摘要
通过分析Castro提出的CLONALG算法在优化多峰值函数时存在峰值搜索能力弱、最优解易退化、收敛效率低等问题的根源,提出了一种基于记忆库小生境自适应克隆选择算法(MNACSA)。该算法首先采用小生境机制将种群分成若干类、分别从每个类中选出最优个体组成新种群;其次建立记忆库和自适应的高频变异率、且在库中引入最佳抗体抑制操作。对算法进行了分析和仿真实验,证明了该算法可以防止优秀个体退化、自动调节种群个体数目、提高优化效率、增强多峰搜索能力。
Through the analysis of the source of the problems that when CLONALG algorithm which Castro raised optimizes the multi-peak function there is weak peak search capability, the easily-degraded optimal solution and the low efficiency of the convergence, this paper provides a self-adaptive clone selection algorithm based on memory niche （MNACSA）. First of all, the algorithm uses mechanism for niche to divide the population into several categories, and the best individuals are selected from each category to form a new population; then the memory and the self-adaptive high frequency of mutation rate should be established and the best antibody inhibition operation should be introduced in the memory. In the paper, algorithm is analyzed and simulated, which proves that the algorithm can prevent the degradation of the outstanding individuals, adjust the number of individuals automatically, improve optimized efficiency and enhance the capacity of multi-peak search .
出处
《贵州大学学报（自然科学版）》
2009年第2期90-93,共4页
Journal of Guizhou University:Natural Sciences
基金
识别天然地震与人工爆炸的分类决策支持系统项目(200808003)
广西研究生创新计划项目(2007106020812M73)
关键词
多峰值函数优化
记忆库
小生境
克隆选择算法
抗体抑制
multi-peak function optimization
memory
niche
clone selection algorithm
antibody inhibition
