一种求解多峰函数优化问题的演化算法

New Evolutionary Algorithm for Solving Multimodal Function Optimization Problems

针对演化计算产生新解无序的问题,提出了基于相似性的邻域搜索策略.利用邻域搜索,可以方便地建立自适应的新解产生机制.针对演化算法设计中存在的搜索效果和效率平衡问题,提出了利用适应值对个体进行分级的搜索策略.通过对个体的分级,可以区分个体在搜索过程中的职能:优秀的个体进行局部极小值的开采;其他的个体进行搜索空间的探索,以发现新的局部极小值.数值实验表明,新算法能有效处理低维多峰函数,能找到所有的全局最优解.对高维多峰函数,也能找到全局最优解.

A similarity based neighborhood exploring strategy is proposed to deal with the mechanism of generating new individuals randomly in evolutionary computation. The new algorithm is called Similarity Based Evolutionary Algorithm （SBEA）. Neighborhood exploring brings the ability to self adaptively generate new individuals easily. Aiming at balancing search results and search speed, we adopt the search strategy to classify the individuals by their fitness. Individuals＇ classification differentiate respective func tion in search process, that is the excellent individuals mine the local optimal solution and others explore the search domain to find new local optimal solution. The experiment results indicate the new algorithm is very efficient for the optimization of low-dimension multi-modal functions. We can obtain all the global opritual solutions easily and quickly. As to high-dimension multi-modal functions, the results are also satisfactory.