一种新的求解多目标优化问题的遗传算法  

A New Genetic Algorithm in Multi-objective Optimization Problem Solving

作　　者：梅羚[1,2] 李元香[1,2] 郑波尽[1,2] 

机构地区：[1]武汉大学计算机学院,武汉430072 [2]武汉大学软件工程国家重点实验室,武汉430072

出　　处：《计算机工程与应用》2005年第30期40-42,共3页Computer Engineering and Applications

基　　金：国家自然科学基金(编号:60473014);国家博士学科点科研基金(编号:20030486049)

摘　　要：提出一种新的求解多目标优化问题的算法-GGGA。该算法运用几何斜率Pareto选择的精英策略,多个子种群从求解目标的不同方向进行区域演化,并借鉴了郭涛算法的多父体杂交算子。数据实验表明这是一种可行的有效算法。算法避免了基于Pareto占优比较的复杂性,在解空间的多样性和快速收敛性方面也显示出优越性。GGGA is a new kind of algorithm for multi-objective optimization problem.This algorithm uses elite strategy based on geometry pareto selection （briefly called GPS）;multiple subpopulations evolve from different aspects of the objectives;in every subpopulation,use multi-father crossover operator in GTA.The numerical experiments show that GGGA is feasible and effective.It avoids the complexity of non-dominated set of solutions based on Pareto front.It also provides superiority in terms of diversity and convergence..

关 键 词：多目标优化 精英空间 郭涛算法 多种群 

分 类 号：TP301.6[自动化与计算机技术—计算机系统结构]