Modulus of elasticity prediction model on sugi ( Cryptomeria japonica ) lumber using online near-infrared (NIR) spectroscopic system

Abstract

In this study, static bending measurements and online NIR spectra acquisitions were combined to construct modulus of elasticity (MOE) prediction model for sugi lumber. NIR spectra were acquired from tangential surface of sugi lumbers at a speed of 120 m min⁻¹ to assess its effectiveness in the wood industry. Cross-validation partial least squares regression (CV-PLSR) and test-set-validation partial least squares regression (TSV-PLSR) analyses were employed for analysing the data. The second derivative (2d) spectra with 19 smoothing points (Savitzky–Golay algorithm, second polynomial) gave the best result as spectral pre-processing treatment with the lowest root mean square error of cross-validation and the highest coefficient of determination for cross-validation based on the optimum number of latent variables as assessed from the minimum validation residual variance value in the CV-PLSR analysis. These 2d spectra were then used in the TSV-PLSR analysis for 100 repetitions to check the robustness of the calibration.