高效求解Pareto最优前沿的多目标进化算法  

A Multi-objective Evolutionary Algorithm for Efficiently Solving Pareto Optimal Front

作　　者：童晶[1] 赵明旺[2] 

机构地区：[1]武汉科技大学计算机科学与技术学院,湖北武汉430081 [2]武汉科技大学信息科学与工程学院,湖北武汉430081

出　　处：《计算机仿真》2009年第6期216-219,共4页Computer Simulation

摘　　要：设计了一种新的求解均匀分布的Pareto最优解集的多目标进化算法(MOEA),其主要的特点是使用了一种新的个体适应值的计算方式,方法是通过群体中某一个体与群体的最优非劣解集的最小距离来刻画个体的适应值的。算法还结合了遗传算法中的精英策略以及NSGA-Ⅱ中的拥挤距离[12],提高了非劣解向Pareto最优前沿收敛的速度,并且保证了Pareto最优解集的多样性。仿真结果表明,算法不仅能够获得分布良好的Pareto最优前沿,而且能够极大地简化计算,减少了算法的运行时间,其计算复杂度为ο(mn2)(m表示的是目标函数的个数,n是种群的规模)。This paper proposes a novel multi -objective evolutionary algorithm for obtaining even distributed Pareto non -dominated solutions. This algorithm is characterized by a new fitness function that uses the minimum distance between an individual and optimal non - dominated solutions to compute the individual fitness in a population. And, the algorithm uses elitism of Genetic Algorithm and crowding distance of NSGA - Ⅱ to quicken further the convergence rate of solutions to Pareto optimal front, and to improve diversity of solutions in Pareto optimal front. The simulation results indicate that this algorithm can not only get well distributed pareto optimal front, but also can simplify its computation and decrease sharply its runtime. The computational complexity of the algorithm is o （mn^2 ） （ where m is the number of objectives and n is the population size）.

关 键 词：多目标优化问题 多目标进化算法 最优非劣解 精英策略 拥挤距离 

分 类 号：TP301.6[自动化与计算机技术—计算机系统结构]