| About this Abstract |
| Meeting |
2025 TMS Annual Meeting & Exhibition
|
| Symposium
|
AI/Data Informatics: Computational Model Development, Verification, Validation, and Uncertainty Quantification
|
| Presentation Title |
Predicting Microstructure-Property Linkage in Alloys Using Graph Neural Network |
| Author(s) |
Abigail Hogue, Benjamin Rhoads, Samrat Choudhury |
| On-Site Speaker (Planned) |
Abigail Hogue |
| Abstract Scope |
Deep learning-based machine learning tools have recently shown significant potential to accelerate the prediction of microstructure evolution. While deep neural networks like convolution neural networks (CNN) can extract information from 3-D microstructure images, they often require a large network architecture and substantial training time. In this research, we trained a graph neural network (GNN) with phase-field generated microstructures of Ni-Al alloys to predict the evolution of mechanical properties. We found that unlike 3-D CNN, GNN requires significantly less training time and can predict yield strength with higher accuracy compared to a 3-D CNN. Further, it was observed that GNN can be trained on graphs derived from microstructure images with different resolutions and dimensions, which cannot otherwise be done with a CNN. Overall, our work demonstrates the ability of the GNN to accurately and efficiently extract relevant information from materials microstructures without having restrictions on data resolution or dimension. |
| Proceedings Inclusion? |
Planned: |
| Keywords |
Characterization, Machine Learning, Mechanical Properties |