Optimal control of an open channel irrigation system based on nonlinear models

We apply an optimal control approach to a real irrigation system based on a nonlinear blackbox model. Open channel systems are nonlinear, and their dynamics are based on the Saint-Venant equations, and have nonlinear component structures. For these reasons, linear models are not well adapted even less complex concerning the models number of parameters. Water waste in large systems are not negligible, so we have an energy waste by pumping useless water. Our goal is to find the control trajectories by minimizing the energy consumption subject to different constraints like the physical limits of the system and also the water systems dynamics. The optimal solution is generated by some sequential quadratic programming algorithm. We apply an optimal control approach on a real irrigation system, the Canal de la Bourne (France) using a nonlinear black-box model NARX.