改进Zerilli-Armstrong、Arrhenius和GWO-BPNN模型对HAl61-4-3-1合金高温流变应力的预测  

Prediction of high temperature flow stress of HAl61-4-3-1 alloy by modified Zerilli-Armstrong, Arrhenius and GWO-BPNN models

作　　者：梁强 张贤明[1] 李平 李永亮 徐永航 LIANG Qiang;ZHANG Xian-ming;LI Ping;LI Yong-liang;XU Yong-hang(Engineering Research Center for Waste Oil Recovery Technology and Equipment,Ministry of Education,Chongqing Technology and Business University,Chongqing 400067,China;College of Mechanical Engineering,Chongqing Technology and Business University,Chongqing 400067,China)

机构地区：[1]重庆工商大学废油资源化技术与装备教育部工程中心,重庆400067 [2]重庆工商大学机械工程学院,重庆400067

出　　处：《材料热处理学报》2022年第9期193-204,共12页Transactions of Materials and Heat Treatment

基　　金：重庆市自然科学基金面上项目(cstc2020jcyj-msxmX0276);2021年度校内资助项目计划(2152026)。

摘　　要：精确的本构模型是铝黄铜合金热精锻工艺方案设计及数值模拟仿真的关键。基于挤压态HAl61-4-3-1合金在变形温度为873~1073 K、应变速率为0.01~10 s^(-1)区间的等温热压缩实验数据,分析热压缩过程中摩擦和温升效应对流变应力的影响并对流变应力曲线进行修正。基于修正数据,构建了HAl61-4-3-1合金的改进Zerilli-Armstrong、Arrhenius本构模型及GWO-BPNN本构模型。结果表明:GWO-BPNN模型的均方根误差RMSE和平均绝对百分比误差MAPE分别为0.444 MPa和1.078%,而改进Zerilli-Armstrong模型的分别为2.567 MPa和5.470%,改进Arrhenius模型的分别为1.202 MPa和3.163%,表明GWO-BPNN模型的预测精度更高。并且,通过引入GWO算法对BPNN的初始权值和阈值进行寻优,使得该模型具有较高的预测精度和较优的稳定性,能更好地描述HAl61-4-3-1合金的高温流变行为。Accurate constitutive model is the key to the process design and numerical simulation of hot precision forging of aluminum brass alloy.Based on the isothermal thermal compression experimental data of extruded HAl61-4-3-1 alloy at the deformation temperature of 873-1073 K and strain rate of 0.01-10 s^(-1),the effects of friction and temperature rise on flow stress during thermal compression were analyzed,and the flow stress curves were corrected.Based on the modified data,the modified Zerilli-Armstrong,Arrhenius constitutive models and GWO-BPNN constitutive model of the HAl61-4-3-1 alloy were constructed.The results show that the root mean square error RMSE and mean absolute percentage error MAPE of the GWO-BPNN model are 0.444 MPa and 1.078%respectively,while those of modified Zerilli-Armstrong model are 2.567 MPa and 5.470%respectively,and those of modified Arrhenius model are 1.202 MPa and 3.163%respectively,indicating that the prediction accuracy of the GWO-BPNN model is higher.Moreover,GWO algorithm is introduced to optimize the initial weight and threshold of BPNN,which makes the model have higher prediction accuracy and better stability,and can better describe the high-temperature flow behavior of the HAl61-4-3-1 alloy.

关 键 词：HAl61-4-3-1合金 流变应力 本构模型 BP神经网络 灰狼优化算法 

分 类 号：TG146.1[一般工业技术—材料科学与工程]