基于Elman神经网络的炉卷轧机的轧制力预报  

Rolling Force Prediction of Steckel Mill Based on Elman Neural Network

作　　者：韩庆 周石光 

机构地区：[1]中国钢研科技集团公司新冶电气股份有限公司,北京100081

出　　处：《钢铁》2009年第6期56-59,共4页Iron and Steel

摘　　要：为了改善国内某钢铁厂炉卷轧机的轧制力模型的预报精度,提出将结合热模拟实验建立的传统轧制力模型计算值作为Elman神经网络的一个输入项,将传统数学模型预报的轧制力与实测轧制力的相对误差作为此神经网络输出项的方式构建网络模型,通过大量的在线数据分析,这种将神经网络与传统数学模型相结合的方法明显地改善了轧制力的预报精度。该神经网络模型可为以轧制力为主要控制目标的炉卷轧机的过程自动化系统提供可靠的模型参数。In order to improve the rolling force prediction accuracy of steckel mill, the rolling force predicted by traditional rolling force model which is associated with thermal simulation is as an input of Elman neural network. The predicted roiling force error between traditional math model and measured value is as an output and then we establish the network model. Through plenty of on-line data, this method which associate neural network with traditional math model greatly improves rolling force prediction. This neural network could be the reliable model parameter for steckel mill automation system control according to the roiling force.

关 键 词：炉卷轧机 热模拟 轧制力 ELMAN神经网络 

分 类 号：TG333[金属学及工艺—金属压力加工]