基于NIR分析和模式识别技术的葛根品种及产地判别

采用模式识别技术对不同品种(柴葛及粉葛)及不同产地(云南、安徽、广西、湖北、四川、重庆、湖南)的葛根进行判定。采集12个产地共120个葛根样品的近红外光谱数据,对光谱进行预处理并建立共有模式后,进行相似度及PLS判别分析,多元统计分析结果显示,除安徽柴葛外,其他组样品之间的相似度较高。分别选择不同的样品为测试集和训练集,基于PLS-DA对葛根种类粉葛和柴葛进行模式识别,对种类识别率为100%,对产地识别率为84.44%。采用kNN分析对葛根产地和品种同时进行模式识别,样品识别率达100%。实验结果表明,采用kNN模式识别可以很好识别不同产地和类别的葛根样品,方法具有可行性和有效性,为利用近红外光谱结合模式识别技术进行葛根品种真伪优劣鉴别、道地性及产地可追溯研究提供了理论依据和实用方法。

Recognition of Radix Puerariae Varieties and Origin Based on Pattern Recognition and Near Infrared Spectroscopy Technology

The pattern recognition method was used to determine the Radix Puerariae of different varieties(Pueraria lobata Ohwi and Pueraria thomsonii Benth)and origins(Yunnan,Anhui,Guangxi,Hubei,Sichuan,Chongqing,Hunan). 120 samples from 12 producing areas were collected,then established common patterns after preprocessed NIR spectra. The similarity and PLS discriminant analysis of Radix Puerariae samples were preliminary,the result showed that the sample groups were high similar excepting Pueraria lobata Ohwi from Anhui. Different samples were selected as test set and training set,PLS-DA was used to identify the types of Pueraria lobata Ohwi and Pueraria thomsonii Benth,the identification rate for varieties was 100% and identification rate for origin was 84.44%. The kNN was also used for pattern recognition of varieties and origin at the same time. The identification rate of the samples was 100%. The result indicates that the proposed methods were feasible and effective.Moreover,this investigation provided the theoretical support and practical method for recognition of Radix Puerariae varieties and origin utilizing near infrared spectral data.