STRUCTURE DETERMINATION OF A SEVERE NONLINEAR PROCESS: Application of Techniques to a Fed-batch Bakers' Yeast Process

Abstract

The fed-batch Bakers' yeast fermentation is a nonlinear and time-varying process. A simulation model of the process has been made. Three nonlinear structure determination techniques have been tested on the simulated process: an extended Wiener-Hammerstein cascade model described by Volterra kernels (HABER, 1989), a nonlinear model, which is linear in the parameters (BILLINGS et al., 1989) and the Group Method of Data Handling (FARLOW, 1984). The objective of the work was to determine the structure and parameters of a SIS0 model of the glucose flow (input)- respiratory quotient (output) dynamic relation of the nonlinear process. The technique of Billings is found to be most flexible with respect to the tuning knobs, the choice of model structure and the estimation of the parameters. Note that all techniques can be applied to any other nonlinear process.