ADAPTIVE ONLINE OPTIMIZING CONTROL OF BIOREACTORS - A 2-PHASE APPROACH

A two-phase approach of adaptive on-line optimizing control is applied for the first time to continuous fermentation processes, biomass productivity in a chemostat and penicillin productivity in an immobilized cell fluidized bed reactor. The results of dynamic simulation study show that the two-phase approach using nonlinear state-space models is more promising for continuous bioreactors than earlier approaches based on linear input-output models. The first phase of identification is able to cope with the plant/model mismatches through estimation of disturbances and adjustment of model parameters. The second phase of optimization is capable of reaching the optimal operating conditions. Adaptive on-line optimizing control is recommended for complex industrial biochemical processes due to its robustness, reliability, and immunity to measurement noises in the presence of plant/model mismatch.