多维多极值函数优化的和声退火算法

Harmony Annealing Algorithm for Multi-dimensional Function Optimization

针对多极值实函数优化问题,本文结合和声搜索与模拟退火算法,提出了一种新的搜索算法,即和声退火算法。新算法保留了和声搜索的搜索机理,但对和声搜索中于和声记忆库外的搜索方法用超快速模拟退火算法作了改进,对和声记忆库内新解产生方法也作了相应的调整,从而提高了对多维问题的搜索效率。数值实验结果表明算法对和声搜索有明显的改进,收敛速度更快,跳出局部极值点的能力较强。新算法在解决多维多极值优化问题方面比遗传算法更具效率,值得进一步研究与推广应用。

Based on the combination of harmony search and simulated annealing, a novel optimizing method, harmony annealing algorithm (HAA), for solving multi-dimensional real optimization problems is presented in this paper. With the main search mechanism of harmony search , the solution generation outside harmony memory is improved by using very fast simulated annealing and the solution combination inside harmony memory is modified. So that the search efficiency for solving multi-dimensional optimizing problems is improved. Experimental numerical results show that the search efficiency of the new method is much higher than that of harmony search, especially for solving multi-dimensional and multimodal optimization problems. Further exploration and application are proposed.