摘要
为提高求解多目标优化问题效率,对通用差异演化(GDE)算法及其自适应参数控制问题进行了研究。首先,分析了GDE3算法的编码、交叉、变异、选择等原理和算法流程;然后,利用个体的适应度作为参数调整的依据,并结合一定的调整概率提出一种新的对缩放因子和交叉概率参数自适应控制策略,提高算法的搜索能力;最后,通过典型的多目标函数对自适应控制参数的通用演化算法(selfGDE3)、GDE3和非劣分层遗传算法2(NS-GA-Ⅱ)的性能进行比较分析,结果表明,selfGDE3算法具有良好的搜索性能。
In order to solve the multi-objective optimization problem efficiently, this paer researched on generalized differential evolution algorithm and the method of adaptive controlling parameter. Firstly, it analyzed the principle and process of generalized differential evolution algorithm 3, including coding, crossover, mutation. Secondly, the algorithm put forward adaptive controlling strategy to crossover and mutation parameter based on the fitness of individual and the adjusting probability, which improved the performance of algorithm. Finally, it compared the performance of selfGDE3, GDE3 and NSGA- II through testing some benchmark functions. The results show the feasibility of selfGDE3.
出处
《计算机应用研究》
CSCD
北大核心
2012年第8期2899-2902,共4页
Application Research of Computers
基金
装备预先研究资助项目
关键词
多目标优化
非劣分层遗传算法2
通用差异演化算法
自适应控制参数
multi-objective optimization
NSGA-I1
GDE （generalized differential evolution）algorithm
adaptive controlling parameter
