Models for sensory evaluation of beer and the beer brewing process were con
structed using a fuzzy neural network (FNN). A new method for optimal model
selection using a genetic algorithm and a SWEEP operator method was compar
ed with a conventional method using the parameter increasing method. As the
result, the new method was useful for the optimal model selection by simpl
ifying the model structure, improving the reliability of fuzzy rules, and a
ccelerating the calculation speed (about 10 times as fast as conventional m
ethod) for constructing the model with high accuracy. The percentage of cor
rect answers of the sensory evaluation model is 92%. The important variable
s are selected as the input variables, and the obtained fuzzy rules in mode
ling coincide well with knowledge data bases acquired by process operators,
and it is proven that the obtained FNN models are adequate.