摘要
为了克服基本蝙蝠算法后期收敛速度慢、易陷入局部最优的缺点,在原始算法中引入小生境技术并进行改进。在改进算法中,将小生境半径设置为自适应变化的动态函数;在单个小生境群体中采用信息共享机制,对相似蝙蝠数量的过度增长进行抑制;采用优质蝙蝠邻域搜索及存储策略对每一代每个小生境群体的优质蝙蝠进行储存。对某21节点系统进行了无功优化,并与遗传算法、基本蝙蝠算法进行比较,结果表明改进的算法具有更好的全局搜索能力和收敛性能。
In order to overcome the shortcomings of bat algorithm(BA)such as low convergence speed in the later peri-od of optimization and vulnerability to falling into local optimums,niche technique is introduced to improve the originalBA. The improvements are as follows:niche radius is set as a dynamic function,which varies adaptively;informa-tion sharing mechanism is adopted in a single niche community to restrict the excessive growth of similar bats;the strat-egies of neighborhood search and storage with valued bats are used to preserve the potential optimal bats in each genera-tion of each community. The reactive power optimization of a 21-node system is carried out,whose result is comparedwith those obtained by using genetic algorithm(GA)and BA,indicating that the improved algorithm has better globalsearch and convergence performance.
出处
《电力系统及其自动化学报》
CSCD
北大核心
2017年第10期35-39,51,共6页
Proceedings of the CSU-EPSA
基金
国家自然科学基金资助项目(61102039)
湖南省自然科学基金资助项目(14JJ7029)
中央高校基金资助项目
湖南省教改课题资助项目
关键词
无功优化
小生境
蝙蝠算法
小生境半径动态划分
信息共享机制
reactive power optimization
niche
bat algorithm(BA)
dynamic partitioning of niche radius
information sharing mechanism
