摘要
对于含复杂约束条件的多目标优化问题,提出了一种基于群体分类的遗传算法。其分类方法是:首先将种群分为不可行群体和可行群体,又将可行群体分为可行非Pareto群体和可行Pareto群体,然后再用k-均值聚类将可行Pareto群体划分为非聚类Pareto群体和聚类Pareto群体,最后对上述4个群体分别赋以适当的R适应值。数值计算表明,这种新的算法不仅能得到分布广泛、均匀的Pareto最优解,而且进化速度很快。
The paper presents a constraint-handling approach for muhiobjective optimization.The general idea is shown as follow：Firstly, the population was classified into two groups ：feasible population and infeasible population.Secondly, feasible population was classified into Pareto population and un-Pareto population. Thirdly, the Pareto population was defied with k-average classify approach into colony Pareto population and in-colony Pareto population.lastly,R-fitness was given to each population.Simulation results show that the algorithm not only improves the rate of convergence but also can find feasible Pareto solutions distribute abroad and even.
出处
《软件导刊》
2009年第12期38-41,共4页
Software Guide
基金
安徽省优秀青年人才基金(2009Sqrz054)
关键词
遗传算法
多目标优化
约束条件
聚类分析
Genetic Algorithm
Muhiobjective Optimization
Constraint Condition
Clustering Analysis
