Identification of a Simulated Continuous Yeast Fermentation

This work deals with multivariable adaptive identification of a simulated continuous yeast fermentation. The purpose of the identification is to obtain a model which can be used in a multivariable adaptive controller. A dynamic, structured, non-segregated model with eight state variables is used to simulate the process. A linear model (ARMAX-type) is identified by a Recursive Extended Least Squares method with a variable forgetting factor. Perturbations in two input variables are applied. Validation of the identified models is based on singular value analysis in the frequency domain.