Using Conditional Generative Adversarial Networks in Computational Mechanics -- Code

Abstract

We use cGANs to solve computational mechanics. We model a structure using FEA and generate simulations results (deflection and stress). We then transform the FEA results into images where the colour channels correspond to the loading, material and geometric properties of the structure under study. We then use the images as input to the cGAN which is trained to infer deflections from forces, stresses from deflections and stresses from forces. The submission includes the data and the code used in this work.