| About this Abstract |
| Meeting |
2024 TMS Annual Meeting & Exhibition
|
| Symposium
|
Aluminum Alloys: Development and Manufacturing
|
| Presentation Title |
Machine Learning Assisted Development of Aluminum Alloys With High Strength at Elevated Temperatures |
| Author(s) |
Jinshian Huang, Daisuke Ando, Yuji Sutou |
| On-Site Speaker (Planned) |
Jinshian Huang |
| Abstract Scope |
Enhancing the high-temperature strength of aluminum alloys holds great importance across various industries. We propose a novel approach utilizing machine learning to develop aluminum alloys with high strength at high temperatures in this work. By employing a machine learning strategy that combines correlation-based screening and genetic algorithms, we conducted feature selection on composition-derived descriptors. Through Bayesian optimization within a quaternary alloy system, we successfully identified four distinct alloys that exhibit exceptional high-temperature strength even without heat treatment. Moreover, we elucidated the model's output using the Shapley Additive exPlanations (SHAP) method. The results underscore the crucial role of specific elements in augmenting the high-temperature strength of aluminum alloys, offering invaluable insights for developing aluminum alloys in demanding thermal environments. |
| Proceedings Inclusion? |
Planned: Light Metals |
| Keywords |
Aluminum, High-Temperature Materials, Machine Learning |