Strain rate affects the deformation mechanism of a Ti-55511 titanium alloy: Modeling of constitutive model and 3D processing map using machine learning

The constitutive model and processing map are effective tools for analyzing the hot deformation behavior of titanium alloys. The stress-strain curves of the Ti-5Al-5Mo-5V-1Cr-1Fe alloy were obtained through the Gleeble thermal simulation experiment. The deformation microstructures at different strain rates were characterized by EBSD and TEM. According to the stress-strain curve, the constitutive models were established by using the Arrhenius equation and the BP artificial neural network (BP-ANN). The constitutive model constructed by BP-ANN has higher fitting accuracy, with the Pearson’s R of 0.99902 and the MAPE of 5.499 %. The 3D processing map of the Ti-55511 alloy was constructed by the BP-ANN constitutive model, which takes temperature, strain, and strain rate as variables. In the processing map, the deformation mechanism is dynamic recovery when the power dissipation efficiency is 0.2–0.33, and the deformation mechanism is dynamic recrystallization when the efficiency is 0.33–0.5. At temperatures of 1123 K and 1153 K and the strain of 0.4, the deformation mechanism is dynamic recovery at the strain rate of 1 s−1. At the strain rate of 0.1 s−1, the deformation mechanism is dynamic recrystallization. This is due to the fact that the higher strain rate prevents dislocations from rearranging, thereby inhibiting dynamic recrystallization behavior. [Display omitted] •The ANN constitutive model of a Ti-55511 titanium alloy has higher accuracy.•3D processing map was constructed based on the ANN constitutive model.•Deformation mechanism is judged according to the power dissipation efficiency.•The judgment is confirmed by the experimental microstructure morphology.