基于近红外的多相偏最小二乘模型组合分析实现茶叶原料品种鉴定与溯源的研究

Study on Identification and Traceability of Tea Material Cultivar by Combined Analysis of Multi-Partial Least Squares Models Based on Near Infrared Spectroscopy

为实现茶叶的品种溯源,以近红外光谱分析技术为基础,尝试采用PLS、欧氏距离等方法的组合来实现茶叶的特殊原料品种的鉴别、具体品种识别分析。结果表明:研究中以4个品种(龙井43,群体种、迎霜和乌牛早)的茶树鲜叶为制成的茶叶样本为材料,分别建立了4个不同原料加工成的茶叶样本的PLS鉴定模型,对定标集样本识别的准确率分别为89.8%,90.9%,96.1%和99.5%,验证集的识别准确率分别为87.1%,84.2%,96.1%和97.5%;而采用4个品种鉴定模型的组合分析"初次识别"并且结合欧氏距离法进行"二次识别"后,最终对4种不同品种茶叶样本的原料品种的识别准确率分别为90.3%(定标集)和83.5%(验证集)。该研究为实现特殊品种加工成茶叶的鉴定以及成品茶的品种溯源提供了一种参考方法。

The present study attempted to achieve the identification and traceability of tea material cultivar through combination of multi-partial least squares models and Euclidean distance,etc.The results indicate that with the samples manufactured with tea fresh leaves of cultivar Longjing 43,Qunti,Yingshuang and Wuniuzao as the analysis objects,4 models were established in this study so as to identify the tea with material cultivar being tea fresh leaves of cultivar Longjing 43,Qunti,Yingshuang and Wuniuzao separately by PLS.Their accuracy rate of identification of samples in the calibration set were 89.8%,90.9%,96.1% and 99.5%,respectively,while those in test set were 87.1%,84.2%,96.1% and 97.5%,respectively.After the ＂first identification＂ through the combined analysis of the four models for identification of tea material cultivar and the ＂second identification＂ adopting the Euclidean distance,the accuracy rate of material cultivar recognition for the tea samples was 90.3%（calibration set） and 83.5%（test set）,respectively.This study provided a reference method for the identification of tea manufactured with a specific material cultivar and the material cultivar traceability of the manufactured tea.