Deepening Unknown Solution Exploration Algorithms in Globally Multimodal Search Spaces

2022 Fiscal Year Final Research Report

• Project/Area Number: 20K11986
• Research Category: Grant-in-Aid for Scientific Research (C)
• Allocation Type: Multi-year Fund
• Section: 一般
• Review Section: Basic Section 61040:Soft computing-related
• Research Institution: Tokyo Institute of Technology
• Principal Investigator: Ono Isao 東京工業大学, 情報理工学院, 教授 (00304551)
• Project Period (FY): 2020-04-01 – 2023-03-31
• Keywords: 進化計算 / ブラックボックス最適化 / 大域的多峰性 / 関数最適化 / 組合せ最適化 / シミュレーションベース最適化 / データ同化

Outline of Final Research Achievements

By extending the evolutionary computation methods for globally multimodal black-box optimization proposed so far, we developed methods that can find better solutions more efficiently than conventional methods for more difficult problems, and confirmed its effectiveness through numerical experiments using benchmark problems and real-world problems. In particular, we succeeded in finding several solutions with performance comparable to or better than solutions found by experts that conventional methods could not found in the 2-group 7-element zoom lens system design problem. Furthermore, we proposed evolutionary computation methods for various difficult problems such as large-scale traveling salesman problems, vehicle routing problems with time windows, discrete black-box function optimization problems with dependencies among variables, and symbolic regression, and confirmed their effectiveness through numerical experiments.

Free Research Field

進化計算

Academic Significance and Societal Importance of the Research Achievements

本研究で提案したブラックボックス最適化手法は，手法の性能評価に用いたレンズ系設計問題だけでなく，環境・エネルギー分野，社会・経済分野，航空・宇宙分野，材料分野などのさまざまな分野のシステム最適化や未知解発見に適用可能な汎用性の高い手法であり，これまで未解決だった各分野の難問の解決へとつながることが期待される．また，巡回セールスマン問題や時間枠制約付き配送計画問題のための最適化手法は流通業界における問題解決への応用が期待され，データモデリングのためのシンボリック回帰手法および状態・パラメータの逐次推定手法は様々な分野におけるデータ解析への応用が期待される．