The aroma quality and quantity of organic green tea are closely related to its storage time.In this study,organic green tea samples stored for 1 to 16 years at room temperature were analyzed using an electronic nose(E-nose)to assess their aroma profiles.Chemical chemometrics methods and machine learning techniques such as Kernel Ridge Regression(KRR),Support Vector Regression(SVR),and Back Propagation Neural Networks(BPNN)multivariate regression analysis methods were used to establish regression models between electronic nose sensing signals and tea storage time,and the performance of each model was evaluated.The results showed that the BPNN model exhibited strong nonlinear fitting capabilities,and the generalization performance of the BPNN model were further confirmed through validation in the test set.The research results indicate that the electronic nose analysis method based on Metal Oxide Semiconductor(MOS)sensors can be used as an alternative method for chemical analysis in evaluating the storage time of organic green tea.Combined with the BPNN model,it can serve as an effective predictive model to provide reference for the quality identification of long-term stored green tea in practical production.
关键词
绿茶/电子鼻/贮存期/回归模型/反向传播神经网络
Key words
Green tea/Electric nose/Storage time/Regression model/BPNN
维普
万方数据