神经网络辅助理论本构模型预测高熵合金高温流动应力行为

A Neural Network-Assisted Theoretical Constitutive Model to Predict the High Temperature Flow Behavior of High-Entropy Alloys

本文建立了一套确定双曲正弦Arrhenius型方程系数的神经网络模型,选取了高熵合金不同高温和应变速率下的流动应力预测来验证模型。首先采用Al_(0.3)CoCrFeNi高熵合金进行检验,并与传统方法进行比较,结果表明,神经网络方法在高应变速率和低温条件下得到的系数能够更好地描述试验热流应力。进一步采用均方根误差(RMSE)和相关系数(R)对模型结果和试验结果进行评估,神经网络方法在整体数据下的RMSE和R分别为27.7和0.985,优于传统方法的33.1和0.979。最后,利用该神经网络模型研究了其他高熵合金,如(CoCrNi)94Ti3Al3、FeCrCuNi2Mn2和AlCrCuFeNi的热变形行为,神经网络预测结果与试验结果吻合好,表明该神经网络模型具有较好的普遍适用性。

Metals and alloys are widely used in industry due to their excellent mechanical properties,and researchers are committed to finding new materials with better properties or mechanisms to enhance properties of materials.In the forming process of metal and alloy materials,the hot deformation can refine the grain effectively to improve the mechanical properties such as yield strength and tensile strength.Therefore,it is necessary to study the deformation behavior of metal and alloy materials at high temperature.Hyperbolic–sinusoidal Arrhenius-type model is widely used by researchers because of its good simulation effect at high temperatures.This paper studies the building process of the model and optimizes the modeling process with the help of neural network model.A neural network model is constructed to efficiently determine the hyperbolic–sinusoidal Arrhenius-type equations,based on which the flow stress of high-entropy alloys(HEAs)for different high temperatures and strain rates can be well predicted.In this study,the reported hot deformation behaviors of Al_(0.3)CoCrFeNi HEAs are examined by current model.The results show that the coefficients obtained by the neural network method can better describe the experimental hot flow stress,especially at high strain rate or low temperature conditions.The root mean square error(RMSE)and the correlation coefficient(R)are used to assess the degree of difference between the results.The RMSE and R of the neural network method at total data are 27.7 and 0.985,respectively,which are better than 33.1 and 0.979 of the traditional method.To show the general applicability of the model,the hot deformation behaviors of(CoCrNi)94Ti3Al3,FeCrCuNi2Mn2 and AlCrCuFeNi are analyzed by the model.The research work in this paper can improve the efficiency and accuracy of hyperbolic–sinusoidal Arrhenius-type model,reduce the difficulty of establishing the model,and has positive significance for the wide use of the model.