Intelligent Acoustic-Based IoT-Enabled Sensing and Monitoring System for Automated Real-Time Leak Detection, Localization, and Pinpointing in Water Distribution Infrastructures

Water distribution infrastructures (WDIs) are a vital part of the economy, as they provide and transport water to consumers, which is essential for daily activities. These infrastructure systems are comprised of a network of aging pipes that are prone to leaks and breaks. The interruption of a water supply during leak rectifications can lead to inconveniences to consumers/users, and erroneous leak assessments can result in substantial repair costs. Therefore, the early and precise identification of leaks in WDIs, coupled with prompt rectifications, is pivotal for enhancing water distribution and supply efficiency. Recently, there has been growing interest in acoustic leak detection and localization methods using hydrophones due to their various advantages over traditional and other methods. While various acoustic-based methods have been proposed by existing research efforts, they have some limitations, including most of them being focused on either leak detection and/or localization rather than pinpointing; most of them lack the real-time internet-of-things (IoT)-enabled capabilities; and most of them are limited in their application to either straightforward, simple 1D representation or noncomplicated or specific 2D representation of water distribution networks (WDNs). This paper addresses these limitations by developing an intelligent acoustic-based IoT-enabled water monitoring sensing system for automated real-time water leak detection, localization, and pinpointing for 2D WDNs. First, a novel water leak detection, localization, and pinpointing algorithm was developed and demonstrated for a simulated WDN, where data were collected, prepared, and processed to extract the leak noise signatures from the acoustic energy within the low-frequency range, and the time delays of arrivals were calculated to identify the potential leak location. Second, an IoT-enabled sensor-fusion system was developed based on four acoustic hydrophones, an Arduino microcontroller, and an NVIDIA Jetson series microprocessor board on the edge. Third, the proposed approach was validated in lab experiments where an actual water distribution system was built, and a graphical user interface was also developed to visualize the detected leak, display the localized leak zone, and pinpoint the leak location in the WDI system. The proposed approach achieved network normalized accuracy up to 99.88%, with mean absolute localization error as low as 0.075 m and a 100% success rate within 0.20 m across all tested network topologies while achieving up to 83% success within 0.10 m. This paper adds to the body of knowledge by proposing a novel water leak pinpointing algorithm that is integrated into an intelligent IoT-enabled real-time water management and monitoring system that could be used for any 2D WDN and any complicated WDI system architecture. This allows not only the detection of a leak but also the identification of its exact location, thus enabling quicker/faster management and more efficient mitigation actions to be implemented and directed to the right location in the WDI system to ensure a reliable water supply to consumers, less potential damage to surrounding properties, and reduced volumes of nonrevenue water.