AerMet100超高强钢热变形行为及3种流变应力模型对比

Hot deformation behavior and comparison of three rheological stress models on AerMet100 ultra-high strength steel

对Aer Met100超高强钢试样进行了变形温度为1173~1473K、应变速率为0.01~10s^(-1)、变形量为60%的热压缩实验。结果表明:随着应变的增加,Aer Met100超高强钢试样的真应力先迅速增大,然后增长速率减小,直至趋于动态平稳。在实验的变形条件内,真应力-真应变曲线呈现出动态回复型与动态再结晶型这2种曲线形式。基于实验结果,分别构建了应变补偿型Arrhenius本构模型、优化型Johnson-Cook模型和BP神经网络模型,分析对比了3种模型对Aer Met100超高强钢高温变形行为的预测精度,得到3种模型的线性相关系数R分别为0.99461、0.98694和0.99998;平均相对误差绝对值AARE分别为3.029%、5.220%和0.129%。其中,BP神经网络模型预测的流变应力线性相关强度最高,模型预测精度最高。

The hot compression tests of AerMet100 ultra-high strength steel samples were carried out at the deformation temperature of 1173-1473 K,the strain rate of 0.01-10 s^(-1) and the deformation amount of 60%.The results show that with the increasing of strain,the true stress of AerMet100 ultra-high strength steel samples first rapidly increases,and then the growth rate decreases until it tends to be dy-namically stable.Under the deformation conditions of the test,the true stress-true strain curve exhibits two kinds of curve types wit dy-namic recovery type and dynamic recrystallization type.Based on the test results,the strain-compensated Arrhenius constitutive model,the optimized Johnson-Cook model and the BP neural network model are constructed respectively,and the prediction accuracy of the three models on the high temperature deformation behavior of AerMet100 ultra-high strength steel were analyzed and compared.The linear corre-lation coefficients R of the three models are 0.99461,0.98694 and 0.99998,respectively,and the average relative error absolute values AARE are 3.029%,5.220%and 0.129%,respectively.Among them,the BP neural network model predicts the highest linear correla-tion strength of rheological stress and the highest prediction accuracy.