Optimization of Koji Making Process Using FNN and Cfga

摘要

Fuzzy neural network(FNN) models for estimation of enzyme activity in product koji from process variables of koji making process was contructed. Uisng this model and the genetic algorithm(GA), optimal temprture and humidiy orbits for objective value of enzyme activity were calculated. In conventional GA, the orbit is defined by only fitness function, whichis based on the difference between objective value and estimated one calculated using the using the models, although the accuracy of the models varied in any small data space. The confidence of the solution is never considered yet.