Estimativa da umidade em cavacos de madeira por espectroscopia no infravermelho próximo

Abstract

The moisture of the wood chips for industrial use is one of the most important proprieties. To evaluate the characteristics of its inputs the industries of forest base needs easily applicable techniques because traditional method gravimetric or resistive are generally slow and onerous which make the process impracticable. In this context the goal of this study was to apply near infrared spectroscopy (NIR) to estimate the moisture in wood chips by independent cross-validations, as well verify to what extend the spectral acquisition path, and the face of wood affect the statistics associated with partial least squares regressions. Spectrum was obtained in a spectrometer endowed with an integrating sphere and optical fiber in the longitudinal and transverse faces of wood chip clones of Eucalyptus urophylla x Eucalyptus grandis at six years old. During the drying of the wood chips samples previously saturated, the test bodies had their masses determined and the spectrum in the NIR was measured in 10 steps of drying since the saturated condition until the anhydrous condition. The values of moisture of the wood chips were related with the NIR spectrum. From the information were performed an analysis of main components (PCA), partial least square regression (PLS-R) and partial least squares discriminant analysis (PLS-DA). The best model for estimating moisture in wood chips was generated from spectra obtained on the longitudinal face viapartial least square regression (R²cv = 0,96 e RMSECV = 7,15% ) and longitudinal face via optical fiber (R²cv = 0,90 e RMSECV =11,86% ). The integration sphere and the longitudinal face are the most suitable for measuring NIR spectra and generate robust models for estimating moisture in Eucalyptus woods chips.