Unlocking biomass valorization: Machine Learning insights for Reductive Catalytic Fractionation of cotton stalks

Abstract

Lignocellulosic biomass valorization has become an intensive area of research due to the importance of renewable nature and availability of biomass. However, biomass fractionation and depolymerization produce numerous datasets, that are difficult to visualise and interpret for the scale-up of the process. Therefore, machine learning algorithms, which can discover hidden patterns in data are applied to these datasets. Reductive Catalytic Fractionation (RCF) of lignocellulosic biomass is an emerging methodology to valorize biomass completely and effectively. Herein, the present work includes the Correlation Analysis and the Principal Component Analysis (PCA) of product distribution obtained from RCF of cotton stalks. Interactions between process variables and delignification (DL), sugar retention (SR), total phenolic monomers (PM), and individual phenolic monomers yield were evaluated. Correlations among DL, SR, and PM yields were also evaluated at different reaction conditions through PCA, which were explained using the reaction mechanism and molecular chemistry of lignin.