基于近红外特征谱段相关系数法鉴别真伪龙齿

Rapid Identification of Dens Draconis and its Adulterants by NIR Correlation Coefficient Method

目的：利用近红外特征谱段相关系数法建立定性模型对真伪龙齿进行快速鉴别。方法：在传统性状鉴别的基础上，使用近红外光谱仪的光纤附件采集光谱，以正品龙齿的近红外光谱为参照光谱，选择特征谱段，设定合适阈值，建立真伪龙齿定性鉴别模型，通过模型快速判断待测龙齿样品的真伪。结果：选定5000~4200 cm-1谱段为特征谱段，计算训练集样品中真伪龙齿的相关系数，设定阈值为92.67%，用10批验证集样品对模型进行验证，预测正确率为90%。结论：利用近红外特征谱段相关系数法建立的定性鉴别模型具有较好的预测能力，可以做为快速鉴别真伪龙齿的方法。

Objective：To develop a qualitative analysis model for the fast identification of Dens Draconis and its adulterants by NIR correlation coefficient method. Methods：On the basis of the traditional morphological identification, the spectra were collected u-sing the fiber accessory of a near-infrared spectroradiometer. The reference spectra were set up using the NIR spectra of certified Dens Draconis. The characteristic spectral section was chosen and the appropriate threshold was set to establish a qualitative analysis model for the rapid identification of Dens Draconis and its adulterant. Results：The spectral section of 5 000-4 200 cm-1 was selected as the characteristic spectral section, the correlation coefficient of Dens Draconis and its adulterant was calculated in training set samples, and 92. 67% was used as the threshold. Totally 10 batches of validation set samples were validated the qualitative analysis model, and the prediction accuracy was 90%. Conclusion：The method has good prediction ability, and can be used in the rapid identification of Dens Draconis and its adulterant.