Prediction of Hot Deformation Behavior for Inconel 740H Alloy Based on Ensemble Learning

An ensemble constitutive equation model was proposed to predict the hot deformation behavior of Inconel 740H alloy during hot compression tests. The material parameters in the Arrhenius-type equation were compensated for strain, strain rate and hot deformation temperature. A material parameter self-learning method and ensemble learning idea were introduced to enhance the model’s robustness and generalization ability. The effectiveness of the model was verified by using the data within and beyond modeling data range of the hot deformation conditions. The results show that, for data within the modeling data range, the artificial neural network model and ensemble model have smaller errors than that of the Arrhenius-type equation compensation of strain model. For the data beyond the modeling data range, the artificial neural network model achieves a large prediction error due to the lack of physical law guidance. Compared with the Arrhenius-type equation compensation of strain model, the prediction accuracy of the ensemble model increases from 90.25