Pyrolysis kinetic analysis and model constructions of different ranks of coal and validation by GA–BP neural networks

Abstract

In the current steel industry, environmental pressures are mounting. If the reducing gases generated from coal pyrolysis can be utilized for reducing iron ore, the goal of low-carbon production with controllable costs can be achieved. However, the release of volatile content in coal has a significant impact on the pyrolysis process and final products. Therefore, in this study, the pyrolysis characteristics and gaseous products of four types of coal with different volatile matter content was investigated by thermogravimetry-mass spectrometry (TG-MS) technology. The results showed that coals with higher volatile matter content demonstrated stronger reactivity during pyrolysis. As the volatile matter content in coal increased, the starting temperature for releasing the reducing gases decreased, leading to a significant increase in the released gas volume. In addition, the three kinetic factors were calculated by isoconversional method, distribution of the activation energy model (DAEM) and Coats-Redfern (CR) method respectively, and the pyrolysis reaction mechanism function suitable for all coal types was reconstructed based on common solid reaction models. Finally, a predictive model based on a GA-improved back-propagation neural network (BPNN) was developed, which useing 21 input parameters to predict the TG curves, activation energy (E), lnA and mechanism function (fα) and achieving R2 values above 0.95. This study contributes to a comprehensive understanding of the pyrolysis mechanism of coals with different volatile matter content, providing scientific guidance for effective utilization.