近红外光谱信息筛选在玛咖产地鉴别中的应用

发布时间：2018-01-27 22:29

本文关键词： 玛咖近红外光谱鉴别光谱信息筛选模型集群分析　出处：《光谱学与光谱分析》2016年02期 　论文类型：期刊论文

【摘要】：食药植物玛咖富含多种营养成分,极具药用价值。采用近红外漫反射光谱,对采自秘鲁及云南共139份玛咖样品进行产地鉴别。采用多元信号校正结合二阶导数和Norris平滑预处理光谱,利用光谱标准偏差初步选择光谱波段(7 500~4 061cm~(-1)),结合主成分-马氏距离(principal component analysis-mahalanobis distance,PCA-MD)筛选出适宜的主成分数为5。基于所筛选的光谱波段及主成分数,利用"模群迭代奇异样本诊断"方法剔除2个异常样品后,分别采用竞争自适应重加权法(competitive adaptive reweighted sampling,CARS)、蒙特卡洛-无信息变量消除法(monte carlo-uninformative variable elimination,MC-UVE)、遗传算法(genetic algorithm,GA)和子窗口重排(subwindow permutation analysis,SPA)四种方法筛选光谱变量信息,利用模型集群分析(model population analysis,MPA)思想对所筛选的光谱变量信息进行评价。结果显示,RMSECV(SPA)RMSECV(CARS)RMSECV(MC-UVE)RMSECV(GA),分别为2.14,2.05,2.02,1.98,光谱变量数分别为250,240,250和70。采用偏最小二乘判别分析法(partial least squares discriminant analysis,PLS-DA)对四种方法筛选的光谱变量建立判别模型,随机选择97份样品作为建模集,其余40份样品作为验证集。通过R2,RMSEC和RMSEP分析可知,R2:GAMC-UVECARSSPA,RMSEC和RMSEP:GAMC-UVECARSSPA,且GA,MC-UVE,CARS和SPA四种方法筛选的光谱信息所建立的产地判别模型预测正确率分别为95.0%,92.5%,90.0%和85.0%。四种方法筛选的光谱信息所建立的产地判别模型均具有较好的预测性能,其中GA法所筛选的光谱信息建立的判别模型更准确。该方法的建立旨在为中药材鉴别和品质评价奠定基础。
[Abstract]:The medicinal plant Maca is rich in many kinds of nutrients and has great medicinal value. The near infrared diffuse reflectance spectroscopy (NIR) is used. A total of 139 Maca samples from Peru and Yunnan were identified by multivariate signal correction combined with second-order derivative and Norris smoothing pretreatment spectra. The spectral standard deviation was used to select the spectral band (7 500 ~ 4 061 cm ~ (-1) ~ (-1)). Principal component analysis-mahalanobis distance. Based on the selected spectral bands and principal fractions, two abnormal samples were eliminated by the method of "mode group iterative singular sample diagnosis". Competitive adaptive reweighted sampling (CARSs) was used respectively. Monte Carlo-Non-information variable elimination method monte carlo-uninformative variable elimination MC-UVE. Genetic algorithm (GA) and subwindow permutation analysis. The spectral variable information was screened by spa and the model population analysis was analyzed by model cluster analysis. MPA) thought to evaluate the selected spectral variable information. The results show that. RMSECV / SPAN RMSECV / CARSv / RMSECV / MC-UVEN RMSECVGA, 2.14 / 2.05 / 2.02, respectively. 1.98, the number of spectral variables is 250,240. The partial least squares discriminant analysis was determined by partial least squares discriminant analysis. The discriminant model of spectral variables screened by four methods was established by PLS-DA.The 97 samples were randomly selected as modeling set and the other 40 samples were selected as validation set. RMSEC and RMSEP analysis showed that R2: GAMC-UVECARPARPAE RMSEC and RMSEP: GAMC-UVECARSSPA, and GA. The prediction accuracy of the model based on the spectral information of MC-UVECARS and SPA was 95.0% and 92.5%, respectively. Four methods were used to screen the spectral information. The model of origin discrimination established by the four methods had good prediction performance. The discriminant model of the spectral information screened by GA method is more accurate, and the establishment of this method aims to lay a foundation for the identification and quality evaluation of Chinese medicinal materials.
【作者单位】：云南省农业科学院药用植物研究所;
【基金】：国家自然科学基金项目(31460538,81260608) 云南省自然科学基金项目(2013FD066,2013FZ150)资助
【分类号】：S567.2;O657.33
【正文快照】： 引言玛咖(Lepidium meyenii Walp.)为十字花科(Cruciferae)独荇菜属(Lepidium)植物,常用名Maca(音译“玛咖”),又名Maka,Maca-maca,Peruvian ginseng,Maino,Ayak-willku,Ayakchichira等,原产于海拔3 500~4 500m的南美洲安第斯山区,现主要分布于秘鲁中部的Puna生态区和秘鲁东南

【相似文献】