粒子群优化
多群优化
群体行为
水准点(测量)
元启发式
职位(财务)
粒子(生态学)
趋同(经济学)
数学优化
计算机科学
集合(抽象数据类型)
中心(范畴论)
算法
数学
生物
程序设计语言
地理
生态学
大地测量学
财务
经济增长
经济
化学
结晶学
作者
Lei Yu,Qinghua Zheng,Zhongqi Shi,Lu Jiang
出处
期刊:Neurocomputing
[Elsevier]
日期:2007-01-01
卷期号:70 (4-6): 672-679
被引量:111
标识
DOI:10.1016/j.neucom.2006.10.002
摘要
Center particle swarm optimization algorithm (CenterPSO) is proposed where a center particle is incorporated into linearly decreasing weight particle swarm optimization (LDWPSO). Unlike other ordinary particles in LDWPSO, the center particle has no explicit velocity, and is set to the center of the swarm at every iteration. Other aspects of the center particle are the same as that of the ordinary particle, such as fitness evaluation and competition for the best particle of the swarm. Because the center of the swarm is a promising position, the center particle generally gets good fitness value. More importantly, due to frequent appearance as the best particle of swarm, it often attracts other particles and guides the search direction of the whole swarm. CenterPSO and LDWPSO are extensively compared on three well-known benchmark functions with 10, 20, 30 dimensions. Experimental results show that CenterPSO achieves not only better solutions but also faster convergence. Furthermore, CenterPSO and LDWPSO are compared as neural network training algorithms. The results show that CenterPSO achieves better performance than LDWPSO.
科研通智能强力驱动
Strongly Powered by AbleSci AI