Multi-leak detection and isolation in water distribution networks

Abstract

Water Distribution Networks (WDNs) are complex systems that faces the challenge of detecting and locating water leaks in the system as quickly as possible due to the need for an efficient operation that satisfies the growing world demand for water. This paper introduces an entirely data-driven leak detection and localization method based on flow and pressure analysis. The method can be divided into leak detection when the fusion data of the flow and pressure measurements are studied, thus obtaining the instant where the leak starts and if there is more than one simultaneous leak (multi-leak) occurring in the network. The second part is the leak localization using the fusion of the pressure residues by applying the radial base function (RBF) interpolation to obtain the network zone with the highest leak probability. The method is validated using the L-TOWN benchmark proposed at the Battle of the Leakage Detection and Isolation Methods (BattLeDIM) 2020 challenge.

Postprint (published version)