基于立队竞争的演化算法

基于立队竞争的演化算法 Evolutionary Algorithms Based on Sort-Competition 刘良兵 LIU Liang-bing1 1 2 3 4 first-author 吴方才 WU Fang-cai2 1 2 3 4 黄樟灿 HUANG Zhang-can3 1 2 3 4 华中师范大学数学系 华中师范大学数学系 Department of Mathematics,Normal University of China Center Department of Mathematics,Normal University of China Center 武汉理工大学自动化学院 武汉理工大学自动化学院 School of Automation, Wuhan University of Technology School of Automation, Wuhan University of Technology 武汉大学软件工程国家重点实验室,湖北,武汉,430070 武汉大学软件工程国家重点实验室,湖北,武汉,430070 State Key Laboratory of Software Engineering,Wuhan University State Key Laboratory of Software Engineering,Wuhan University ,湖北,武汉,430072 ,湖北,武汉,430072 将竞争机制引入到变异算子中 ,从而使变异算子具有了很强的群体特性 ,同时为使算子更加简练和更具通用性 ,在这种竞争中作者更加注重个体适应值的相对次序而不是它们的绝对大小 .基于这种思想提出了一种新的演化算法 ,在该算法中变异的作用已不仅仅是简单地恢复群体中损失的多样性 ,此时它已成为一种主要的搜索算子 .通过实验发现新算法具有很高的搜索效率和求解精度 . In this paper, competition mechanism is drawn into mutation operator through which the mutation operator shows its high population property and the mutation operator itself becomes less complicated and general in use. In the competition proposed, more importance is attached to the relative order of the fitness than to their accurate values. Based on the above, a new evolutionary algorithm is presented in which the act of mutation is not the simple resumption of the lost population variety. Moreover, it becomes a main searching operator. The conclusion drawn through experiments is that the searching efficiency and solving accuracy of the new algorithm are quite high. 竞争 自适应性变异 自聚集 competition adaptive mutation self-congregate O229 国家自然科学基金资助项目 （ 60 2 0 40 0 1） 2003-03-01 2021-04-01 3