Modeling the correlation between texture characteristics and tensile properties of AZ31 magnesium alloy based on the artificial neural networks

This work is aimed to investigate the relationship between the texture and tensile properties of the AZ31 Mg alloy by the machine learning method. The texture characteristics parameters, namely the maximum pole intensity (Imax), texture dispersion (D), and texture directivities along the longitudinal direction (PLD) and transverse direction (PTD), are extracted from the (0002) pole figures of the AZ31 Mg alloy. An artificial neural networks (ANN) model to describe the relationship between the texture characteristic parameters and tensile properties is constructed and trained by the data collected from the literature. To validate the reliability and generalization performance of the ANN, 6 samples with different texture characteristics are prepared, and their textures and tensile properties are evaluated through electron backscattered diffraction (EBSD) measurement and uniaxial tensile test, respectively. The results indicate that the ANN model exhibits good prediction performance in yield strength and elongation of the AZ31 Mg alloy when it is applied to the new cases. The correlations between the texture characteristics and tensile properties are analyzed according to the ANN-predicted results. The maximum pole intensity and texture dispersion significantly influence the tensile properties of the AZ31 Mg alloy. With increasing the Imax or decreasing the D, the strength is increased but the elongation is reduced. As increasing the texture directivity along the LD, the tensile properties of the AZ31 Mg alloy show non-monotonic changes. This research presents a correlation model