107 領域遺伝型遺伝的アルゴリズムを用いた多目的最適設計

概要

論文の詳細を見る
These days, requirements of functions tend to spread and the designers need to treat problems as multi-objective optimization. However, decision making through multi-objective optimization is not that simple and we need to give preference of the designer to give a solutions, which has been done through local information called trade off ratio. Usually, trade-off ratio and the results does not have linear relationships, thus it is quite hard to give desired trade-off ratio in a few steps. In the case there are a few objective functions, we can visualize so called Pareto solutions. Then, the designer can grasp the whole relationships of objective functions and it is useful information, something like a map in the whole relationships of objective functions and it is useful information, something like a map in exploring. For that purpose, multi-objective genetic algorithms are quite powerful tools, and there are tons of studies for it. We have been proposing a method to use Data Envelopment Analysis as estimating Pareto Optimality. In this study, we use Genetic Range Genetic Algorithms, and give a new range around obtained Pareto solution and between them. In order to show the effectiveness of the proposed method, we showed some simple example with and without ill setting.
社団法人日本機械学会の論文
2002-10-04