摘要
采用单轴拉伸实验测试了不同温度下Mg-Gd-Y稀土镁合金的力学响应,表征了其塑性流动行为。结果表明,随着温度的升高,Mg-Gd-Y稀土镁合金呈现了不同程度的热软化效应,该合金的力学行为存在应变与温度的耦合效应。基于Johnson-Cook、Lim-Huh以及一种改进的多项式温度项模型,描述了Mg-Gd-Y稀土镁合金不同温度下的非线性塑性流动行为。系统地研究了激励函数、隐含层、优化率法和神经元数量等因素对人工神经网络模型拟合精度的影响,并比较了遗传算法和粒子群算法的优化效果。基于实验数据,综合比较了3种数学模型以及ANN的预测精度,发现ANN模型预测精度较高,可以准确描述温度对Mg-Gd-Y稀土镁合金力学行为的非线性影响规律。
The uniaxial tension experiment was used to test the mechanical response of Mg-Gd-Y rare earth magnesium alloy at different temperatures,and the plastic flow behavior was characterized.The results show that with the increase of temperature,the Mg-Gd-Y rare earth magnesium alloy presents thermal softening effect in different degree.The mechanical behavior of this alloy has a coupling effect of strain and temperature.Based on Johnson-Cook,Lim-Huh and an improved polynomial temperature term model,the nonlinear plastic flow behavior of Mg-Gd-Y rare earth magnesium alloy at different temperatures was described.The effect of excitation function,hidden layer,optimization algorithm and the number of neurons on the fitting accuracy of artificial neural network(ANN)model was studied systematically,and the optimization effect of genetic algorithm and particle swarm algorithm was compared.Based on the experimental data,the prediction accuracy of the three mathematical models and ANN was compared comprehensively.It is found that the ANN model has higher prediction accuracy and can accurately describe the nonlinear effect law of temperature on mechanical behavior of Mg-Gd-Y rare earth magnesium alloy.
作者
武鹏飞
肖旭彤
娄燕山
陈强
宁海青
WU Peng-fei;XIAO Xu-tong;LOU Yan-shan;CHEN Qiang;NING Hai-qing(School of Mechanical Engineering,Xi′an JiaoTong University,Xi′an 710049,China;Southwest Technology and Engineering Research Institute,Chongqing 400039,China)
出处
《塑性工程学报》
CAS
CSCD
北大核心
2022年第6期134-140,共7页
Journal of Plasticity Engineering
基金
国家自然科学基金资助项目(52075423,U2141214)
机械系统与振动国家重点实验室项目(MSV202009)
高性能复杂制造国家重点实验室项目(Kfkt2019-02)。
关键词
稀土镁合金
温度效应
塑性流动行为
本构方程
人工神经网络
rare earth magnesium alloy
temperature effect
plastic flow behavior
constitutive equation
artificial neural network
