高级搜索

不同鲜叶等级制成的福安金牡丹红茶化学元素特征分析

Chemicals in Fu'an Jinmudan Black Teas Made of Different Grades of Leaves

    摘要
  • 摘要:
    目的 探究福建福安金牡丹红茶中化学元素差异，及其与鲜叶等级的关联性，从而构建不同等级红茶的判别模型，为福安金牡丹红茶的等级判别提供基础数据。
    方法 采集福安市主要茶区35份金牡丹红茶样品（特级 21 份、一级 14 份），采用电感耦合等离子体质谱仪（Inductively coupled plasma mass spectrometry, ICP-MS）测定样品中的15种矿质元素和15种稀土元素含量。结合相关性分析、线性判别分析（Linear discriminant analysis, LDA）等多元统计方法，筛选特征元素构建多层感知神经网络（Multilayer perceptron, MLP）判别模型。
    结果 福安金牡丹红茶的矿质元素含量为钾（K）＞钙（Ca）＞镁（Mg）＞锰（Mn）＞铝（Al）＞钠（Na）＞铁（Fe）＞锌（Zn）＞铜（Cu）＞铅（Pb）＞锡（Sn）＞汞（Hg）＞砷（As）＞硒（Se）＞镉（Cd），其中K、Ca、Mg、Mn为主要元素（占总量的96.09%）；稀土元素的含量为铈（Ce）＞镧（La）＞钇（Y）＞钕（Nd）＞镨（Pr）＞钐（Sm）＞钆（Gd）＞镝（Dy）＞镱（Yb）＞铒（Er）＞钬（Ho）＞铽（Tb）＞铕（Eu）＞铥（Tm）＞镥（Lu），其中Ce的含量最高。相关性热图分析结果表明，矿质元素间存在显著正相关（Al与Ca、Fe与Zn 等）或负相关（Cu与部分稀土元素）；稀土元素（除Ce外）两两呈显著正相关。LDA结果显示，线性判别模型判别正确率为91.4%。并结合受试者工作特征曲线（Receiver operating characteristic curve, ROC）及显著性分析显示，筛选出Al、As和Ca［曲线下面积（Area under the curve, AUC）＞0.7，P＜0.05］为不同等级福安金牡丹红茶的特征元素，具有独特的元素指纹，其中Ca元素贡献最大。MLP结果显示，训练组特级准确率为93.8%，一级准确率为77.8%；检验组特级准确率为100%，一级准确率为60%，该模型能够较好将不同等级金牡丹红茶进行区分。
    结论 福安金牡丹红茶的矿质元素和稀土元素分布具有等级差异性，其中 Al、Ca、As 3种矿质元素可作为金牡丹红茶等级区分的关键指标。

     

    Abstract:
    Objective Difference in chemicals of black teas made of leaves of varied grades plucked from Camellia sinensis (L.) O. Kuntze cv. Jinmudan was determined, and a discriminant model to differentiate the products established.
    Method Jinmudan tea leaves of 21 super-grade and 14 1st grade were collected from the main tea producing areas of Fu'an city in Fujian province. Contents of 15 minerals and 15 rare earth elements in the teas made were determined by inductively coupled plasma mass spectrometry. A multilayer perceptron (MLP), or neural network, discriminant analysis was performed to construct a regression model on the analytical results with the accuracy of differentiating the teas made from different grades of leaves evaluated by correlation analysis, linear discriminant analysis (Fisher), and multivariate statistical methods.
    Result The minerals in the black teas ranked in content as K＞Ca＞Mg＞Mn＞Al＞Na＞Fe＞Zn＞Cu＞Pb＞Sn＞Hg＞As＞Se＞Cd, with the 1st four minerals constituting 96.09% of the total. The contents of rare earth elements ranked as Ce＞La＞Y＞Nd＞Pr＞Sm＞Gd＞Dy＞Yb＞Er＞Ho＞Tb＞Eu＞Tm＞Lu, with Ce being on the top of the list. There were significant positive correlations between Al and Ca and between Fe and Zn, and negative correlations between Cu and some rare earth elements. Among the rare earth, except Ce, significant positive correlations were found between two elements. The linear discriminant analysis (Fisher) showed that the accuracy of the prediction model was 91.4%. Combined with the receiver operating characteristic curve (ROC) and significance analysis, Al, As, and Ca at area under curve ＞0.7 and P＜0.05 (Ca contributed the most) were selected as the characteristic elements with unique fingerprints of the various black teas. The MLP discriminant analysis showed the accuracy on the training group to be 93.8% with the 1st level at 77.8%, and on the test group, 100% with the 1st level at 60%. Thus, the secured model appeared to be capable of adequately differentiating the Jinmudan black teas made of varied grades of leaves.
    Conclusion The minerals and rare earth elements contained in the Fu'an Jinmudan black teas indicated that Al, Ca, and As were the key indicators that could effectively distinguish the teas made of different grades of leaves. A means to facilitate product quality verification of Jinmudan teas on the market was made available.

     

    • HTML全文
    • 参考文献(29)
    • 相关文章
    • 施引文献
  • 资源附件(0)

/

返回文章
返回