摘要
在研究和声搜索对多维函数优化问题的基础上,结合传统的模拟退火算法,提出一种混合优化算法——和声退火算法。该算法改进了和声的搜索机制,选取合理的取值概率HMCR以及动态的微调概率PAR,在和声记忆库内随机搜索,获得较高质量的新和声;然后对新和声执行一次Metropolis算法,从而增强了全局探索能力,减小了陷入局部极小值的机会。仿真实验数据表明,算法明显优于和声搜索和模拟退火算法,具有较高的求解质量和效率。
This paper put forward a kind of hybrid optimization algorithm : harmony annealing algorithm, which was based on harmony search and simulated annealing algorithm for multidimensional function optimization problem. Improved the algorithm the search mechanism of HS, selected the reasonable values about harmony memory considering rate (HMCR) and dynamic pitch adjusting rate (PAR). Searched the new vector randomly in the HM, then executed a Metropolis algorithm. It could improve the efficiency of the exploration and reduce the probability of trapped by local minimum value. The simulation results demonstrate the effectiveness and good quality of the proposed procedure, and better than those of harmony search and simulated annealing algorithm.
出处
《计算机应用研究》
CSCD
北大核心
2010年第3期853-855,859,共4页
Application Research of Computers
基金
国家自然科学基金资助项目(60874075)
中国博士后科学基金资助项目(20070410791)
华中科技大学数字制造装备与技术国家重点实验室开放课题
关键词
和声搜索
模拟退火
和声退火
函数优化
harmony search(HS)
simulated annealing(SA)
global harmony annealing
function optimization