Contrastive learning method for leak detection in water distribution networks

Abstract Detecting and mitigating leaks in water distribution networks are vital for water conservation. Machine-learning-based (ML) acoustic leak detection models were introduced as effective alternatives for leak management. However, ML model training requires sufficient labeled data, which hinder...

Full description

Saved in:
Bibliographic Details
Main Authors: Rongsheng Liu, Tarek Zayed, Rui Xiao
Format: Article
Language:English
Published: Nature Portfolio 2024-11-01
Series:npj Clean Water
Online Access:https://doi.org/10.1038/s41545-024-00406-6
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1850056030336057344
author Rongsheng Liu
Tarek Zayed
Rui Xiao
author_facet Rongsheng Liu
Tarek Zayed
Rui Xiao
author_sort Rongsheng Liu
collection DOAJ
description Abstract Detecting and mitigating leaks in water distribution networks are vital for water conservation. Machine-learning-based (ML) acoustic leak detection models were introduced as effective alternatives for leak management. However, ML model training requires sufficient labeled data, which hinders related development. To address this challenge, this study employed contrastive learning (CL) for leak detection using limited labeled signals. Experimental results indicate that flip-x and amplitude scaling are optimal combinations for contrastive learning. Besides, ablation and t-distributed stochastic neighbor embedding (t-SNE) results reveal that increasing the model depth does not always yield performance improvement, and five convolutional blocks are more suitable for the leak detection problem in this study. Comparison experiments demonstrate that contrastive learning outperforms supervised learning (SL) when trained with insufficient labeled data. The out-of-sample validation results indicate that the proposed leak detection model is robust and effective in unexplored pipelines. The proposed framework significantly advances ML-based leak detection research and supports sustainable water management practices.
format Article
id doaj-art-aee6d3f20a72428faed4e7f208a75c68
institution DOAJ
issn 2059-7037
language English
publishDate 2024-11-01
publisher Nature Portfolio
record_format Article
series npj Clean Water
spelling doaj-art-aee6d3f20a72428faed4e7f208a75c682025-08-20T02:51:48ZengNature Portfolionpj Clean Water2059-70372024-11-017111310.1038/s41545-024-00406-6Contrastive learning method for leak detection in water distribution networksRongsheng Liu0Tarek Zayed1Rui Xiao2Department of Building and Real Estate, The Hong Kong Polytechnic University, Hung HomDepartment of Building and Real Estate, The Hong Kong Polytechnic University, Hung HomDepartment of Building and Real Estate, The Hong Kong Polytechnic University, Hung HomAbstract Detecting and mitigating leaks in water distribution networks are vital for water conservation. Machine-learning-based (ML) acoustic leak detection models were introduced as effective alternatives for leak management. However, ML model training requires sufficient labeled data, which hinders related development. To address this challenge, this study employed contrastive learning (CL) for leak detection using limited labeled signals. Experimental results indicate that flip-x and amplitude scaling are optimal combinations for contrastive learning. Besides, ablation and t-distributed stochastic neighbor embedding (t-SNE) results reveal that increasing the model depth does not always yield performance improvement, and five convolutional blocks are more suitable for the leak detection problem in this study. Comparison experiments demonstrate that contrastive learning outperforms supervised learning (SL) when trained with insufficient labeled data. The out-of-sample validation results indicate that the proposed leak detection model is robust and effective in unexplored pipelines. The proposed framework significantly advances ML-based leak detection research and supports sustainable water management practices.https://doi.org/10.1038/s41545-024-00406-6
spellingShingle Rongsheng Liu
Tarek Zayed
Rui Xiao
Contrastive learning method for leak detection in water distribution networks
npj Clean Water
title Contrastive learning method for leak detection in water distribution networks
title_full Contrastive learning method for leak detection in water distribution networks
title_fullStr Contrastive learning method for leak detection in water distribution networks
title_full_unstemmed Contrastive learning method for leak detection in water distribution networks
title_short Contrastive learning method for leak detection in water distribution networks
title_sort contrastive learning method for leak detection in water distribution networks
url https://doi.org/10.1038/s41545-024-00406-6
work_keys_str_mv AT rongshengliu contrastivelearningmethodforleakdetectioninwaterdistributionnetworks
AT tarekzayed contrastivelearningmethodforleakdetectioninwaterdistributionnetworks
AT ruixiao contrastivelearningmethodforleakdetectioninwaterdistributionnetworks