Automatic fuzzy modeling for Ginijo sake brewing process using fuzzy neural networks

Abstract

Automatic fuzzy modeling was studied for Ginjo sake brewing process using a fuzzy neural network (FNN). From the analysis of data for 25 Ginjo sake brewings, the control period was separated into 4 regions. We constructed 4 FNN models for fuzzy control in each control region. Acquired models could estimate the set temperature precisely, and acquired rules coincided well with the experience of Toji. The suitability of acquired models was confirmed by the simulation proposed by us.