Multi-Guider and Cross-Searching Approach in Multi-Objective Particle Swarm Optimization for Electromagnetic Problems

The main difference between single and multi-objective optimizations using particle swarm optimization is how the guider to locate the global optimal and Pareto optimal solutions are defined in corresponding optimization problems. In general multi-objective particle swarm optimization, only one guid...

Full description

Saved in:
Bibliographic Details
Published in:IEEE transactions on magnetics 2012-02, Vol.48 (2), p.539-542
Main Authors: Pham, Minh-Trien, Zhang, Diahai, Koh, Chang Seop
Format: Article
Language:English
Subjects:
Algorithm design and analysis
Convergence
Cross-disciplinary physics: materials science
rheology
Cross-searching
Exact sciences and technology
Genetic algorithms
Materials science
Measurement
multi-guiders
multi-objective optimization
Other topics in materials science
Pareto optimization
Particle swarm optimization
Physics
TEAM Problem 22
Citations: Items that this one cites
Items that cite this one
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:The main difference between single and multi-objective optimizations using particle swarm optimization is how the guider to locate the global optimal and Pareto optimal solutions are defined in corresponding optimization problems. In general multi-objective particle swarm optimization, only one guider is selected, and, in order to reduce the non-dominated solution for more diversity in the external archive, only crowding distance in objective space is considered. This paper presents a new approach of selecting multiple guiders to lead a swarm toward a Pareto-front. Additionally, in order to overcome the local Pareto front, mutation operator is applied for not only particles but also members in an external archive. Furthermore, aside from considering the crowding distance of solutions in objective space to maintain the diversity of solutions, the crowding distance in variable space is also taken into account. The proposed algorithm is compared with recent approaches of multi-objective optimizer in solving a multi-objective version of the TEAM 22 benchmark optimization problem with three and eight design variables.
ISSN:0018-9464
1941-0069
DOI:10.1109/TMAG.2011.2173559