摘要
采用Gleeble-1500型热/力试验机对25CrMo4钢进行了等温恒应变速率压缩试验,研究了其在变形温度1050~1150℃,应变速率0.1~10 s^(-1)范围内的热变形行为,通过试验确定了层数为3×12×1的BP神经网络结构形式的本构关系模型,并在BP神经网络本构模型的基础上采用鲸鱼优化算法构建了WOA-BP神经网络本构模型。结果表明,25CrMo4钢的流变应力对变形温度和应变速率较为敏感,降低变形温度和增大应变速率均可以提升流变应力。在高温和低应变速率条件下,流变曲线大多呈现稳态流动特征。经过计算预测值与试验值的误差得出,WOA-BP神经网络本构模型的相关系数和平均相对误差分别为0.99927和0.8915%;采用BP神经网络建立的本构模型的相关系数和平均相对误差分别为0.99677和2.1764%。WOA-BP神经网络本构模型具有更高的精度,能更加准确地预测25CrMo4钢的高温流变应力。
Gleeble-1500 thermal/mechanical testing machine was used to carry out the isothermal compression tests with constant strain rate of 25CrMo4 steel,and its thermal deformation behaviors at the deformation temperature of 1050-1150℃ and strain rate of 0.1-10 s^(-1) were studied.The constitutive relationship model of BP neural network structure with layer number of 3×12×1 was determined through tests,and the whale optimization algorithm(WOA)was adopted on the basis of BP neural network constitutive model to construct the WOA-BP neural network constitutive model.The results show that the flow stress of 25CrMo4 steel is sensitive to the deformation temperature and strain rate,and the flow stress can be increased by decreasing the deformation temperature and increasing the strain rate.At high temperatures and low strain rates,the flow curves mostly exhibit steady-state flow characteristics.After calculating the error between predicted values and tested values,it is obtained that the correlation coefficient and average relative error of the WOA-BP neural network constitutive model are 0.99927 and 0.8915%,respectively.The correlation coefficient and average relative error of constitutive model established by BP neural network are 0.99677 and 2.1764%,respectively.The WOA-BP neural network constitutive model has higher accuracy and can predict the high-temperature flow stress of 25CrMo4 steel more accurately.
作者
张习康
江洋
王迪
李冠锋
ZHANG Xi-kang;JIANG Yang;WANG Di;LI Guan-feng(Transportation Institute,Inner Mongolia University,Hohhot 010070,China)
出处
《塑性工程学报》
CAS
CSCD
北大核心
2023年第8期182-187,共6页
Journal of Plasticity Engineering
基金
国家自然科学基金资助项目(52161022)
内蒙古自然科学基金资助项目(2022MS05051)。