Independent Component Analysis Theory And It's Application

As a new signal processing method, independent component analysis appeared and developed with the development of blind signal separation technology in the later 1980s, because of its demands for very little prior knowledge of source signals and better signal separation performance, it has obtained successful applications in a more and more wide range of fields. In recent years, many scholars and experts used it in the seismic information processing area, and also have made many achievements. The purpose of this paper is mainly to explore the application of ICA in seismic information processing field.Traditional de-noising method will damage the effective signal frequency components while eliminating the noise, yet independent component analysis method is based on statistical independence principle, and can remove noise while retaining the effective signal frequency components. The handling objects of ICA are non-Gaussian signals, according to the high-order statistical analysis knowledge and the information entropy theory, ICA can be used to extract independent components and find the hidden information ingredients from observed data based on the extraction criteria of the independence between the implicit variables, which makes the separation results have more physical meaning. Compared with the traditional multi-dimensional signal analysis method, the components separated from ICA are not only irrelevant to each other, but also mutually statistically independent. Therefore, ICA can reveal the nature of the data structure more efficiently. Because of its important breakthrough in many ways, in this paper, ICA will be applied to purify seismic attributes, de-noise seismic data and identify gas-liquid two-phase flow.By assuming that the effective signal and random noise are statistical independent, ICA is used to purify the seismic attributes data, so as to eliminate the interference and obtain the independent seismic attributes. Because of the merit that ICA can retain the effective signal frequency components while removing noise, it is used to filter out the random noise in seismic data, so as to keeping bandwidth of the seismic data while improving the signal to noise ratio. In view of the characteristic that ICA can extract independent components and find the information hidden in the data, ICA is used to separate the detected signals of gas-liquid two-phase flow, in order to obtain the signals characterizing gas and water from the mixed signals, and then can facilitate the further processing and analysis of the observed signals.