We found a match
Your institution may have access to this item. Find your institution then sign in to continue.
- Title
基于 GC-IMS 技术的福建白茶产地判别.
- Authors
罗玉琴; 韦燕菊; 林 琳; 林馥茗; 苏 峰; 孙威江
- Abstract
White tea is one of the six categories of tea. Fresh leaf picking, withering and drying are the three basic processing technology of white tea, which are relatively simple. White tea originated in Fujian Province, mainly produced in Fuding City, Zhenghe County, Jianyang county and Songxi County. Aroma is one of the important factors that determine the quality of tea. The main aroma components of Yunnan Yueyue white tea and Fujian Baihao Yinzhen tea were reported, but the differences of volatile aroma components of white tea from different main producing areas in Fujian Province were not clear. Gas Chromatography Ion Mobility Spectrometry (GC-IMS) is a new gas phase separation and detection technology in recent years, which has high resolution of gas chromatography and low detection limit of ion mobility spectrometry. In order to reveal the different volatile aroma components of white tea from different areas in Fujian Province, and to realize the rapid identification of white tea producing areas, GC-IMS technology was used to detect the volatile components of white tea from different areas in Fujian Province. Meanwhile, Linear Discriminant Analysis (LDA) was carried out to reduce the dimension of aroma data, and established a discrimination model of white tea producing areas combined with chemometrics method. The results showed that the contents of volatile compounds in white tea among the producing areas of Fuding, Fu’an, Zhenghe, Jianyang and Songxi were different. The white tea samples of Zhenghe, Jianyang and Songxi had higher similarity, and lower content of volatile aroma substances. Both GC-IMS spectrum data and 241 kinds of labeled aroma compounds data could be used to distinguish the origin of white tea, and LDA based on marker material data was better than it based on GC-IMS spectrum data. The discriminant rates of K Near Neighbor Linear Discriminant Analysis (LDA-KNN), Multi-Layer Perceptron Linear Discriminant Analysis (LDA-MLP) and Support Vector Machine Linear Discriminant Analysis (LDA-SVM) model based on the GC-IMS spectrum data were 91.84%,93.88% and 93.88%, respectively. By comparing the three patterns of misjudgment samples, it was found that the origin misjudgment occurred between Zhenghe white tea and Songxi Jianyang white tea, which was related to the small difference of volatile aroma components and high similarity of samples. The results showed that the discriminant rates of Adaboost Linear Discriminant Analysis (LDA-Adaboost), Decision Tree Linear Discriminant Analysis (LDA-Decison Tree), LDA-KNN, LDA-MLP, Random Forest Linear Discriminant Analysis (LDA-Random Forest) and LDA-SVM were 100%. The positive discrimination rate of the origin model based on the marker substance was higher than that based on the GC-IMS spectrum data. All six discriminant models based on the labelled substances data could effectively distinguish the origin of white tea. The results of this study can provide technical support for the origin protection of Fujian white tea.
- Subjects
FUJIAN Sheng (China); FISHER discriminant analysis; SUPPORT vector machines; ION exchange chromatography; RANDOM forest algorithms; SEPARATION (Technology); ION mobility spectroscopy; FREEZE-drying
- Publication
Transactions of the Chinese Society of Agricultural Engineering, 2021, Vol 37, Issue 6, p264
- ISSN
1002-6819
- Publication type
Article
- DOI
10.11975/j.issn.1002-6819.2021.06.032