Prediction of Hot Rolling Machine Running States Based on Neural Network

In continuous hot mill production lines, there are several rolling machines that are usually classified into rough rolling, finished rolling, and so on. To ensure the quality of steel products, the parameters for rolling force and the width/thickness control system should be set according to the rolling technology used. However, during actual production, such parameters often deviate from the set points due to various disturbances. It is therefore important to adjust such control system parameters dynamically whenever the system running states changes from normal area. This in turn requires that the system running states be predicted correctly. This paper makes a full analysis of the rolling states by applying data fusion methods based on neural network and database of the distributed data acquisition system. The results indicate that the prediction model is correct and provides an important reference to optimize farther the rolling parameters.