The Predicted System Of Si Content In Molten Iron Based On GA-BP Network

In the process of puddling, holding reasonable temperature of blast furnace is realize the produce the direct guarantee that the blast furnace can work steadily to aim—high quality, high yield, long life, low cost. With the complexity of process of the blast furnace, the content of Si in the molten iron is used to reflect the change of temperature. In the natural condition, the Si content of molten iron becomes direct ratio with the temperature. In the process of puddling, if the Si content of molten iron and its direction of movement can be predominated duly and estimated accurately, it is advantaged to stabilize heat system, reduce the condition fluctuate, depress the Si content of molten iron and advance iron quality.Neural network technique is active in the development of artificial intelligence, which has the ability of self-adaption , self-study, self-form, associate and memory, parallel calculate. Especially it is the same with describing uncertain estimation, identification and classification, which should take into account many factors at one time. But its learn speed is slow and it may constringe local extremum point. So this paper combines GA with BP to establish the model. The model makes use of GA to modify the power value of network to construct anagenesis model, shorten time of practice and study, improve the forecast precision of model.This paper uses GA-BP model to construct three-layer NN model. When the error of the Si content is 0.05%, the system shoots straight to 90%. The result shows: GA-BP model can gain higher precision than traditional BP network model.