Blind Source Separation Based On Improved Particle Swarm Optimization Algorithm

As a new research field developed in late 1980 s,blind source separation(BSS)has always been favored by many scholars.BSS refers to the estimation and separation the source signals only through the observed signals without knowledge of the transmission system and the source signals.After decades of development,the scope of BSS applications has become more and more extensive and the application scenarios of blind source separation can be seen in different fields,such as the separation and recognition of speech signals,image signal processing,mechanical signal processing,and wireless In communications and other fields.Independent component analysis is one of the most mature technologies in blind source separation.The Independent Component Analysis(ICA)algorithm is a signal processing technology emerging from the 1990 s.It calculates the high-order statistical characteristics of signal data,ICA is the process of separating individual source signals from multiple mixed signals without other prior knowledge.ICA is developed along with BSS subject and is an important method in BSS.There are problems for current ICA such as poor robustness and low convergence precision,hence in this dissertation,the Particle Swarm Optimization(PSO)algorithm is applied to ICA algorithm.The improved PSO algorithm improves the effectiveness and accuracy of the algorithm.The main work includes the following aspects:1.The historical origin of the BSS and the current research status of the BSS at home and abroad are reviewed.The basic principles of the BSS are analyzed in detail and several algorithms commonly used in the blind source separation are summarized.The basical conditions for blind source separation and several commonly used methods for performance comparison of separated signals are given.2.The basic principle and characteristics of particle swarm optimization(PSO)algorithm are introduced in detail.Considering the problems of traditional BSS methods,such as poor robustness and low convergence accuracy,the combination with PSO is proposed,which can traverse to find the optimal values.A preliminary simulation experiment was conducted.3.However,The blind source separation of traditional particle swarm optimization algorithm has the disadvantage that it is easy to fall into the local optimality.For this reason,the independent component analysis algorithm combing dreaming particle swarm optimization(DPSO)algorithm is proposed.Simulation results proves that the ICA combining DPSO further improves the convergence accuracy and enhances the effectiveness and stability.4.The improved particle swarm optimization algorithm is applied to the blind source separation of speech signals and the blind source separation of multiple grayscale image aliasing.In the experiment,the improved algorithm was compared with the traditional ICA algorithm and the Fast ICA algorithm.The experimental results show that the dream particle swarm algorithm not only separates the mixed signals well,but also has a better convergence effect.