软件导刊
軟件導刊
연건도간
SOFT WARE GUIDE
2014年
11期
39-42
,共4页
多目标优化%协同进化%聚集密度%分布性
多目標優化%協同進化%聚集密度%分佈性
다목표우화%협동진화%취집밀도%분포성
Multi-objective Optimization%Co-evolution%Crowding-density%Distribution
为了改善协同进化多目标优化算法性能，引入了聚集密度对超级个体集合进行更新。其基本思想是：首先计算种群中各个体的聚集密度，再定义一个偏序集，然后根据一定的比例依次从偏序集中选择个体更新。根据数值试验和量化指标测试了新算法的收敛性与分布性。结果表明，新算法在收敛性方面与常规协同进化多目标算法相当，但其分布性获得了一定程度的改善。
為瞭改善協同進化多目標優化算法性能，引入瞭聚集密度對超級箇體集閤進行更新。其基本思想是：首先計算種群中各箇體的聚集密度，再定義一箇偏序集，然後根據一定的比例依次從偏序集中選擇箇體更新。根據數值試驗和量化指標測試瞭新算法的收斂性與分佈性。結果錶明，新算法在收斂性方麵與常規協同進化多目標算法相噹，但其分佈性穫得瞭一定程度的改善。
위료개선협동진화다목표우화산법성능，인입료취집밀도대초급개체집합진행경신。기기본사상시：수선계산충군중각개체적취집밀도，재정의일개편서집，연후근거일정적비례의차종편서집중선택개체경신。근거수치시험화양화지표측시료신산법적수렴성여분포성。결과표명，신산법재수렴성방면여상규협동진화다목표산법상당，단기분포성획득료일정정도적개선。
In order to improve the performance of co-evolutionary multi-objective optimization algorithm ,the crowding-density is put into to algorithm for updating the super individual collection .The basic idea is:First ,the crowding-density of each individual in the group is calculated ,and then a partial order set is set up according to the objective function value and crowding-density .Finally ,individuals are selected from the partial order set according to the principle of proportional selection ,and the elite set is updated .The convergence and distribution of improved algorithm are studied by means of nu-merical experiments ,and results show that :The convergence of improved algorithm is roughly equal with the conventional co-evolutionary multi-objective optimization algorithm ,but the distribution of improved algorithm has been significantly improved .
