Treffer: Unsupervised Graph-GAN model for stress–strain field prediction in a composite.

Out of an astronomical number of combinations in the materials design space, the quest for high-performance novel composites is pacing up. Composite designing needs predictive advanced tools and techniques to complement advanced manufacturing techniques and avoid costly experimental trials, complex modeling, and simulations. Current work focuses on developing an unsupervised material science-informed deep learning model architecture, Graph-GAN, for end-to-end prediction of deformation fields from the unseen composite designs accurately without any spatial data loss. The rate-independent J2 plasticity model for small deformations (5%) is modified for computing deformation fields in synthetically generated biphasic 2D microstructures in a python-based platform for ground-truth dataset generation. Graph-GAN performed impeccably well over typical traditional generative adversarial networks (image-to-image translations) with RMSE < 0.06 for prediction in the test dataset with regular circular geometries and RMSE < 0.07 for unseen microstructures with irregular and arbitrary geometries for the secondary phase. For the first time, this work presents the integration of graph convolutional networks (GCNs) into generative adversarial networks (GANs) for predictive materials science using machine learning. This approach paves the way for numerous opportunities for experts across various disciplines to explore similar methodologies. [ABSTRACT FROM AUTHOR]

Copyright of Journal of Materials Science is the property of Springer Nature and its content may not be copied or emailed to multiple sites without the copyright holder's express written permission. Additionally, content may not be used with any artificial intelligence tools or machine learning technologies. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)