the seismic signal response to different sedimentary patterns,to achieve the prediction of underground lithology,physical properties,oil-bearing,and the most critical step is the seismic signal classification.According to the characteristics of the signal itself,it can be divided into different types to realize the structural property analysis of underground strata,which is more conducive to guiding oil and gas exploration.In order to reduce the difficulty of data processing,seismic phase analysis technology is basically based on post-stack seismic signals.However,the post-stack signal comes from the rough weighted superposition of the pre-stack signal,and both the detail information and the data volume have been compressed and blurred,so that the accuracy of seismic phase analysis results cannot be guaranteed.So the main research of this paper is based on the technology of pre-stack signal classification.From the perspective of machine learning,this paper constructs the learning dictionary atomic library by Ksvd,and then combines the principle of signal sparse representation to extract the features of pre-stack seismic signals.Finally,unsupervised clustering algorithm is used to realize classification and recognition.So this paper is mainly about the pre-stack seismic signal feature extraction,classification and recognition algorithms.