Determination of Quality Parameters in Moist Wood Chips by Near Infrared Spectroscopy Combining PLS-DA and Support Vector Machines

Abstract

In this work, studies are described using near infrared spectroscopy and chemometrics for determination of quality parameters in moist wood chips, such as basic density, lignin content, and extractives. A classification model using partial least squares—discriminant analysis (PLS-DA) was developed to determine the level of moisture in the samples. Then, for each moisture level, a calibration model was built for quality parameter predictions using least squares support vector machines (LS-SVM). Multivariate calibration was performed for a data set of 92 wood chip samples. The PLS-DA algorithm was able to classify the samples in the correct class with a small error (lower than 6%) and it was possible to develop a LS-SVM model for quality parameter determination for each class of moisture content with only a few samples and with average relative errors comparable to those obtained by conventional analysis.