摘要
针对和声搜索算法在处理复杂函数优化问题时容易陷入局部最优、收敛精度低的缺点,提出了一种改进的和声搜索算法,不同于已有的HS算法.整个和声记忆库被划分为一些小的子和声记忆库,每个子库适时地更新内部信息,然后将各子库中的最优解构成一个较优记忆库并进行搜索,这些子记忆库通过重组周期被反复重组,信息在这些子库中被交换,在算法的最后搜索阶段,为了表现一个更好的局部搜索能力,所有和声形成一个和声记忆库.同目前提出的一些HS算法相比,新算法有更好的优化性能.
The Harmony Search（HS） algorithm traps into local optima easily,and it has low convergence accuracy when used for the optimization of complex functions.In order to overcome these shortcomings,an improved HS algorithm is proposed.Different from the existing HS algorithm,the whole harmony memory is divided into some small sub-harmony memories,and each sub-harmony memory updates information timely.Then the optimal solution of sub-harmony memory constitutes a better harmony memory,which is used for search.The sub-harmony memories are regrouped frequently by using the regrouping strategy,and information is exchanged among the sub-harmony memories.In the final search stage,to have a better local search ability,all the harmonies form a harmony memory.Experiments are conducted on five benchmark functions,and the results show that the proposed HS algorithm has a better optimization performance,as compared with some recent HS variants.
出处
《甘肃科学学报》
2010年第3期40-43,共4页
Journal of Gansu Sciences
基金
国家自然科学基金项目(60674108)
商洛学院科研基金项目(09sky011)
关键词
和声搜索
元启发式算法
智能优化
复杂函数
重组周期
harmony search
meta-heuristic algorithm
intelligent optimization
complex function
regrouping period
