采用BP神经网络方法建立了镍基合金薄壁筒胀形回弹率预测的神经网络模型。模型的输入参数包括胀形温度、胀形压力、摩擦系数、薄壁筒与模具间隙等,模型的输出为薄壁筒中心截面平均回弹率。与传统回归拟合公式相比,该模型具有容错性好、通用性强等优点。该模型可以预测镍基合金薄壁筒在不同胀形工艺参数匹配下的回弹率,也可以用于优化胀形工艺参数。 A BP neural network method was used to establish the neural network model for predicting the bullet rebound rate of nickel-based alloy thin-walled tube. The input parameters of the model include bulging temperature, bulging pressure, friction coefficient, thin-walled tube and die gap, etc. The output of the model is the average rebound rate of the center section of the thin-walled tube. Compared with the traditional regression fitting formula, the model has the advantages of good fault tolerance and versatility. The model can predict the rebound rate of nickel-based alloy thin-walled tube under different bulging process parameters, and can also be used to optimize bulging process parameters.