Machine learning based prediction of nitrogenous product yield in biomass pyrolysis oil

Hong Tian, Yuqian Zou, Shan Cheng, Dangzhen Lv, Zhijie Wang, Zhangjun Huang, Jiwen Fu,
Machine learning based prediction of nitrogenous product yield in biomass pyrolysis oil,
Journal of the Energy Institute,
Volume 123,
2025,
102291,
ISSN 1743-9671,
https://doi.org/10.1016/j.joei.2025.102291.
(https://www.sciencedirect.com/science/article/pii/S1743967125003198)
Abstract: A crucial technology for achieving its high value conversion in oil production is biomass pyrolysis. The amount of nitrogen-containing products in bio-oil has a negative impact on its quality and has become a barrier to its industrial production as fuel. Conditions during pyrolysis and the kind of biomass feedstock have a direct impact on the production of nitrogen-containing chemicals in bio-oil, and accurate prediction of nitrogen-containing yields is important for optimising the pyrolysis process and achieving its high-value conversion. The nitrogen-containing product output of bio-oil during biomass pyrolysis was predicted in this study using a machine learning technique. By collecting and analysing 195 sets of experimental data of different types of biomass under different pyrolysis conditions, various machine learning models, such as support vector regression machine (SVR), random forest (RF) and neural network, were constructed with the aim of achieving accurate prediction of nitrogen-containing product yield in bio-oil. According to the experimental data, the neural network performs the best among the suggested models in predicting the trend of nitrogen-containing components in bio-oil, with an R2 of 0.9357. This study provides a new method for product optimisation in biomass pyrolysis process, which has a wide range of application prospects.
Keywords: Pyrolysis; Bio-oil; Nitrogen-containing; Machine learning