Decentralized ellipsoidal state estimation for Model Predictive Control in irrigation canals

Control strategies allow steering water levels towards their desired values in irrigation canals. The knowledge of the system state is essential to apply any control strategy. Centralized ellipsoidal estimation techniques show adequate performance to estimate unmeasured state variables in small systems, but may became difficult to apply in large scale ones due to the increasing computation burden. To achieve this issue, in this work is presented a decentralized ellipsoidal estimation technique that maintains the quality of estimation with respect to a centralized strategy, but considerably reduces the computation time requirements by exploiting the system structure. An adaptation of the irrigation canal developed by the ASCE Task Committee on Canal Automation Algorithms is used as a case study to show the performance of the proposed methodology.