深層学習を用いた塑性変形シミュレーションの代理モデル構築

抄録

In the field of computational mechanics, surrogate models, which replace part or all of numerical calculations with machine learning such as deep learning, are attracting attention. A wide variety of numerical calculations have been studied for surrogate models, but it is difficult to model phenomena involving plastic deformation, especially buckling, and there are still few examples. In this study, a surrogate model was constructed for the simulation of the bending test of hat-shaped steel with such a bucking. We have constructed a model that can predict the bending position, amount of deformation, and stress distribution.