基于BP人工神经网络喷射成形7055铝合金的本构模型

基于Gleeble热力模拟技术对喷射成形7055铝合金的高温流变应力特征规律进行研究,并构建耦合应变量的唯象型Arrhenius本构方程用以预测合金的流变应力,同时基于BP人工神经网络构建该材料的神经网络型本构方程对比预测流变行为。结果表明:喷射成形7055铝合金的流变应力状况受变形参数的影响较为显著,与变形温度呈负相关,并与应变速率呈正相关。利用两类本构模型预测该合金的流变应力,其中唯象型Arrhenius本构方程的平均相对误差δ值大于2%,该模型的预测误差随变形温度升高呈上升趋势,且在热加工温度区间下(450℃左右),平均绝对误差及平均相对误差达到峰值,较难精准预测该变形区间内合金的流变应力特征。而BP人工神经网络模型的预测准确度更高,平均相对误差δ值仅为0.813%,且具有较高的温度稳定性。

Investigation of constitutive model of as-extruded spray-forming 7055 aluminum alloy based on BP artificial neural network

Gleeble physical simulation technique was employed to investigate the high-temperature flow stress characteristics of the studied spray forming 7055 aluminum alloy.Simultaneously,the Arrhenius constitutive model which couples the parameter of true strain and the BP artificial neural network constitutive model were contrastingly utilized to predict the flow stress behavior of the experimental alloy.The result shows that the flow stress of spray forming 7075 aluminum alloy is significantly affected by deformation parameter,which is negative correlated with deformation temperature and positively correlated with strain rate.Through the comparison of the two models,the average relative error of the Arrhenius constitutive model lies over 2%.And the error of the model tends to increase with the rising temperature.Moreover,the average absolute error and the average relative error reach the maximum at hot processing temperature(around 450℃).It is difficult to precisely predict the flow stress characteristics of the alloy.However,BP artificial neural network constitutive model has higher prediction accuracy,the average relative errorδvalue is only 0.813%and has higher temperature stability.