标题：Novel chaotic grouping particle swarm optimization with a dynamic regrouping strategy for solving numerical optimization tasks
作者：Chen K.; Xue B.; Zhang M.; Zhou F.
作者机构：[Chen, K] School of Control Science and Engineering, Shandong University, Jinan, Shandong, 250061, China, School of Engineering and Computer Science, 更多
通讯作者地址：[Zhou, F] School of Control Science and Engineering, Shandong University, Jinan, China;
关键词：Chaotic grouping mechanism; Chaotic map; Dynamic regrouping strategy; Particle swarm optimization
摘要：Particle swarm optimization (PSO) has been widely applied to address various optimization problems, since it is easy to implement and only has a few control parameters. However, PSO often suffers from a lack of population diversity during the search process and is ineffective in balancing exploration and exploitation, especially in solving complex numerical optimization tasks. To overcome these disadvantages of PSO for complex numerical optimization problems, a new chaotic grouping PSO algorithm with a dynamic regrouping strategy (CGPSO-DRS) is proposed in this paper. The newly proposed CGPSO-DRS is based on a dynamic multiswarm PSO framework that cooperates with the chaotic grouping mechanism (CGM) and the dynamic regrouping strategy (DRS). First, the CGM divides the entire population into many subswarms via a chaotic sequence. The CGM not only improves the population grouping quality in the search process but also increases the diversity of the population. Second, the DRS is used to guide the regrouping of the population, and the population starts searching with a new configuration. Here, the DRS changes with the number of function evaluations. The DRS facilitates the effective utilization of information to balance the early exploration and the later exploitation performances. In addition, the DRS can increase the diversity of the population in the search process. Experiments have been conducted on 41 benchmark functions, and the numerical results demonstrate that the proposed CGPSO-DRS method outperforms similar population-based approaches and state-of-the-art PSO variants in accelerating the convergence speed and finding the global optimum. © 2020 Elsevier B.V.