Closed loop identification of an irrigation channel

We consider the closed loop identification of an irrigation channel. A first order nonlinear model of the water level is used, and previous work by Weyer (2000) has shown that this model gives a good description of the dynamics. Earlier, such models were obtained using open loop system identification which caused disruptions to the normal operations of the irrigation channel. These disruptions can be reduced by using closed loop identification. The gates in the irrigation channel are controlled using feedback from water levels, and the excitations are provided by set point changes. Only relatively small set point changes and short experiment times are needed in order to obtain a reliable model. The obtained models are validated against data sets with large variations in flow, and there is very good agreement between the simulated model output and the observed water levels. Hence, models of irrigation channels can be obtained using closed loop identification which causes only minor disruptions to the normal operation of the channel