基于三维色谱指纹图谱的茶叶质量等级评定新方法研究

Tea, as the China's competitive agricultural products, however, suffers from a number of shoddy or fake goods in the trade market. Nowadays, the identification of tea quality mainly depends on human sensory evaluation, which is a widely used method with outstanding advantages but also has some deficiencies such as strong subjectivity and difficult quantification. Chromatographic fingerprints can provide a new approach for the comprehensive evaluation of tea quality. But the traditional chromatographic fingerprints analysis is susceptible to factors such as peak overlap, time drifts or baseline drifts. What's more, the information mining in the traditional chromatographic fingerprints analysis is usually not enough. To address these problems, this project plans to select green tea samples of different quality grades as the research object, and construct three-way chromatographic fingerprints (retention time × spectrum/mass spectrum × sample) of tea by liquid chromatography coupled with diode array detection or mass spectrometry. The three-way data analysis methods based on second-order tensor decomposition in combination with the chemical pattern recognition methods are employed to establish the quality discrimination model of green tea and then screen key quality-related chemical components. Qualitative and quantitative analysis are carried out to identify these key quality-related components for the establishment of quantitative relationship model between micro-chemical composition and macro-quality of green tea. This study can lay a theoretical foundation for the development of objective and quantifiable quality evaluation system of tea quality, provide theoretical guidance for the quality control of tea, and play a positive role in promoting the healthy development of tea industry.

茶叶是我国的优势农产品，市场上以次充好、虚标质量等级等问题严重。现行的茶叶质量等级感官审评方法优点突出、应用广泛，但主观性强、难以量化。色谱指纹图谱为综合评定茶叶的质量提供了新思路，但常规色谱指纹图谱易受色谱峰重叠、时间/基线漂移等因素影响，且信息挖掘不足。本项目拟以绿茶为研究对象，选取不同质量等级的绿茶样本，采用液相色谱联用二级管阵列或质谱检测器构建茶叶的三维色谱指纹图谱（保留时间×光谱/质谱×样本），借助基于二阶张量分解的三维数据分析方法与化学模式识别方法相结合建立绿茶的质量判别模型，筛选出关键的品质相关成分，并对其进行定性定量分析，探明关键品质成分组成与绿茶质量等级间的关系，建立绿茶微观化学物质组成与其宏观品质间的定量关系模型，最后用于绿茶宏观品质的预测。本研究可为开发客观、可量化的茶叶质量等级评定系统奠定理论基础，为茶叶的质量控制提供理论指导，对茶产业的健康发展起到积极的促进作用。

茶叶市场上以次充好、虚标质量等级等问题严重。目前常用的感官评定方法存在主观性强、难以量化的问题。本项目基于三维色谱指纹图谱结合化学计量学多维校正与分辨、多元统计分析方法开展了绿茶质量等级评定方法的研究。主要内容如下：1）多维数据分析理论方法研究:①建立了有峰重叠和时间漂移存在情况下指纹谱信息提取方法，并证实了交替三线性分解方法的抗轻微时间漂移的能力；②建立了食品真实性判别的多维模式识别模型，判别准确率可达100%。2）基于三维色谱指纹图谱的茶叶质量品质评定方法研究。①建立了基于特定目标组分的茶叶等级评定方法。采用HPLC-DAD结合基于交替三线性分解算法的多维校正方法实现了茶叶中10种主要成分快速定量分析，基于此实现了西湖龙井茶的等级分类。②采集了茶叶样品的HPLC-DAD和UPLC-DAD三维色谱指纹图谱，结合多元曲线分辨-交替最小二乘建立了三维色谱指纹图谱中色谱、光谱信息提取方法，基于指纹图谱中提取的色谱信息建立了茶叶采摘季节、产地和等级判别模型，对茶叶采摘季节正确识别率可达91.3%，对浙江和山东两个产地的正确判别率可达92.8%，通过变量筛选发现并鉴定了相应的特征标志物；基于色谱指纹图谱进行绿茶等级判别研究发现二级和三级茶的识别率较低，存在二级和三级间的误判。3）基于代谢指纹图谱的茶叶质量等级评定方法研究。①采用LC-QTOF-MS建立了不同等级绿茶非挥发物代谢指纹图谱，并结合化学计量学多元统计分析方法建立了绿茶等级的判别模型，OPLS-DA两两分类模型R2和Q2值均大于0.6。最终筛选并鉴定得到正离子模式下30种差异代谢组分，负离子模式下12种差异代谢组分，可作为绿茶等级区分和品质预测的标志物；②采用GC-MS建立了不同等级宜昌大叶种绿茶中的挥发性成分代谢指纹图谱，共鉴定到了94种挥发性成分，最终筛选得到了16种共有的差异挥发性代谢组分，可用于高等级和低等级绿茶区分。本研究为开发客观、可量化的茶叶质量等级评定系统奠定了理论基础，对茶叶品质评价的标准化具有重要现实意义。