A Machine Learning Approach to Optimize Heat Treatment for Increased SCC Resistance in Al-Mg and Al-Zn-Mg Alloys

Stress corrosion cracking (SCC) poses a significant threat to the structural integrity of Al- Mg and Al-Zn-Mg alloys used in aerospace, automotive and marine applications. Traditionally heat treatment methods have called for over-aging heat treatment to improve SCC resistance of Al-Zn-Mg alloys, however, this is accompanied by a reduction in the yield strength. While traditional manufacturing methods offer solutions that improve SCC resistance it is often detrimental to critical material properties. To address this challenge, we propose a machine learning (ML) model that leverages existing databases encompassing comprehensive information on heat treatment profiles, microstructure, and material property relationships. The main objective of this study is to harness the capabilities of ML algorithms in identifying optimal heat treatment schedules that maximize SCC resistance in Al-Mg and Al-Zn-Mg alloys. By analyzing these databases, the model establishes correlations between heat treatment parameters, microstructure characteristics, and SCC resistance, enabling the prediction of the alloys’ SCC performance under various heat treatment conditions. The ML model employs advanced algorithms, such as random forest, support vector machines, and neural networks, to establish relationships between input heat treatment parameters and SCC resistance. Through an iterative learning process, the model determines the optimal heat treatment schedule that maximizes SCC resistance. Rigorous training and validation using cross-validation techniques ensure the accuracy and generalizability of the model across a range of compositions relevant to Al-Mg and Al-Zn-Mg alloys. The optimized heat treatment schedule achieved through the ML model results in a favorable combination of microstructural features, including precipitate size, distribution, and morphology, which contributes to improved corrosion resistance. The ML model enables the prediction of SCC performance for untested heat treatment conditions, facilitating the design of tailored heat treatment protocols.