基于BP神经网络的轧机油膜厚度补偿的测试与建模

Measurement and modeling of rolling mill oil film thickness compensation based on BP neural network

针对单机架热轧中厚板轧机天铁2500mm中厚板生产线的油膜补偿问题进行了研究,用基于BP神经网络的方法建立了轧机油膜厚度补偿模型,并与已成熟应用的基于Reynolds方程的轧机相对油膜厚度补偿方法进行比较分析。结果表明,BP神经网络模型比基于Reynolds方程的轧机相对油膜厚度补偿方法具有预测精度高、样本学习时间快、收敛速度快的优点,BP神经网络模型可以对轧机油膜厚度进行良好的补偿。

The problem of oil film compensation in single-stand hot strip rolling mill was researched. The model of roll- ing mill oil film compensation based on BP neural network was established and compared with the method based on oil film thickness compensation in rolling mill Reynolds equation. The results show that BP neural network model has the advantages of high model forecast precision, quick sample learning and fast convergence rate compared to the method based on oil film compensation in rolling mill Reynolds equation. So the rolling mill oil film can be compensated efficiently by BP neural network model.