Modeling And Optimization Of The Blast Furnace Iron Temperature Based On Data Mining

Blast furnace temperature guarantees anterograde run for the blast furnace, it alsorestricts a main factor of the blast furnace smelting cost. Reasonable blast furnacetemperature is the important indexes that blast furnace running smoothly. At present, theblast furnace iron silicon content is mainly used to predict the furnace temperature. But itappears positively relation rather than a strict linear relationship between the blast furnacetemperature and the iron silicon content in blast furnace, and then the application of blastfurnace hot metal silicon content for the single index prediction of blast furnace temperaturehas certain defects.With the rapid development of the information and technology, the blast furnacesmelting process accumulated a large amount of data, how to look for low carbon, highefficiency and high-quality production of blast furnace model is imminent. It is imminentthat how to look for low carbon, high efficiency and high quality of the blast furnaceproduction mode. At the same time, the further research of intelligent control theory, whichprovides the theoretical basis and method support for the research of the blast furnacetemperature prediction model established based on data mining technology. Under the datamining technology, through the blast furnace temperature data feature extraction to establishimplicit mathematical relationship among different variables, it has caused highly attentionand becomes the frontier of metallurgy science development today.In view of the huge amounts of data in the blast furnace smelting process, theinaccuracy of blast furnace iron silicon content forecasting furnace temperature hot metalsilicon content in blast furnace smelting exist certain disadvantages as to the object of studyand the subjectivity of the smelting mechanism modeling, the paper uses the data miningtheory to establish the statistical model based on the blast furnace iron temperatureforecasting. In view of the big inertia and nonlinear system in the blast furnace smeltingprocess, and then consider the effect of the time lag to the blast furnace iron temperaturemodeling. It uses the multivariate time-series to establish the multivariable time series modelof the iron temperature, and through the variable time series analysis revealed that the development of blast furnace temperature change rule. And through the variable time seriesanalysis reveals the development of the blast furnace temperature change rule.Blast furnace temperature prediction model is established according to a specificenvironment conditions. Namely each prediction optimization model of the blast furnacetemperature is established at the special constraint conditions, the model cannot directly orgm to other blast furnace temperature prediction. This article conducts the statisticalmodeling and optimization in the same import library conditions such as the same baiyunobo ore composition, load, speeding and air volume, etc relatively consistent case for thefurnace temperature data of a certain steel mill large. The research of the blast furnace irontemperature forecasting model based on data mining blast provides the important theoreticalguidance significance for furnace temperature accurately forecast.This paper firstly introduces the multivariate mathematical modeling and basic theoryof fuzzy mathematics modeling. Then the input and output variables of the blast furnacetemperature prediction model are analyzed by the pretreatment of the data. It mainly includesoutlier inspection, defect value supplement, data correlation analysis and the each variable ofthe multivariate time series model determined the time delay. Each variable lag step of themultivariate time series model is determined by the generalized partial autocorrelationcoefficient method. Then the intelligent algorithm is applied to the multiple regression modeland multivariate time series model. Among them, the two models all adopt the multi-inputsingle output system. Finally, the application variable data by blast furnace smelting onlinedetection system collection completes the modeling and simulation check content. Throughthe error figure，forecast figure performance and evaluation indicators of each model, whichanalysis comparatively the advantages and disadvantages of each algorithm and theapplicable scope of different models. The conclusion: the multivariate time based on t-sfuzzy neural network whose tracking result and model accuracy is optimal.This article forms the blast furnace operation optimization control system based on datadriven, and analyses the rich modeling and optimization theory based on the data mining.The research of the blast furnace iron temperature forecasting optimal model provides a goodtheoretical basis for the cost accounting and the operation of blast furnace foreman control inthe blast furnace smelting process. The continuously exploration of the blast furnace temperature prediction optimization model plays an important role in saving energy andreducing consumption for blast furnace of metallurgical industry, and also reduces the degreeof Chinese steel industry dependence on imported iron ore.