Seismic Signal Fast Decomposition And Application By Multichannel Matching Pursuit With Genetic Algorithm

The frequency of seismic wave in different time has been changing when it propagates indifferent medium of underground strata, and the time-varying frequency can reveal importantinformation of geology, such as the characteristics of geological structures, lithology and fault,thus, decomposing seismic signals, which characterizes the frequency components of seismicsignals in different time completely, can extract geological information implied in the seismicsignal. Matching pursuit (MP) is widely used in seismic signal processing field with itssparsity and flexible adaptability, however, a problem exists in single-channel matchingpursuit (SCMP): the nonuniqueness of decomposition, and the reconstructed profile arelacking in spatial continuation between seismic traces because of the data noise. What’s more,single-channel matching pursuit involves enormous computational cost, while MultichannelMatching Pursuit based on Genetic Algorithm introduced in the paper can overcomes theshortcomings of traditional signal-channel matching pursuit algorithm.Firstly, this paper introduces the significance and domestic and overseas development ofseismic signal processing.Then, many seismic signal time-frequency decomposition methods are proposed andanalysed. Matching pursuit (MP) algorithm can overcome the shortcomings of traditionalsignal decomposition methods. Then the paper introduces the principle and application ofMatching Pursuit algorithm in detail. Yet the computational burden of this algorithm inseismic signal decomposition is so huge that many scholars put forward different improvedalgorithms.Finally, this paper describes multichannel matching pursuit (MCMP) algorithmdetailedly, it is to decompose all seismic signals at the same time and get the besttime-frequency atoms that be shared by all the seismic traces through changing the criterionfunction of searching the best matching atoms so that make decomposed atom stable andreduce the influence of data noise to some extent, so there is lateral continuation along aprofile. Combining with Genetic Algorithm, multichannel matching pursuit can reduce thecomplexity of algorithm. Furthermore, the algorithm is applied to fractional frequency displayof actual seismic profile data. There are simulation experiments for both synthetic and actual seismic data onsingle-channel matching pursuit (SCMP) and multichannel matching pursuit algorithm, andthe results are analyzed in detail. These simulation experiments have verified the efficiency ofthe algorithm introduced in this paper.