Prediction of Wear Rate of Glass-Filled PTFE Composites Based on Machine Learning Approaches

Wear is induced when two surfaces are in relative motion. The wear phenomenon is mostly data-driven and affected by various parameters such as load, sliding velocity, sliding distance, interface temperature, surface roughness, etc. Hence, it is difficult to predict the wear rate of interacting surfaces from fundamental physics principles. The machine learning (ML) approach has not only made it possible to establish the relation between the operating parameters and wear but also helps in predicting the behavior of the material in polymer tribological applications. In this study, an attempt is made to apply different machine learning algorithms to the experimental data for the prediction of the specific wear rate of glass-filled PTFE (Polytetrafluoroethylene) composite. Orthogonal array L25 is used for experimentation for evaluating the specific wear rate of glass-filled PTFE with variations in the operating parameters such as applied load, sliding velocity, and sliding distance. The experimental data are analysed using ML algorithms such as linear regression (LR), gradient boosting (GB), and random forest (RF). The R² value is obtained as 0.91, 0.97, and 0.94 for LR, GB, and RF, respectively. The R² value of the GB model is the highest among the models, close to 1.0, indicating an almost perfect fit on the experimental data. Pearson's correlation analysis reveals that load and sliding distance have a considerable impact on specific wear rate as compared to sliding velocity.