适用于多峰函数优化的改进顺序生境遗传算法  

Improved sequential niche genetic algorithm for multimodal optimization

作　　者：于歆杰[1] 王赞基[1] 

机构地区：[1]清华大学电机工程与应用电子技术系,北京100084

出　　处：《清华大学学报（自然科学版）》2001年第3期17-20,共4页Journal of Tsinghua University(Science and Technology)

基　　金：清华大学博士论文基金项目

摘　　要：针对标准顺序生境 (SN )算法在求解复杂优化问题时遇到的困难 ,提出了两种改进策略。一种策略是新的保存最优解策略 ,即在遗传算法的选择和杂交两个阶段之间 ,用已经找到的某些峰替代当前种群中最差的部分个体 ;另一种策略是对那些与已找到的峰相近或相似的个体采用较高概率进行变异 ,而不是将其适应值降低。将标准 SN算法与以上两种策略相结合 ,提出了 4种新 SN算法 ,还采用一个复杂欺骗性问题作为验证算法有效性的手段 ,结果说明 ,这些算法均能有效地找到该问题所有Two improved strategies are presented for the standard sequential niche (SN) algorithm which faced difficulties when it is used to solve the sophisticated optimization problems. The first method is the new elitist method, which replaces the worst individuals with the recorded peaks between the selection and the crossover of the genetic algorithm. The second method mutates the individual near the recorded peaks with relatively high probability, without decreasing their fitness. Four new SNs are presented by combining the standard SN with these two strategies. The massively deceptive problem is used to verify the strategies and algorithms. The results show that the methods effectively locate the total 32 global peaks.

关 键 词：多峰函数优化 顺序生境技术 保存最优解策略 概率 遗传算法 适应值函数 峰半径 距离度量方式 

分 类 号：O242.23[理学—计算数学]