人工神经网络对管材张减精度预测

无缝钢管张力减径是多机架机组对荒管进行无芯棒连轧过程。它具有复杂的金属流动状态，影响其精度的工艺数较多，难以用有轧制理论进行分析。应用人工神经网络的ＢＰ算法，对不同坏料样本进行训练学习，确定其权值和阈值，建立起预测管材精度的数学模型。经试验证明该模型具有较高的可靠性和精度。

PREDICTION OF ACCURACY OF STRETCH REDUCTION BY ARTIFICIAL NEURAL NETWORKS

Stretch reduction of seamless tubes is a continuous rolling process for blank tubes without mandral It is a complicated process of metal flows and influenced by many technical parameters, so is difficult to be analyzed This paper describes the prediction of accuracy of stretch reduction with use of the back propagation algorithm of neural networks Connection weights and thresholds are determined through learning and modifying each samples It is proved that the models established have certain reliability and precision