Distributed Adaptive Predictive Control of Water Delivery Networks

The paper proposes a structure based on distributed adaptive predictive algorithms for control of water delivery open canal networks. Canals are decomposed in pools, the level of each is manipulated by a gate. To each gate/pool, a local controller is associated. In order to improve the overall as well as the local performance, communication between local controllers is provided through feedforward terms from adjacent pools. The comparison of the base line (without any communication among local controllers) performance with different configurations of the controller network proposed is made through simulation. In addition it is shown that the algorithm considered yields the best performance to each local controller, as measured by a quadratic cost, given the information pattern available in terms of sensor measurements.