Blast Furnace Semlting Process Optimization Of Pulverized Coal Injection Based On Data-driven

As a typical complex industry control process, the blast furnace smelting processhave with more variables,strong couplingbetween variables,nonlinear and delay for along time.The control of blast furnace smelting process is complex,is influenced bymany factors,this kind of control is random,and is not simply a parameter or a group ofparameter of linear systems or feedback control of nonlinear systems,The operation ofthe blast furnace smelting process depends on the experienced commander,Due to thelack of experience、complex furnace condition and lack of awareness to blast furnacesmelting process,which greatly,lead to large fluctuation of blast furnace smeltingprocess、large coke consumption、higher cost.Blast furnace ironmaking process without coke,to solve the problem of theshortage of coke and its higher cost,as an indispensable important adjust technicalmeasures in contemporary blast furnace iron making,that makes coal injection to replacepart of metallurgical coke strategy arises,it is important to coke saving、improving therationality of metallurgical coal resourceand improving economic efficiency ofenterprises.Topic at baotou steel features under the background of blast furnace smeltingprocess,based on data mining technology combining genetic optimized BP neuralnetworks to predict coal injection,on consulting a lot of related references at home andabroad on the basis of the specific research content is as follows:1.At the scene of the blast furnace data collected by missing values and outliers,due to the blast furnace is a big time lag system, affecting the furnace temperature andtechnological parameters of coal injection quantity most with time delay.the process ofdata from the field to the missing value directly repair,for outliers eliminated with alinear interpolation method to make up the time series.Take the correlation analysis ofvarious process parameters calculated correlation coefficient,were selected andtemperature with great influence on the coal injection quantity of parameters as theinput variables of the model,and considering the time lag of output variables.2.In this paper,the theory of ironmaking process and blast furnace expertexperience,aiming at the particularity of baiyun obo ore smelting,The optimization ofthe data selected,the using of screening optimization performance inherent in the global search using genetic algorithm to optimize BP neural network model of the weights andthresholds,respectively established based on genetic algorithm to optimize the BP neuralnetwork predictive model of blast furnace coal injection quantity optimization andprocess indicators(molten iron content of [Si] and charging coke rate) prediction model.To optimize the use of data that the above model can output according to the furnacecondition current blast furnace coal injection amount the best optimization Settings,andpredicted the development trend of the corresponding technical index.3.To predict coal injection volume ratio is not high and the predicted values followsex is bad,using support vector classification data,on the basis of blast furnacetemperature trend is divided into hot to cold,normal,to three kinds of data. Based onsupport vector machine (SVM) to cool the hot furnace temperature data classificationmodel,for the blast furnace field data collected by classification,and prediction of BPneural network was optimized by using genetic algorithm to optimize coal injectionquantity value,model of overall performance is greatly improved.Field data collected by data classification to build the classification model topredict coal injection,the results show that the prediction accuracy in data classificationof coal injection prediction based on obtained greatly improved,but also can judgeaccording to the blast furnace temperature situation coal injection amount, when thefurnace condition to hot appropriate decrease pulverized coal and furnace condition tothe amount of heat is increasing coal injection best coal injection quantity is givenfinally.Two hours every time increase and decrease coal injection can’t more than2tons.Practical application shows that this method can provide instructions to the siteoperating personnel,realize the purpose of the blast furnace anterograde stable,improvethe economic benefit.