Research On Rolling Force's Forecast Pattern Of Hot Rolling Mill Based On Neural Network

The quality of product in hot rolling and the security of the rolling set are highly due to the quality of hot rolling, and the of quality hot rolling also influence the consumed energy of product source. Consequently, how to the forecast the rolling force by using the temperature of bullet when leaving furnace, has play a important role in prolonging the life of rolling machine and reducing the consumed energy of the whole process.Because most of the existing models of rolling force are deduced according to a series of simplification and predigesting which have many inaccuracy factors, traditional models can't give us satisfied results.The technology of neural networks has been provided as a effective tool. But the method of modeling of rolling force, which based on neural network, is still in the phase of research in the world. In this paper, a neural network model which is used for prediction of rolling force is set-up based on the data of the hot rolling mill in Steel Corp. On the way of integrated control for steel's reheating process, the work use the optimum temperature when steel leave the stove, build a BP network model of rolling force's prediction.The model gives two BP algorithms as traditional BP network and improved BP network, in order to get better iterative speed and high precision, the work optimize the unit of couch, the beginning value, and the study speed. Comparing with the two algorithms, the result of this improving model greatly approaches to the set values of rolling schedule. The model can predict the rolling force with high precision and is capable of replacing the former model used to calculate the rolling schedule. It shows that the model can be used in calculation of engineering application. By the given model, the work gives a scheme used for online control, finally.