Intelligent load distribution of hot rolling based on data driving

Load distribution of assembling unit in hot rolling is to determine reasonable rolling force between every machine rack, which makes thickness and shape of steel appropriate. This paper gives a method based on double-layer BP neural network, and readjusts the form of activation function Sigmoid, moreover, this new method is applied in forecast of rolling force between seven machine racks in some steel mill. Experiment results show that this method not only increases learning speed but improves generalization ability in learning process. Method and its application given in this paper provide a new way for load scheduling of machine racks.