摘要
为了改进原始和声搜索算法的全局搜索性能,提出了基于迭代局部搜索的和声搜索算法。该算法在充分利用和声记忆库中信息和提高搜索效率的同时,对于不满足停止准则的新和声采用基于改进kick策略移动的迭代局部搜索算法进行寻优,从而使新算法具有较强的"爬山"能力。针对4个benchmark函数对新算法做了测试,并与粒子群优化算法以及已有的几个算法进行了比较,结果表明该算法跳出局部极值点的能力较强、收敛速度更快、寻优精度较高;最后将新算法应用到焊接梁的优化设计问题中,仿真结果验证了该算法在求解焊接梁最小造价问题时优于原始的和声搜索算法、遗传算法等方法。
To improve the performance of the original harmony search（HS） algorithm,an HS algorithm based on iterated local search is developed.The information of harmony memory is fully used and the efficiency of searching is improved.If the new harmony is not satisfied the stopping condition,the new harmony will be moved with iterated local search based on the random kick strategy.Hence,it has a stronger ability of ＂hill climbing＂.Experiments on four benchmark functions show that the proposed algorithm outperforms particle swarm optimization and other algorithms in converging speed,precision and global searching ability.The proposed algorithm is applied to the optimization design of welded beam,and the comparison results show that the proposed algorithm is better than other algorithms.
出处
《控制工程》
CSCD
北大核心
2010年第5期665-668,共4页
Control Engineering of China
基金
国家自然科学基金资助项目(60774097)
关键词
和声搜索算法
粒子群优化算法
迭代局部搜索
函数优化
harmony search algorithm
particle swarm optimization
iterated local search
function optimization
