Flow Stress Model of Hot Deformation for CoNiV Medium Entropy Alloy

Hot compression experiments were performed to characterize the high-temperature deformation behavior of CoNiV medium-entropy alloy (MEA). Hot compression tests were carried out using a Gleeble-3500 thermomechanical simulator over strain rates of 0.001 s^-1 to 1 s^-1 and temperatures ranging between 950 °C and 1100 °C. Based on the experimentally determined hot compression data, three models for predicting the flow stress of CoNiV MEA were established: the Zerilli-Armstrong (Z-A) constitutive model, an artificial neural network (ANN) model, and a gated recurrent unit (GRU) model. This study comprehensively evaluated the prediction accuracy of each model using the coefficient of determination (R²), mean absolute error (MAE), and root mean square error (RMSE). The results show that, compared with the other two models, the Z-A model cannot accurately predict the flow behavior of CoNiV MEA in the studied hot-working regime. The R² value of the ANN model is 0.98974, while the GRU model exhibits the highest predictive capability, with an R² value of 0.98981, an MAE of 6.29621, and an RMSE of 13.10832. The proposed model demonstrates superior prediction accuracy compared with other models, enabling high-precision characterization of the high-temperature evolution of the flow stress in the CoNiV MEA. This study provides a theoretical foundation for the design and optimization of hot working parameters for the CoNiV MEA.