Predicting thermal stress in binary composites through advanced generative adversarial networks

This paper presents a novel approach using Generative Adversarial Networks (GANs) to predict thermal stress distribution in binary composites. The composite plates, generated randomly, serve as the basis for finite element simulations that produce datasets for training the GANs. To enhance the training process, binary images representing composite architecture are combined with corresponding thermal stress distribution images, creating a unified dataset. Remarkably, the trained GANs exhibit a high level of accuracy in predicting thermal stress within the composites. The results underscore the potential of GANs in accurately predicting thermal stress, opening avenues for advancements in materials science.