基于排挤机制改进的多目标进化算法  

An Improved Multi-objective Evolutionary Algorithm Based on Crowing Mechanism

作　　者：侯中喜[1] 陈小庆[1] 郭良民[1] 

机构地区：[1]国防科技大学航天与材料工程学院,湖南长沙410073

出　　处：《国防科技大学学报》2006年第4期18-21,共4页Journal of National University of Defense Technology

基　　金：国家863基金资助项目(2005AA756050)

摘　　要：进化算法是求解多目标优化问题(MOP)重要而有效的方法。为加快收敛速度,提高收敛精度,在已有算法(NSGA-Ⅱ)的基础上,引进小生境思想,提出了更为合理的排挤机制。通过典型应用函数的计算测试,结果表明:上述改进不仅具有较高的计算效率,而且能够得到分布更为合理的解,且能保持解的多样性分布。Evolutionary algorithms are the main and effective methods in solving multi-objective optimization problems （MOP）.Based on the NSGA-Ⅱ algorithm studied and analyzed, we improved its crowding mechanism by introducing the Niche theory to expedite its convergence velocity and improve its convergence precision. The representative test functions show that the improvements have higher computational efficiency and can obtain a reasonable distributing solution; it can also maintain the solutions＇ diversity.

关 键 词：多目标优化 进化算法 排挤机制 

分 类 号：TP301[自动化与计算机技术—计算机系统结构]