%0 Journal Article
%T 基于电感耦合等离子体质谱法区分不同产地蜂蜜<br>Differentiating Honey from Different Origins Based on Inductively Coupled Plasma Mass Spectrometry
%A 候玉侠
%A 林立群
%A 田晓婷
%A 王瑞雪
%J Hans Journal of Agricultural Sciences
%P 919-927
%@ 2164-5523
%D 2020
%I Hans Publishing
%R 10.12677/HJAS.2020.1011141
%X <div style=\"text-align:justify;\">
	<span style=\"font-size:14px;\">为了更好地辨别蜂蜜的产地，保证蜂蜜质量和真实性，本次实验利用电感耦合等离子体质谱法(ICP-MS)对84个A地区蜂蜜样品和70个B地区蜂蜜样品进行铅、硼同位素检测。采用主成分分析法对实验数据进行分析处理，发现样品可以较好地分散开。依据实验数据通过Mass Profiler Professional软件建立5种预测模型，分别为决策树(Decision Tree)、朴素贝叶斯(Naive Bayes)、神经网络(Neural Network)、偏最小二乘判别分析法(Partial Least Squares Discrimination)和支持向量机模型(Support Vector Machine)，并随机选择30个蜂蜜样品对其进行验证。结果表明，决策树(Decision Tree)和朴素贝叶斯(Naive Bayese)模型综合预测水平比较好，总体样品预测准确率分别为93.3%和83.3%，具有较高准确性。因此，利用电感耦合等离子体质谱法测定蜂蜜中同位素铅、硼同位素比值来进行产地溯源的方式是可行的。<br />
In order to better identify the origin of honey and ensure the quality and authenticity of honey, the Pb and B isotopes of eighty-four honey samples from area A and seventy honey samples from area B which were tested by ICP-MS. The principal component analysis method was used to analyze the experimental data and found that the samples could be well dispersed. According to the experimental data, five prediction models were established through Mass Profiler Professional, namely Decision Tree, Naive Bayes, Neural Network, Partial Least Squares Discriminant analysis method and Support Vector Machine model, and thirty honey samples were randomly selected for authen-ticating. The results show that the comprehensive prediction levels of Decision Tree and Naive Bayese models were better, and the overall prediction accuracy of the samples was 93.3% and 83.3%, respectively, with high accuracy. Therefore, it is feasible to use Inductively Coupled Plasma Mass Spectrometry to determine the isotope ratio of Pb and B in honey to trace the origin.</span> 
</div>
%K 电感耦合等离子体质谱法(ICP-MS)，同位素，蜂蜜<br>Inductively Coupled Plasma Mass Spectrometry (ICP-MS)
%K Isotope
%K Honey
%U http://www.hanspub.org/journal/PaperInformation.aspx?PaperID=38758