Research On Prediction System Of Coke Quality Driven By Intelligent Model

In recent years,with the large-scale blast furnace and the development of ironmaking technology,iron and steel enterprises have higher requirements on the quality of coke.In order to meet this requirement,each coking enterprise has established a coke quality prediction model suitable for its own production demand,which can predict the coke quality index according to the coal quality index in a short time.However,as time goes on,especially when the coal quality index changes,the prediction of the model is often inaccurate and needs to be adjusted.Therefore,it is of great significance for the prediction of coke quality to establish a dynamic adaptive intelligent model based on the quality data of coal and coke.Therefore,an intelligent model for coke quality prediction is proposed in this paper.Before establishing the prediction model,the production data should be cleaned with outliers.Secondly,through correlation analysis or forward stepwise regression method,the coal quality indexes participating in the modeling are selected,and a static prediction model of coke quality is established according to the coal quality indexes.After the static prediction model has been predicted for a period,once the deviation between the predicted value and the actual value reaches the preset threshold value,the model adjustment module will be started immediately.The system will automatically select the coal quality data and coke quality data involved in the prediction,and generate the coke quality prediction model that meets the current requirements.Based on building the intelligent model,this paper adopts the Python language and Django framework and combines the quality data of coal and coke to design and implement the coke quality prediction system,including five functional modules: data display,outlier cleaning,data modeling,model prediction and model adjustment.The practice shows that the designed system can intelligently adjust the model according to the actual production data and meet the set expectation,which has certain practical guiding significance for the prediction of coke quality.