Tea-quality Identification Method Based On Spectral Analysis Techniques

The tea consumption in our country is large, with the development of thecountry, the requirements to the tea quality are increasing. People distinguish it interms of gas, smell and pattern of the tea based on experience of the specialist. But itis subjective and ambiguity in some scene, and not form a scientific point of view toillustrate the problem, thus a persuasive method, both theoretically and experimentallyis need urgently. In recent years spectroscopy method development rapidly in China,applied widely in various industries, and, the characteristics of fast, non-destructive,reproducibility, etc. gradually affected in various industries. With the intensive andrapid development of modern theory of spectral analysis and chemical metrologysoftware, a analysis method of the tea quality is suggested combined both above,and a tea component analysis model in view of theoretical and experimental meaningis established, for the potential application on nondestructive testing of the tea quality.The primary work of this paper, base on the experimental opportunities nearinfrared spectrometer instrument, build the tea quality detection platform, under thenormal condition of the experiment and operation, the parameters of tea polyphenolsand caffeine content of four kinds of tea such as BiLuoChun, Tippy, Tieguanyin,Longjing are obtained nondestructively respectively. Experimental process is mainlymeasured for each tea polyphones and caffeine absorption spectra degree, in the caseof different concentrations of caffeine and polyphones, they correspond to the spectralcurve showing a certain gradient, individual mutations, but overall is not affected.The spectrum curve graph measured are often accompanied by noise in theexperiment, thus a method of smoothing preprocessing optimize spectrum is adoptedto eliminate the influence so the smoother curve found finally, and the peaks becomesmore apparent. In the experiment, it found that the presence of significant teapolyphones absorption peak at1872nm and2218nm, caffeine obvious absorptionpeak at1900nm and2212nm, and the establishment of each of the polyphones in teaafter smoothing at1872nm and2218nm via correction model content and degree ofabsorption, found a linear relationship, the conclusion that polyphone content andanalysis model absorbance at1872nm exacter than that of2218nm. At1872nm the teapolyphones of such Biluochun get correlation coefficient0.9378and root mean squareerror0.008015via correction model, then in the same condition0.9211and0.003173 are obtained at2218nm. And, the factor of0.9317and0.07489at1900nm,0.9864and0.00208at2212nm are obtained for the term of caffeine.Finally, via matlab software a variety of tea polyphones correlation coefficientmap relevant to the real and predicted values is established to correct the experimentalmodel, and tend to close in general, so it is better to explain that tea polyphones andcaffeine ingredients detection model is accurate. It is hoped the study may offer avaluable reference for subsequent research on detection of tea quality.