摘要
用近红外光谱仪对绿茶毛峰、粗信阳毛尖、铁观音、峨眉雪芽等4种茶叶,120个样品进行扫描,获得近红外光谱数据.先随机选取2/3样品数据为模型集,其余为验证集,用支持向量机(SVM)将模型集构建模型,经验证集验证,最终获得了97.5%的识别率.又用主成分分析(PCA)的方法对数据降维,用前两个主成分画出4种茶叶分类图,并与SVM结果比较,表明SVM的方法更佳.
NIR spectra of 120 samples of four kinds of tea (Lvchamaofeng, Cuxinyangmaojian, Tieguanyin, Emeixueya were scanned by a NIR spectrometer. The spectral data of 2/3 samples were selected randomly as a model set, and the spectral data of the rest samples were validation set. A support vector machine was used to build a model based on the model set; the model was validated by the validation set, and the final identification rate was 97.5%. The spectral data was subjected to dimension reduction treatment by principal component analysis (PCA); the classification map of the four kinds of tea was mapped by using the first two principal component, and then was compared with the SVM result. The comparison result showed that the SVM method was superior.
出处
《河南工业大学学报(自然科学版)》
CAS
北大核心
2013年第5期53-57,共5页
Journal of Henan University of Technology:Natural Science Edition
关键词
近红外
支持向量机
主成分分析
茶叶
分类
near-infrared spectroscopy
support vector machine
principal component analysis
tea
classification