Intelligent control of green tea fixation with Microwave Processing

A microwave system for green tea fixation was built in this study. A machine vision was used to capture the tea leaves' images during the fixation process. The RGB, L*a*b*, dissimilarity, entropy, contrast, homogeneity, correlation, and energy were online calculated based on the original image information. In a ‘forward selection’ method, the G, B, correlation, and the absolute value of ‘a*’ were used to predict the moisture content with genetic algorithm combined with back propagation neural network. The predicted moisture content was used to terminate the fixation process automatically. To optimize the whole fixation process, a fuzzy logic controller was designed to control the temperature and microwave power continuously. The control was based on the physicochemical changes of the tea leaves, which was predicted with the extracted image information. With the intelligently controlled fixation, the final product quality was greatly improved, with the tea polyphenol content of 16.23%, amino acid of 6.67%, phenol ammonia ratio of 2.43, chlorophyll content of 9.33 mg/g, polyphenol oxidase activity of 4.98 U/mL, and sensory score of 91.5.