クラスタリングを用いた湾曲繊維CFRPの最適繊維配向モデルの探索手法の提案

抄録

In recent years, 3D-printed Carbon Fiber Reinforced Plastic (CFRP) composites is receiving a lot of attention in additive manufacturing field. This method can make CFRP composites containing curved fiber by 3D printer^[1]. In order to apply it to mechanical structure, it is important to reveal mechanical properties of 3D-printed CFRP composites. In this study, we conduct modal and buckling analysis by FEM in a lot of plate models containing curved fiber and investigate the similarity of mapping data in 2-dimensional data by the statistical method with results obtained from buckling and modal analysis. Specifically, we reduced data such as buckling loads and natural frequencies onto 2-dimensional features by machine learning algorithm and it is mapped by clustering on 2-dimensional plane. It demonstrates we can explore the model having desired properties when we design the curved fiber CFRP composites plate by considering the relation between the mapping visualization and each model.