Identification of green tea grade using different feature of response signal from E-nose sensors

Abstract

Detection of tea grade by a human taste panel is affected by external factors and usually inaccurate, but it might be promising to use an electronic nose (E-nose). In this paper an investigation has been made to determine the grade of different tea samples using an E-nose. Feature vectors of the teas with different quality grade (Labeled: T120, T600, T800, T1200 and T1800) were extracted from the E-nose response signals, and the data were processed by using the principle components analysis (PCA) and linear discriminant analysis (LDA). Using the average and integrated value of feature vectors, 100% correct classification by LDA was achieved for five different tea samples with different qualities. The results indicated that the E-nose was capable of discriminating different grades of green teas.