Electrochemical fingerprints identification of tea based on one-dimensional convolutional neural network

Rapid identification of tea leaves is an important problem in food analysis. Electrochemical fingerprinting is a new analytical technique which is particularly good at identifying plant products. The work involved electrochemical fingerprinting of black, white and green tea. A one-dimensional convolutional neural network (CNN) structure suitable for electrochemical fingerprint classification is constructed through simulation experiments. The size and number of convolution cores and the structure of fully connected layers are determined. The classification effect of this CNN model is compared with the traditional classification methods and traditional classifiers. The results showed that the combination of electrochemical fingerprint and CNN could effectively identify the tea species.