Constitutive model of 3Cr23Ni8Mn3N heat-resistant steel based on back propagation (BP) neural network (NN)

Abstract

The 3Cr23Ni8Mn3N heat-resistant steel was subjected to isothermal constant strain rate compression experiments using a Gleeble - 1 500D thermal simulator. The thermal deformation behavior in the range of deformation temperature 1 000 - 1 180 °C and strain rate 0,01 - 10 s-1 was studied. Based on experimental data, the stress-strain curves of 3Cr23Ni8Mn3N were established. And the constitutive relation of BP neural network (3 × 10 × 10 × 1) was constructed. The flow stress was predicted and compared by the ANN constitutive model. The correlation coefficient (R) is 0,999, and the average relative error (AARE) is 0,697 %. The results show that the ANN constitutive model has high accuracy for predicting the thermal deformation behavior of 3Cr23Ni8Mn3N. The model can provide a good reference value for thermal processing.

Keywords

• 3Cr23Ni8Mn3N
• artificial neural network
• constitutive model
• heat - resistant
• stress - strain curve