利用傅里叶变换红外光谱结合判别分析对三七的道地性及产地进行鉴别研究。测试了11个县13个种植点的136株三七主根样品的红外光谱, 利用Omnic8.0软件计算了每个样品红外光谱的二阶导数光谱。分别采用1 800～700 cm-1光谱范围的红外光谱数据和二阶导数光谱数据, 运用逐步判别分析法建立模型对三七的道地性进行判别研究, 二阶导数光谱数据建立的模型对三七道地性的识别效果更好, 回判正确率为100%, 预测正确率为93.4%。采用交叉验证法检验了模型的稳定性, 并对此方法进行了外推性验证。用二阶导数光谱数据结合相同的判别方法对三七的产地进行识别, 比较了不同光谱范围和不同训练样本数建立的模型判别效果, 每个种植点选择8个样本作为训练样本, 采用1 500～1 200 cm-1光谱范围的数据建立的模型判别效果较好, 回判正确率为99.0%, 预测正确率为76.5%。结果表明, 红外光谱结合判别分析对三七道地性的识别效果好, 有望成为实际中鉴别三七道地性的新方法;对三七产地的识别有一定的效果, 可作为三七产地鉴别的一种新思路。

The genuineness and producing area of Panax notoginseng were studied based on infrared spectroscopy combined with discriminant analysis. The infrared spectra of 136 taproots of P. notoginseng from 13 planting point in 11 counties were collected and the second derivate spectra were calculated by Omnic 8.0 software. The infrared spectra and their second derivate spectra in the range 1 800～700 cm-1 were used to build model by stepwise discriminant analysis, which was in order to distinguish study on the genuineness of P. notoginseng. The model built based on the second derivate spectra showed the better recognition effect for the genuineness of P. notoginseng. The correct rate of returned classification reached to 100%, and the prediction accuracy was 93.4%. The stability of model was tested by cross validation and the method was performed extrapolation validation. The second derivate spectra combined with the same discriminant analysis method were used to distinguish the producing area of P. notoginseng. The recognition effect of models built based on different range of spectrum and different numbers of samples were compared and found that when the model was built by collecting 8 samples from each planting point as training sample and the spectrum in the range 1 500～1 200 cm-1, the recognition effect was better, with the correct rate of returned classification reached to 99.0%, and the prediction accuracy was 76.5%. The results indicated that infrared spectroscopy combined with discriminant analysis showed good recognition effect for the genuineness of P. notoginseng. The method might be a hopeful new method for identification of genuineness of P. notoginseng in practice. The method could recognize the producing area of P. notoginseng to some extent and could be a new thought for identification of the producing area of P. notoginseng.