摘要
为了提高协同进化多目标进化算法的全局收敛性,提出了一种调用协同进化算子的自适应方法。其基本思想是:根据目标函数的变化率自动调用协同进化算子;当种群进化正常时,调用合作算子和吞并算子;当种群进化接近停滞时,调用分裂算子。通过数值实验用量化指标研究了新算法的收敛性和分布性,结果表明,与常规协同进化多目标进化算法相比,新算法不仅具有良好的分布性,而且全局收敛性有了明显的提高。
In order to improve the global convergence of co-evolutionary multi-objective evolutionary algorithm, an adaptive method to call co-evolution operator is proposed. The basic idea of method is that co-evolution operators are dynamically called according to the change rate of objective function. When the evolution is normal, cooperation operator and merging operator are called, and otherwise division operator is called. The convergence and distribution of improved algorithm are studied by means of numerical experiments, and results show that the new algorithm not only has good distribution, but also global convergence has been significantly improved compared with the conventional co-evolutionary multi-objective evolutionary algorithm.
出处
《计算机工程与应用》
CSCD
北大核心
2016年第6期26-30,73,共6页
Computer Engineering and Applications
基金
安徽省教育厅自然科学基金项目(No.2012kb236)
关键词
多目标进化算法
协同进化
自适应
收敛性
分布性
multi-objective evolutionary algorithm
co-evolution
adaptive
convergence
distribution
