A Proposal of Robust Leak Localization in Water Distribution Networks Using Differential Evolution

Abstract

The performance of popular leak location strategies that depend on the use of a leak sensitivity matrix (LSM) is highly dependent on the conditions used for generating such matrix. Since noise is present in real Water Distribution Networks (WDNs), the location results can deteriorate seriously when using the former methods. A robust model-based leak localization approach by using Differential Evolution (DE) algorithm is presented in this article. The proposal uses the topological information of the water distribution network in the location of the leak and it does not depend on the sensitivity matrix. The proposed method demonstrates its robustness compared with the LSM approach by using the Hanoi network as case study. The performance achieved by the proposal improves the leak location accuracy by 20% with respect to LSM-based leak location method in the presence of different noise magnitudes.