SA765Gr.Ⅱ合金钢热拉伸本构模型的参数反求

Hot tensile constitutive model parameters reverse engineering of SA765Gr.II alloy steel

在温度为950~1150℃,应变速率为0.01~5 s^(-1)的条件下,使用Gleeble-1500D热模拟实验机对SA765Gr.Ⅱ合金钢进行了等温热拉伸实验以研究其热拉伸变形行为。首先通过线性回归方法推导了SA765Gr.Ⅱ合金钢的Norton-Hoff模型参数,之后提出了一种基于自适应模拟退火(ASA)算法求解本构模型参数的方法(反求方法)。结果表明:相比于回归方法,反求方法构建的模型预测相关系数R从0.9831提高到0.9958、均方根误差RMAE由6.392降低至3.603、平均相对误差AARE由5.38%降低至3.69%。线性回归方法构建的模型预测误差期望与标准偏差分别为0.97和8.76,反求方法构建的模型预测误差期望与标准偏差分别为0.13和5.14。通过反求方法构建的Norton-Hoff模型预测精度得到了提高。

Under the conditions of temperature range of 950℃to 1150℃and strain rate range of 0.01 s^(-1)to 5 s^(-1),isothermal tensile experiments were conducted on SA765Gr.Ⅱalloy steel by the Gleeble-1500D thermal simulation machine to investigate its hot tensile deformation behavior.Firstly,the Norton-Hoff model parameters of the SA765Gr.II alloy steel were derived using linear regression method.Then,a method based on adaptive simulated annealing(ASA)algorithm was proposed to solve the constitutive model parameters(reverse method).The results show that compared to the regression method,the correlation coefficient R of the model constructed by the reverse engineering method increases from 0.9831 to 0.9958,the root mean square error RMAE decreases from 6.392 to 3.603,and the average relative error AARE decreases from 5.38%to 3.69%.The expected prediction error and standard deviation of the model constructed by linear regression method are 0.97 and 8.76,respectively,while the expected prediction error and standard deviation of the model constructed by reverse engineering method are 0.13 and 5.14,respectively.The prediction accuracy of the Norton-Hoff model constructed through reverse engineering method has been improved.