【摘要】 基于傅里叶变换红外光谱技术,利用逐步判别分析法对金银花品种和产地进行鉴别研究。采集、测试了5个产地9个品种150份金银花样本的红外光谱,并计算了各样品红外光谱的一阶导数光谱和二阶导数光谱。分别选用不同的样本组成训练集和检验集,以1800～900 cm^-1、1500～700 cm^-1和1200～700 cm^-1波数范围的红外光谱、一阶导数光谱和二阶导数光谱数据为判别变量建立判别模型对金银花的品种和产地进行鉴别。判别结果显示,以1800～900 cm^-1波数范围的二阶导数光谱数据为判别变量建立的模型鉴别效果相对较好,对品种和产地的鉴别正确率依次达93.20%和96.13%。研究结果表明,采用逐步判别模式识别可以很好地鉴别不同品种和产地的金银花,方法可行有效,可为金银花品种和产地朔源提供方法。更多还原

【Abstract】 The Fourier Transform Infrared Spectroscopy of 150 Honeysuckle specimens of nine varieties collected from five regions were obtained to discriminate the sample variety and geographical origin.All specimens were randomly divided into a training set and a validation set.The discrimination analysis based on the original spectra,first-derivate spectra and second-derivate spectra revealed that the model based on the second-derivate spectra has the highest accuracy,and the model based on the waveband of 1800～900 cm^-1 was better than that of 1500～700 cm^-1 and 1200～700 cm^-1,which yielded correct rate of 93.20% for distinguishing variety and 96.13% for identifying origin.The study demonstrated that Fourier Transform Infrared Spectroscopy combining with discrimination analysis might be used as an effective way to distinguish Honeysuckle variety and origins.更多还原