基于模式识别和遗传神经网络算法的醋香附近红外光谱等级评价和含量预测模型研究  

Study on near infrared spectrum grade evaluation and content prediction model of vinegar-processed Cyperi Rhizoma based on pattern recognition and GA-BPNN

作　　者：邱丽媛 梁泽华[1] 吴鑫雨 潘颖洁 方剑文[1] QIU Li-yuan;LIANG Ze-hua;WU Xin-yu;PAN Ying-jie;FANG Jian-wen(School of Pharmacy,Zhejiang Chinese Medical University,Hangzhou 311400,China)

机构地区：[1]浙江中医药大学药学院,浙江杭州311400

出　　处：《中草药》2021年第13期3818-3830,共13页Chinese Traditional and Herbal Drugs

基　　金：国家重点研发计划——中药饮片质量识别关键技术研究(2018YFC1707001)。

摘　　要：目的基于近红外光谱(near infrared spectrum,NIRS)技术建立一种能快速准确识别醋香附饮片等级并预测其挥发油中α-香附酮、香附烯酮含量的质量评价模型,为其他中药材或中药饮片的质量评价提供参考。方法采集醋香附的NIRS信息,并建立39批醋香附挥发油气相色谱-质谱联用(GC-MS)指纹图谱,对挥发油中的α-香附酮、香附烯酮进行定量,采用相似度分析、多元统计分析、主成分分析(principal component analysis,PCA)、聚类分析、偏最小二乘-判别分析(partial least squares-discriminant analysis,PLS-DA)、Logistic回归分析等方法处理数据,划分等级;利用遗传神经网络算法(GA-BPNN)将等级划分结果、α-香附酮含量、香附烯酮含量分别与NIRS信息进行拟合,建立等级预测模型和含量预测模型。结果根据主成分聚类分析法可以将醋香附划分为3个等级,其中一等品6批,二等品8批,三等品25批,PLS-DA分析结果与主成分聚类分析结果一致。采用多元Logistic回归分析建立了饮片等级分类经验公式P一等=exp(G1)/[1+exp(G1)]、P二等=exp(G2)/[1+exp(G2)]、P三等=1-P二等,等级预测结果和主成分聚类分析结果一致。利用GA-BPNN建立的醋香附饮片等级预测模型预测准确率达89.74%,模型准确性较好;α-香附酮、香附烯酮回归模型预测集决定系数分别为0.9923、0.9697,能很好地预测醋香附挥发油中α-香附酮、香附烯酮含量。结论采用GA-BPNN所建立的基于近红外技术的醋香附饮片等级评价模型能快速准确地预测醋香附饮片等级,为醋香附及其他中药材或中药饮片质量标准的制定和等级评价模型的研究提供了参考。Objective A quality evaluation model was established based on near infrared spectroscopy(NIRS),which could quickly and accurately identify the grade of Cyperi Rhizoma processed with vinegar and predict the content ofα-cyperone and cyperenone in its volatile oil,so as to provide reference for the quality evaluation of other Chinese medicinal materials or Chinese herbal pieces.Methods The near infrared spectrum information of vinegar-processed Cyperi Rhizoma was collected,and the GC-MS fingerprint of 39 batches of Cyperus officinalis volatile oil was established.Theα-cyperone and cyperenone in the volatile oil were quantified.The data were processed by similarity analysis,multivariate statistical analysis,PCA,HCA,PLS-DA,logistic regression analysis and other methods,and the grades were classified.Genetic neural network algorithm(GA-BPNN)was used to fit the grade classification results,the content ofα-cyperone and cyperenone with the near infrared spectral information,respectively,to establish the grade prediction model and the content prediction model.Results According to the principal component cluster analysis,vinegar-processed Cyperi Rhizoma can be divided into three grades,and the results of PLS-DA analysis were consistent with those of principal component cluster analysis.Multivariate logistic regression analysis was used to establish the empirical formula for the classification of decoction pieces.The results of grade prediction were consistent with those of principal component cluster analysis.The prediction accuracy of GA-BPNN model was 89.74%,and the model accuracy was good;The determination coefficients ofα-cyperone and cyperenone regression model were 0.9923 and 0.9697,respectively,which could well predict the content ofα-cyperone and cyperenone in volatile oil of vinegar-processed Cyperi Rhizoma.Conclusion The grade evaluation model of vinegar-processed Cyperi Rhizoma based on near infrared technology established by GA-BPNN can quickly and accurately predict the grade of vinegar-processed Cyperi Rh

关 键 词：醋香附 挥发油 等级评价 气相色谱-质谱联用 近红外光谱 模式识别 遗传神经网络算法 Α-香附酮 香附烯酮 质量评价 相似度分析 多元统计分析 主成分分析 聚类分析 偏最小二乘-判别分析 LOGISTIC回归分析 

分 类 号：R283.6[医药卫生—中药学]