基于改进粒子群的独立分量分析算法研究

摘要

In order to solve the problems such as easy falling into local optimum particle and slow convergence speed in traditional particle swarm optimization(PSO)algorithm,an independent component analysis(ICA) algorithm based on the improved PSO algorithm was proposed.The method chose the value of the inertia weight factor ωrandomly in the section to make the particle have adaptive ability.Because of this,the improved PSO algorithm could search the optical particle quickly.Meanwhile,it used the mutual information in ICA as the objective function,and the improved PSO algorithm to optimize the objective function,which made the compo-nents to be independent among each other.Simulation results showed the proposed method inproved the global search ability,could separate the mixed signal effectively and improved the result of the blind source separation.%针对传统粒子群优化（PSO）算法对目标函数进行优化时，粒子容易陷入局部最优及收敛速度慢的缺陷，提出了一种基于改进 PSO 算法的独立分量分析（ICA）算法。该算法通过随机分段选择调节 PSO 算法中的惯性因子ω，使粒子具有一定的自适应能力，以快速找到最优粒子；然后，将 ICA 中的互信息作为目标函数，通过改进的 PSO 算法优化 ICA 中的目标函数，使独立分量中的各个成分相互统计独立。仿真实验结果表明，本算法可明显提高全局搜索能力，有效地实现混合信号的分离，改善盲源信号的分离效果。