Research On Optimization Of Blind Signal Separation Technique

Blind source separation (BSS) has been a hot issue in the signal field in recent years. It is a means aimed at retrieving and extracting unknown sources from the observed signals which are independent with each other. Independent component analysis (ICA) is one of the most effective methods of solving blind source separation problem at present.Independent component analysis is a new signal separating technique with BSS developing. It is a newly developed and powerful technique for recovering latent independent sources extracted from their mixtures. The data analyzed by ICA could originate from many different kinds of application fields, including voice signal and digital images, as well as document databases and so on.This article focuses on Independent Component Analysis Algorithm. We discuss some algorithms based on non-Gaussian and propose new ideas to optimize. The main works in this thesis can be introduced as follows:(1)Introduce the theoretic base of BSS and introduce in detail the status of independent component analysis in the aspects of algorithms and applications.(2)Give the definition of simple mathematical preliminaries in ICA technique. We try to discuss main problems of the research of ICA including the mathematical definition of ICA, the assumptions made about ICA problems and the mathematical theory and methods commonly used in ICA. This article mainly studies ICA based on the non-Gaussian maximum principle. The conjugate gradient method to analysis algorithm was used to optimize the large computation, because the independent variable analysis algorithm in the source signal recovery process exists large computation. Finally, computer simulations illustrate that the optimized algorithm can well separate mixed voice signal.(3)For the natural gradient based on negative entropy slow convergence, we use the principle of simulated annealing to optimize it. Finally, computer simulations illustrate that the optimized algorithm can well separate mixed voice signal.(4)Quantum genetic algorithm (QGA) is a kind of the emerging global optimization algorithms which combine quantum computation theory and genetic algorithm. This article puts forward a blind source separation method based on QGA and ICA. The simulation result shows the effectiveness of the proposed methods.