Data‐Driven Modeling for Predicting Tribo‐Performance of Flax‐Steel‐Epoxy Hybrid Composites Using Machine Learning Approach

Machine learning (ML) models have recently become popular for their ability to replicate human decision‐making in different fields. This study presents a data‐driven machine learning method for the analysis and prediction of sliding wear behavior of flax‐epoxy hybrid composites, both with and without steel mesh. Three hybrid composites with different stacking sequences of reinforcing fibers are prepared and subsequently well‐planned sliding wear trials are conducted according to the design of experiments. The experimental analysis examined the influence of sliding velocity, sliding distance, and normal load on the specific wear rate, revealing that wear rate increases with higher sliding velocity and normal load but decreases with higher steel mesh content. The collected data is then utilized to train four ML models, whose performances are measured using five different metrics. Among them, the extreme gradient boosting (XGB) model showed the best accuracy, with coefficient of determination (R 2 ) values of 0.93651 for composite F11, 0.98544 for composite F7S4, and 0.97546 for composite F4S7. Feature importance analysis revealed sliding velocity as the most significant factor. Electron microscopy is further utilized to study predominant wear mechanisms.