基于粗糙集理论的改进ε-支配多目标进化算法  

Improved ε-dominance multi-objective evolutionary algorithm based on rough set theory

作　　者：过晓芳[1,2] 

机构地区：[1]西安工业大学数理系,西安710032 [2]西安电子科技大学计算机学院,西安710071

出　　处：《计算机应用研究》2011年第3期948-950,956,共4页Application Research of Computers

摘　　要：基于epsilon支配概念的epsilon-MOEA(-εMOEA)算法具有良好的收敛性和分布性,但是存在epsilon值不易设置,解集中边界个体容易丢失等缺陷。通过结合粗糙集理论中边界域的概念,提出了基于粗糙集理论的改进epsilon-MOEA算法,从而改善解集中部分个体丢失等现象。实验结果表明,新算法相比传统epsilon-MOEA算法在解集分布性和收敛性上具有较好的改善。The epsilon-MOEA,which based on the epsilon dominance,had the good performance on convergence and diversity,however,it also had some disadvantages.Such as,the value of epsilon was difficult to set,many extreme or representative individuals was easy to lose.In order to solve this problem,introduced boundary region in rough set theory to improve the spread and quality of the initial approximation of the Pareto front in epsilon-MOEA,then proposed a improved epsilon-MOEA based on rough set theory.The experimental results illustrate epsilon-MOEA/RST has the good performance,which is much better at convergence and diversity than epsilon-MOEA.

关 键 词：多目标优化 epsilon支配 粗糙集理论 边界域 

分 类 号：TP301.6[自动化与计算机技术—计算机系统结构]