基于粒子群算法的盲源信号分离

Blind Source Separation Based on Particle Swarm Optimization

当源信号个数大于2,联合对角化(JADE)算法在盲源信号分离时效果不理想.提出了一种基于粒子群算法(PSO)的盲源信号分离(BSS)算法.该算法利用PSO算法代替JADE算法中的联合对角化操作,以混合信号的峭度为目标函数,采用独立分量分析的方法,对瞬时混合的信号进行了盲分离,理论分析和仿真结果表明了该算法的可行性和有效性.

When the number of signal is more than 2,the results of BSS（blind source separation） by JADE（joint approximate decomposition of eigen matrices） are not desirable.In this paper,a new algorithm of BSS based on PSO（particle swarm optimization） is proposed.The operation of joint diagonalization is instead of PSO.Using kurtosis of the mixed signal to the target function of BSS,it succeeds to separate the instantaneous mixed signal by the method of independent component analysis.The analysis and simulations suggest that the scheme is feasible and effective.