The Research Of Seismic Data De-noise Which Based On The Modified Fast Independent Component Analysis Algorithm

Independent Component Analysis (ICA) is a blind source separation methodwhich from multivariate data for its intrinsic statistical independence and non-Gauss’sfactor. ICA can separate and extract source signals from the mixed signal which don’thave any the information that the source signal and the hybrid matrix and need thesource signals are independent. In many practical situation, the assumption, which thesource signals are independent, is reasonable.Based on those reasons, ICA methodsare widely applied in many signal processing fields such as communication, patternrecognition, image processing, data compression, economic and so on.The background of the thesis and the development of ICA that include itsapplication and research situation are briefly introduced by this paper. The ICAprinciple that include its based mathematical model and the relevant mathematicalknowledge and the realization process and detailed discussed. In terms of the theoryof fast independent component analysis algorithm and practice circumstance, the theNewton iterative convergence speed of ICA is only two order up to three order. Inorder to further accelerate the convergence rate, the five order convergence rate of theimproved Newton-Raphson method, which can greatly improve the convergencespeed, is proposed.This new method that reduced the number of iterations andavoided the iterative oscillation condition make its have better convergenceperformance. Due to the algorithm of the FastICA depends on the initial value, andconsider on the damping factor, the new algorithm that add the the damping factor inFastICA is propounded.This new method improved dependence of initial weight inthe algorithm. The signal simulation to de-noise seismic data is operated and verifiedthe feasibility of the new algorithm.Seismic data is an important source of information that revealed the formation ofgeologic formation, and the noise is often accompanied by seismic signals in seismicrecords. When this random noise energy is too forceful, serious influence seismicimaging authenticity and reliability, interference in seismic data interpretation,geological exploration cost to increase, among them, the denoising problem is themost important problem in seismic exploration. Based on the careful analysis of seismic signal based on the characteristics, the improved FastICA method was appliedin the real seismic signal de-noising. The simulation results show that the improvedFastICA can well eliminate noise in the recording of seismic signals, so that theprofile quality are greatly improved, therefore the independent component analysis inseismic data processing application foreground is very wide.