Bridging uncertainty gaps with artificial intelligence-assisted syngas precise prediction in coal gasification

Coupling of green hydrogen with the coal chemical industry is pivotal for clean coal utilization and low-carbon transition. This study aims to predict syngas composition efficiently using artificial intelligence-assisted machine learning models, particularly the BP-MLPNN model, addressing raw material diversity and process uncertainties. BP-MLPNN model demonstrates superior reliability and robustness in syngas component prediction, as indicated by significantly lower MSE and RMSE values ranging from 0.002 to 11.61 and 0.05 to 3.41, respectively, along with R² values ranging from 0.84 to 1.00. This performance surpasses other models without overfitting. Subsequently, the BP-MLPNN model underwent SHAP analysis to elucidate the internal mechanism of “black box” model. A simple interface input APP was developed to achieve human–machine interaction. This model can mitigate uncertainties in analyzing the integrated coal chemical industry and green hydrogen production system, providing technical guidance and references to quantify its advantages and potential in producing various chemical products.