| About this Abstract |
| Meeting |
2025 TMS Annual Meeting & Exhibition
|
| Symposium
|
Neutron and X-ray Scattering in Materials Science and Engineering
|
| Presentation Title |
Machine Learning Models For X-ray Diffraction Temperature Inference |
| Author(s) |
Griffin Hess, Georgios Zipitis, Sachith Dissanayake, Chenliang Xu, Niaz Abdolrahim |
| On-Site Speaker (Planned) |
Griffin Hess |
| Abstract Scope |
Measuring temperature in dynamic compression experiments is crucial for understanding material behavior but methods are currently lacking. Physics-based synthetic diffraction patterns were used to train convolutional neural nets to predict temperature in copper based on the Debye-Waller Factor. A variety of diffraction patterns were generated for copper based on Bragg's law, Gaussian mosaicity, and the Debye approximation. Peak intensity changes due to pressure or temperature were accounted for using the Grüneisen parameter from the Mie-Grüneisen equation of state. Models show good performance on temperature prediction for a variety of conditions. Classical atomistic simulations of dynamic high compression experiments will be used to generate realistic patterns to mix with synthetic data to improve performance for dynamic experiments. We will discuss challenges in measuring temperature directly such as texture evolution, phase changes, and defect generation in dynamic high compression experiments. |
| Proceedings Inclusion? |
Planned: |
| Keywords |
Characterization, Copper / Nickel / Cobalt, Machine Learning |