Tea quality evaluation by applying E-nose combined with chemometrics methods.

Area of Science:

• Food Science
• Analytical Chemistry
• Sensory Science

Background:

• Tea quality control is crucial due to its global popularity and distinct flavor profiles.
• Objective evaluation methods are needed to complement traditional sensory assessments.
• Volatile organic compounds significantly contribute to tea's aroma and quality.

Purpose of the Study:

• To evaluate the feasibility of using an electronic nose (E-nose) for assessing tea quality grades.
• To develop and compare classification models for tea quality based on E-nose data.
• To investigate the quantitative prediction of specific volatile compounds in tea using E-nose technology.

Main Methods:

• An electronic nose (E-nose) was employed to detect volatile components in tea leaves and infusions.
• Feature extraction involved "35th s value", "70th s value", and "average differential value".
• Data reduction techniques (PCA, MDS, LDA) and classification algorithms (LR, SVM) were utilized.
• Quantitative analysis employed Support Vector Machine (SVM) with Linear Discriminant Analysis (LDA) for volatile compound prediction.

Main Results:

• Linear Discriminant Analysis (LDA) improved classification efficiency compared to PCA and MDS.
• Support Vector Machine (SVM) outperformed Logistic Regression (LR) in classification models.
• SVM achieved 100% classification accuracy for tea infusion grades using LDA-processed data.
• Satisfactory prediction R² values (0.8980–0.9617) were obtained for linalool, nonanal, and geraniol using SVM with LDA.

Conclusions:

• The electronic nose (E-nose) is a feasible tool for both qualitative and quantitative analysis of tea quality.
• LDA combined with SVM provides a robust approach for accurate tea quality classification.
• E-nose technology offers a promising avenue for objective and efficient tea quality assessment.