Study of Geometric Design Parameters Range Optimization
Hiroyuki Sano, Nicolas Schneider, Kazuki Semba, Yusaku Suzuki, Takashi Yamada · 2023
A determination method of design parameter ranges to avoid geometry collapses is discussed toward large-scale optimization. It is known that parametric models with a large number of geometric parameters tend to collapse due to geometrical conflicts, and the geometry collapse leads to a poor solution in optimizations. A technique, which is a filtering process composed of range optimization and a predictive model, has been proposed to avoid geometry collapses. Although the technique works well, there is an argument that the technique over-eliminates possible cases. To address this issue, the behavior of the filtering process is investigated. The results show that a sufficient population in the range optimization can mitigate missing possible cases.