Detection, Isolation, and Magnitude Estimation of Unknown Flows in Open-Channel Irrigation Systems

Abstract

The development of modeling and estimation strategies, useful for determining the magnitude and location of unknown flows such as seepage and leaks, appears as a valuable tool to increase the efficiency of the open-channel irrigation systems (OCIS). However, it has been identified that in OCIS, most of the strategies reported on detection, isolation, and magnitude estimation of unknown flows (DIMEUF) have been developed from linear models that do not include information about energy balances along the channels, where these balances are fundamental to differentiate changes of levels due to conduction effects, from changes of levels due to unknown flows. Therefore, in this work, a recent OCIS modeling approach, which includes mass and energy balances for each channel and non-linear hydraulic descriptions of the flows, is explored in the development of two strategies for DIMEUF based on the moving horizon estimation (MHE) approach. The first strategy is deterministic, designed under the assumption that by filtering of the measurements, the noise can be sufficiently attenuated. Therefore, the noise information is not included in the design process. On the other hand, the second strategy is stochastic, and includes remaining noise information in the design process. The developed strategies have been tested using data from a testbed implemented in a specialized software, and the results show that, in a large operation region, the proposed strategies are capable of accurately describe the channel behavior and unknown flows, and that the inclusion of the remaining noise information increases the performance of the strategies for DIMEUF.