Fuzzy Logic and Artificial Neural Network for Quality Improvement of Black Tea

Abstract

This paper discussed quality improvement of black tea using fuzzy approach on quality functions deployment end the development of backpropagation neural network using the software NWPH plus. The research was conducted at PTPN VIII tea Industry, Goalpara plantation. Result of the study showed that, parameter first priority based on customer evaluation was tea flavour. The Important process parameter of black tea based on result of fuzzy relationship matrix was the withering process. Based on the test of "trial and error' of network training process, the best network architecture for withering process monitoring [3-15-1] was obtained, that is 3 neurons in input layer, 15 neurons in hidden layer end 1 neuron in output layer. Three inputs ant one output consist of time. How, temperature and moisture content. The result suggests that development of backpropagation neural network can be used for process evaluation of withering processes.