Prediction of Gas Yield in the Co-Pyrolysis Process of Biomass and Polyethylene by Machine Learning Method
SUN Qidian1, FU Zhe1, HUA Fang1, JI Ye2, CHENG Yi1
1. Department of Chemical Engineering, Tsinghua University, Beijing 100084, China;
2. China Petroleum Planning and Engineering Institute (CPPEI), China National Petroleum Corporation, Beijing 100083, China
Abstract：In order to better predict the gas yield during co-pyrolysis of biomass and polyethylene (PE), this study collected datasets of gas yields from the literature on co-pyrolysis of biomass and PE, and developed a prediction model for predicting PE-biomass co-pyrolysis gas yields by using a machine learning approach, and quantified the relationship between the input factors and the gas yields by employing the feature significance and bias-dependence analysis methods. With different input datasets, including the gradient boosting decision tree model (GBDT), K-nearest neighbor model (KNN), and random forest model (RF) showed good generalization and accuracy in both training and test data. The training results of the GBDT model showed that the root-mean-square error (RMSE) was 1.19, and the coefficient of determination (R2) was 0.99, and in predicting the gas production rate, the RMSE was 3.86 and R2 was 0.91. Nitrogen content and ash content of the biomass played the most important role in gas production during co-pyrolysis, followed by the heating temperature and the plastic-to-biomass mass ratio.
SUN Qidian,FU Zhe,HUA Fang et al. Prediction of Gas Yield in the Co-Pyrolysis Process of Biomass and Polyethylene by Machine Learning Method[J]. Chemical Reaction Engineering and Technology, 2025, 41(1): 163-171.
