Process metallurgy and data driven prediction and feedback of blast furnace furnace heat indicators

The prediction and control of furnace heat indicator was of great significance to improve the furnace hot level and furnace condition for the complex and difficult to operate the hour-class delay blast furnace (BF) system. In this work, a prediction and feedback model of furnace heat indicator based on the fusion of data driven and BF ironmaking process was proposed. The data of raw and fuel materials, process operation, smelting state and slag and iron discharge during the whole BF process were comprehensively analyzed, a total of 171 variables, 9223 groups of data. A novel method of delay analysis of furnace heat indicator was established. and the extracted delay variables had played an important role in the modeling. Compared with the traditional machine learning algorithm, the method that combined the Genetic algorithm (GA) and Stacking was efficient to improve the performance. The hit rate for the predicting the temperature of hot metal in the error range of plus or minus 10 ℃ was 92.4%, and that for [Si] in the error range of plus or minus 0.1% was 93.3%. On the basis of the furnace heat prediction model and the expert experience, a feedback model of furnace heat operation was established to push the quantitative operation suggestions to stabilize the BF heat level, which had been highly recognized by the BF operators. Finally, this comprehensive and dynamic model had been successfully applied in the practical BF, and the BF temperature level was improved remarkably with the furnace temperature stability rate increasing from 54.88% to 84.89%, which had achieved significant economic benefits.