Machine Learning / Composite Material Design

Design of Planar Composite Materials Using CVAE

This study combines the newly developed deep learning technique and traditional finite element analysis method. We use Conditional Variational Autoencoder (CVAE) to learn the relationship between planar composite materials’ topology and resulting mechanical properties (stiffness and toughness in particular). The resulting model can be used for real life composite design problem and to find optimal designs. We also validate our simulation by performing tensile tests with 3D-printed specimens

Design of Grid-like Mask via 4D Printing and ML

This is a ongoing collaborative work with Yi-Xian Xu. We try to build a algorithm that can inversely design a planar composite material which permanently deforms to mask like geometry under hot water. This study combines finite element analysis , machine learning and 4D printing.