Machine Learning Methods for Predicting Mechanical Behavior of Aluminum Alloys

Abstract

Predicting the mechanical behavior of metallic materials, such as stress-strain curves, is important for studying the plastic behavior of materials. This paper intends to investigate machine learning methods’ capacity to predict the aluminum alloy’s stress-strain curves under different temperature levels. Specifically, three machine learning methods (Gaussian process regression (GPR), neural network (NN), and boosted trees (BT) were employed to predict the stress-strain response of Al6061- T6 at different temperatures, including 25°C, 100°C, 200°C, and 300°C. The performance of the studied machine learning methods has been verified using actual strain-stress measurements collected using uniaxial tensile testing on Al6061-T6. Four statistical scores have been adopted to evaluate the prediction accuracy. Results revealed the potential of machine learning methods in predicting strain-stress measurements. Furthermore, results showed that the NN model dominates the other models by providing a prediction with an averaged mean absolute error percentage of 0.213.