Dynamic Model Fuzzy Transition in DMC Type Controllers for Varying Dynamics in Bioreactors

This research presents an approach to implement industrial type DMC (Dynamic Matrix Control) controllers in biological process where a fed-batch operation is required. In this type of operation a substantial change in the process dynamics occurs due to changes in volume, reaction rate, pressure, temperature, concentration, and other process variables. To compensate for changes in the dynamic, several matrix models are identified, and using a fuzzy supervisor a soft model transition is implemented, such that the controller and the process performance can adapt to the new operating conditions without affecting the process integrity. The control system design, its identification, and implementation are presented, focusing in the articulation between MPC (Model Predictive Control) and Fuzzy Supervision. The overall strategy is demonstrated at a simulation level using a benchmark fermentation process for Penicillin production in Fed-Batch mode. In this process glucose is used as a substrate, and a fungus, penicillium chrysogenum, is used in a fed-batch reactor to obtain penicillin. The model presents an unstructured approach to the reactor dynamic system, and the initial steps for this research required the development of an identification procedure for such zones.