Blind Source Separation Algorithm Baced On Apso Algorithm Of Research And Application

Blind source separation (BSS) algorithm is an important branch in the field of modern signal processing,which can only separate source signals from observed signals under the condition of hybrid system and unknown source signals. In recent years the independent component analysis(ICA) technology as an important part of BSS technology gets more attention,which has become a research direction of the signal processing and has been applied widely in speech signal processing, image processing, biomedicine, communication system,economic data analysis, etc. and this technology has important research significance and application value.ICA is a kind of methods which is used by the data calculation through the characteristics as its judge standard, and it separates unknown factors mixed independent source signals from observed signal. The iterative particle swarm algorithm was used to find the optimal separation matrix to separate the independent source signals. This paper improves particle swarm optimization,and puts forward a new BSS algorithm baced on APSO, and applies this arithmetic to mixed voice signal and fuzzy gray image’s blind source separation. Paper’s main contents include:1. The research analyzes the principle of BSS technology and ICA, mainly discuss the assumptions of BSS and solvability, related information theory knowledge, statistics and several kinds of objective function, etc.2. The research analyzes the several common BSS algorithm, including the gradient descent algorithm, fixed point algorithm and neural network algorithm, etc, and carries on the comparative analysis.3. The research puts forward a kind of BSS algorithm based on improved adaptive particle swarm optimization (APSO), using two fitness function to particle evolution direction which are kurtosis and negative entropy respectively as the first and second fitness function of particle swarm algorithm according to the principle of their common Gaussian discriminant standard as independence to adaptive update of separation matrix, The simulation verifies its validity.4. The improvement of BSS algorithm based on APSO is applied to mixed voice signal blind source separation, and multiple fuzzy gray-level images blind source separation of aliasing. The results of simulation show that the improved APSO algorithm can well separate mixed observation signal, the calculation of performance index is superior, the convergence effect is good compared with the FastICA algorithm, the traditional APSO algorithm and the proposed algorithm.This paper detailed describes the blind source separation step, the selection of the objective function, the advantages and disadvantages of all kinds of commonly used algorithms, the proposed method based on improved adaptive particle swarm algorithm of blind source separation algorithm is a detailed description and computer simulation.