Discriminant Analysis of Wood-Based Materials with Weathering Damage by near Infrared Spectroscopy

Abstract

This research was aimed at finding a suitable discriminate technique for wood-based materials which had suffered from weathering damage, as an analogue of waste wood condition, by means of near infrared (NIR) spectroscopy and several kinds of chemometrics. The feasibility of Mahalanobis generalised distance, K nearest neighbours (KNN) or soft independent modelling of class analogy (SIMCA) as the classification method of wooden materials was examined in detail. The differences in the accuracy of classification with the spectrophotometer (for example, laboratory use and field use), pre-treatment of NIR spectra or the wavelength range as the explanatory variables were also taken into account. Five categories of wood-based materials (solid wood, laminated wood, particle or fibreboard, impregnated wood and overlaid wood) were exposed to outdoor sunlight for up to six months. It was difficult to apply Mahalanobis generalised distance to the classification of five material types where the NIR spectra varied greatly with the sample categories. Both of KNN and SIMCA presented the highest correct classification of over 93%, independent of the spectrophotometer. These results support the applicability of NIR spectroscopy to the classification of waste wood at the actual factory and job site.