Biochar production from agricultural biomass through microwave-assisted pyrolysis: predictive modelling and experimental validation of biochar yield

Abstract

A machine learning model for microwave-assisted pyrolysis technology was developed in this study. The data set of 112 unique experiments was created by analysing the published literature based on biochar. The linear, interactive and quadratic regression models were trained with the selected data set. Out of three regressions models, the quadratic model ${\text{Biochar Yield}} = - 601.78 + 17.336 \times {\text{VM}} + 25.338 \times {\text{AC}} - 0.26367 \times T - 0.293 \times {\text{VM}} \times {\text{AC}} - 0.10305 \times {\text{VM}}^{2} - 0.1893 \times {\text{AC}}^{2} + 1.59 \times 10^{ - 3} \times T^{2}$ was found to have highest R² value of 0.894. The predicted model developed based on the data from literature was validated with laboratory experimental results. Prediction and validation results showed that data prediction can be a useful tool for the preview of selected feedstock and process parameters. Volatile matter, ash content and temperature were found to be the prominent factors in the trained model. Biochar yield was predicted with a minimum root mean square error of 7 when validated with experimental results. Thus, the predicted model can be used as empirical equation for future experiments to predict biochar yield.