中药皂角刺质量的化学模式识别研究

Quality assessment of Gleditsiae Spina by chemical pattern recognition

目的建立皂角刺的化学模式识别方法,用于区分皂角刺正品及伪品,为完善皂角刺的质量评价方法提供依据。方法采用近红外光谱法采集43批样品的近红外光谱信息,利用连续投影算法提取特征变量,运用判别分析法、聚类分析法和反向传播(back propagation,BP)神经网络方法对皂角刺及其伪品进行分类。结果通过判别分析、聚类分析和BP神经网络可将皂角刺及其伪品准确分类,分类准确率为100%,判别结果与性状鉴别结果一致。结论首次建立了皂角刺质量的化学模式识别方法,可以准确区分皂角刺及其伪品,为皂角刺的质量评价提供科学依据。

Objective To set up the chemical pattern recognition method for the discrimination of G leditsiae spina. Methods Near infrared spectrum information of 43 batches of G leditsiae Spina and its adulterants were collected and the NIR data were preprocessed using software OPUS 6. 5. The software M atlab R2014 a was employed to extract characteristic variables,divide data set and establish BP neural network model. The discriminant analysis and cluster analysis were performed by SPSS 21. 0. Results The results showed that the cluster analysis,discriminant analysis and BP neural network accurately classified G leditsiae Spina and its adulterants. Conclusion Chemical pattern recognition method was established for the identification of G leditsiae Spina and it provided a reference for discrimination and quality assessment of G leditsiae Spina.