A Comparative Analysis of Machine Learning Techniques for Predicting the Wear Rate of Ceramic Coated Steel

Ceramic coatings are necessary for steel as they offer resistance to corrosion, high-temperature degradation, and abrasion, thereby enhancing the wear characteristics of steel structures. Evaluating the wear rate of Ceramic-Coated (CC) steels is crucial for enhancing the reliability and longevity of steel components. Various factors such as microstructural features and operating conditions complicate the wear analysis of CC steel. To overcome this obstacle, the present study employs several Machine Learning (ML) models such as Elastic Net Regressor (ENR), Robust Regressor (RR), Extreme Gradient Boosting Regressors (XGBoost), and Bagging Regressor (BR), to predict the wear rate of CC steel. Pearson Correlation Coefficient (PCC) revealed that the hardness of the coating greatly affects the wear rate. Among various ML regressor models, the BR model exhibited the optimum performance with the R2 of 0.93 with ENR, RR, and XGBoost exhibiting lower R2 values of 0.79, 0.84, and 0.89 respectively. Eventually, the BR model is used to predict the wear rate of TiN and Al2O3-coated steel, and the experimental results of the same are compared. The comparison of results revealed an error percentage of ± 7.78% between the experimental and predicted wear rate.