Ti-25Nb合金BP神经网络本构模型的建立

欧阳旭; 王克鲁; 吴俊慷; 彭辉权

doi:10.15980/j.tzzz.2021.04.007

研究·计算机应用 | 浏览量 : 0 下载量: 5 CSCD: 0

PDF
导出
分享
收藏
专辑

Ti-25Nb合金BP神经网络本构模型的建立
Construction of BP Neural Network Constitutive Model for Ti-25Nb Alloy
2021年41卷第4期页码：425-428
纸质出版日期： 2021 ，
DOI： 10.15980/j.tzzz.2021.04.007

扫描看全文

[1]欧阳旭,王克鲁,吴俊慷,彭辉权.Ti-25Nb合金BP神经网络本构模型的建立[J].特种铸造及有色合金,2021,41(04):425-428.

Ouyang Xu, Wang Kexu, Wu Junkang, et al. Construction of BP Neural Network Constitutive Model for Ti-25Nb Alloy[J]. Special Casting & Nonferrous Alloys, 2021,41(4):425-428.
[1]欧阳旭,王克鲁,吴俊慷,彭辉权.Ti-25Nb合金BP神经网络本构模型的建立[J].特种铸造及有色合金,2021,41(04):425-428. DOI： 10.15980/j.tzzz.2021.04.007.

Ouyang Xu, Wang Kexu, Wu Junkang, et al. Construction of BP Neural Network Constitutive Model for Ti-25Nb Alloy[J]. Special Casting & Nonferrous Alloys, 2021,41(4):425-428. DOI： 10.15980/j.tzzz.2021.04.007.

摘要

采用Gleeble-3500型热模拟试验机对Ti-25Nb合金进行等温恒应变速率压缩试验

研究合金在470～620℃

0.001～10s

-1

条件下的流动应力行为

并基于BP神经网络建立合金的本构模型。结果表明

Ti-25Nb合金的真应力具有负温度、正应变速率敏感性

合金在试验变形条件下均呈现流动软化型特征。基于BP神经网络所建立的本构模型计算精度较高

其相关系数为0.999 6

相对误差为4.27%

平均绝对相对误差为1.33%

能够较好地描述并预测Ti-25Nb合金的流动应力行为。

Abstract

The isothermal constant strain rate compression test was carried out by a Gleeble-3500 thermal simulation tester.The flow stress behavior of Ti-25Nb alloy at 470～620℃and 0.001～10 s

-1

was investigated and constitutive model was established based on BP neural network.The results reveal that the true stress of Ti-25Nb alloy has negative temperature and positive strain rate sensibility

showing the characteristics of flow softening under any deformation condition.The constitutive model based on BP neural network has a high calculation accuracy with correlation coefficient of 0.999 6

relative error of 4.27% and average absolute relative error of 1.33%

which can well describe and predict the flow stress behavior of Ti-25Nb alloy.