Near Infrared Spectral Data Establish The Predictable Model Of Populus Tomentosa-holocellulose Content

Near infrared spectrum (NIR) data can be used for the qualitative and quantitative analysis. through summarizing the traditional NIR pretreatment methods, the applicability of modeling method for the data of material properties can be discussed. This paper mainly try to establish a model for Populus tomentosa, according to its material property data and spectral data.As the research objects, Populus tomentosa has the superior performance in pulp and paper field. We attempts to put forward a steadily mathematical model based on Lambert-Beer Law. It has practical meanings for the analytical modeling establishment to predict rate of pulp. The theoretical model can predict Populus tomentosa-holocellulose contents. The first step of modeling method is to take pretreatments on near infrared spectral data, which can be divided into many groups according to the wavelength, then build up 50 group nonlinear mathematical model.Finally the holocellulose contents ensemble prediction model is given by the weighted average formula. The correlation coefficient r=0.9455, the average relative error=0.0060, the goodness of fit index R2=0.8940.This model establishment as a new method to predict the Populus tomentosa-holocellulose contents, may benefit future pulp performance prediction of Populus tomentosa.