基于近红外光谱的野生与栽培红芪鉴别研究

目的：建立基于近红外光谱的野生与栽培红芪判别模型，对野生与栽培红芪进行鉴别研究。方法：采用多元散射校正法、标准正态变换、一阶导数、二阶导数、Savitzky-Golay filter及Norris derivative filter过滤的组合方法对原始光谱进行预处理，应用光标法及变量重要性图选取光谱波段，建立野生与栽培红芪主成分分析-马氏距离判别模型及偏最小二乘判别模型，并验证模型。结果：多元散射校正法、标准正态变换预处理方法适宜于红芪近红外光谱的处理；所建立的二次偏最小二乘判别（PLS-DA）模型稳定、有效且无过度拟合，可用于鉴别野生与栽培红芪；野生与栽培红芪的主要差异性光谱波段为4 000.000~5 418.991 cm^-1。结论：近红外光谱可鉴别野生与栽培红芪，为其他野生型与栽培型中药材的鉴别提供研究思路。

Objective: To establish the discrimination model of wild and cultivated Hedysari Radix based on near-infrared spectroscopy and identify the wild and cultivated Hedysari Radix. Methods: Multiplication signal correction,standard normal variate,first derivative,second derivative,Savitzky-Golay filter and Norris derivative filter were used to preprocess the original spectra of Hedysari Radix. The cursor method and variable importance plot were applied to select the spectral range. The model of principal component analysis-Mahalanobis distance and partial least square discrimination analysis were established and validated for identifying wild and cultivated Hedysari Radix. Results: The method of multiplication signal correction and standard normal variate were suitable for the treatment of near-infrared spectra of Hedysari Radix. The model of secondary partial least square discrimination analysis(PLS-DA) was stable,effectivel and unprotected,which could be used to identify wild and cultivated Hedysari Radix. The discriminative spectral range of wild and cultivated Hedysari Radix were from 4 000.000 cm^-1 to 5 418.991 cm^-1. Conclusion: Near infrared spectroscopy was proved to be feasible to identify wild and cultivated Hedysari Radix,which provided the ideas for identifying other wild and cultivated Chinese herbal medicines.