Achieving offset free model predictive control of irrigation canals

Boran Ekin Aydin, Peter-Jules van Overloop

Friday 3 july 2015

12:39 - 12:42h at Central America (level 0)

Themes: (T) Water resources and hydro informatics (WRHI), (ST) Management support systems and serious gaming, Poster pitches

Parallel session: Poster pitches: 15H.WRHI - SeriousGaming

Model predictive control (MPC) is a powerful control option which is increasingly used by operational water managers for managing water systems. However, these applications on water systems have offset problem because of the mismatch between the models used in MPC and the actual system. The main reasons of this mismatch are the unknown disturbances and other uncertainties present in the system, which cannot be modelled. Mismatch results in offset which prevents MPC to achieve its goal of reaching the set point for the controlled variable. This article shows three different methods to achieve offset free MPC of the first pool of the laboratory canal at Technical University of Catalonia, Barcelona. First method is adding a kalman filter to the internal model used to update the states of the system by using the measurements. Second method is modelling the disturbance dynamics of the system by means of a disturbance model. An integrating disturbance is augmented into the internal model as an additional state. The last method, Model Based Observer (MBO) is developed and tested by the writers as an alternative offset free MPC. Working principle of MBO is observing the past simulation results and measurements of the system to improve the simulation results for the following predictions. Simulation results of the three methods are provided and the results are used to compare the three methods. Especially, the results of the last method have key importance to test whether MBO will be a competitive alternative offset free MPC method