Espectroscopia no infravermelho próximo (nirs) aplicada na avaliação da superfície de pisos de madeira

The aim of this study was to verify the applicability of near infrared spectroscopy (NIRS) in assessing the quality of surface samples of wood floors under continuous surface treatment with sandpaper and varnishes. Wood samples of Myroxylon balsamum, Peru Balsam, were used due to their quality for making floors and other related applications. Ten samples of wood (wood flooring pattern dimensions (WxLxT) of 8, 50, 1.8 cm) underwent ten treatments combining sandpaper and varnish applications, since the original sample (untreated) to the final product. Using Bruker MPA with integrating sphere in the near-infrared range (NIR) from 800 to 2500 nm, ten spectra were obtained from the surface of each timber specimen. NIR spectra results were explored by Principal Component Analysis (PCA) as a descriptive analysis method. The NIR spectral data were used to develop classification models using partial least squares discrimination analysis (PLS-DA) regression method. The scheme defined by PC1 and PC2, explained 80% of PCA's variability. It showed different treatment groups regarding the original samples (without treatment), to the three initial and the final (final product) treatments. PLS-DA cross-validation indicated that NIR models are suitable for discrimination of the ten M. balsamum samples treatments, with an overall error of 5 %. Thus, it is concluded that the NIRS method has the potential to predict the surface treatment of wood M. balsamum in the floor making industrial process. For routine analysis it is necessary to develop NIRS classification models based on sampling expansion, including natural variations of wood and its surface treatments.