Hokkaido Forest Products Research Institute, Asahikawa, Hokkaido, 071-0198, Japan.
Appl Spectrosc. 2010 Jan;64(1):92-9. doi: 10.1366/000370210790572016.
Near-infrared (NIR) spectroscopy, coupled with multivariate analysis, has been used to evaluate the wood properties of sawn lumber of Japanese larch (Larix kaempferi), whose diffuse reflection spectra were acquired under static and moving conditions. Prediction models of the dynamic modulus of elasticity (E(fr)), the modulus of elasticity in bending tests (E(b)), the bending strength (F(b)), the wood density (DEN), and the moisture content (MC) were developed using partial least squares (PLS) analysis. For all wood properties, models obtained from data collected under the moving condition as an analogue of on-line measurement were superior to those from the static condition data. The regression coefficients for the PLS models predicting the mechanical properties in both static and moving conditions showed clear peaks at the absorption bands due to the three major polymers of wood, i.e., cellulose, hemicellulose, and lignin. NIR spectroscopy has high potential for the on-line grading of sawn lumber.
近红外（NIR）光谱分析结合多元分析已被用于评估日本落叶松（Larix kaempferi）锯材的木材特性，其漫反射光谱是在静态和动态条件下采集的。采用偏最小二乘法（PLS）分析，建立了动态弹性模量（E（fr））、抗弯弹性模量（E（b））、抗弯强度（F（b））、木材密度（DEN）和含水率（MC）的预测模型。对于所有木材性能，从模拟在线测量的动态条件下采集的数据获得的模型均优于从静态条件下数据获得的模型。在静态和动态条件下预测力学性能的 PLS 模型的回归系数在木材的三种主要聚合物（纤维素、半纤维素和木质素）的吸收带处显示出明显的峰值。近红外光谱在锯材的在线分级方面具有很大的潜力。
