Enhanced Water Leak Detection with Convolutional Neural Networks and One-Class Support Vector Machine
Daniele Ugo Leonzio, Paolo Bestagini, Marco Marcon, Stefano Tubaro
TL;DR
The study tackles leak detection in water distribution networks using a data-driven approach that relies on pressure measurements from multiple nodes. It introduces a pipeline that uses an autoencoder to extract compact embeddings from no-leak data, followed by a one-class SVM to detect leaks as anomalies within sliding windows of pressure data. On the Modena benchmark, the method achieves about 0.92 detection accuracy with an average detection delay of 40.21 hours, while demonstrating robustness to Gaussian noise and the ability to generalize to Hanoi and Pescara networks. This approach reduces dependence on detailed pipe parameters by leveraging topology and historical no-leak data, offering a practical, scalable solution for water utilities.
Abstract
Water is a critical resource that must be managed efficiently. However, a substantial amount of water is lost each year due to leaks in Water Distribution Networks (WDNs). This underscores the need for reliable and effective leak detection and localization systems. In recent years, various solutions have been proposed, with data-driven approaches gaining increasing attention due to their superior performance. In this paper, we propose a new method for leak detection. The method is based on water pressure measurements acquired at a series of nodes of a WDN. Our technique is a fully data-driven solution that makes only use of the knowledge of the WDN topology, and a series of pressure data acquisitions obtained in absence of leaks. The proposed solution is based on an feature extractor and a one-class Support Vector Machines (SVM) trained on no-leak data, so that leaks are detected as anomalies. The results achieved on a simulate dataset using the Modena WDN demonstrate that the proposed solution outperforms recent methods for leak detection.