Modelling, estimation, and control of baker's yeast production

Abstract

Conventional control of biotechnological processes is often difficult because of the inherent complexity of the processes and the lack of measurement equipment. The purpose of this paper is to demonstrate that one such process---- Baker's yeast fermentation --- can be controlled by a linear-quadratic controller based on a nonlinear dynamic model and ordinary industrial measurement equipment. The model is established mainly from fundamental laws, and unknown parameters are estimated from experiment input and output data. A Kalman filter is designed to estimate unmeasurable states from measured output. The states are then controlled to a prespecifled trajectory by the LQ-controller. The state estimation is tested against the experiment data. The controller behavior is investigated by simulation. All theories used here are simple, well-understood and easy to implement on digital computers. The measurement equipment Is available on the market.