Model predictive control tuning considering l1 norm for disturbance rejection

In this paper the inclusion of some classical measures for disturbance rejection [4] in a general procedure for tuning multivariable model predictive controllers (MPC) with constraints is presented. These measures are stated as l₁ norms of the system transfer functions, providing a framework for different sets of disturbances. The case of asymmetrical bounded signals (disturbances, inputs) is also studied and included in the procedure. The methodology has been applied to tune an MPC for the activated sludge process in a simulated wastewater treatment plant (WWTP). The problem is stated as a non linear programming problem with constraints over the system norms. The mathematical optimization for tuning all controller parameters is tackled in two iterative steps due to the existence of integer and real numbers (horizons and weights).