Research On Analysis Method For Major Components In Green Tea Based On Near Infrared Spectroscopy

Green tea is a kind of tea with the highest yield and the maximum export volume in China,and it is one of the three temperance drinks liked by people at home and abroad not only for its health effectiveness of thirst,but also for a variety of medical and health care functions of it.Moisture content,tea polyphenols,caffeine and free amino acids are four important kinds of quality components in green tea,whose content values have directly impact on the color,smell and taste of green tea leaves.Traditional methods of chemical analysis of these contents have defects,such as high cost,long time consuming, easy to produce pollution,destructive and poor repeatability.Therefore it is necessary to explore new,fast and nondestructive analysis method of quality components in green tea.In this paper,158 comminuted and non-comminuted green tea samples were scanned respectively by two different types of NIR spectrometer(NIRSystems6500-type and InfraXact Lab-type) based on modern Near Infrared Spectroscopy(NIRS) analysis technology.The quantitative analysis models for the four kinds of major components content in green tea were established respectively by Principal Components Regression (PCR),Partial Least Squares(PLS) and Modified Partial Least Squares(MPLS) methods. And the optimal calibration models were evaluated by objective function method.Then the prediction performance of calibration models was validated by independent validation sets.The main research results are as follows:(1) By comparison and analysis,the experimental results show that all the optimal models were obtained by MPLS calibration method except for the moisture content of comminuted samples which was obtained by PLS calibration method using the NIRSystems6500-type spectrometer.(2) The spectrum pretreatment method of the optimal calibration models for moisture content,tea polyphenols,caffeine and free amino acids of comminuted samples using the NIRSystems6500-type spectrometer was MSC,2,4,1,1,MSC,2,4,4,1,MSC,1,4,1,1, SNV+D,1,8,8,1 respectively.And the comprehensive evaluation index of the optimal calibration models(value of f) was 99.05%,96.07%,95.52%,93.42%;the standard error of calibration(SEC) was 0.1591%,0.9314%,0.1626%,0.0993%;and the determination coefficient of calibration(R_C~2) was 0.9856,0.9473,0.9167,0.8751.Using independent validation sets to validate the optimal models,the result is that the relative standard deviation(RSD) was 5.083%,3.092%,5.509%,5.181%respectively;the relative prediction deviation(RPD) was 3.580,3.684,2.521,2.093;the standard error of prediction(SEP) was 0.338%,0.999%,0.238%,0.129%;and determination coefficient of prediction(R_P~2) was 0.9261,0.9219,0.8116,0.7773.While using the InfraXact Lab-type spectrometer,the optimal spectrum pretreatment method was SNV+D,2,4,4,1,MSC,2,4,1,1,SNV+D,2,4,1,1,SNV+D,2,8,8,1 respectively; the value of f was 99.03%,95.96%,87.83%,92.17%;SEC was 0.1648%,0.9214%, 0.2525%,0.1020%;and R_C~2 was 0.9849,0.9410,0.7762,0.8521.The validation result is that RSD was 3.598%,3.835%,7.355%,6.040%;RPD was 5.042,2.904,1.861,1.854; SEP was 0.236%,1.243%,0.317%,0.151%;and R_P~2 was 0.9605,0.8159,0.6434,0.7416.(3) The spectrum pretreatment method of the optimal calibration models for moisture content,tea polyphenols,caffeine and free amino acids of non-comminuted samples using the NIRSystems6500-type spectrometer was MSC,2,4,4,1,MSC,2,8,4,1,SNV+D,1,8,1,1, MSC,1,8,4,1 respectively.And the value of f was 97.84%,94.86%,96.35%,90.20%;SEC was 0.2505%,1.0068%,0.1500%,0.1148%;and R_C~2 was 0.9638,0.9224,0.9320,0.8160. And the validation result is that RSD was 4.199%,4.138%,6.814%,5.737%;RPD was 4.353,2.733,1.980,1.857;SEP was 0.278%,1.339%,0.293%,0.144%;R_P~2 was 0.9508, 0.8830,0.7340,0.7174.While using the InfraXact Lab-type spectrometer,the optimal spectrum pretreatment method was MSC,1,4,4,1,SNV+D,1,4,1,1,MSC,2,8,1,1,SNV+D,2,8,4,1 respectively; the value of f was 96.60%,93.39%,94.69%,88.65%;SEC was 0.3238%,1.2122%, 0.1962%,0.1250%;and R_C~2 was 0.9400,0.8969,0.9030,0.7882.And the validation result is that RSD was 6.248%,5.657%,10.562%,6.320%;RPD was 2.954,1.930,1.197, 1.646;SEP was 0.413%,1.829%,0.451%,0.158%;and R_P~2 was 0.8912,0.8146,0.6780, 0.6243.The results show that quantitative analysis of the moisture content,tea polyphenols, caffeine and the total free amino acids in green tea is feasible by using NIRS.The determination of Moisture content,tea polyphenols and caffeine in comminuted green tea samples can be achieved high prediction accuracy and fast NIRS analysis,but the determination of the total free amino acids can be only achieved rudeness estimation.And the determination of Moisture content and tea polyphenols in non-comminuted green tea samples can be achieved high prediction accuracy,fast and nondestructive NIRS analysis, but the determination of caffeine and the total free amino acids can be only achieved rudeness nondestructive estimation.