基于种群分类解决遗传算法的“早熟”与“漂移”问题
Solution of Premature Convergence and Genetic Drift Based on Population Cluster
作者:李世伦(四川大学 数学学院,四川 成都610064);罗懋康(四川大学 数学学院,四川 成都610064);何小勇(四川大学 数学学院,四川 成都610064)
Author:(School of Mathematics,Sichuan Univ.,Chengdu 610064,China);(School of Mathematics,Sichuan Univ.,Chengdu 610064,China);(School of Mathematics,Sichuan Univ.,Chengdu 610064,China)
收稿日期:2006-04-20 年卷(期)页码:2006,38(6):127-130
期刊名称:工程科学与技术
Journal Name:Advanced Engineering Sciences
关键字:遗传算法; 早熟收敛; 遗传漂移; 模糊聚类
Key words:genetic algorithms; premature convergence; genetic drift ; fuzzy cluster
基金项目:国家自然科学基金资助项目(10331010); 博士点基金资助项目(20030610020)
中文摘要
为了有效解决遗传算法在实际应用过程中经常面临的早熟收敛和遗传漂移问题,分析了导致早熟收敛和遗传漂移这两种现象出现的原因,针对其主要原因提出了基于模糊聚类的种群分类改进的遗传算法,避免近亲繁殖导致早熟,并将模糊聚类的结果与各种遗传操作有效结合,提高了算法向最优解收敛的准确性和稳定性。最后,仿真结果显示新的改进算法比标准遗传算法更有效。
英文摘要
For solving effectively premature convergence and genetic drift that often occur in applying Gas to practice, the reason for the premature and genetic drift was analyzed. A new aproach based on fuzzy cluster was proposed to effectively overcome the two phenomenon. The covergence of the new algorithm was discussed. Experimental results showed that the new algorithm is more effective than classical genetic algorithms.
【关闭】