Prediction of cellulose micro/nanofiber aspect ratio and yield of nanofibrillation using machine learning techniques

Abstract Predictive monitoring of two key properties nanocellulose, aspect ratio and yield nanofibrillation, would help manufacturers control optimize production processes, given the uncertainty that still surrounds their influential factors. For that, 20 different types cellulosic lignocellulosic micro/nanofibers produced from spruce pine softwoods, by pre-treatment fibrillation techniques, were used as training testing datasets aiming at development evaluation three machine learning models. The models Random Forests (RF), Linear Regression (LR) Artificial Neural Networks (ANN), broadening scope our previous work (Santos et al. in Cellulose 29:5609–5622, 2022. https://doi.org/10.1007/s10570-022-04631-5 ). Performance these evaluated comparing statistical parameters such Mean Absolute Percentage Error (MAPE) R². inputs chosen among easily controlled or measured variables: Total lignin (wt%), Hemicellulose Extractives HPH Energy Consumption (kWh/kg), Cationic Demand (µeq/g), Transmittance 600 nm Consistency index (Ostwald-De Waele’s k In both cases, ANN trained here provided satisfactory estimates (MAPE = 4.54% R 2 0.96) nanofibrillation 6.74% 0.98), being able to capture effect applied energy along process. RF LR resulted correlation coefficients 0.93 0.95, respectively, for ratio, while 0.87 0.92.