Data‐Driven Optimization of Microstructure for Enhanced Fluoride Corrosion Resistance of High‐Purity 63/37 Brass

A machine learning‐assisted framework was developed to investigate the relationship between microstructure and corrosion behavior of high‐purity 63/37 brass in fluoride containing solution. A dataset was established based on microstructural characterization and potentiostatic polarization tests. Among nine machine learning algorithms, XGBoost achieved the best predictive performance. SHAP and ICE analyses identified α‐phase size differential and α‐phase size as the dominant factors controlling the stable current density, whereas β‐phase content showed limited influence. Exhaustive enumeration and PSO identified favorable microstructural regions and a candidate configuration with low predicted I stable . The proposed framework provides quantitative guidance for designing corrosion‐resistant brass microstructures in fluoride‐containing environments.