Leakage localization in city water distribution networkBogataj, Mark (Avtor) Šubelj, Lovro (Mentor) water distribution networksleakage localizationsensor placementoptimizationThis thesis addresses the challenge of water leak localization within water distribution networks (WDNs), focusing on the strategic placement of sensors to enhance leak localization accuracy. We explore a range of algorithms to tackle this problem, beginning with basic graph theory-based approaches that offer preliminary solutions without a detailed WDN model. For more sophisticated sensor placement strategies, we investigate advanced algorithms, including genetic algorithms (GA), mutual information quadratic unconstrained binary optimization (MIQUBO), and cost-effective lazy forward (CELF) algorithms. An exhaustive search methodology is employed to benchmark the optimal solution. Among these, the GA emerges as the most cost-effective method, balancing performance and investment. Our evaluation, conducted across four distinct network sizes and three leak magnitudes, demonstrates that while most algorithms achieve high performance, GA consistently delivers leak localization accuracy of around 90% or higher in some cases. This work identifies GA as a superior strategy for sensor placement from a cost-performance perspective and contributes a comprehensive framework for enhancing leak localization capabilities in WDNs.20242024-04-15 10:45:01Magistrsko delo/naloga155749VisID: 37104COBISS_ID: 193117955sl