摘要
首先基于一些实例研究了差异演化(DE)的参数选择问题;然后在分析DE特点的基础上,将缩放因子F由固定数值设为随机函数,实现了一个简化的DE版本(SDE).该方法不仅减少了需调整的参数,而且对CR的参数选择更为宽松.与已有文献中遗传算法的带约束型数值优化问题的实验结果对比,表明SDE能在较少的计算次数内获得较好的结果.
The parameter selection of differential evolution (DE) is studied by experiments on some benchmark examples. A simplified DE version (SDE) is realized with randomized scaling factor F based on the analysis for the features of DE, which not only reduces a parameter, but also is flexible for the selection of parameter CR. The experiments by comparing with genetic algorithm (GA) on some constrained numerical optimization problems show that SDE can get better results in much less evaluation time.
出处
《控制与决策》
EI
CSCD
北大核心
2004年第1期49-52,56,共5页
Control and Decision
关键词
差异演化
演化计算
数值优化
计算机算法
参数设置
Differential equations
Genetic algorithms
Optimization
Parameter estimation
