PREDICTION OF BLAST FURNACE IRON OUTPUT BASED ON SUPPORT VECTOR REGRESSION
Blast furnace ironmaking is an important link in steel production.Based on the amount of iron water flowing out of the blast furnace,selecting a specified number of iron bags for loading and transportation can improve production efficiency and reduce the overall consumption of the scheduling line.Accurately predicting the amount of iron is of great significance for subsequent production scheduling.On the one hand,blast furnace ironmaking involves a large number of physical and chemical reactions and parameter changes,and the ironmaking process cannot be observed in real-time from the outside,making it difficult to achieve accurate automatic control through direct mechanism analysis;On the other hand,the rich measurement data of blast parameters,coke ratio,slag composition and other parameters recorded during the ironmaking process can be used for data-driven modeling and analysis.This article aims to analyze the ideal flow rate of molten iron through a mechanistic model and design a machine learning model based on support vector regression to predict the iron output of blast furnaces.The experimental results of modeling and analyzing the iron production data of a blast furnace with a daily output of 8000 tons show that the average error of the support vector regression model in predicting the iron production is within 200 tons,and the average error,prediction standard deviation and other indicators are better than other common machine learning models,demonstrating the accuracy of the data-driven model.It can provide guidance for actual blast furnace ironmaking analysis and modeling,thereby reducing resource consumption and improving the overall production efficiency of the steel production line.
万方数据
维普
