采用位置混沌重构的入侵杂草优化在盲源分离的应用

传统盲源分离（blind source separation，BSS）优化算法的应用场合非常有限，而且分离性能不高，为此提出了一种新的采用位置混沌重构的入侵杂草优化算法（invasive weed optimization，IWO），并对其在盲源分离的应用进行了研究。新算法在每轮更新的初期驱动选出的较优个体向此时种群的最优个体做适当距离的移动，这样不仅会增加种群的多样性，避免算法出现早熟，而且收敛速度也较快。盲信号分离仿真实验证实，与标准IWO、粒子群优化算法（particle swarm optimization，PSO）和自然梯度算法（natural gradient，NG）相比，新算法的性能优势明显，收敛速度较快，分离精度较高。

Application of invasive weed optimization based on location chaos reconstruction to blind source separation

Aiming at the limitation of traditional optimization algorithm for blind source separation (BSS), this paper proposed a new invasive weed optimization (IWO) based on location chaos reconstruction, and explored its application to BSS. The new algorithm drove the superior individuals towards the optimal individual of the population at the early stage of each update. It could increase the population diversity, avoid the premature phenomenon and accelerate the convergence speed. Simulation experiments for BSS show that the proposed algorithm is more effective than the basic IWO, particle swarm optimization (PSO) and natural gradient (NG), and has faster convergence speed and higher separation accuracy.