Welcome to the InfoSci Platform
Reference Hub2
A Dynamic Multi-Swarm Particle Swarm Optimization With Global Detection Mechanism

A Dynamic Multi-Swarm Particle Swarm Optimization With Global Detection Mechanism

Bo Wei, Yichao Tang, Xiao Jin, Mingfeng Jiang, Zuohua Ding, Yanrong Huang
Copyright: © 2021 |Volume: 15 |Issue: 4 |Pages: 23
ISSN: 1557-3958|EISSN: 1557-3966|EISBN13: 9781799859857|DOI: 10.4018/IJCINI.294566
Cite Article Cite Article

MLA

Wei, Bo, et al. "A Dynamic Multi-Swarm Particle Swarm Optimization With Global Detection Mechanism." IJCINI vol.15, no.4 2021: pp.1-23. http://doi.org/10.4018/IJCINI.294566

APA

Wei, B., Tang, Y., Jin, X., Jiang, M., Ding, Z., & Huang, Y. (2021). A Dynamic Multi-Swarm Particle Swarm Optimization With Global Detection Mechanism. International Journal of Cognitive Informatics and Natural Intelligence (IJCINI), 15(4), 1-23. http://doi.org/10.4018/IJCINI.294566

Chicago

Wei, Bo, et al. "A Dynamic Multi-Swarm Particle Swarm Optimization With Global Detection Mechanism," International Journal of Cognitive Informatics and Natural Intelligence (IJCINI) 15, no.4: 1-23. http://doi.org/10.4018/IJCINI.294566

Export Reference

Mendeley
Favorite Full-Issue Download

Abstract

To overcome the shortcomings of the standard particle swarm optimization algorithm (PSO), such as premature convergence and low precision, a dynamic multi-swarm PSO with global detection mechanism (DMS-PSO-GD) is proposed. In DMS-PSO-GD, the whole population is divided into two kinds of sub-swarms: several same-sized dynamic sub-swarms and a global sub-swarm. The dynamic sub-swarms achieve information interaction and sharing among themselves through the randomly regrouping strategy. The global sub-swarm evolves independently and learns from the optimal individuals of the dynamic sub-swarm with dominant characteristics. During the evolution process of the population, the variances and average fitness values of dynamic sub-swarms are used for measuring the distribution of the particles, by which the dominant one and the optimal individual can be detected easily. The comparison results among DMS-PSO-GD and other 5 well-known algorithms suggest that it demonstrates superior performance for solving different types of functions.