Predicting xylose yield from prehydrolysis of hardwoods: A machine learning approach

Hemicelluloses are amorphous polymers of sugar molecules that make up a major fraction lignocellulosic biomasses. They have applications in the bioenergy, textile, mining, cosmetic, and pharmaceutical industries. Industrial use hemicellulose often requires polymer be hydrolyzed into constituent oligomers monomers. Traditional models degradation kinetic, usually only appropriate for limited operating regimes specific species. The study hydrolysis has yielded substantial data literature, enabling diverse set to collected general widely applicable machine learning models. In this paper, dataset containing 1955 experimental points on batch hardwood was from 71 published papers dated 1985 2019. Three (ridge regression, support vector regression artificial neural networks) assessed their ability predict xylose yield compared kinetic model. Although performance ridge unsatisfactory, both networks outperformed simple network reducing mean absolute error predicting soluble test 6.18%. results suggest trained historical may used supplement data, number experiments needed.