马氏体时效硬化不锈钢高温流变应力本构模型

Constitutive model for high-temperature flow stress of martensitic age hardening stainless steel

利用Gleeble 1500D热模拟试验机对15-5PH马氏体时效硬化不锈钢在变形温度为950～1150℃和应变速率为0. 01～10 s-1条件下进行了等温压缩实验。从获得的20组高温流变应力曲线中随机选取15组数据,分别建立了考虑应变参量的Arrhenius本构模型和人工神经网络模型,并通过这两种模型对剩余5组高温流变应力进行了预测。结果表明,两种模型都能准确反映15-5PH钢的高温动态软化规律。与考虑应变参量的Arrhenius本构模型相比,人工神经网络模型预测出的应力值与实验测量的应力值吻合度更好,其相关性系数和相对误差分别为0. 993和6. 63%。这说明所建立的人工神经网络模型可以很好的描述15-5PH不锈钢的流变应力,其预测值与实验值吻合较好。

Using Gleeble 1500 D thermal-mechanical simulator,the thermal compression tests of 15-5 PH martensitic age hardening stainless steel were carried out at the temperature range of 950-1150 ℃ and the strain rate range of 0. 01-10 s-1. The artificial neural network（ ANN） model and the Arrhenius constitutive model considering strain parameters were established based on the randomly selected fifteen sets of high-temperature flow stress curves from the twenty sets of ones,and the rest five sets of flow stress curves were predicted using the two models. The result shows that both the two models can accurately predict the dynamic softening behavior of 15-5 PH stainless steel at high temperatures. Compared with the Arrhenius constitutive model,the flow stresses predicted by ANN model better agree with the experimental ones. The correlation factor and the relative error of ANN model are 0. 993 and 6. 63%,respectively,which indicates that the proposed ANN model can accurately describe the flow stress of 15-5 PH stainless steel,a good agreement between predicted results and experimental ones can be obtained.