About this Abstract |
| Meeting |
2027 TMS Annual Meeting & Exhibition
|
| Symposium
|
Magnesium Technology 2027
|
| Presentation Title |
Machine-Learning-Assisted Development of Mg-Gd-Y Alloys with Improved Mechanical Properties |
| Author(s) |
Zhi Yang, Haiyan Fu, Perova Sharlin, Zhihong Xia, Zhihua Dong, Bin Jiang |
| On-Site Speaker (Planned) |
Zhi Yang |
| Abstract Scope |
Machine learning methods are playing an increasingly important role in the efficient design of advanced materials. In the present work, a machine learning model is established based on the optimal training of different algorithms to efficiently describe the relationship between mechanical properties, chemical composition, and processing parameters for Mg-Gd-Y alloys. The cross-validation and independent test sets were adopted in the optimal evolution of the models. Among the implemented machine learning models, including .Random Forest(RF), Support Vector Regression(SVR), XGBoost, and Gradient Boosting Regression(GBR), the ensemble model including SVR and XGBoost is demonstrated to show the high accuracy for the studied Mg-Gd-Y alloys, with the test R² value of 0.914 for ultimate tensile strength, 0.895 for tensile yield strength, and 0.835 for elongation. With the model, new Mg-Gd-Y alloys with improved mechanical properties are developed efficiently and validated with the experimental measurements. |
| Proceedings Inclusion? |
Planned: |
| Keywords |
Magnesium, Machine Learning, Mechanical Properties |