基于人工神经网络的一种板形反馈控制

A feedback control of sheet shape based on artificial neural network

针对影响板形的 6个主要因素，利用人工神经网络技术，通过对现场采集的大量数据样本的学习，建立了一种优化的 6－ 10－ 12－ 1结构 BP前馈型网络板形预测和控制模型。仿真与实验结果表明：该模型预测和控制板形精度的平均误差绝对值均在 5％之内，具有较好的效果，可应用于板形在线预测与控制。

With considering six major influence factors on sheet shape and the artificial neural network technology, an optimized 6－ 10－ 12－ 1 feed－ front BP network structure model for sheet shape prediction and control was obtained, after great deals of sample data were trained. The experimental results show that the absolute value of mean error of such prediction and control precision is within 5％ , which gets good effect on sheet shape control. The BP network model can be used in predicting and controlling sheet shape on line.