基于数据融合的信阳毛尖茶品质检测技术研究

The aroma of tea is an important factor to determine the quality and flavor of tea. The microscopic aroma characteristics and characteristic aroma component of tea were analyzed using gas chromatography mass spectrometry coupled with different extracting method of aroma. Aroma intensity of tea were analyzed by electronic nose technology. Aroma quality discrimation of tea by gas chromatography mass spectrometry coupled with electronic nose technology is in the stage of exploration. . In order to study the mechanism of characteristic aroma components and aroma formation of Xinyang Maojian tea, the whole chemical volatile profile, composition and content of compound characteristics are analysed and evaluated by electronic nose and GC/MS. The characteristic aroma of tea are analysed by electronic nose technology coupled with GC/MS is a new method. The sensor array were optimized by variance and principal component analysis. The best sensing feature vectors representing the characteristics of the tea aroma are studied by feedback neural network and genetic network. In order to establish the relationship between sensing feature vector and tea microinformation from results of GC/MS, the Bayes and fuzzy theory were used. The characteristic aroma components and quality of Xinyang Maojian tea were analysed by three level data fusion. The micro and macro information of tea aroma were analyzed by electronic nose technology.This research will provide detection method for analysis and evaluation of tea aroma quality.

茶叶的香气是决定茶叶品质与风味的重要因素。气谱质谱结合不同的香气提取方法，可分析茶叶香气中一些微观的气味特征和特征香气组分。电子鼻技术可对茶叶的综合特性香气进行分析。把二者结合起来对茶叶香气进行系统研究还处于探索阶段。. 本项目将电子鼻和GC/MS系统联合起来进行信阳毛尖茶的挥发性成份分析，以期得到特征香气物质成分和香气形成的机理，是一种检测茶叶特征香气的新思路、新方法。项目通过方差和主成分分析对电子鼻气体传感器阵列进行优化，用反馈神经网络和遗传网络等方法寻觅电子鼻最能表征茶叶特征香气的传感特征向量，用贝叶斯和模糊理论等建立传感特征向量与GC/MS分析的微观信息之间的关系，对数据进行三级融合，分析信阳毛尖茶特征香气成分及其品质。从而能通过电子鼻技术来检测茶叶香气的微观和宏观信息。项目的完成将得到一种基于电子鼻和GC/MS融合检测茶叶香气方法。

茶叶的香气是决定茶叶品质与风味的重要因素。气谱质谱结合不同的香气提取方法，可分析茶叶香气中一些微观的气味特征和特征香气组分。本项目以信阳毛尖明前茶为研究对象，利用电子鼻技术测定了其香气并进行了识别分析，检测了不同等级的茶叶理化成分，利用GC/MS技术分析研究了信阳毛尖茶的香气成分。主要结论如下：. 通过电子鼻系统对不同等级的信阳毛尖明前茶进行检测。主成分分析和线性判别分析识别结果表明定性识别模式可以根据等级和品质对不同的茶叶进行分类检测。研究结果表明，氨基酸、茶多酚和咖啡碱含量与其品质之间存在明显的正相关。基于电子鼻技术的信阳毛尖理化成分预测模型研究表明预测模型能够对信阳毛尖茶的氨基酸、茶多酚和咖啡碱含量进行很好的预测。. 对不同等级的信阳毛尖茶的香气进行研究，结果显示信阳毛尖茶香气成分中酸的含量较高，与信阳毛尖茶品质呈正相关的香气物质为香叶醇、橙花叔醇、顺-茉莉酮和β-紫罗酮。利用定量分析方法建立信阳毛尖茶香气物质含量与电子鼻信号之间相关关系的数学模型，将该模型对茶叶香气物质的预测有很高的精度。. 对电子鼻传感器阵列进行多特征数据融合优化。对多特征向量矩阵进行优化，最后采用单因素方差分析，缩小不同品质茶叶组内间距，增大组间间距，更利于实现茶叶品质的区分。结果显示，主成分分析(PCA)可有效区分3种不同等级茶叶，因子载荷优化使各品质茶叶组内间距减小，单因素方差优化使一级与二级茶叶区分效果更明显，线性判别分析(LDA)效果要优于PCA分析3个不同等级的茶叶可得到极为明显的区分，研究结果表明，用多特征优化融合可有效提取电子鼻对茶叶的响应信息，有利于对不同品质茶叶进行识别。