Leakage Detection, Graph-Based Modeling, Graph Neural Networks (GNN), Water Distribution Network Simulation
Abstract
The water distribution network (WDN) is crucial for ensuring reliable access to potable water in urban and rural communities. However, water wastage due to leaks in the infrastructure is a significant challenge, affecting operational efficiency and causing environmental and financial impacts. Effective leak detection is complex and requires proper data analysis, especially when considering the correlation between them. Graph-based modeling offers a robust representation of the structure of DERs, while Graph Neural Networks (GNNs) have the potential to improve accuracy in detecting and locating leaks. This article aims to continue studies into the use of GNNs to locate leaks in WDNs, proposing the use of the GraphSAGE algorithm. This article is based on a previous study, which demonstrated the viability of the graph-based approach to the task, and uses it as a reference for evaluating the performance of the GraphSAGE algorithm when it comes to locating leaks in WDNs.
