Development of AI-assisted MPC controllers for industrial bioreactors

Bioreactors are complex nonlinear systems governed by numerous interacting parameters making their control particularly challenging.Measurement noise and hard constraints on process variables are critical considerations that further complicate the design and performance of control strategies.Traditional control techniques, such as PID, often struggle to compensate for nonlinear dynamics, reject disturbances, and enforce operational constraints.To address these limitations, this thesis investigates Model Predictive Control (MPC) strategies and evaluates the benefits of integrating AI-driven techniques into the control framework.The control objective is to regulate the biomass of a fed-batch bioreactor, by manipulating the substrate's feed rate.