Multivariable predictive control of irrigation canals. design and evaluation on a 2-pool model

This paper presents the formulation of a multivariable controller for centralized control of canals based on the methodology of predictive control. The formulation relies on a linear state space model derived from Saint-Venant's equations discretized through the Preissmann implicit scheme. Predictive control uses this model to compute the control action that verifies a performance criterion defined over a finite prediction interval. The controller is combined with a Kalman filter to reconstruct the state variables and the unknown perturbations from a reduced number of measured variables, which are water levels at the upstream and downstream end of each pool. The control system is numerically tested on a 2-pool canal using a full non-linear simulation package (SIC)