Feasibility-based Weighted MOEA/D in Constrained Optimization

カテゴリ: 部門大会

論文No: SS1-1

グループ名: 【C】2022年電気学会電子・情報・システム部門大会

発行日: 2022/08/24

タイトル(英語): Feasibility-based Weighted MOEA/D in Constrained Optimization

著者名: 安田 雄佑（東京都立大学）,熊谷 渉（横河電機）,田村 健一（東京都立大学）,安田 恵一郎（東京都立大学）

著者名(英語): Yusuke Yasuda (Tokyo Metropolitan University),Wataru Kumagai (Yokogawa Electric Corporation, Ltd),Kenichi Tamura (Tokyo Metropolitan University),Keiichiro Yasuda (Tokyo Metropolitan University)

キーワード: 有制約最適化|メタヒューリスティクス|制約対処法|MOEA/D|実行可能性|Constrained Optimization|Metaheuristics|Constraint Handling Technique|MOEA/D|Feasibility

要約(日本語): Metaheuristics are mainly intended for unconstrained optimization, but real-world applications call for extending them to constrained optimization. Constraint handling techniques that use constraint violation as an additional objective function are known, but since they use Pareto ranking, there are issues regarding the feasibility and convergence of the obtained solutions. In this paper, we extend MOEA/D, a leading multi-objective optimization technique, to constrained optimization and propose an adaptive weight adjustment method using individual feasibility. The usefulness of the proposed method is verified.