Quality Detection Of Xinyang Maojian Tea Based On Electronic Nose And Electronic Tongue

Xinyang Maojian tea is one of China's top ten famous teas,and is also a famous specialty in Henan province.With bright color,it has a refreshing taste.In this study,the quality of Xinyang Maojian tea was detected by using electronic nose and electronic tongue.The odor information of Xinyang Maojian tea was collected by electronic nose detection;the basic taste and aftertaste of Xinyang Maojian tea was evaluated by electronic tongue;chemical detection method was used to detect the three important flavour materials affecting the taste of Xinyang Maojian tea,which include tea polyphenols,caffeine and Amino acid.This study used factor loading analysis,correlation analysis,and single factor analysis of variance etc.to optimize the sensor arrays,compared the effects of sensor array before and after optimization by using two dimensionality reduction methods including principal component analysis(PCA)and linear discriminant analysis(LDA),classified the pattern of Xinyang Maojian tea by using electronic nose and electronic tongue multi-sensor data fusion,and quickly predict the flavor materials of Xinyang Maojian tea.The main contents and conclusions are as follows:(1)The quality of Xinyang Maojian tea was identified by odor through the electronic nose sensor array multi-feature optimization fusion method.The normalization processing,factor loading optimization and single factor analysis of variance are used to optimize and extract the multi-feature information,and eliminate the redundant variables in the multi-feature information.The PCA and LDA diagrams show that the tea pattern recognition effect is improved after feature optimization and extraction.The electronic nose can evaluate the tea grade by collecting odor information.(2)The electronic tongue was used to detect the basic taste and aftertaste of Xinyang Maojian tea,and the information of the two was combined to identify the quality of Xinyang Maojian tea from the perspective of taste.Machine learning was performed on the fusion information by support vector machine,and the support vector machine model was optimized by grid search method and particle swarm optimization.The result shows that the supports vector machine model optimized by particle swarm algorithm have better ability of identifying tea quality.(3)The bitterness of tea was quickly predicted by an electronic nose sensor array.Correlation analysis was used to optimize the electronic nose sensor array,and the multiple linear regression(MLR),partial least squares regression(PLSR)and BP neural network(BPNN)recognition models of tea bitterness were established.The results of comparative analysis show that the BPNN model,which has the self-learning and self-adaptive ability,has stronger ability of predicting bitterness of tea.(4)The electronic nose and tongue multi-sensor information were combined to identify the tea quality and the detection of important flavour materials of tea.First,it extracted the features of the electronic nose and electronic tongue response value by PCA analysis,and then combined the extracted features,and finally it established multiple linear regression,multiple linear stepwise regression(MLSR)and quadratic polynomial stepwise regression(QPSR)model of the important flavour materials of tea.The results show that the QPSR model has higher precision in the prediction of tea polyphenols,caffeine and amino acid content,thus is suitable for the prediction of flavour material contents in tea.