基于BP神经网络的AZ61镁合金流变性能的预测与评估

Prediction and Assessment of Rheological Properties of AZ61 Magnesium Alloy Based on BP Neural Network

采用热模拟机对AZ61镁合金进行等温压缩试验,研究变形温度和应变速率对流变应力的影响。结果表明:温度越低、应变速率越高,都使流变应力增大。根据实验数据,建立了BP神经网络(BP-ANN)模型。该模型预测值与实验值吻合良好,其相关系数达到0.999 6,平均误差为0.11%。

The isothermal compression testing of AZ61 magnesium alloy was conducted using thermal simulation machine, and the effect of deformation temperature and strain rate on flow stress was investigated. The results show that flow stress increase with decreasing deformation temperature or accelerate deformation. Moreover, a BP-ANN model was established according to the experiment results. The predicted values in the model were well agree with experimental values. The correlation coefficient of the both values is 0.999 6, and the average error is 0.11%.