基于近红外光谱不同波段的红松木材含水率预测分析

Moisture Content Prediction of Korean Pine Wood Based on Near Infrared Spectroscopy at Varied Wave Lengths

运用近红外光谱技术结合偏最小二乘法（PLS）,对所采集光谱进行一阶导数和二阶导数处理,并对未处理原始光谱、一阶导数处理光谱和二阶导数处理光谱分别在7个不同波段范围内建立红松含水率预测模型。结果表明红松样本近红外光谱经一阶导数处理,波段在1 000~2 100 nm范围内所建模型最优,其校正集相关性系数为0.992 5,校正标准偏差和校正均方根误差分别为0.025 9和0.025 7,验证集相关系数为0.991 7,预测标准误差与预测均方根误差分别为0.031 8和0.031 7。研究表明,结合样本特性选取特定光谱波段范围建立预测模型可大幅度减少建模时间、降低建模成本,同时可以提高模型的预测精度。

A study was conducted to examine the feasibility of using near infrared（NIR） spectroscopy analysis combinded with parial least squares regression（PLS） analysis to predict moisture content（MC） of Korean pine wood samples.The NIR spectra were pretreated with the first and the second derivatives.MC prediction models for Korean pine wood were developed based on the original NIR spectra,the first derivated NIR spectra and the second derivated NIR spectra at seven varied wave lengths.The optimal model was shown between wave lengths of 1 000-2 100 nm of the first derivated NIR spectra with correlation coefficient（R1） of 0.992 5,standard error of calibration and root mean square error of calibration of 0.025 9 and 0.025 7,respectively.The correlation coefficient for model validation was 0.991 7 with standard error of prediction and root mean square error of prediction of 0.025 9 and 0.025 7,respectively.It indicates that to restrict the model development in some typical wave lengths could significantly reduce time and cost of model develpment and improve the model accuracy.