Study On Temperature Prediction Model Of Blast Furnace Hot Metal Based On Data Preprocessing And Intelligent Optimization

Furnace temperature is stable and has a direct impact on blast furnace anterograde, so thetemperature is one of the most important indexes of blast furnace, the furnace temperature atthe same time also can indirectly reflect the cost of the blast furnace smelting. Usually, we usedto call molten iron temperature of furnace temperature, and the content of molten ironcontaining silicon to a great extent, can reflect the hot metal temperature. Many experts andscholars study found that the hot metal temperature and silicon content has a very closerelationship between them, but do not belong to strictly online sex, just rely on silicon contentalso to measure molten iron temperature obviously is not science, have to find a more scientificand reasonable way.Now in the period of rapid development of science and technology, measuring thetemperature of the molten iron, there are many ways so accurately measure the temperature ofthe molten iron is not a difficult. So that it can be measured by the molten iron temperature toestablish a corresponding database, through the data to choose a most suitable prediction model.Currently there are a lot of intelligence theories in academia, and the number is growing, theheight of the theory can be used to guide the blast furnace hot metal temperature forecastingmodel of intelligent optimization, make the construction of a model more reasonable. To builda reasonable hot metal temperature forecasting model, need data preprocessing of blast furnace,this is a basic prerequisite, this requires database has the characteristics of the summary andanalysis, and then establish the corresponding mathematical relationship. The basicmetallurgical industry are using this method, the future predicted molten iron temperature willstill choose this way.Blast furnace temperature prediction model building is a complex process, involving a lotof factors, each model is aimed at a specific environment and conditions, in this particularenvironment, the restriction factor is certain, so the model is only aimed at these constraints,the prediction of other blast furnace temperature can’t be pure to replicate this model, need tomake corresponding adjustment. Import library blast furnace condition of this paper isbasically consistent, such as Baiyunebo ore composition is the same, and is expected to speedand load conditions are more stable, Baotou6#blast furnace temperature data statistical modeling and optimization.Blast furnace temperature prediction model is firstly analyzed pretreatment of input, suchas outlier test and correlation analysis of the data, and so on. Then particle swarm optimization(pso) is adopted in the blast furnace stability forecast model of ant colony algorithm andextreme learning machine as well as the echo state network (ESN). Finally, through thevariable data for calibration modeling simulation of molten iron temperature, comparison andanalysis on the performance of the various models, and then a model of targeted to find out themost reasonable, use this model to guide the on-site production of blast furnace.This paper constructed a system of blast furnace temperature forecast, the system is basedon data preprocessing, the related model and the optimization theory is analyzed. Choose amost reasonable blast furnace hot metal temperature forecasting model, to achieve theaccounting and goals to reduce the cost of blast furnace smelting process, also can help theblast furnace foreman to strengthen the control of blast furnace smelting process. Explorationand optimization of the model has proved to be the blast furnace metallurgy industry to reduceenergy consumption, improve the utilization rate of iron ore, so that to a certain extent to cansolve the present situation of our country’s dependence on imported iron ore is too.