基于电子鼻与多元统计分析判别三七品质  

Quality assessment of Panax notoginseng using electronic nose and multivariate statistical analysis

作　　者：刘元林[1] 龙鸣 张希 田晓静[1,2] 柏家林 宋礼[4] LIU Yuan-lin;LONG Ming;ZHANG Xi;TIAN Xiao-jing;BAI Jia-lin;SONG Li(College of Life Science and Engineering,Northwest Minzu University,Lanzhou 730124,China;Biomedical Research Center,China-Malaysia National Joint Lab,Lanzhou 730030,China;College of Traditional Chinese Medicine,Yunnan University of Chinese Medicine,Kunming 650500;Gannan Research Institute of Yak Milk,Hezuo 747000,China)

机构地区：[1]西北民族大学生命科学与工程学院,甘肃兰州730124 [2]西北民族大学生物医学研究中心,中国-马来西亚国家联合实验室,甘肃兰州730030 [3]云南中医药大学中药学院,云南昆明650500 [4]甘南牦牛乳研究院,甘肃合作747000

出　　处：《中成药》2021年第3期700-707,共8页Chinese Traditional Patent Medicine

基　　金：甘肃省自然科学基金(18JR3RA371);西北民族大学中央高校基本科研业务费资金资助项目(31920190022);教育部动物医学生物工程创新团队(IRT_17R88);西北民族大学“双一流”和特色发展引导专项-生物工程特色学科(10018703,1001070204)。

摘　　要：目的基于电子鼻与多元统计分析判别三七Panax notoginseng(Burk.)F.H.Chen的品质。方法在优化电子鼻检测条件基础上,对传感器响应信号进行多元统计与神经网络分析。结果电子鼻检测三七较佳条件为样品量1.5 g;顶空生成时间15 min;顶空体积250 mL;载气体积流量400 mL/min。多元统计表明主成分分析和典则判别分析均能区分三七主根与支根,但后者效果优于前者;利用三七主根和支根气味信息结合典则判别分析,可实现对三七产地的定性判别,其中主根气味信息的判别效果更好。多层感知器神经网络分析可以实现对三七主根、支根及产地的定量判别,主根与支根分类准确率达99.49%;主根产地判别准确率为99.49%;支根产地判别准确率为95.95%。结论电子鼻结合多元统计与神经网络分析可以实现对三七品质的判别,且该方法高效快速可用于实际生产。AIM To identify the quality of Panax notoginseng(Burk.)F.H.Chen(P.notoginseng)using electronic nose(E⁃nose)and multivariate statistical analysis.METHODS With the optimized detection conditions for E⁃nose,multivariate statistics and neural network analysis were carried out on the sensor response signals.RESULTS The best conditions for E⁃nose detection of notoginseng were identified as the follows:1.5 g sample weight,250 mL volume headspace generated within 15 min,with the loaded gas flowing at rate of 400 mL/min.Although multivariate statistics revealed the success of both principle component analysis and canonical discriminant analysis in distinguishing the taproot and rootlet of P.notoginseng.The latter worked more efficiently.Multilayer perceptron neural network realized the discrimination of notoginseng taproot and rootlet quantitaty,and their geographical origin,ensuring a 99.49%accuracy for classification of taproot and rootlet,and taproot geographical origin,a 95.95%accuracy of rootlet origin as well.CONCLUSION With the combinative use of multivariate statistics and neural network analysis,E⁃nose can identify the quality of P.notoginseng.

关 键 词：三七 电子鼻 多元统计 MLP神经网络分析 

分 类 号：R282.5[医药卫生—中药学]