【摘要】 探讨了近红外光谱法快速测定木材纤维素、戊聚糖和Klason木质素含量的可行性。用常规实验室方法测定106种木材样品的纤维素、戊聚糖和Klason木质素含量,用近红外光谱仪采集相应样品的光谱。利用偏最小二乘法(PLS)和完全交叉验证方式,对原始光谱进行一阶导数+矢量归一化(SNV)和一阶导数+多元散射校正(MSC)预处理后,选择7502~5446 cm-1、4601~4247cm-1和6102~4597 cm-1区间建立样品纤维素、戊聚糖、Klason木质素含量和光谱数据之间的相关性模型。模型的决定系数(R2)高,分别为0.9705、0.9927和0.9827;交叉验证均方根偏差(RMSECV)低,分别为0.40%、0.38%和0.49%;残留预测偏差(RPD)分别为5.83、11.7和7.61。采用3个模型分别对样品进行预测,预测偏差分别为-0.36%~0.34%、-0.40%~0.32%和-0.36%~0.39%。结果表明,可以利用近红外光谱法快速测定木材纤维素、戊聚糖和Klason木质素含量。更多还原

【Abstract】 The feasibility of Fourier transform near infrared spectroscopy(FT-NIR) to rapidly determine cellulose,pentosan,and Klason lignin contents of wood was investigated.The cellulose,pentosan,and Klason lignin contents of 106 samples were analyzed with traditional chemical method,the FT-NIR spectra of which were collected from wood meal using integrating sphere.Partial Least Square(PLS) regression analyses were carried out to describe the relationships between the data sets of laboratory chemical analysis and the FT-NIR spectra.The prediction models for the cellulose,pentosan,and Klason lignin contents of wood samples were established using full cross validation in the wavenumbers ranges of 7502~5446 cm-1,4601~4247 cm-1 and 6102~4597 cm-1 by the spectral data preprocessing methods of the 1.DerMSC and 1.DerVNorm.The high coefficients of determination(R2) between the predicted NIR values and those obtained from traditional chemical methods were 0.9705,0.9927,and 0.9827,respectively.The low root mean square errors of cross validation(RMSECV) values were 0.40%,0.38%,and 0.49%,respectively.Residual prediction deviation(RPD) values were 5.83,11.7,and 7.61,respectively.The models derived from wood meal spectra were very precise.The predictions were very good,standard errors of prediction ranges of-0.36%~0.34%,-0.40%~0.32%,and-0.36%~0.39%,respectively.Overall,NIR spectroscopy has proven to be an easy to facilitate,reliable,accurate and rapid method for non-destructive wood cellulose,pentosan,and Klason lignin determination.更多还原