IRMA-International.org: Creator of Knowledge
Information Resources Management Association
Advancing the Concepts & Practices of Information Resources Management in Modern Organizations

Experimental Study on Boundary Constraint Handling in Particle Swarm Optimization: From Population Diversity Perspective

Experimental Study on Boundary Constraint Handling in Particle Swarm Optimization: From Population Diversity Perspective
View Sample PDF
Author(s): Shi Cheng (University of Liverpool, UK), Yuhui Shi (Southern University of Science and Technology (SUSTech), China)and Quande Qin (Shenzhen University, China)
Copyright: 2013
Pages: 29
Source title: Recent Algorithms and Applications in Swarm Intelligence Research
Source Author(s)/Editor(s): Yuhui Shi (Southern University of Science and Technology (SUSTech), China)
DOI: 10.4018/978-1-4666-2479-5.ch006

Purchase


Abstract

Premature convergence happens in Particle Swarm Optimization (PSO) for solving both multimodal problems and unimodal problems. With an improper boundary constraints handling method, particles may get “stuck in” the boundary. Premature convergence means that an algorithm has lost its ability of exploration. Population diversity is an effective way to monitor an algorithm’s ability of exploration and exploitation. Through the population diversity measurement, useful search information can be obtained. PSO with a different topology structure and a different boundary constraints handling strategy will have a different impact on particles’ exploration and exploitation ability. In this paper, the phenomenon of particles gets “stuck in” the boundary in PSO is experimentally studied and reported. The authors observe the position diversity time-changing curves of PSOs with different topologies and different boundary constraints handling techniques, and analyze the impact of these setting on the algorithm’s ability of exploration and exploitation. From these experimental studies, an algorithm’s ability of exploration and exploitation can be observed and the search information obtained; therefore, more effective algorithms can be designed to solve problems.

Related Content

Rafael Martí, Juan-José Pantrigo, Abraham Duarte, Vicente Campos, Fred Glover. © 2013. 21 pages.
Peng-Yeng Yin, Fred Glover, Manuel Laguna, Jia-Xian Zhu. © 2013. 20 pages.
Volodymyr P. Shylo, Oleg V. Shylo. © 2013. 10 pages.
Tabitha James, Cesar Rego. © 2013. 19 pages.
Gary G. Yen, Wen-Fung Leong. © 2013. 25 pages.
Shi Cheng, Yuhui Shi, Quande Qin. © 2013. 29 pages.
Xin-She Yang. © 2013. 12 pages.
Body Bottom