Abstract
Researches of multi-objective optimization using PSO are starting recently. It has a character of multi-point search. Its search engine is simple, as to aim interior division point of personal best and global best which has achieved among groups. It can be applied to indifferentiable problems because it does not use gradient. We proposed Agent PSO in other times. It is able to set parameters for each individual. It can express biotic diversity and reduce difficulty of parameter setting that is fault of existing PSO. Then we proposed Multi-Swarm Agent PSO. It has multi-swarm to set various parameters more than ordinary Agent PSO. We confirmed improved performance. From our past numerical experiment, we understand that biotic diversity is a key to improve performance. In addition, designers require diversity of Pareto solution of Multi-Objective Optimization. In this study, we propose Multi-Objective Agent PSO that is Agent PSO applied to multi-objective problems. We carried out the proposed method in some bench mark tests, and show the effectiveness of the results and from comparisons.
