基于粒子群算法的盲源分离算法  

Blind Source Separation Based on Particle Swarm Optimization

作　　者：张文爱[1] 刘俊豪[1] 

机构地区：[1]太原理工大学信息工程学院,山西太原030024

出　　处：《太原理工大学学报》2006年第2期169-172,共4页Journal of Taiyuan University of Technology

摘　　要：简要地介绍了粒子群算法(PSO)及其改进算法和盲源分离算法(BSS),改进的粒子群算法具有并行性、易实现等优点。将改进的粒子群算法与盲源分离算法相结合,提出了基于粒子群算法的盲源分离算法。该算法以混合信号的峰度为目标函数,采用独立分量分析的方法,用改进的粒子群算法代替常规的最陡梯度下降法,对瞬时混合的信号进行盲分离,解决了梯度算法收敛速度慢的问题。实验仿真表明:该算法具有收敛速度快、分离效果好等特点。In this paper, blind source separation, particle swarm optimization and modified particle swarm optimization are introduced and a new algorithm of blind source separation based on particle swarm optimization is proposed. Using kurtosis of the mixed signal to the target function of blind source separation, replacing steepest gradient descent method by modified particle swarm optimization, it succeeds to separate the instantaneous mixed signal by the method of independent component analysis. It solves the problem of low searching speed in steepest gradient descent method. The computer simulation results showed its character of fast convergence speed and stable performance, etc.

关 键 词：盲源分离 粒子群算法 群集智能 

分 类 号：TN911.7[电子电信—通信与信息系统]