The Study Of Predictive Model Of Silicon Content Of Hot Metal In Blast Furnace

The silicon content of hot metal is not only an important symbol to estimate the quality of product, but also reflect the energy consumption of the blast furnace. Therefore, it is available to steady thermal system, reduce the fluctuation of blast furnace performance, decrease the silicon content of hot metal and improve the quality of raw iron if the silicon content and its change trend can be measured in time, it is convenience to take the adjustment measure. To develop a real-time online prediction system of silicon content of hot metal is very important to direct blast furnace operation and improve the level of blast furnace process control. In the course of systematic modeling, the artificial neural networks method is studied and three-layer former network structure is confirmed. In allusion to the defect of grads descension of traditional back propagation network algorithms, some improving measures such as adaptive learning and additive momentum has been taken, and better application result is acquired.The prediction system of silicon content of hot metal is realized by adopting the mix the programming strategy of MATLAB and VC++ and using ADO database technology. The software is developed on base of the production data of XiangTan Iron and Steel Plant, the prediction rate is over 85%, and it lays the foundations of online control of silicon content of hot metal.