Publication

Abstract : AZ91 magnesium alloy-based hybrid surface composites reinforced with aluminium diboride (AlB2) and graphene were fabricated via multi-pass Friction Stir Processing (FSP). Varying reinforcement levels (5–15 vol.%) led to significant grain refinement (60 µm to 12 µm) and improved mechanical properties, including a rise in microhardness from 63 HV to 98 HV and an increase in tensile strength from 124 MPa to 212 MPa. Wear rate and coefficient of friction were reduced by 68.6% and 41.7%, respectively, due to the synergistic effects of AlB2 and graphene. Machine learning models were applied to predict specific wear rate, with the Gradient Boosting Regressor achieving the highest accuracy (R2 = 0.974, MSE = 0.000384). SHAP forecasts align with the predictions of the ML model feature significance, indicating that the model's predictions are accurate. The integration of experimental and ML approaches demonstrates an efficient pathway for optimizing composite performance.