摘要
针对传统差分进化算法在求解问题中种群易收敛、易早熟的问题,提出了一种基于混合策略的差分进化算法.该算法根据粒子适应度、适应度标准差和粒子间距离标准差,将种群分为3个不同大小、不同功能的子种群,每个子种群采用不同策略和控制参数来实现自己被指定的功能.算法在搜索过程中既增强了种群的全局搜索能力,又增加了收敛精度.通过对4个标准函数的测试,仿真结果表明该算法比其他算法具有更好的寻优能力.
In this paper, a differential evolution traditional differential evolution algorithm which algorithm based on hybrid strategy was proposed to solve the 1S sional problems. This algorithm divided the popula i ent functions according to the fitness,standard devi tion used different strategies and parameters to achi easy to convergence and premature in solving high-dimen- tion into three sub-populations of different sizes and differ- ation of fitness and distance of particles. Each sub-popula- eve their specific functions. It not only enhances the global search ability of the population, but also increases the precision of convergence during the search process. Having tasted four classic benchmarks problems, the experiment results show that the proposed algorithm is an effective method for different optimization problems.
出处
《郑州大学学报（工学版）》
CAS
北大核心
2013年第5期59-62,共4页
Journal of Zhengzhou University（Engineering Science）
基金
国家自然科学基金资助项目(211000531605)
中国博士后科学基金特别资助项目(2012T50639)
教育部高等学校博士学科点专项科研基金资助项目(20114101110005)
河南省科技公关资助项目(132102210521)
关键词
差分进化算法
多种群
混合策略
differential evolution algorithm
multi-population
hybrid strategy
