A characterization for the constitutive relationships of 42CrMo high strength steel by Artificial Neural Network and its application in isothermal deformation

• Autor(a) principal: Quan,Guo-zheng
• Data de Publicação: 2014
• Outros Autores: Liang,Jian-ting, Lv,Wen-quan, Wu,Dong-sen, Liu,Ying-ying, Luo,Gui-chang, Zhou,Jie
• Tipo de documento: Artigo
• Idioma: eng
• Título da fonte: Materials research (São Carlos. Online)
• Texto Completo: http://old.scielo.br/scielo.php?script=sci_arttext&pid=S1516-14392014000500002
• topic: artificial neural network; 42CrMo high strength steel; dynamic recrystallization; prediction potentiality; FEM simulation

In hot working process, the prediction of material constitutive relationship can improve the optimization design process. Recently, the artificial neural network models are considered as a powerful tool to describe the elevated temperature deformation behavior of materials. Based on the experimental data from the isothermal compressions of 42CrMo high strength steel, an artificial neural network (ANN) was trained with standard back-propagation learning algorithm to predict the elevated temperature deformation behavior of 42CrMo steel. The inputs of the ANN model are strain, strain rate and temperature, whereas flow stress is the output. According to the predicted and experimental results, it indicates that the developed ANN model shows a good capacity of modeling complex hot deformation behavior and can accurately tracks the experimental data in a wide temperature range and strain rate range. In addition, the predicted data outside of experimental conditions were obtained, indicating good prediction potentiality of the developed ANN model. The θ -σ curves outside of experimental conditions indicate that the predicted strain-stress curves exhibit a typical dynamic recrystallization softening characteristic of high temperature deformation behavior. Through the coupling of the ANN model and finite element model, the hot compression simulations at the temperature of 1273 K and strain rates of 0.01~10 s-1 were conducted. The results show that the predicted constitutive data outside the experimental conditions successfully improved the prediction accuracy of forming load during the FEM simulation.