研究目的
To develop calibrations for determining the compressive strength of wood using NIR spectroscopy and to evaluate the predictive ability of the calibrations.
研究成果
The study demonstrated that near-infrared spectroscopy can be used to accurately measure the compressive strength of wood. The BiPLS-GA-PLS model provided the best prediction results, indicating its robustness and simplicity for such applications. Future studies could explore the method's applicability to multiple species of wood.
研究不足
The study focused on a single species of wood, Xylosma racemosum, and the applicability of the method to other wood species was not investigated. The method's performance under varying environmental conditions was also not explored.
1：Experimental Design and Method Selection:
The study used near-infrared (NIR) spectroscopy to predict the compressive strength of wood. The methodology included preprocessing the NIR spectra with multiplicative scatter correction (MSC) and Savitzky-Golay (SG) smoothing, selecting optimal intervals using backward interval partial least squares (BiPLS), and selecting featured wavelengths with a genetic algorithm (GA). A partial least squares (PLS) regression model was then established.
2：Sample Selection and Data Sources:
Thirty green logs of Xylosma racemosum wood were collected, and 180 samples were prepared without defects or significant differences in color. The samples were divided into calibration and prediction sets.
3：List of Experimental Equipment and Materials:
An ultra-compact NIR fiber optic spectrometer (Insion Co., GmbH, Heilbronn, Germany) was used to acquire NIR spectra from 900 to 1900 nm.
4：Experimental Procedures and Operational Workflow:
NIR spectra were acquired, preprocessed, and analyzed using BiPLS and GA to select optimal intervals and wavelengths. The compressive strength was determined according to national standards.
5：Data Analysis Methods:
The quality of the models was assessed using the linear coefficient of determination (R2), the root mean square error of the calibration (RMSEC), and the root mean square error of the prediction (RMSEP).
独家科研数据包，助您复现前沿成果，加速创新突破
获取完整内容
