基于简化粒子群优化的卷积混叠盲源分离算法  

A Blind Source Separation Algorithm for Convolved Mixed Signals Based on Simple Particle Swarm Optimization

作　　者：王娜娜 付强 曹新贞 

机构地区：[1]63626部队 [2]63607部队

出　　处：《遥测遥控》2016年第5期28-34,共7页Journal of Telemetry,Tracking and Command

摘　　要：针对现有的卷积混叠盲源分离算法存在的算法复杂和分离精度低等问题,提出一种新的基于简化粒子群优化的卷积混叠盲源分离算法。算法将不同时间点分离信号的互累积量消失作为分离准则,将信号的四阶互累积量作为目标函数,采用简化粒子群优化算法代替基本粒子群算法对目标函数进行全局优化,解决了基本粒子群算法容易陷入局部极值的问题。Matlab仿真结果表明,新算法可以有效实现对卷积混叠信号的盲源分离。对比基本粒子群算法,新算法分离精度更高、收敛速度更快。A new blind source separation method based on simple particle swarm optimization is proposed to solve the problems that the current blind source separation method for convolved mixed signals is complex and has low separation accuracy. Firstly, fourth-order cross cumulant is used as objective function based on separation criterion of separated signal＇s cumulant disappearing at different time points. Secondly, the simplified particle swarm optimization algorithm is used to replace the basic particle swarm optimization algorithm for global optimization of the objective function, which solve the problem that the basic particle swarm optimization algorithm is easy to fall into local extremum. Matlab simulation results show that the new method can achieve the blind source separation for convolved mixed signals effectively. Compared with the basic particle swarm optimization method, the new algorithm has higher convergence speed and separation accuracy.

关 键 词：简化粒子群优化 卷积混叠 四阶互累积量 盲源分离 

分 类 号：TN911.7[电子电信—通信与信息系统]