Prediction of Carbon Emission Concentrations in Blast Furnaces Based on Digital Twin

Abstract

Accurately predicting CO2 and CO emission concentrations in blast furnaces is essential for the sustainable development of the steel industry and achieving carbon neutrality. This article proposes a novel method for predicting blast furnace carbon emission concentrations based on digital twin. First, a gray-level co-occurrence matrix (GLCM) is used to select unobstructed flame portions from top thermal images, and features are extracted from these images using ResNet50 and an autoencoder. Independent component analysis (ICA) is then applied to extract non-Gaussian independent source features from the image data, which serve as input to a random forest (RF) model. Second, a mechanistic model is used to construct the temperature field within the blast furnace. Using ICA inverse transformation to reconstruct the important features of the images, and combining them with operational parameters and temperature field features to establish a feature-sharing matrix. Principal component analysis (PCA) is then used to extract principal components, which serve as input to an ensemble of random vector functional-link networks (RVFLNs). Finally, particle swarm optimization (PSO) is used to fuse the outputs. The results show that the proposed model achieves mean absolute percentage errors (MAPEs) of 0.8742% for CO2 and 0.8396% for CO concentrations, indicating high accuracy and practical application value.