A Knowledge-Guided XGBoost Model for Predicting Aluminum Alloy Corrosion Rates in Atmospheric Environments

Accurate prediction of aluminum alloy corrosion rates in atmospheric environments is crucial for material performance optimization and engineering safety. Traditional data-driven machine learning models often face challenges such as poor robustness and interpretability, especially when dealing with small sample datasets. To address these issues, this paper proposes a Knowledge-Guided XGBoost (KG-XGBoost) model that integrates domain knowledge into the machine learning process. The model incorporates new features, such as work function (WF) and chlorine adsorption energy (CIAE), calculated based on first-principles calculations, and introduces a composite loss function that combines data-driven and physics-based constraints. Experimental results demonstrate that the KG-XGBoost model significantly improves prediction accuracy and robustness compared to traditional XGBoost models.