Abstract
This study examines the effects of mechanical behavior, thermal characteristics, and tribological variables (sliding frequency, normal load, and temperature) on the tribological performance of carbon nanotube (CNT)-coated aramid fabric-reinforced epoxy composites using a computational and data-driven machine learning (ML) approach. Predictive models for the coefficient of friction (COF) were developed based on previous tribological, mechanical, and thermal data, employing three ML algorithms: artificial neural network (ANN), gradient boosting machine (GBM), and random forest (RF). The models showed the following results—ANN: R2 = 0.9088, GBM: R2 = 0.92807, and RF: R2 = 0.85294, with the GBM model providing the best predictions. The dataset with the best performance had an error percentage of 0.003658%, while the poorest performance showed 13.56625%. Feature score analysis highlighted load, sliding frequency, and CNT content as key factors influencing COF. This data-driven ML analysis offers significant insights into the tribological behavior of fiber-reinforced polymer composites, aiding in material design and performance optimization.