Identification Of Tea Varieties By Infrared Spectroscopy Combined With Pattern Recognition

There are several methods for the division of Chinese tea,which can be divided into six varieties according to the method of making and the degree of the oxidative fermentation of tea polyphenols : green tea,white tea,yellow tea,oolong tea,black tea and dark tea.At present,the application of infrared spectroscopy has developed into an emerging modern detection and identification method,which has certain scientific significance for the research of tea classification.The traditional identification methods of Chinese tea include sensory evaluation,chemical fingerprint verification technology and biometric fingerprint verification technology.These methods are subject to subjective and external environmental factors,so that the identification process has some limitations.Mid-infrared spectroscopy is a simple,fast and accurate identification method.By taking all the spectral information of tea,the mathematical model of spectral data was established by statistical method.Therefore,the samples can be classified quickly and accurately and objectively.This paper was studied based on Fourier Transform Infrared Spectroscopy（FTIR）combined with the dentification mode for the classification tea from six varieties.The main finished work and achievements are summarized as follows:1.Six tea varieties were characterized by Infrared Spectroscopy.In this paper,Fourier Transform Infrared Spectroscopy was used for qualitative analysis of different tea varieties.The infrared spectra of 42 tea samples were measured.The functional groups of chemical compositions were determined by the absorption peak position,absorption intensity and shape of the original spectrum,and the differences of infrared spectrum of six tea varieties were analyzed.2.The infrared spectral characteristic bands of different tea varieties were found.The similarity analysis of six tea varieties was carried out.According to the principle of tea fermentation and the original spectrum analysis,characteristics of spectral band were choose.Their second derivate spectra in the range 1400-900cm^-1 were calculated and the infrared spectral data was compressed.Their absorption peaks in the fingerprint area were analyzed.3.The principal component analysis（PCA）was used to identify six varieties of tea.The principal component analysis of the spectral data of 1400-900cm^-1 was made by SPSS20.0.The accuracy rate of the principal component model was 100% and the accuracy rate was 96.7%.The result shows that it is feasible to classify tea from six varieties by mid-infrared spectroscopy and realize the rapid identification of six tea varieties.4.The cluster analysis（HCA）of 42 tea samples was conducted.Their second derivate spectra in the range 1400-900cm^-1 were used to build model by systematic cluster in this work,and dark tea,black tea,oolong tea,yellow tea,white tea and green tea are each grouped together in a distance of 5.The result shows that cluster analysis is effective for tea classification,which is consistent with the similarity analysis and principal component analysis.