Sensory And Chemical Quality Evaluation Of Longjing Tea Using Chemometrics And Near-infrared Spectroscopy Technique

Tea is one of the popular hobby beverages,and its quality evaluation is always concerned in the industry and scientific researches.To date,the assessment of tea quality is mainly based on panel sensory by experts throughout the production and consumption of tea.The sensory evaluation demands restricted operating environment and professional experts,and are easily affected by multiple factors.This method is difficult to meet the needs of production and consumer-oriented development.The present thesis was focused on one of the most famous green tea in China– Longjing tea.Food sensory analysis,chemometrics and Near-infrared spectroscopy(NIR)technique were carried out for systematic studies on sensory and chemical quality evaluation of Longjing tea,including consumer-sensory assessment,chemical supporting of sensory quality,rapid and non-destructive assessment of sensory quality and determination of quality-related chemicals.The results provided novelty idea of consumer-sensory assessment for other kinds of tea products,and interpreted the underlying mechanism of tea quality chemistry and spectrochimica.The rapid and non-destructive assessment of sensory quality and determination of quality-related chemicals of Longjing tea using NIR technique might be applied in production and consumption.Here are the main achievements of this thesis.1.Eighteen consumer-lexicons were developed to evaluate sensory flavor of Longjing tea using the optimal Check-All-That-Apply test.The results of multiple factor analysis and other chemometrics shown that the lexicons could be enough to evaluate the quality of Longjing tea,and the most frequently used lexicons were “astringency” and “umami”.2.With the sensory evaluation and determination of quality-related chemicals of Longjing tea,partial least squares(PLS)were applied to establish relationship between chemicals and sensory quality,umami taste and astringency taste,respectively.The coefficients of determination were more than 0.81.The results of variable importance for projection shown that the key chemicals of total sensory quality of Longjing tea were epigallocatechin gallate(EGCG), epigallocatechin(EGC),total catechins(T-cat),theanine(Thea),glutamic acid(Glu),myricetin-glucoside(Myr-glc),myricetin-galactoside(Myr-gal),kaempferol-glucoside(Kae-glc),kaempferol-rutinoside(Kae-rut).3.Non-destructive and quantitative assessment of sensory quality for Longjing tea was developed by utility of NIR combining with PLS and synergy interval-PLS.The best performance in this thesis was achieved by the application of synergy interval-PLS and multiplicative scatter correction plus mean centering pretreatments,and correlation coefficients were more than 0.90.Chemometrics results shown that specific spectral regions obtained by synergy interval were more likely correlated to the presence of catechins,flavonol glycosides and alkaloids.4.Non-destructive and qualitative analysis of umami and astringency for Longjing tea was carried out by application of NIR combining with optimal support vector machine.The prediction accuracy was 75.18%.The best performance in this thesis of umami model was provided by first derivation combining with particle swarm optimization-support vector machine.The best performance of astringency model was provided by mean centering combining with grid search-support vector machine.5.An NIR based model to predict concentrations of 10 key chemicals simultaneously was established using back propagation neural network,including EGCG,EGC,T-cat,caffeine,Thea,Glu,Myr-glc,Myr-gal,Kae-glc and Kae-rut.The model had higher prediction accuracy than the models developed by each compound,with correlation coefficient of 0.92.When more than 200 samples were treated,this model performed well and tended to be stable.