Research On Sparse Representation Method Of Seismic Data Based On Multi-objective Evolutionary Algorithm

As my country's oil demand increases year by year,oil imports show a rapid upward trend,but my country's oil production remains relatively stable.In order to increase oil production,seismic exploration is the main means of detecting oil resources,the seismic data generated in the conventional large-scale seismic exploration process is difficult to recover in real time due to limitations in communication bandwidth and power consumption.In addition,the field exploration environment is complicated,when the geophone is deployed,it may encounter mountains and rivers that the geophone cannot be placed,which affects the integrity of seismic data acquisition.Compressed sensing theory is combined with large-scale seismic exploration,and the original seismic data is compressed and transmitted by the measurement matrix,which reduces the amount of seismic data transmitted and improves the transmission efficiency.Moreover,for locations where it is difficult to place geophones,the missing original seismic data can be recovered by using a reasonable arrangement of geophones and later using reconstruction algorithms.The sparseness of seismic data is a prerequisite for being able to be compressed,according to the theory of compressed sensing,the sparser the seismic data,the higher the reconstruction accuracy.Different sparse representation methods have different sparseness of seismic data after sparse representation,in order to make the reconstruction of seismic data more accurate,a sparse representation method of seismic data based on multi-objective evolutionary algorithm is studied.The main research contents are as follows:(1)The sparsity of seismic data and sparse representation theory are studied.The necessity of sparse representation of seismic data is analyzed in detail,the wave equation of elastic waves is studied,and the sparseness of seismic data generated in the process of geological propagation of seismic waves is proved,the theory of sparse representation of seismic data is studied by combining the formula.(2)The sparse representation methods of seismic data are studied,and the shortcomings of each method are analyzed in detail.The sparse representation method based on fixed transform domain and the sparse representation method based on dictionary learning are studied,the principle of each method is studied in detail,and each method is used to sparse the seismic data,and then the effect of their sparse representation is studied,and the shortcomings of each method for the sparse representation of seismic data are analyzed in detail.(3)The SPEA2 algorithm for sparse representation of seismic data is studied.In this paper,entropy and mutual information are studied,based on entropy and mutual information,the two objective functions of SPEA2 algorithm are studied,the genetic operator and initial dictionary are designed,and the sparse representation method of seismic data based on SPEA2 algorithm is proposed.(4)Combined with the measured seismic data,the sparse representation method is studied,and the research results are summarized in detail.In order to evaluate the effect of the sparse representation method on the sparse representation of seismic data,the reconstruction index is studied,according to the reconstruction index,the sparse representation method of fixed transform domain and the dictionary learning sparse representation method are studied.The SPEA2 algorithm is used to represent seismic data sparsely,the sparse representation effect of SPEA2 algorithm is studied,and compared with the sparse representation method of fixed transform domain and dictionary learning sparse representation method,the research results are summarized in detail.In summary,this article explains the relationship between dictionary and seismic data from the perspective of relevance and redundancy,studies entropy and mutual information,designs genetic operators and initial dictionaries,and proposes a multi-objective evolutionary algorithm for sparse representation of seismic data.The dictionary obtained by the multi-objective evolutionary algorithm is tested,under the same conditions,after the sparse representation of the seismic data,the reconstructed seismic data has a higher signal to noise ratio and the relative error of the reconstruction is smaller.