期刊
IEEE TRANSACTIONS ON CYBERNETICS
卷 44, 期 7, 页码 1127-1140出版社
IEEE-INST ELECTRICAL ELECTRONICS ENGINEERS INC
DOI: 10.1109/TCYB.2013.2279802
关键词
Exemplar pool (EP); function optimization; local searcher; particle swarm optimization (PSO); roulette wheel rule; scatter learning PSO algorithm (SLPSOA)
类别
资金
- National Nature Science Foundation of China [61105126, 51177126, 61203350]
- Ph.D. Programs Foundation of the Ministry of Education, China [20100201110031]
Particle swarm optimization (PSO) has been proved to be an effective tool for function optimization. Its performance depends heavily on the characteristics of the employed exemplars. This necessitates considering both the fitness and the distribution of exemplars in designing PSO algorithms. Following this idea, we propose a novel PSO variant, called scatter learning PSO algorithm (SLPSOA) for multimodal problems. SLPSOA contains some new algorithmic features while following the basic framework of PSO. It constructs an exemplar pool (EP) that is composed of a certain number of relatively high-quality solutions scattered in the solution space, and requires particles to select their exemplars from EP using the roulette wheel rule. By this means, more promising solution regions can be found. In addition, SLPSOA employs Solis and Wets' algorithm as a local searcher to enhance its fine search ability in the newfound solution regions. To verify the efficiency of the proposed algorithm, we test it on a set of 16 benchmark functions and compare it with six existing typical PSO algorithms. Computational results demonstrate that SLPSOA can prevent premature convergence and produce competitive solutions.
作者
我是这篇论文的作者
点击您的名字以认领此论文并将其添加到您的个人资料中。
推荐
暂无数据