Open AccessNear Infrared (NIR) Spectroscopy to Predict Physical Properties of Acacia mangium at Three Different Age Classes
Author(s) -
Lina Karlinasari,
Merry Sabed
Publication year - 2017
Publication title -
wood research journal
Language(s) - English
Resource type - Journals
ISSN - 2774-9320
DOI - 10.51850/wrj.2013.4.1.7-12
Subject(s) - calibration , acacia mangium , near infrared spectroscopy , partial least squares regression , materials science , water content , environmental science , correlation coefficient , moisture , solid wood , coefficient of determination , analytical chemistry (journal) , mathematics , chemistry , composite material , statistics , environmental chemistry , geology , optics , physics , biochemistry , geotechnical engineering , layer (electronics)
Near Infrared (NIR) spectroscopy has been used to predict several properties of wood. This is one of the nondestructive testing (NDT) methods providing fast and reliable wood characterization analysis which can be applied in various manufacture industry, included forest sector, in control and process monitoring task. Moisture content and wood density are important properties related to strength properties. The aim of this study was to evaluate NIR technique in obtaining calibration models for determining moisture content and wood density of Acacia mangium in the age of 5, 6, 7 years-old. Spectra were measured in both solid and ground wood samples. Laboratory testing of physical properties were determined by volumetric and gravimetric methods. The laboratory values were correlated with the NIR spectra using multivariate analysis statistic of Partial Least Square (PLS). The calibration-validation model of this relationship was evaluated by using the coefficient of determination (R2), root means square error of calibration (RMSEC) and cross-validation (RMSECV) values. Generally, a better accuracy was obtained by using calibration model of ground wood compared to that of solid wood samples. At age of 7 years-old, the R2 allowed the use of NIR spectra of solid samples to develop calibration and validation model, especially for wood density. Based on ratio of performance to deviation (RPD) and RMSE, ground samples demonstrated a higher value of RPD, RMSEC, and RMSECV compared to solid wood for all properties.