【摘要】 为提高GH4706合金热加工数值模拟精度，在变形温度为900～1150℃和应变速率为0.01～5 s^-1的条件下，采用等温压缩实验研究了铸态GH4706合金的热变形行为，构建了其应变补偿Arrhenius模型、修正Zerilli-Armstrong（Z-A）模型和反向传播人工神经网络（BP-ANN）模型。结果表明：BP-ANN模型的预测精度最优，其相关系数（r²=0.99982）显著高于应变补偿Arrhenius模型（r²=0.97252）和修正Z-A模型（r²=0.96233）;基于动态材料模型理论，开发Python程序自动生成热加工图，确定铸态GH4706合金最佳热加工区间为：变形温度为1050～1120℃、应变速率为0.1～1 s^-1。更多还原

【Abstract】 To improve the accuracy of numerical simulation of hot processing of GH4706 alloy, the hot deformation behavior of as cast GH4706 alloy was studied using isothermal compression experiments at the conditions of deformation temperatures of 900-1150 ℃ and strain rates of 0.01-5 s^-1. The strain-compensated Arrhenius model, modified Zerilli-Armstrong（Z-A） model, and backpropagation artificial neural network（BP-ANN） model were constructed. The results show that the BP-ANN model has the best prediction accuracy, with a significantly higher correlation coefficient（r²=0.99982） than that of the strain-compensated Arrhenius model（r²=0.97252） and the modified Z-A model（r²=0.96233）. Based on the theory of dynamic material models, a Python program is developed to automatically generate hot processing maps. The optimal hot working range for the as cast GH4706 alloy is determined to be: deformation temperature of 1050-1120 ℃, strain rate of 0.1-1 s^-1.更多还原