摘要
针对目前多模态优化存在无法找到全部局部极值解的问题,提出了一种基于拥挤模型的差分进化算法,利用差分进化算法的全局搜索策略和内在的并行方式,通过拥挤模型的高群集因子(crowding factor,CF)搜索,避免了取代错误,保持了物种的多样性,可准确定位多模态函数的最优解和全部极值解.同时,该算法具有参数少、操作算子简单、收敛速度快等特点.实验结果表明,提出的拥挤差分进化算法处理多模态优化问题时在收敛速度、收敛精度上皆明显优于拥挤遗传算法.
This paper presents a crowding differential evolution(CDE) algorithm applied to multimodal function optimization for finding all the extreme solutions,using DE's(differential evolution,DE) global search strategy and internal parallel pattern.The high crowding factor(CF) value search avoids the replacement error,maintains diversity of species,and can accurately locate all the multimodal function's optimal solutions and extreme solutions.Meanwhile,this algorithm has a lot of advantages such as less parameters,simple operator and swift convergence rate.The algorithm is compared with crowding genetic algorithm and simulation experiment results show that crowding differential evolution is better than crowding genetic algorithm(CGA) in both convergence rate and convergence accuracy.
出处
《哈尔滨工程大学学报》
EI
CAS
CSCD
北大核心
2011年第2期223-227,共5页
Journal of Harbin Engineering University
关键词
多模态函数优化
拥挤模型
差分进化算法
群集因子
multimodal function optimization
crowding model
differential evolution
crowding factor