Application of wavelength selection strategy based on SWCSS-UVE algorithm in improving model transfer for near infrared analysis of holocellulose and lignin
Honghong Wang, Yunchao Hu, Zhijian Liu, Zhixin Xiong, Liang Long
Abstract
Model transfer techniques in near infrared spectroscopy are important to avoid repetitive modeling, reduce the human and material consumption required for modeling. The use of a method without standards for model transfer by screening stable consistent wavelength (SWCSS) is a new strategy worth investigating. The SWCSS-UVE combination algorithm is proposed to realize the sharing of the near infrared analysis model of holocellulose and lignin content in three different spectroscopic instruments. The UVE algorithm is used to reduce the adverse effect of invalid wavelength in SWCSS method to improve model transfer accuracy. The results show that the 333 and 483 wavelengths finally selected by the SWCSS-UVE method are used to establish the holocellulose and lignin models respectively. The RPDs of the holocellulose and lignin content in the spectra measured by the two target instruments are both greater than 3.5. The transfer result of holocellulose is better than that of lignin, and both are better than the transfer result of SWCSS algorithm. This indicates that the SWCSS-UVE algorithm can eliminate the uninformative and unimportant wavelengths contained in the SWCSS method, greatly simplify the model transfer process, significantly improve the universality of the master instrument model and its transfer accuracy, which provides a new method for the wide application of NIR analysis models.