摘要
在工程实践的多目标优化问题中,当已知目标空间中的一个或几个最优解时,往往需要在某一特定的区域搜索到比较稠密的Pareto解集合.本文提出了基于精英种子策略的多目标遗传算法,把已知的最优解信息加到优化过程中,利用最近邻方法来识别进化个体的所属的Pareto支配性类别,引导优化方向.仿真实验结果表明,该算法在特定区域内表现了比NSGA-Ⅱ更优的局部搜索能力,并且能够搜索到零散分布的Pareto最优解子集.
When one or more optimal solutions are known in multi-objective optimization in engineering practice, it is often required to obtain some highly dense Pareto optimal solutions in some specific regions. This paper proposes a multi-objective genetic algorithm based on the elite seed strategy. The strategy incorporates known elitist solutions into optimization process and utilizes the nearest neighbor method to recognize the Pareto dominance between solutions, consequently guiding the optimization direction. Experimental results show the algorithm is more effective than NSGA-II in local areas, and it is also capable of finding out the dispersively distributed Pareto optimal subsets.
出处
《湖南理工学院学报（自然科学版）》
CAS
2011年第4期29-32,共4页
Journal of Hunan Institute of Science and Technology(Natural Sciences)
基金
国家自然科学基金项目(60975049
30971540)
湖南省自然科学基金重点项目(11JJ2037)
湖南省大学生研究性学习和创新性实验项目
关键词
多目标优化
Pareto支配性
最近邻方法
分类
精英种子
multi-objective optimization
Pareto dominance
nearest neighbor method
classification
elite seed
