论文部分内容阅读
Green tea is one of the six major teas in China with the longest history, the highest output and the widestsphere of consumption.Among all the sensory evaluation indexes of green tea,color, the most intuitive impression, plays a particularlyimportant role.Various kinds of components in green tea have an impact on the color, among which, chlorophylls are the principal pigments for green tea.As for made green tea, chlorophyllsare often accompanied by the presence of their various derivatives, such as pheophytins, chlorophyllides, and many others.Currently, the separation, identification andquantitative determination of chlorophylls and their derivatives from photosynthetic organisms, including tea leaves have been studied by instrument based approaches such as high performance liquid chromatography, thin-layer chromatographyandcolorimetric method forquality monitoring.However, these methods areexpensive,laborious,time consumingand destructive, which cannot mect the requirements of rapid online detection for large scale samples.Therefore, to overcome the disadvantages of the current detection methods, NIR spectroscopy was proposed as an alternative to conventional techniques.In this research, NIR spectroscopy coupled with chemometrics methods was applied for the determinations of chlorophylls and their derivatives in green tea.15 kinds of teas were purchased from the market, and 10 repetitions in each kind were prepared, thus a total of 150 samples were investigated in this research.NIR spectra of the samples were acquired on a FOSS NIR Systems 6500 spectrometer and the spectra were collected in the wavelength range of 400-1100 nm.The corresponding chemical values of chlorophylls, pheophytins andchlorophyllides were measured by colorimetric methods.Then,wavelettransformation (WT) was adopted to extract information in different time and frequency domains from NIRspectra.The spectra were decomposedinto five parts (cdl, cd2, cd3, cd4 and ca4) on four levels and partial least squares (PLS) regression models were respectively built for the five parts.Themodeling results indicatedthat the low-frequency approximation signal (ca4) was the most important information forthe establishment of the measurement models, and the corresponding PLSregression models obtained good performance in prediction sets with all Rpabove 0.85 and RMSEP below1.35.To further explore the important wavelengths closely related to chlorophylls and their derivatives in green tea, two dimensional(2D) correlation algorithms was performedbased on the reconstructed spectra and the assignments of the selected characteristic wavelengths were also analyzed, which deepened the relation between NIR spectroscopy and the inner structures of chlorophylls and their derivatives.For thefirst time, we present the use of NIR spectroscopy for thedetermination of chlorophylls and their derivatives as the main pigments indried, green teas and the development of wavelettransformationin removing irrelevant information for good prediction of pigments content.Additionally, the advantage of twodimensional correlation algorithms for reliable band assignmentand an improved quality analysis is demonstrated.