Unknown Inputs Observer-Based Output Feedback Predictive Controller for an Activated Sludge Process

paper investigates the control problem of an activated sludge process despite the non-measurable states and the unknown inputs (UIs). Precisely, our objective is to control the global nitrogen concentration (S-N) at the settler output, with partial state measurement and UIs. For this MIMO system, a nonlinear predictive controller based on a high-gain observer is proposed. First a state feedback (full information) predictive control low is designed in order to maintain the global nitrogen concentration in accordance with the standard imposed by the European standards for water quality. Then, a high-gain observer-based UI is designed whose aim is to estimate the non-measurable states (the concentrations of the readily biodegradable substrate S-s and the ammonia S-NH4) and the UIs (the concentrations of substrate soluble in water and the ammoniacal nitrogen entering the reactor: S-sin and S-NH4in) despite their variable behavior. Finally, estimated states and inputs are used to generate the output feedback predictive control. Simulation results validate the objective of the paper.