2025-07-082025-07-082025Robles-Velasco, A., Onieva, L., Guadix Martín, J. y Cortés, P. (2025). A methodology for incident detection in sectorized waterdistribution networks based on pressure and flow data. Computer-Aided Civil and Infrastructure Engineering, 1-20. https://doi.org/10.1111/mice.13493.1093-96871467-8667https://hdl.handle.net/11441/175139This is an open access article under the terms of the Creative Commons Attribution-NonCommercial-NoDerivs License, which permits use and distribution in any medium,provided the original work is properly cited, the use is non-commercial and no modifications or adaptations are made.This study presents an intelligent system for predicting incident reports (IRs) insectorized water distribution networks, such as drains in sidewalks, lack of pres-sure, lack of water, leaks, or others, based on pressure and flow data. Currently,incident detection in the industry is highly inefficient, as it is always performedreactively—only after an incident has already occurred and its negative conse-quences are visible to users. Since these data are recorded at 5- to 15-min intervals,a methodology is proposed to integrate them with daily IRs. After processing thedata, a supervised classification learning system is developed with a binary out-put variable indicating the likelihood of an incident at a specific time step. Themethodology is validated using 2 years of data from a real network divided intoeight sectors. The system predicts 51.3% of IRs, with 78.9% accuracy, highlightingthe strong influence of daily mean and maximum flows on incidents.application/pdf20 p.engAttribution-NonCommercial-NoDerivatives 4.0 Internationalhttp://creativecommons.org/licenses/by-nc-nd/4.0/info:eu-repo/semantics/articleinfo:eu-repo/semantics/openAccesshttps://doi.org/10.1111/mice.13493