Hot Metal Silicon Content Prediction System Of Blast Furnace Based On The Stove Heat-index And BP Network

That Practice of the production showing, temperature higher or lower in the blast furnace ironmaking is harmful to the normal production of blast furnace. Only in a stable temperature conditions, It will be guaranteed that the iron furnace slag flow and the permeability of charging, stability of gas flow distribution, and the uniformity of the charge droppedin BF. These are blast furnace with a basic stable condition. It have a very strong correlation between the silicon content in the hot metal BF with the stability and Energy Consumption, The quality of hot metal of blast furnace smelting process. So the silicon content of hot metal used as a blast furnace temperature's signs.Thus this topic selects the silicon content of hot metal as a model of Blast furnace temperature prediction in Production of ironmaking for research.This paper uses BF production data of a large domestic steel companies as the background and the Silicon content in the molten iron as the main predicting basis.It establishes the static state model of [Si] prediction With BF material balance and the heat balance calculated heat index and the time series Offline forecasting model of hot metal silicon content with BP neural network. Both models combine to forecast a more effective temperature.The furnace heat index static model forecasting blast furnace temperature originated in the 1950s, the hitting rate generally is not high with these model predicted the temperature, because the parameters that model used are empirical data that it has some gap with the actual application BF.The paper uses the three furnace heat index of history Application better: Tc, Tf, Tq.It uses the mathematical model based on BF online Collection parameters to forecast Silicon content and temperature of hot metal and to ensure which furnace heat index has a better correlation and a higher hit rate.The static model is difficult to reflect the dynamics of the BF process, and dynamic data system's method is mainly based on linear consider. Actual blast furnace smelting process is a non-uniform, non-linear and the noise high-temperature process. Its dynamic process shows complex behavior. The neural network that it has a self-learning, self-organizing, adaptive and nonlinear dynamic handling characteristics and has good anti-noise capability and capacity of Lenovo is a mathematical model to imitate the human nervous system.This paper uses BP neural network to establish the prediction model of hot metal silicon content for time series prediction. Finally,this paper uses Matlab to establish the simulation model based on a furnace heat index forecasting and neural networks time series forecasting. After test, we find that the stove heat-index model has a good effect when the furnace conditions is stable and has obvious error when the furnace conditions is undulation. Neural Network time series forecasting model studies forecasting knowledge from the historical data of accumulation, and continually amends With the production of blast furnace.This model can reflect BF production process dynamic characteristics. The model is a promising methond to forecast hot metal silicon content,and has a better relationship between hot metal silicon content forecasted values with measured values. Taking the advantages of the two methodes at the same time, the content of silicon predicting in molten iron and the heat state of blast furnace are gained, which provide the fine condition to improve the quality of steel and stabilize manufacturing craftwork.