Research On Terminal Temperature Prediction Model Of RH

RH is an important equipment in steel refining, especially in reducing carbon, gas and nonmetal impurities. RH refining follow-up process has strict requirements for its tapping temperature. Because of the vaccum characteristic of RH refining process, it is hard for us to get actual time parameters of steel temperature and we meet a challenge in precisely controling the terminal temperature of RH refining. Based on reasons above, this article carries out the research on terminal temperature prediction model of RH.Based on the RH smelting process and energy budget, this article determines the main factors influencing the terminal temperature of the molten steel. Through qualitative analysis and quantitative calculations, this article determines the degree of influence of each factor on the terminal temperature and builds the mechanistic model of terminal temperature of the molten steel. Mechanism model is simplification of the actual process, therefore, it is difficult to meet the requirements of the site.This artical introduces incremental model which is based on the principle of reproducibility and forecasts terminal temperature of molten steel using incremental computing. Choosing reference furnace based on similarity calculation and determing weights of reference furnace based on the calculated similarity values, we can get the current smelting furnace terminal target by linear weighted sum.RH smelting process is a complex physical and chemical process which is multivariable, nonlinear and strong coupling. Linear increment is bound to have a lot of errors and forecasting accuracy has not yet reached a satisfactory level. Support vector machine (SVM) is based on structural risk minimization principle and has good generalization performance. This article selects regression SVM and forecasts end temperature of molten steel using nonlinear incremental compensating target of reference furnace.Grid search is usually used in SVM parameter selection and it requires people to determine the range of parameters, step length changes and forecast accuracy is not necessarily ideal. In order to avoid the blindness of parameter selection, we use genetic algorithm to optimize the parameters of support vector machine to build incremental model based on GA-SVM compensation, forecasting end temperature of molten steel.Through simulation, incremental model based on GA-SVM compensation used in this paper improves predictive accuracy of terminal temperature of molten steel and can provide reliable guidance on precisely controling the terminal temperature.