No abstract

The concentrations of biomass, substrate and product are very important state variables of almost every bioprocess and generally unable to be measured directly in situ due to the lack of reliable sensors. In this paper, an adaptive observer of the biomass concentration is proposed for an anaerobic fermentation process where only the measurement of the acid product is available on-line. The observer was tested to be effective by several experiments under various operating conditions. In this experimental system, an auto-sampling device was connected between the bioreactor for the fermentation of Zymomonas mobilis and a HPLC so that the concentrations of glucose and ethanol could be directly measured through such implementation.

SUMMARYTwo adaptive root-locus control schemes for processes with time delays have been developed. The control design methodologies are based on classical feedback control and Smith predictor structures with simple pole placement. The overall behaviour of both control systems is adapted along the root loci of the system in each control horizon. Since the Smith predictor structure includes time delay compensation, the tuning gain can be enlarged to improve the performance of the control system. The optimal tuning gain is repeatedly updated by an optimization technique when the estimated model is confirmed. A comparison with simplified model predictive control (SMPC) is also given. SMPC in SISO systems is a special case of the proposed control laws.