动态分群的微粒群优化算法  

Particle Swarm Optimization Algorithm with Dynamic Sub-swarms

作　　者：卜艳萍[1] 俞金寿[1] 

机构地区：[1]华东理工大学自动化研究所

出　　处：《华东理工大学学报（自然科学版）》2007年第6期846-849,共4页Journal of East China University of Science and Technology

摘　　要：在分析基本微粒群优化算法的基础上,引进分群思想,提出了一种动态分群的微粒群优化算法(DPSO)。根据适应值的大小将微粒群分成两个或多个分群,然后,每个分群采用不同的策略分别搜索,得到输出最优值。将动态分群的微粒群优化算法用于一些常用测试函数的优化问题,实例计算表明:DPSO具有较强的全局寻优能力。将DPSO用于延迟焦化装置粗汽油干点软测量,所建模型的泛化性较好,模型具有较高的精度。On the basis of analyzing the particle swarm optimization and introducing the idea of subswarms, a particle swarm optimization algorithm with dynamic sub-swarms（DESO） is proposed. The particle swarm is divided into two or more sub-swarms according to the fitness value during searching. Then, the sub-swarms use different searching strategy, respectively. The best fitness output value is obtained. The DESO algorithm is used to solve the optimization problems of several widely used test functions, and results indicate that the DESO has powerful ability of global searching. The DESO algorithm is also applied to construct a practical soft-sensor of gasoline endpoint of delayed coking plant. The model has effective generalization performance and higher precision.

关 键 词：微粒群优化算法 动态分群 全局优化 软测量 

分 类 号：TP273[自动化与计算机技术—检测技术与自动化装置]