Rolling Force Prediction of Steckel Mill Based on Elman Neural Network

In order to improve the steckel mill rolling force prediction accuracy, the rolling force predicted by traditional rolling force model which is associated with thermal simulation is an input of Elman neural network. At the same time, the predicted rolling force error between traditional math model and measured value is the output of neural network and then we begin to train the network. 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 rolling force.