基于多策略红尾鹰算法的盲源分离研究  

Blind Source Separation Based on Multi-strategy Red-tailed Hawk Algorithm

作　　者：李著成[1] LI Zhucheng(Business College,Beijing Union University,Beijing 100025,China)

机构地区：[1]北京联合大学商务学院,北京100025

出　　处：《探测与控制学报》2024年第6期51-57,共7页Journal of Detection & Control

摘　　要：传统盲源分离优化方法容易导致算法陷入局部最优解,对参数的设定也较为敏感,而且往往缺乏有效的自适应调整机制,难以应对问题的动态特性,针对这些问题,提出一种基于多策略红尾鹰算法(MSRTH)的盲源分离方法。该方法将拉丁超立方抽样法应用在原始红尾鹰(RTH)算法红尾鹰个体位置的初始化上,提升种群的多样性和算法的全局搜索能力以及收敛速度,同时引进高斯变异去打破当前解的局部结构,使得算法有机会跳出当前的局部最优解。仿真实验结果表明,相比原RTH算法、遗传算法(GA),该算法具有更高的分离精度和更快的收敛速度。The traditional optimization method for blind source separation is prone to getting trapped in local optimal solutions,sensitive to parameter settings,and often lacks an effective adaptive adjustment mechanism for the dynamic characteristics of the problem.To address these issues,a blind source separation method based on multi-strategy red-tailed hawk algorithm(MSRTH)was proposed,which applied the Latin hypercube sampling to initialize the positions of red-tailed hawks in the original RTH algorithm.This approach enhanced the diversity of the population and the global search capability of the algorithm,as well as improving its convergence speed.Furthermore,Gaussian mutation was introduced to provide the opportunity for the proposed algorithm to escape from the current local optimum solution.Simulation results showed that in comparation with the original RTH algorithm and genetic algorithm,this method had higher separation accuracy and faster convergence speed.

关 键 词：红尾鹰算法 盲源分离 优化算法 拉丁超立方抽样 高斯变异 

分 类 号：TN911.7[电子电信—通信与信息系统] TP18[电子电信—信息与通信工程]