Hybrid multistep modeling for calculation of carbon efficiency of iron ore sintering process based on yield prediction

Abstract

Iron ore sintering is the second most energy-consuming process in steelmaking. The main source of energy for it is the combustion of carbon. To find ways of reducing the energy consumption, it is necessary to predict the carbon efficiency. In this study, the comprehensive carbon ratio (CCR) was taken to be a measure of carbon efficiency, and a hybrid multistep model (HMSM) was built to calculate it. First, the sintering process was analyzed, and the key characteristics of the process parameters were extracted. Next, an HMSM that combines mechanism modeling, data-driven modeling, and integrated modeling was constructed based on the characteristics of the process parameters. The model has three levels: the prediction of key state parameters, yield prediction, and mechanism modeling. First, an integrated fuzzy predictive model predicts the key state parameters based on the evaluation of current operating conditions. Next, predicted values of the state parameters along with key material parameters are used as inputs for a particle swarm optimization-based backpropagation neural network predictive model that predicts the yield. Finally, the predicted yield is fed into the mechanism model, which calculates the CCR. Mechanism and data correlation analyses were used to determine the most appropriate inputs for the three levels. Model verification using actual process data showed that the HMSM accurately predicted the CCR. More specifically, the relative error was in the range (0 %, 2 %] for 91 % of the test samples, and the maximum error was only 5 %. This model lays the groundwork for increasing the carbon efficiency of iron ore sintering.