Multi-level robust solution search for evolutionary multi-objective design optimization

2015 Fiscal Year Final Research Report

• Project/Area Number: 26730129
• Research Category: Grant-in-Aid for Young Scientists (B)
• Allocation Type: Multi-year Fund
• Research Field: Soft computing
• Research Institution: The University of Electro-Communications
• Principal Investigator: Sato Hiroyuki 電気通信大学, 情報理工学(系)研究科, 助教 (60550978)
• Project Period (FY): 2014-04-01 – 2016-03-31
• Keywords: 多目的最適化 / ロバスト最適化 / 進化計算

Outline of Final Research Achievements

In engineering design optimization, robust solutions (designs) with a small influence of noises are often employed even if their evaluation values are slightly worse than other candidate solutions. For noisy multi-objective optimization problems, this work designed an evolutionary optimization framework which the decision maker can consider not only evaluation values of each solution but its noise levels to select the final solution from the obtained solution set. First, this work proposed a method to approximate the optimal trade-off among multiple objectives and their noise levels. Next, since the number of noises should be minimized is increased with increasing the number of objectives, this work also proposed evolutionary many-objective optimization techniques. Furthermore, a decision making support user interface to effectively show high-dimensional vectors involving multiple objective values and their noises was proposed.

Free Research Field

進化計算