Machine learning approach for the prediction of biomass pyrolysis kinetics from preliminary analysis

The pyrolytic behavior of lignocellulosic biomass is highly complex, and its kinetic varies with operating conditions the type biomass. To reduce timescales, cost rigorous calculations associated new set experimentation used for estimation parameters, model-based predictions are recommended. In present work, Artificial Neural Network (ANN) based machine learning models developed to predict pyrolysis kinetics. Data sets thermogravimetric analysis feedstock characterization from a diverse range were develop test networks. Four in this study on proximate (ANN-1), ultimate (ANN-2), combined (ANN-3) proximate, ultimate, biochemical (ANN-4). A total 704 datasets extracted recalculated Coats-Redfern Method which 662, 585, 465 133 sequentially. models, particular ANN-3 ANN-4 have shown competitive prediction capability (R2 ~ 0.99, RRMSE <10.0%, MAE < 0.071). Relative importance each input (biomass properties & heating rate) outputs (kinetic parameters) was also studied. Biochemical found higher contribution (~38%) comparison (~29%) followed by (~22%) kinetics be affected rate extent ~10%. accurate enough predicting any feedstocks preliminary analysis.