Detailed Leak Localization in Water Distribution Networks Using Random Forest Classifier and Pipe Segmentation

In this paper, a Random Forest classifier was used to predict leak locations for two differently sized water distribution networks based on pressure sensor measurements. The prediction model is trained simulated scenarios with randomly chosen parameters - location, size, and base node demand uncertainty. Leak localization methods found in literature that rely numerical simulations can only network nodes as nodes; however, since occur at any point along pipe segment, additional spatial discretization of suspect proposed paper. It observed segmentation the whole non-feasible approach it rapidly increases number potential locations, consequently increasing complexity model. Therefore, novel proposed, which leaks occurring original only, but its accuracy assessed against measurements from points between nodes. successfully narrow down area and, followed by subsequent prediction, precise be achieved. enables incorporation various uncertainties simulating under different conditions. Investigation size uncertainty variation showed several produce similar makes difficult unambiguously determine location using future approaches coupling modeling optimization are proposed.