データベースからの知識発見に基づくトポロジー最適化問題の定式化

抄録

In topology optimization, it is significantly important to appropriately formulate a structural design problem as a mathematical programming problem. However, determining the formulation is sometimes difficult because design requirements are often ambiguous and they do not directly correspond to mathematical indicators. Therefore, in this paper, the author presents a new framework for determining formulations of topology optimization problems. The proposed framework is based on knowledge discovery in databases (KDD), that is, the author proposes to incorporate topology optimization and KDD for determining appropriate formulations. In the proposed framework, various material distributions obtained by solving various topology optimization problems are collected as data records, and useful knowledges for determining formulations are extracted from the data records on the bases of KDD. The author also presents a numerical example to demonstrate the feasibility of the proposed framework and discusses some issues that should be resolved in future works.