About this Abstract |
| Meeting |
2026 TMS Annual Meeting & Exhibition
|
| Symposium
|
Artificial Intelligence Applications in Integrated Computational Materials Engineering (AI-ICME)
|
| Presentation Title |
Integrating Microstructural, Diffraction, and Compositional Data for Predictive Materials Modeling |
| Author(s) |
Jiwon Park, Chang-Seok Oh, Su-Hyeon Kim |
| On-Site Speaker (Planned) |
Jiwon Park |
| Abstract Scope |
Materials characterization methods generate heterogeneous data types, making unified analysis both difficult and time-intensive. While each characterization technique captures distinct aspects of material structure and properties, conventional approaches analyze these data streams in isolation rather than leveraging their complementary nature. In this study, we present a framework that integrates microstructural images, X-ray diffraction patterns, compositional data, and processing parameters to predict the mechanical properties of aluminum alloys. Our model architecture employs parallel neural network branches that process each data modality independently before merging at strategic points within the network trunk. This fused representation enables both accurate property prediction and interpretable feature importance analysis across all modalities. We validated our approach by connecting the model predictions and associated neural network gradients to their corresponding characterization modalities, confirming that the model accurately captures established strengthening mechanisms. This work demonstrates the significant potential of multi-modal learning to harness complementary materials characterization techniques. |
| Proceedings Inclusion? |
Planned: |
| Keywords |
Machine Learning, Aluminum, Characterization |