Summary
International Symposium on Nonlinear Theory and Its Applications
2015
Session Number:B4L-E
Session:
Number:B4L-E-3
A Study of Robustness of PSO for Non-Separable Evaluation Functions
Yosuke Hariya, Takuya Kurihara, Takuya Shindo, Kenya Jin'no,
pp.724-727
Publication Date:2015/12/1
Online ISSN:2188-5079
DOI:10.34385/proc.47.B4L-E-3
PDF download (1016.8KB)
Summary:
Recently, many researchers paid attention to the studies of meta-heuristics for the continuous value optimization problems. Especially, Artificial Bee Colony algorithm ( abbr. ABC ), Differential Evolution ( abbr. DE ), and Particle Swarm Optimization ( abbr. PSO ) are applied various optimization problems widely. In general, such meta-heuristics can obtain good solutions of multi-modal functions rapidly. However, solution search performance becomes worse in the case of non-separable functions. In this article, we focus on PSO to search an optimum solution of non-separable functions. We clarify the cause of the degradation of the search performance. Based on the above consideration, we propose a Norm linked PSO which solution search performance is robust for the dependencies among each design variable. We confirm the search performance of the norm linked PSO by numerical simulations.