支持向量机微量元素分析法判别乌龙茶,红茶与绿茶.doc

支持向量机－微量元素分析法判别乌龙茶，红茶与绿茶 陈念贻 1，陆文聪1 ， 陈瑞兰1， 叶晨洲 2 ，李国正2 （1.上海大学化学系计算机化学研究室 ，上海 200436 ； 2.上海交通大学图象及模式识别研究所，上海 200030） 摘要: 茶叶中的微量元素含量与产地土壤中的微量元素含量有关，故可用多种微量元素的分析数据结合模式识别计算判别茶叶产地和品牌。本工作根据三类茶叶中Zn, Mn, Mg, Cu, Al, Ca, Ba, K八种元素的分析数据，用支持向量机算法判别乌龙茶、红茶或绿茶，并用留一法检验其预报能力。计算表明：支持向量机算法结果优于Fisher 法和 KNN法。采用福建安溪铁观音乌龙茶样品的元素分析结果作测试样品,检验本工作所得数学模型，判别结果亦与实际符合。 关键词: 乌龙茶判别，微量元素，支持向量机算法 中图分类号：O 06-04 ( Support Vector Machine Applied to Differentiation of Oolong Tea and Black Tea or Green Tea Chen Nianyi1 ， Lu Wencong 1 ， Chen Ruilan1，Ye Cheng-zhou2， Li Guo-zhen2 （1．Laboratory of Chemical Data Mining, Department of Chemistry, Shanghai University, Shanghai 200436，china; 2．Institute of Image and Pattern Recognition, JiaoTong University, Shanghai 200030, china） Abstract: Since the content of trace elements of tea is related to the local soil composition, it seems possible to use the results of trace element analysis to differentiate special brand of commercial tea products. In this work, the contents of eight trace elements: Zn, Mn, Mg, Cu, Al, Ca, Ba and K in tea samples have been used for the differentiation of Oolong Tea from black tea or green tea. The cross validation by leaving-one method has been used to compare the prediction ability of support vector machine method with KNN and Fisher method. It has been found that the prediction result by support vector machine is better than that of KNN or Fisher method. Besides, the mathematical model obtained has been tested by a new sample of oolong tea produced in Anxi county of Fujian province, and the computerized prediction result is in agreement with the fact. Key words: Differentiation of oolong tea, support vector machine, trace element content 1 引言 乌龙茶是我国出产、国际知名的茶叶品牌。乌龙茶罐装饮料以其有一定的防诱变效果已打入国际饮料市场。我国乌龙茶有出口优势。故判别乌龙茶和其他茶叶品种在商品检验中有一定的实用意义。鉴于乌龙茶产地特殊，而茶叶中的微量元素与产地的土壤成分有关，有可能利用茶叶中多种微量元素的分析结果靠模式识别方法区别乌龙茶与它种茶叶。支持向量机(support vector machine，简称SVM) 是Vapnik等近年来根据新发展的统计学习理论建立的新算法。其特点是能最大限度防止统计预报中的过拟合现象，即拟合效果很好而预报效果不佳的现象（此问题在人工神经网络应用中