Data-driven constitutive model of GH4169 alloy within a synergistic high strain rate and elevated temperature

Dynamic compression experiments of GH4169 alloy at deformation temperatures of 293–873 K and strain rates of 1000–6000 s −1 were performed based on the split Hopkinson pressure bar (SHPB) experimental platform to analyze its thermal deformation behavior and constitutive models. The experimental results show that the hot deformation behavior of GH4169 alloy is mainly limited by the combined action of temperature softening and strain hardening, and the true stress decreases with the increase in the deformation temperature or the decrease in the strain rate. Based on the experimental data, an improved artificial neural network (IANN) model and the classical Johnson–Cook (J–C) constitutive model were constructed to describe the deformation behavior of the studied GH4169 alloy, and the mean absolute error (MAE) was used to evaluate the accuracy of the two models. Based on the comparison of the experimental and predicted results, at strain rates of 1000–3000 s −1 , the MAE of the IANN model was 5, which was only 1/20 that of the J–C model, thereby exhibiting perfect accuracy. At strain rates of 4000–6000 s −1 , the MAE of the IANN model was increased to 82 and the accuracy was decreased, but the MAE was still only 1/4 that of the J–C model. Additionally, the IANN method was embedded in a visual operation interface for the construction of constitutive models with high precision and efficiency, and could be extended to constitutive models of other alloy materials.