Prediction of lignocellulosic biomass structural components from ultimate/proximate analysis

In order to reduce time and resource consumption, the mathematical model was developed to predict lignocellulosic biomass structural components including cellulose, hemicellulose and lignin from ultimate/proximate dataset. Self-organizing maps (SOMs) were integrated with a regression model to obtain more precise results than the procedure without data clustering. In SOMs, the 149-biomass dataset from literatures, expressed by the ratios of VM/C, VM/H, VM/O, FC/C, FC/H, FC/O and ASH/O, were employed for training and clustered into 4 groups. The result indicated that each group had its own characteristics. The regression model with pre-analyzed by SOMs provided better results compared to the model without pre-analyzed by SOMs. The model obtained in this study can be applied to further researches in many fields; e.g. biomass characterization and utilization. ? 2021 Elsevier Ltd. All rights reserved.

Automated summary

A mathematical model was developed to predict lignocellulosic biomass structural components, integrating self-organizing maps (SOMs) and regression models for more accurate results, clustering data into 4 groups. Each group showed distinct characteristics, and the regression model pre-analyzed by SOMs outperformed the model without pre-analysis. The model can be applied to biomass characterization and utilization research.