Research On The Principle And Application Of Blind Source Separation

As a kind of important means of data processing, blind source separation has beenwidely applied in geophysical exploration, such as seismic wavelet estimation, seismic datade-noise, seismic attribute extraction, and so on.As a kind of significant and effective method,seismic processing researcher think highly of independent component analysis.On the base ofprobability theory, information theory and estimation theory which are associated with blindsource signal separation, we shows the basic principle, data preprocessing, objective functionsand optimized algorithms of the technique, also emphasized introduced the fast independentcomponent analysis.With the guidance of my teacher, through the simulate experiment,confirm the reliability of the improved H-J algorithm which is used to solving the blind signalseparation problem under the linear mixture condition. At the same time, the emulatedcombine two-phase response by mixed up gas&liquid monophase response signals and thereal gas&liquid two-phase response signals were separated analysis suggests that originalsource signals, which characterized monophase response of gas&liquid, were successfullyseparated from gas&liquid two-phase combine response, and the characteristic of separatedoriginal source signals is contribute for the identify of flow phase variation and computationalflow of the fluid. Aiming at the characteristics of random noise in seismic records, studydeeply the independent component analysis’s applicability both in seismic data de-noise andin seismic attribute extraction. Take advantage of the characteristics of algorithms, research onseismic attribute extraction and optimization. Because of the independence of seismic validattribute which separated by ICA, it shows an important foundation for multiple attributescomprehensive analysis, we can improve the accuracy and reliability of seismic attributeinterpretation. It shows that the independent component analysis method has broadapplication prospects in seismic signal processing.