标题:
基于电感耦合等离子体质谱仪在饶河蜂蜜溯源中的研究与应用The Research and Application of Raohe Honey Tracing Based on Inductively Coupled Plasma Mass Spectrometer
作者:
张海华, 马占峰, 刘志勇
关键字:
饶河蜂蜜, ICP-MS, 主成分分析, 预判模型Raohe Original Honey, ICP-MS, Principal Component Analysis, Forecasting Model
期刊名称:
《Hans Journal of Agricultural Sciences》, Vol.5 No.3, 2015-06-30
摘要:
本研究对102个饶河蜂蜜和31个外地蜂蜜中的硼同位素和锶同位素通过电感耦合等离子体质谱(ICP-MS)进行检测。采用主成分分析法将7个同位素丰度比值变量降维至4个主成分,并选取了主成分1、主成分2和主成分3进行模型的建立。用三个主成分新变量通过Agilent MPP数据分析软件建立Decision Tree、Naive Bayes、Neural Network、Partial Least Square Discriminate和Support Vector Machine五种模型,最终优选出以Component1、Component2和Component3交互图建立的Decision Tree模型作为饶河蜂蜜的产地溯源模型,判别准确率为93.75%。The study detected Boron isotope and Strotium isotope of 102 Raohe origin honeys and 31 nonlocal honeys with ICP-MS. We used principal component analysis (pca) to dimensionally reduct 7 isotope abundance ratio variables to four principal components, and selected three of the four principal components to build models. By Agilent MPP data analysis software, we built five models with the three new principal components. The models are: Decision Tree, Naive Bayes, Neural Network, Partial Least Square Discriminate and Support Vector Machine. Finally, we optimized Decision Tree model as Raohe origin honey traceability model, the resolution accuracy rate is 93.75%.