基于Arrhenius方程和BP神经网络的2024Al/Al18B4O33w复合材料热变形流变应力预测

Prediction on hot deformation flow stress of 2024Al/Al18B4O33w composites based on Arrhenius equation and BP neural network

在350~500℃和应变速率0.01~10 s-1条件下对2024Al/Al18B4O33w复合材料进行等温压缩实验。分析复合材料流变应力曲线,基于应变补偿型Arrhenius方程和BP神经网络模型分别预测其流变应力,通过数据误差分析评估两种模型的精度。通过BP神经网络预测的流变应力数据,建立基于动态材料模型的热加工图,并结合微观组织验证热加工图的准确性。结果表明:BP神经网络模型较应变补偿型Arrhenius方程更能准确地预测2024Al/Al18B4O33w复合材料的流变应力。热加工图预测复合材料热变形合适的工艺参数区域为440~500℃,0.01~0.13 s-1。

Isothermal compression test of 2024Al/Al18B4O33w composites was conducted at 350-500℃and strain rates of 0.01-10 s-1.The flow stress curves of composites were analyzed.The flow stress were predicted based on strain compensated Arrhenius equation and BP neural network model.The accuracies of two models were evaluated by data error analysis.Based on the data of the flow stress predicted by BP neural network,the hot processing map based on dynamic material model was established.The accuracy of hot processing map was verified by the microstructure.The results show that the prediction on flow stress of 2024Al/Al18B4O33w composites based on BP neural network model is more accurate than that based on Arrhenius equation.The optimal processing region in hot deformation of 2024Al/Al18B4O33w composites based on hot processing map were 440-500℃and 0.01-0.13 s-1.