Prediction model of comprehensive coke ratio based on principal component analysis for sintering process

Sintering process is an important step in iron and steel metallurgy, and its product that is sinter is the main raw material for the blast furnace ironmaking. Low efficiency and big high energy consumption are two critic problems in sintering process. Improving carbon efficiency is a significant way to reduce energy consumption of the iron and steel industry as well as the emission of greenhouse gas. In this paper, chemical reactions and physical changes are analyzed clearly to get sintering mechanism, and the comprehensive coke ratio (CCR) is defined as an index measuring the carbon efficiency. By using principal component analysis (PCA) method, the principal components affecting CCR are generated, which serve as the input of back-propagation (BP) neural network model. And then the prediction model of CCR in sintering process is established. Comparing with the conventional BP neural network model, the simulation shows the prediction accuracy of our model is higher, and it can meet the requirements of actual production.