摘要
应用现代近红外光谱分析技术,建立了绿茶中水分、茶多酚、咖啡碱和游离氨基酸4种主要成分的改进偏最小二乘(MPLS)定标模型,用目标函数法对不同光谱预处理方法进行考查,评定出最优模型,并对其进行了验证。各成分最优模型的目标函数值(f)分别为98.84%、95.66%、95.07%和94.25%,相对标准差(RSD)分别为4.53%、4.57%、8.33%和6.39%,预测决定系数(R2)分别为0.95、0.85、0.63和0.71。测定结果表明:应用近红外光谱分析可以实现绿茶中4种成分的快速定量检测,水分、茶多酚和游离氨基酸模型取得了良好的预测效果,可以用于实际检测;咖啡碱模型预测效果较差,需要进一步优化。
Using modern near infrared spectroscopy（NIRS） analytical technology, the modified PLS calibration models of moisture content, tea polyphenol, caffeine and free amino acids in green tea were established. The effects of different spectral data preprocessing methods were studied with object function method. The optimal NIRS models were evaluated and validated. The values of object function for each optimal model are 98. 84%, 95. 66%, 95. 07% and 94. 25% respectively. The relative standard deviations（RSD） are 4. 53%, 4.57%, 8.33% and 6.39 %, and the prediction determination coefficients（R2） are 0. 95, 0. 85, 0. 63 and 0. 71 respectively. The results show that using NIRS technology can achieve rapid quantitative determination of four kinds of composition in green tea. The prediction results of moisture content, tea polyphenol and free amino acids models are good; their models can be used in actual detection. However the prediction result Of caffeine model is poor; the model should be further opti- mized.
出处
《华中农业大学学报》
CAS
CSCD
北大核心
2009年第4期487-490,共4页
Journal of Huazhong Agricultural University
基金
湖北省自然科学基金项目(2007ABA351)资助
关键词
近红外光谱
绿茶
成分
快速测定
near infrared spectroscopy（NIRS）
green tea
composition
rapid determination
