A Physics‐Informed Neural Network Constitutive Model Constrained by SCA Equation for Predicting the Hot Deformation Behavior and Microstructural Evolution of 42CrMo Steel

To accurately predict the rheological behavior of 42CrMo steel during hot working, this study conducted hot compression experiments within a strain rate range of 0.01–10 s ⁻¹ and a temperature range of 1073–1473 K. Metallographic observation and electron backscatter diffraction (EBSD) techniques were combined to investigate the hot deformation behavior and the influence of deformation parameters on dynamic recrystallization (DRX) behavior. Based on the experimental data, strain‐compensated Arrhenius‐type (SCA), artificial neural network (ANN), and physics‐informed neural network constitutive model constrained by SCA equation (SCA_PINN) were constructed. The results show that the coefficients of determination for the three models are 0.9885, 0.9988 and 0.9990, with mean absolute relative errors of 5.02%, 0.25%, and 1.74%, respectively. Predictions for untested working conditions reveal that the SCA_PINN model's 3D stress prediction surface is smooth and continuous, better conforming to the rheological laws of hot deformation. EBSD analysis reveals significant differences in kernel average misorientation between grain interiors and boundaries, and increasing the temperature and decreasing the strain rate are beneficial to the DRX process. Under high strain rate condition, the continuous dynamic recrystallization dominated by subgrain rotation is the main mechanism for the significant refinement of the microstructure of 42CrMo steel.