| Abstract Scope |
This study leverages machine learning methodologies to predict the fatigue crack growth rates (FCGR) of AA7075 alloy under Environmentally Assisted Corrosion Fatigue (EACF) conditions. A comprehensive dataset containing approximately 1800 data points was compiled from diverse experimental sources, addressing the complexities introduced by varied testing environments. Several machine learning algorithms, including Artificial Neural Networks (ANN), CatBoost, and XGBoost, were systematically evaluated for their predictive performance. Initial results identified ANN as particularly effective, demonstrating strong generalization capabilities. Feature importance analysis further revealed that stress intensity factor (K), loading frequency, temperature, and environmental medium are critical factors influencing crack propagation. These findings support the use of machine learning-particularly ANN models as a promising tool for modeling EACF behavior in AA7075 alloy, aiding the design of more durable and corrosion-resistant materials. |