A modular approach to situation identification of the dynamics of bacterial populations synthesizing poly-β-hydroxybutyrate

Biotechnological processes involving bacteria are strongly nonlinear. Therefore, both their productivity and the final product quality may be considerably improved by applying appropriate control strategies to modulate behavior of the bacteria during transitional states. This requires advance identification of indicative signals by off-line investigation (i.e. experimental analysis) and on-line monitoring, (i.e. real time evaluation). A modular scheme is presented for doing this, which incorporates an Extended Kalman Filter and a prediction filter. If this is based on a suitable process-feature vector, which must be chosen in advance, the system can provide sufficient information to trigger appropriate feedback signals. Thus, it can provide a key element in modular situation control, allowing continuously periodic process management. In this publication the individual modules involved, and their assembly into an integrated system are described. Potential problems concerning selection of the feature vector, and experimental results are also discussed.