Experimental and Machine Learning Modeling Study of Tribological performance of T6 Heat-Treated Al 7068/(SiC+Fly ash) Composites

This study investigates the wear behavior of T6 heat-treated Al7068/(SiC + Fly ash) metal matrix composites fabricated using the casting process. The dry sliding wear behavior of Al7068/(SiC + Fly ash) composites in both solution-treated and T6 heat-treated states was investigated using a pin-on-disc tribometer. The influence of critical parameters such as heat treatment condition, sliding distance, and sliding velocity on wear rate(WR) and coefficient of friction (CoF) was systematically examined. The worn surfaces under high-load conditions were further characterized through scanning electron microscopy. The analysis indicated that the dominant wear mechanisms included delamination, abrasive action, oxidative wear, and adhesive interactions. To predict tribological responses various machine learning algorithms includingrandom forest (RF), gradient boosting (GB), support vector machine(SVR), and artificial neural network(ANN) were developed and evaluated. For WR prediction, SVR demonstrated the best performance achieving the highest coefficient of determination (R2) of 97.62% and the lowest RMSE of 0.118901. It was closely followed by the ANN model, which attained an R2 of 97.35% and an RMSE of 0.125644. GB and RF showed comparatively lower accuracy indicating their limited effectiveness for this dataset. In the case of CoF prediction, SVR outperformed all other models with superior accuracy (R2) of 94.95% and minimal error values RMSE of 0.009847, and MAE of 0.008138. While ANN and GB also demonstrated competitive performance.The findings demonstrate that integrating heat treatment approaches with data-driven modeling provides an efficient way to predict and analyze the tribological behavior of Al7068/(SiC + Fly ash) aluminum composites.