摘要
对等通道转角挤压(ECAP)制备的超细晶纯钛,在温度为250～450℃、应变速率为10^(-5)～1s^(-1)的条件下进行热压缩试验。基于真应力和真应变实验数据,分别使用人工神经网络(ANN)和Arrhenius方程建立超细晶纯钛的热变形本构模型,研究其热变形行为。实验结果表明:在变形初期,流变应力随应变的增大而升高,随后趋于平缓,最终达到一个稳定值。人工神经网络训练和预测结果表明:人工神经网络最佳结构为3×12×1,人工神经网络模型预测的平均相对误差(AARE)为2.1%,相关系数(R)为0.9979;Arrhenius方程模型预测的AARE为11.54%,R为0.9464。即人工神经网络模型能够更加精确地描述超细晶纯钛的本构关系。通过对比不同温度下两种模型的误差,发现人工神经网络模型在高温条件下具有更好的稳定性。
Ultrafine grained（UFG） pure titanium was prepared by ECAP up to four passes. The hot compression experiments were conducted at different temperatures（250-450 °C） and strain rates（10-5-1 s-1. The artificial neural network（ANN） and Arrhenius constitutive equation were used for establishing a constitutive model of UFG pure titanium. Experiments show that the flow stress increases with the increase of strain at the beginning of deformation, and then increases slowly. Finally, the stress reaches a stable value. The experimental and predicted values of flow stress show that the average absolute relative errors obtained from the artificial neural network model and the Arrhenius constitutive equations are 2.1% and 11.54%, respectively. The correlation coefficients of the artificial neural network model and the Arrhenius constitutive equation are 0.9979 and 0.9464, respectively. This means that the artificial neural network model can more accurately describe the constitutive relations of UFG pure titanium. By comparing the errors of the two models at different temperatures, it can be concluded that artificial neural network model has better stability under the condition of high temperature.
作者
刘晓燕
杨成
杨西荣
强萌
张欠欠
Liu Xiaoyan1,2, Yang Cheng1, Yang Xirong1,2, Qiang Mengt, Zhang Qianqian1(1. Xi＇an University of Architecture and Technology, Xi＇an 710055, China;2. Metallurgical Engineering Technology Research Center of Shaanxi Province, Xi＇an 710055, China)
出处
《稀有金属材料与工程》
SCIE
EI
CAS
CSCD
北大核心
2018年第10期3038-3044,共7页
Rare Metal Materials and Engineering
基金
国家自然科学基金(51474170)
陕西省自然科学基金(2016JQ5026)
