About this Abstract |
| Meeting |
TMS Specialty Congress 2026
|
| Symposium
|
4th World Congress on Artificial Intelligence in Materials & Manufacturing (AIM 2026)
|
| Presentation Title |
Morphology-Based Estimation of Material Parameters via Physics-Informed Neural Networks |
| Author(s) |
Mahmood Mamivand, Asfandyar Khan |
| On-Site Speaker (Planned) |
Mahmood Mamivand |
| Abstract Scope |
Accurate estimation of material parameters that govern phase transformations remains a critical challenge in materials science community. Traditional methods rely on extensive experiments or computationally expensive inverse modeling. In this work, we present a morphology-based framework for material parameter estimation that leverages recent advances in Physics-Informed Neural Networks and related machine learning models. By integrating physical laws with data-driven learning, our approach infers key material parameters directly from microstructural morphologies obtained during solid-state phase transformations. We demonstrate the effectiveness of this framework using case studies on martensitic transformations and spinodal decomposition, where the trained models successfully recover kinetic and interfacial parameters consistent with known physical trends. These results highlight the potential of physics-informed machine learning as a bridge between microstructural observation and underlying materials physics, paving the way toward data-driven digital twins for microstructure evolution. |
| Proceedings Inclusion? |
Undecided |