Modeling of wetting-state evolution in polypropylene-stearic acid functionalized nanostructured superhydrophobic coatings

Superhydrophobic coatings often lose their water-repellent properties after repeated abrasion and environmental exposure, creating a need for reliable methods to predict coating durability. This study developed a machine learning-based framework to predict the water contact angle (WCA) behaviour of Cu–Zn and polypropylene stearic acid (PP-SA)-modified coatings under ageing and cyclic wear conditions. Linear Regression, Random Forest, Decision Tree, Gradient Boosting, XGBoost, Support Vector Regression, and K-Nearest Neighbours models were evaluated using the coefficient of determination (R 2 ) and mean squared error (MSE). Gaussian data augmentation was applied to improve model robustness and generalization. Although Linear Regression achieved near-perfect fitting on the original dataset, ensemble and kernel-based models demonstrated better generalized performance after augmentation. Random Forest, Gradient Boosting, and Support Vector Regression achieved R 2 values of approximately 0.98, while significant improvements were observed for data-sensitive models such as KNN and SVR. Experimental results showed that PP-SA-modified coatings exhibited higher WCA values, reaching up to 165°, and retained superior hydrophobicity compared with Cu–Zn coatings during prolonged ageing and cyclic abrasion. The close agreement between predicted and experimental degradation trends validates the proposed framework. Overall, integrating machine learning with data augmentation offers an effective and scalable approach for predicting the durability of superhydrophobic coatings and supporting the design of advanced hydrophobic surfaces.