Development of High-Strength Mg–Gd–Y Alloy Based on Machine Learning Method

A machine learning model is established to efficiently describe the relationship between mechanical properties and chemical composition and processing parameters of magnesium alloys. Among the implemented machine learning algorithms, the random forest model is demonstrated to show the high accuracy for the studied Mg–Gd–Y alloy family. By adopting the model, the optimal composition, thermal, and extrusion process parameters of a high-strength Mg–Gd–Y-based alloy are obtained. The ultimate tensile strength and elongation of the designed Mg–Gd–Y alloy are predicted to be 394 MPa and 8.7%, respectively, which is experimentally found to closely correlate to the formation of long period stacking ordered structure. Comparing with the experimental results, the prediction model gives relatively small error of 5.7% and 1.0% for the yield strength and ultimate tensile strength, respectively, and the poor prediction error of elongation is related to the quality of the prepared alloy. The findings are expected to provide helpful guidance for the intelligent design of advanced magnesium alloys.