Research on constitutive model of aluminum alloy 7075 thermal deformation based on deep neural network

The hot deformation behavior of the Al-Zn-Mg-Cu alloy was studied by isothermal tensile tests in the temperature range of 200–350 °C and the strain rate range of 0.001–0.1 s−1. A data-driven deep neural network (DNN) constitutive model and a phenomenological Arrhenius constitutive model were developed for the studied alloy model. The parameters of the DNN model were optimized to improve the prediction accuracy of flow stress. The results show that the accuracy of predictions of the DNN model is better than the Arrhenius model for the hot deformation behavior of 7075 aluminum alloy. The average absolute relative error and correlation coefficient of the DNN model is 1.70 % and 0.9996, respectively. The accuracy of the constitutive model of Arrhenius is relatively low for 7075 aluminum alloy in the range 200–350 °C, 0.001–0.1 s−1. The optimal network depth and the number of neurons per layer for the analytically optimized DNN constitutive model are 6 and 28, respectively. In addition, the developed DNN model can be effectively applied in intelligent manufacturing, such as short-process high-efficiency hot stamping and other plastic-forming technologies.