Microstructural evolution and constitutive analysis of Al-Mg-Zn-Er-Zr based on arrhenius-type and machine-learning algorithm

In this work, artificial neural network (ANN) is employed to predict the hot deformation behavior of Al-Mg-Zn alloys containing small amounts of Er and Zr. A comparative study between the experimental results and the computational results based on Arrhenius constitutive equation and an ANN model was performed, where the theoretical calculation was used to predict the hot deformation behavior of the alloy. The results showed that relative errors obtained from Arrhenius constitutive equation were in the range of -17.7% to + 13.6%, whereas the errors varied from -9.3% to + 9.7% via ANN model. It suggests that the ANN model can avoid some un-certainties of the constitutive equation and predict the thermal deformation behavior of alloys more effectively. The dislocation density has also decreased with an increasing temperature or a decreasing strain rate. The dy-namic aging effect and the dislocation density showed the opposite trend. As hot deformation can induce the intermittent precipitation of Mg-32(Al, Zn)(49) at the grain boundaries, it is expected to improve the corrosion performance of alloy materials.

Automated summary

In this study, an artificial neural network (ANN) was used to predict the hot deformation behavior of Al-Mg-Zn alloys containing small amounts of Er and Zr. A comparative study between experimental and computational results showed that the ANN model outperformed the Arrhenius constitutive equation in predicting the thermal deformation behavior of the alloys.