MIMO LPV state-space identification of an open-flow irrigation canal for control

In this paper a linear parameter varying (LPV) statespace canal control model is obtained by identification in a local way. This LPV identification procedure is based on subspace methods for some operating points of an irrigation canal covering all the operation range. Different subspace algorithms have been used and compared. The model that best represents the canal behavior in a precise way is chosen. This model has been validated by error functions and analysis correlation of residuals in a laboratory multi-reach pilot canal providing satisfactory results.