Prediction model of blast furnace hearth activity based on theoretical analysis and intelligent algorithm
Prolonging the service life,ensuring stable running condition and reducing energy consumption of blast furnace have become the main direction of modern blast furnace production development.In order to achieve these goals,improving the activity of blast furnace hearth is considered to be one of the most important measures.The ac-tivity of the hearth can reveal the chemical reaction and thermodynamic conditions inside the blast furnace and help to understand the basic mechanism of the blast furnace smelting process,and reasonable evaluation of hearth activity is of great significance for guiding blast furnace production.Therefore,firstly,based on the process data collected from the blast furnace production site,the activity index of the blast furnace hearth was obtained after pretreatment and calculation.Then,in order to further improve the prediction accuracy,the feature selection method and redun-dancy analysis were adopted to select the most influential parameter from many parameters as the input parameter.Finally,the Bayesian optimization XGBoost model was used to predict the hearth activity by regression,and the re-gression effect of XGBoost model with default hyperparameters and random forest model was compared.The results show that the Bayesian optimization XGBoost model has excellent performance in predicting the hearth activity,the model has good generalization and nonlinear fitting ability,and the prediction effect is good.The research results provide a strong decision basis for blast furnace production,which can help optimize operating parameters,improve smelting efficiency and reduce energy consumption.
NETL
NSTL
万方数据
