Modeling of a water canal system using a weigthed composition of local linear state space models

Abstract

AbstractThis paper addresses the problem of identifying nonlinear models, directly from input-output measured data, for a benchmark water canal system located in Évora, Portugal. The scarcity of fresh water in many regions of the world demands for a careful management and control of such resource. In order to achieve such objective, techniques from nonlinear modeling and control are being applied and tested in such systems. In this paper, a weighted combination of several local linear models given in state-space is proposed to identify nonlinear models for the canal directly from available data. Such identification methodology can provide a viable way to address the modeling of general nonlinear systems, such as the one proposed in the paper, which is shown by comparing the performance of modeling results with an existing theoretical model.