化学统计学在不同年份和不同产地郁金鉴别中的应用

Discrimination of Radix Curcumae on the Basis of Their Geographical Origin and Batches by Chemometrics

应用高效液相色谱-质谱(HPLC-MS)、顶空-气相色谱-质谱(HS-GC-MS)和近红外光谱(NIRs)技术,对三个产地5个批次中药郁金的水提物和挥发油,分别进行检测并构建郁金部分组分/全组分指纹图谱,采用统计学方法和模式识别方法(KK-Analysis),构建预测模型。结果显示不同批次不同产地的郁金药材存在差异,GC指纹图谱显示:对于挥发油组分而言,郁金药材之间最大的差异来自不同的品种,而相近的批次对它们的影响要小;HPLC指纹图谱则给出了有意义的结果:四川2016和四川2017批次水提物几乎无差别,但广西2016和广西2017批次差别较大,很明显广西2017和浙江2017批次郁金药材的水提物质量相似并优于广西2016批次。NIRS全组分指纹图谱同样显示不同产地不同批次郁金药材存在差异,应用最小二乘-支持向量机(LSSVM)进行识别,测试集识别率为100%,验证集识别率为94.7%。

By high performance liquid chromatography-mass spectroscopy(HPLC-MS),head space-gas chromatography-mass spectroscopy(HS-GC-MS),and near infrared spectroscopy(NIRS),the global and parts(aqueous extracts and essential oils)fingerprints of Radix Curcumae in three geographical origins and five batches were obtained.Chemometric method of pattern recognition(KK-Aanalysis)was applied for resoluting these fingerprints data.It was found that for aqueous extracts there was difference between ' Guangxi 2016' and ' Guangxi 2017',while ' Sichuan 2016' and ' Sichuan 2017' were similar,and obviously,the quality of 'Zhejiang 2017' and ' Guangxi 2017' was better than that of ' Guangxi 2016';moreover,for essential oils,there were differences among Zhejiang,Guangxi,and Sichuan samples,while ' Sichuan 2016' was similar to ' Sichuan 2017',and ' Guangxi 2016' was similar to ' Guangxi 2017'.Furthermore,NIRS global fingerprints were recognized by least square-support vector machine(LSSVM)and it was shown that the geographical origin and batches of the Radix Curcumae samples could be descriminated,and the recognition rates for training and test sets were 100%and 94.7%,respectively.