Abstract
Further efforts are still needed to refine and optimise complex thermochemical pyrolysis processes crucial in waste management and clean energy production. In this work, a comparative artificial intelligence (AI) based modelling study is conducted using four supervised machine learning models, including artificial neural network (ANN), random forest (RF), support vector regression (SVR), and extreme gradient boosting (XGB) to predict the three-phase product yields of pyrolysis. The models were trained using a database of previous experiments focused on continuous pyrolysis in fluidised bed reactors, with biomass feedstock characteristics and pyrolysis conditions as input features. A reactor dimension parameter through H/D (the ratio of the reactor height, H and the reactor diameter, D), for the first time, is also included as an input feature. The models are optimised through feature reduction and 5-fold cross-validation hyperparameter tuning. They show that reducing the organic composition of biomass to include only chemical composition results in the best feature-reduced model. After the comparison of performance scores and total feature importance, the general ranking for AI model accuracy for this study is XGB>RF>ANN>SVR. The H/D ratio also has the highest feature importance scores of 21.71% and 29.52% in predicting the oil and gas yield of the feature-reduced XGB model, confirming the importance of this added parameter. Preliminary contour plot analysis of the database shows that for the considered reactors, optimum oil yields are obtained at H/D ratio< 5, while the optimum gas yields are expected at H/D ratioc closer to 10 for fluidised bed reactors as another indicator of factor importance.
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Disclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.