Summary
International Symposium on Nonlinear Theory and its Applications
2017
Session Number:A3L-D-1
Session:
Number:A3L-D-1-1
Multi-Objective Particle Swarm Optimizer Networks with Tree Topology
Kyosuke Miyano, Hidehiro Nakano, Arata Miyauchi,
pp.283-286
Publication Date:2017/12/4
Online ISSN:2188-5079
DOI:10.34385/proc.29.A3L-D-1-1
PDF download (461.7KB)
Summary:
Multi-Objective Particle Swarm Optimizer (MOPSO) is a kind of metaheuristic algorithms for solving multi-objective optimization problems. In MOPSO, a global best solution set corresponding to the Pareto solution set is stored in an archive memory. Island-model MOPSO (IMOPSO) has a tree topology of sub-swarms; a upper layer sub-swarm search the Pareto solution set in the multi-objective function, while lower layer sub-swarms search the best solutions in each single objective function. IMOPSO can effectively search high-quality Pareto solution set. This paper investigates the performance of some migration strategies. Then, it is shown that a migration strategy between lower layer sub-swarms can provide the good search performance. In the simulation experiments, the results for some benchmark problems are shown.