基于矿质元素含量和支持向量机的茶叶鉴别分析

Identification of tea based on mineral content and support vector machines

为了实现茶叶种类与产地的识别,提出了一种基于矿质元素和支持向量机的茶叶鉴别方法.该方法首先运用ICP光谱仪测定30个茶叶样本中的Mg,Al,P,Ca,Mn,Fe,Cu,Zn,Ba等共16种元素含量,接着对采集到的数据进行标准化处理,随机抽取样本用于设计训练基于支持向量机的多元分类器,然后对测试样本进行种类与产地识别.试验结果表明,采用"一对一"的多分类支持向量机方法比聚类分析具有更好的抗干扰性和更强的分类能力,在小样本的情况下对茶叶种类和产地的识别率均达到91.67%,能有效进行茶叶鉴别.

In order to identify variety and origin of teas,a method was proposed based on mineral content and support vector machines（SVM）.The contents of Mg,Al,P,Ca,Mn,Fe,Cu,Zn and Ba were analyzed by ICP-OES and were normalized.The data were collected randomly as learning samples for designing and training multielement classifier to identify tea variety and origin by SVM.The results show that classification method which is based on ″one versus one″ multi-class support vector machine has better classification ability and stronger anti-jamming capability than that of cluster analysis.For small samples,the tea variety and origin identification accuracy can reach 91.67%,which illuminates that the method is effective for indentifying tea variety and origin.