215 多目的最適設計におけるハイブリッド最適化手法の開発 : 可変メッシュサイズを適用したSA/GA法の検討

概要

論文の詳細を見る
In this report, the improvement of the computation algorithm is tried aiming at a further performance gain of the SA/GA hybrid optimization method. In the problem of treating by the previous method, the square lattice is applied to the design space, and all nodes of the search individual can exist only on the one grid point. Therefore the shape of the individual generated becomes minute when the width of the grating space (width of the mesh) is narrow. On the other hand, it becomes difficult to settle to the Pareto optimum solution because the total of the solution that can be generated becomes huge and localized solution increases. Then, the method, that of widely setting width of mesh at early stage of calculation and narrowing moving range of node and width of mesh gradually, is newly introduced as measures of this problem. It was confirmed to obtain the Pareto optimum solution efficiently evading stagnating in the localized solution by such the operation.
一般社団法人日本機械学会の論文
2011-09-05