近红外光谱技术快速测定木材抽出物含量的研究

探讨了近红外光谱法快速预测木材冷水、热水、1.0%NaOH和苯醇抽出物含量的可行性。四个模型的决定系数（R2）高,分别为0.9804、0.9800、0.9823和0.9648;交叉验证均方根偏差（RMSECV）低,分别为0.21%、0.29%、0.48%和0.24%;冷水、热水、1.0%NaOH抽出物模型的残留预测偏差（RPD）值分别为7.14、7.07和7.51,而苯醇抽出物模型的RPD值仅5.33。采用不同校正模型分别对样品进行预测,四个模型的预测偏差分别为-0.19%～0.20%、-0.29%～0.28%、-0.36%～0.42%和-0.25%～0.14%,基本符合标准方法的误差要求。结果说明可以利用近红外光谱分析技术对木材抽出物进行快速、准确地测定。

A study on rapid prediction of wood extractives content with near infrared spectroscopy

The feasibility of NIR spectroscopy to rapidly determine the extractives content of the wood such as cold water,hot water,1.0%NaOH and benzene ethanol extractive were investigated.FT-NIR spectra were collected from wood powder by using integrating sphere.Partial least squares（PLS） regression analyses were carried out to describe relationships between the data sets of laboratory chemical data and the FT-NIR spectra.High coefficient of determination （R2） of cold water,hot water,1%NaOH and benzene ethanol extractive model were 0.9804,0.9800,0.9823 and 0.9648,respectively.Low root mean square errors of cross-validation （RMSECV） of those were 0.21%,0.29%,0.48% and 0.24%,respectively.Residual prediction deviation （RPD） of cold water,hot water,and 1%NaOH extractive model was 7.14,7.07 and 7.51,respectively.However,for the benzene ethanol extractive,the RPD was 5.33 only.The predictions were good,with prediction deviation of cold water extractive models ranging from-0.19% to 0.20%,those of hot water extractive ranging from-0.29% to 0.28%,those of 1%NaOH extractive ranging from-0.36% to 0.42%,and those of benzene ethanol extractive ranging from-0.25% to 0.14%.The results demonstrate that the extractives content of wood can be rapidly predicted with NIR.