Leakage Detection in Water Distribution Systems Based on Time–Frequency Convolutional Neural Network

Guo, Guancheng; Yu, Xintong; Liu, Shuming; Ma, Ziqing; Wu, Yipeng; Xu, Xiyan; Wang, Xiaoting; Smith, Kate; Wang, Xue

doi:10.1061/(asce)wr.1943-5452.0001317

Cited by 61 publications

(17 citation statements)

References 42 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…A multi-Layer Perceptron (MLP), a class of feedforward artificial neural network (ANN), is used herein. A three-layer ANN is proposed, since the generalization capability of multi-layer perceptron networks is not improved for more than three layers [29,30]. Concerning the number of neurons, several configurations of the ANN should be analyzed and a sensitivity analysis should be carried out to determine the best compromise between ANN simplicity and accuracy of the results.…”

Section: Problem Formulation and Ann Architecture Definitionmentioning

confidence: 99%

Near–Real Time Burst Location and Sizing in Water Distribution Systems Using Artificial Neural Networks

Capelo

Brentan

Monteiro

et al. 2021

Water

View full text Add to dashboard Cite

The current paper proposes a novel methodology for near–real time burst location and sizing in water distribution systems (WDS) by means of Multi–Layer Perceptron (MLP), a class of artificial neural network (ANN). The proposed methodology can be systematized in four steps: (1) construction of the pipe–burst database, (2) problem formulation and ANN architecture definition, (3) ANN training, testing and sensitivity analyses, (4) application based on collected data. A large database needs to be constructed using 24 h pressure–head data collected or numerically generated at different sensor locations during the pipe burst occurrence. The ANN is trained and tested in a real–life network, in Portugal, using artificial data generated by hydraulic extended period simulations. The trained ANN has demonstrated to successfully locate 60–70% of the burst with an accuracy of 100 m and 98% of the burst with an accuracy of 500 m and to determine burst sizes with uncertainties lower than 2 L/s in 90% of tested cases and lower than 0.2 L/s in 70% of the cases. This approach can be used as a daily management tool of water distribution networks (WDN), as long as the ANN is trained with artificial data generated by an accurate and calibrated WDS hydraulic models and/or with reliable pressure–head data collected at different locations of the WDS during the pipe burst occurrence.

show abstract

Section: Problem Formulation and Ann Architecture Definitionmentioning

confidence: 99%

Near–Real Time Burst Location and Sizing in Water Distribution Systems Using Artificial Neural Networks

Capelo

Brentan

Monteiro

et al. 2021

Water

View full text Add to dashboard Cite

show abstract

“…Reducing water losses in water distribution networks (WDNs) is of great significance to control carbon footprint and improve infrastructure sustainability (Wu et al 2011). While water losses comprise of apparent and physical components, it is the latter, due to hidden pipe leakages and/or other anomaly events (Mounce et al 2010;Guo et al 2021), which pose the biggest challenge for water utilities. Hidden leakages in water pipelines contribute significantly to the total non-revenue water (NRW) (Cassidy et al 2020) in WDNs and they can be classified into reported, hidden, and background leaks.…”

Section: Introductionmentioning

confidence: 99%

“…Over the years, different techniques have been developed to assist field engineers to detect and localize hidden leakage events in WDNs. For example, the use of physics-based hydraulic models to analyse flow and pressure data (Wu et al 2010(Wu et al , 2021Pérez et al 2011;Goulet et al 2013;Gerard et al 2016;Gianfredi et al 2017;Chew et al 2022), and devicebased methods which collect leak acoustic signals in water pipelines (Khulief et al 2012;Muntakim et al 2017;Butterfield et al 2018aButterfield et al , 2018bCody et al 2020;Guo et al 2021). Other leak detection methodologies include Bayesian updating, coupled with optimal distribution methods (Alawadhi & Tartakovsky 2020), transient test-based techniques (TTBTs) (Meniconi et al 2021), and trend-change analysis (Xue et al 2022).…”

Section: Introductionmentioning

confidence: 99%

“…Typically, an acoustic signal, as caused by pipe leakages, is due to the complex interaction between the flowing water and interiors of the underground pipe wall(s), hence generating random wave signals with both short-term nonstationary and long-term stationary frequency components (Almeida et al 2014). In the practical field context, the emitted acoustic signals are, however, mixed with a range of environmental noises which can lead to a low signal-to-noise ratio (SNR) (Wen et al 2004;Guo et al 2021). Thus, it is imperative to first pre-process raw acoustic signals, followed by leveraging the processed data to detect possible hidden leakages underground.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Acoustic feature-based leakage event detection in near real-time for large-scale water distribution networks

Chew¹,

Kalfarisi²,

Meng³

et al. 2023

Journal of Hydroinformatics

View full text Add to dashboard Cite

Acoustic sensors are widely deployed to detect hidden leakages in water distribution networks (WDNs). However, few studies have been conducted to quantitatively understand the dominant leakage acoustic characteristics, which are usually mixed with unknown environmental noises, coupled with the constraint of sparse deployment of acoustic sensors. In this paper, a comprehensive approach, that primarily leverages acoustic data feature analysis, is developed to detect pipe leakages in near real-time via a series of systematic analyses, namely: (1) data quality assessment; (2) features identifications; (3) outlier detection and event classification; and finally (4) near real-time leakage detection. The proposed solution methodology has been tested on two major WDNs in Singapore having around 1,000 km of water pipelines installed with 74 permanently installed hydrophone sensors. The leakage detection results obtained from our case study demonstrate that the dominant leakage acoustic characteristics can be captured in lower intrinsic mode functions (IMFs), to within the approximate frequency range of 100–750 Hz, by decomposing the original acoustic signal. Systemwide leakage event detection and classification models are subsequently trained and tested on acoustic datasets collected over 13 historical months, where F1-scores of more than 70% can be obtained from the emulated near real-time leakage detection analysis.

show abstract

“…Instead of feature extraction, some other researchers have transferred 1-D acoustic signals to spectrograms/melspectrograms [14][15][16] or scalograms, 17 and then applied a CNN to detect leaks. A variational autoencoder was used in conjunction with a CNN by Cody, Tolson 16 to eliminate the need for large training data sets.…”

Section: Introductionmentioning

confidence: 99%

A convolutional neural network for pipe crack and leak detection in smart water network

Zhang

Alexander

Stephens

et al. 2022

Structural Health Monitoring

View full text Add to dashboard Cite

The implementation of a smart water network (SWN) is viewed as a strategic approach to address many challenges faced by water utilities, such as pipe leak detection and main break prevention. This paper develops a convolutional neural network (CNN)–based model to classify acoustic wave files collected by the South Australian Water Corporation’s (SA Water’s) SWN over the city of Adelaide. The VGGish model (VGG refers to the team who developed the model—Visual Geometry Group) is selected as a suitable transfer learning model to extract features from wave files. The CNN model classifies an acoustic wave file as an anomaly or other background or environmental noise. Identification of a wave file as an anomaly triggers a Siamese CNN model to determine whether it is related to a regular/irregular scheduled event (for example, irrigation system near public parks or water consumption by large buildings). A field investigation is initiated if a wave file is classified as an anomaly and it is not related to a scheduled event. The developed models have been validated using data that is recorded by SWN in Adelaide. This validation data set comprises 1098 wave files, which are recorded by 34 accelerometers and are associated with 32 known leaks. The validation results shown that accuracy of alarms generated by the developed models is 92.44%. The validations confirm the developed models as an effective tool for water pipeline leak and crack detection, which, in turn, enables proactive management of the pipeline assets.

show abstract

Leakage Detection in Water Distribution Systems Based on Time–Frequency Convolutional Neural Network

Cited by 61 publications

References 42 publications

Near–Real Time Burst Location and Sizing in Water Distribution Systems Using Artificial Neural Networks

Near–Real Time Burst Location and Sizing in Water Distribution Systems Using Artificial Neural Networks

Acoustic feature-based leakage event detection in near real-time for large-scale water distribution networks

A convolutional neural network for pipe crack and leak detection in smart water network

Contact Info

Product

Resources

About