分合粒子群优化算法  

Particle swarm optimization based on division and union strategy

作　　者：陈建超[1] 胡桂武[1,2,3] 

机构地区：[1]广东商学院数学与计算科学学院,广州510320 [2]教育部数据工程与知识工程重点实验室,北京100872 [3]中国人民大学信息学院,北京100872

出　　处：《计算机应用研究》2010年第6期2024-2025,2027,共3页Application Research of Computers

基　　金：国家自然科学基金资助项目(60873017);广东省自然科学基金资助项目(06301003)

摘　　要：基于社会系统中普遍存在"分久必合,合久必分"的现象,提出了基于分合思想的粒子群优化算法。分策略提高了演化群体的多样性,克服了粒子群优化算法局部收敛的缺陷。合策略吸取了不同群体的优良特性,提高了算法的全局搜索能力。函数优化的仿真结果证明了算法的有效性。Basing on a ubiquitous scheme：the world, long divided, must reunite; long united, must divide, which was reported to well represent the nature of social evolution process, this paper proposed divided and united particle swarm optimization. In the new algorithm, division strategy was able to improve the diversity of population, which overcame the local convergence of PSO.Union strategy was able to combine the merit of different population and improved its global research ability. Finally, used the novel algorithm to solve function optimization problem. The result shows that the algorithm is effective.

关 键 词：粒子群优化算法 早熟 函数优化 分合策略 

分 类 号：TP301.6[自动化与计算机技术—计算机系统结构]