基于人工神经网络的Ti-555211合金热变形本构模型研究(英文)  

Modeling Constitutive Relationship of Ti-555211 Alloy by Artificial Neural Network during High-Temperature Deformation

作　　者：安震[1] 李金山[1] 冯勇[1,2] 刘向宏[2] 杜予晅[2] 马凡蛟[2] 王哲[3] 

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 [3]钢铁研究总院,北京100081

出　　处：《稀有金属材料与工程》2015年第1期62-66,共5页Rare Metal Materials and Engineering

基　　金：Project of Introducing Talents of Discipline to Universities("111"Project)(B08040)

摘　　要：利用Gleeble-3800热模拟实验机,在应变速率0.001-1 s^-1以及变形温度750-950℃范围内对Ti-555211合金进行等温恒应变速率压缩实验。基于人工神经网络的方法建立了Ti-555211合金热变形本构模型。模型的可靠性用平均相对误差和相关系数来确定。结果表明,所建立的本构模型与实验值的平均相对误差为1.60%,相关系数为0.99938,表明该模型能很好地预测该合金的本构关系。用神经网络来确定本构关系比传统的数学方程更加具有优势。热模拟实验结果表明,随着变形温度的升高和应变速率的减小,该材料的峰值应力有所减小,不连续屈服现象随着变形温度升高和应变速率的增大变得更加明显。流变曲线在不同的变形参数条件下表现形式也不同。Using experimental data gained from hot compression tests in the temperature range of 750-950 °C and strain rate range of 0.001-1 s^-1, the constitutive relationship of Ti-555211 titanium alloy was investigated based on the back propagation artificial neural network constitutive model（ANN model）. The capability of the model was measured by the average absolute relative error（AARE）, and correlation coefficient（R）. The simulated values were compared with experimental values. The results show that the R and AARE for the ANN model are 0.99938 and 1.60%, respectively, indicating that the hot deformation behavior of Ti-555211 titanium alloy can be predicted by the ANN model efficiently and accurately. Furthermore, the back propagation artificial neural network model is a more efficient quantitative way to predict the deformation behavior of the Ti-555211 titanium alloy compared to the mathematical equation. The results show that the peak stress of the alloy decreases with increasing of temperature and decreasing of strain rate, and the phenomenon of discontinuous yielding is more obvious with the increase of deformation temperature and strain rate. The flow curve characteristics under different deformation parameters show obvious differences.

关 键 词：Ti-555211合金 本构关系 人工神经网络 

分 类 号：TP183[自动化与计算机技术—控制理论与控制工程] TG146.23[自动化与计算机技术—控制科学与工程]