A model transfer strategy based on screening stable wavelength for quantitative analysis of holocellulose and lignin content distribution in pulpwood.

The application of near-infrared spectroscopy (NIR) technology in analyzing the compositional content of wood from pulping raw materials suffers from the problem that the multivariate correction model cannot be shared among different NIR spectrometers. To address this problem, NIR spectra of 84 samples of common pulpwood species (Cunninghamia lanceolata, Eucalyptus robusta Smith, Acacia, Pinus massoniana, Populus L) are analyzed by traditional laboratory methods for the content of their main components (holocellulose and lignin content). Screening wavelengths based on spectrum ratio analysis (SWSRA) combined with UVE algorithm is used to screen stable characteristic bands, clarify the characteristic absorption of holocellulose and lignin, and optimize the model transfer results. The results are compared with those of SWSRA, CARS, piecewise direct correction algorithm (PDS) and slope intercept algorithm (S/B). The results show that for holocellulose, the 653 wavelengths screened by the SWSRA-UVE method contain a large amount of overtone and group frequency absorption information of C–H and O–H of holocellulose. For lignin, the 639 wavelengths screened by the SWSRA-UVE method contain a large amount of overtone and group frequency absorption information of lignin C–H and C=O. The SWSRA-UVE algorithm is optimal in improving the transfer effect of the models. This method can effectively remove the unimportant wavelengths in the SWSRA method and significantly improve the transfer accuracy and efficiency of the master models. This study contributes to the promotion of near-infrared spectroscopy in the rapid analysis and determination of the main components of pulping raw wood. [ABSTRACT FROM AUTHOR]