ROBUST ADAPTIVE PREDICTIVE CONTROL OF THE DISSOLVED GASEOUS ENVIRONMENT OF SUBMERGED MICROBIAL PROCESSES

The partial pressures of dissolved oxygen (pO(2)) and of dissolved car bon dioxide (pCO(2)) in laboratory-scale bioreactors cannot be effecti vely controlled by conventional controllers because of the constantly changing dynamics of the fed-batch processes, the large variety of the micro-organisms to be tested and the disturbances encountered during cultures. A strategy using a decentralized control of the pO(2) by the agitation rate and of the pCO(2) by the air flow rate has been develo ped. Each controller is made of an adaptive predictive algorithm based on the Generalized Predictive Control, and uses a strategy driver to gradually update its smoothness along the fed-batch. This totally dece ntralized control strategy of a partially coupled multi-variable non-l inear system has been implemented on a PC computer and connected to th e existing hardware without modifications. It needs no manual tuning a nd has been successfully used with various microorganisms despite nume rous environmental disturbances. (C) 1998 John Wiley & Sons, Ltd.