Author(s): Wei Zeng; Martin Lambert; Mark Stephens; Xiang Wang; Ruilin Liu; Chengcheng Yin
Linked Author(s): Martin Lambert
Keywords: Smart water network internet of things leak detection machine learning
Abstract: Distributed acoustic sensors have been widely used in water distribution systems for leak detection. Apart from leak noise, acoustic noises generated from pumping/valve operation, pedestrians, traffic, rain and other sources in the water system can be also captured by the acoustic sensors, leading to a large amount of false positive alerts in such IoT-based water asset monitoring systems. In this study, dedicatedly designed signal processing methods have been proposed to extract acoustic features of different types of acoustic noise, including leak noise and non-leak noise that commonly occur in real water networks. These tailored acoustic features are found to be crucial in feature-based leak detection in real water network environments.
Year: 2025