基于BP网络Ti-6Al-3Nb-2Zr-1Mo合金等温压缩流变应力预测  

Isothermal Compression Flow Stress Prediction of Ti-6Al-3Nb-2Zr-1Mo Alloy Based on BP-ANN

作　　者：陈海生[1,2] 冯勇[1,2] 马凡蛟[1,2] 毛友川 刘向宏[1,2] 张平祥[1,2] 寇宏超[1] 傅恒志[1] 

机构地区：[1]西北工业大学凝固技术国家重点实验室,陕西西安710072 [2]西部超导材料科技股份有限公司,陕西西安710018

出　　处：《稀有金属材料与工程》2016年第6期1549-1553,共5页Rare Metal Materials and Engineering

基　　金：国家国际科技合作专项(2013DFB50180);陕西省科技统筹创新工程计划项目(2015KTTSG01-08)

摘　　要：通过对Ti-6Al-3Nb-2Zr-1Mo合金820~970℃,0.001~1 s^(-1)条件下的热模拟压缩试验,得到不同变形条件下的高温变形真应力-真应变曲线。基于此实验数据建立了该合金BP-ANN本构预测模型和传统的回归模型。结果表明:2个模型的最大相对误差分别为4.35%和13.9%,平均绝对误差AARE分别为1.42%和6.53%,说明BP-ANN模型具有较优异的预测能力,此模型可作为Ti-6Al-3Nb-2Zr-1Mo钛合金高温变形本构模型。Elevated-temperature flow behavior of Ti-6Al-3Nb-2Zr-1Mo titanium alloy was investigated by isothermal hot compression tests at strain rate from 0.001 to 1 s^（-1） and in the temperature range of 820~970 °C on Gleeble-3800 simulator. The BP-ANN and regression models were established based on the experimental flow stress. Results show that the absolute value of maximum relative error obtained from the ANN model and the regression method are 4.35% and 13.9%, respectively. The average absolute relative errors are 1.42% and 6.53% corresponding to the ANN model and the regression method, respectively, which demonstrates that BP-ANN has a better prediction precision, and it can be used as the constitutive model of Ti-6Al-3Nb-2Zr-1Mo titanium alloy.

关 键 词：Ti-6Al-3Nb-2Zr-1Mo钛合金 热变形 BP-ANN 本构方程 

分 类 号：TG146.23[一般工业技术—材料科学与工程]