Application Of Independent Component Analysis In Seismic Exploration

Independent component analysis is a novel method, which can separate themulti-dimensional observed signals into independent components via linear transformationmaking the non-Gaussian maximize. The method makes full use of higher order statistics ofsignals, can adaptively decompose the data into statistically independent component, digs outthe main features of the data. This technology has been successfully applied to various aspectsof the seismic exploration. The paper counters the difficulty of the carbonate reservoirsexploration, applying ICA technology to this exploration, supplying new methods includingrandom noise denoising based on ICA, reservoir identification, attribute optimization andreservoir parameters prediction techniques for the difficulty in the exploration.In the aspects of independent component analysis theoretical algorithm, Systematicallydiscuss the basic theory of ICA technique, and modify the Newton iteration formula of fastindependent component analysis based on negentropy, the modified algorithm possesseshigher convergence speed and better robustness, and which is efficient to be realized byhardware. It provides a stable and efficient software environment for appling ICA in theseismic exploration.The paper studies the application of the ICA technology in prestack seismic randomnoise suppression, proves the effectiveness of ICA technology in random noise suppression,and first proposes random noise denoising technology which combines empirical modedecomposition and ICA. This technology applies ICA in random noise denoising of singlechannel seismic signal, which not only avoid the limitation of ICA denoising technology, butalso takes full advantage of the benefits of both technologies. Experimental simulation andreal data prove that this method has good noise suppression effect, greatly improve the signalto noise ratio of prestack seismic data.The paper utilizes the reservoir identification technology which combines JADE algorithm and generalized S transform, combines the conditions of seismic and geological,realizes the reservoir identification of the marine carbonate. It researchs the seismic attributeoptimization technique based on independent component analysis and kernel independentcomponent analysis and the reservoir parameters prediction techniques based on seismicattribute optimization. Through the processing and comparison analysis of experimentalsimulation and real data, it proves that the support vector machines reservoir parameterprediction technique based KICA effectively depicts the spatial distribution characteristics ofthe reef, and accurately predicts the reservoir parameter distributions.