Data-driven intelligent analysis of hydrocarbon generation kinetics and differential characteristics during pyrolysis of coal macerals

Abstract The heterogeneity of macerals represents a key challenge to accurately evaluating the hydrocarbon generation potential of coal. Conventional methods often overlook these differences, leading to biased understanding of its hydrocarbon generation characteristics. Therefore, this study integrates maceral identification, thermal simulation experiments, and machine learning algorithms to develop the extreme gradient boosting (XGBoost) prediction models for the yields of gaseous and liquid hy