Search

• An LSTM-GAN generative model is employed for enriching the leak detection dataset. • The model captures the time-series features of leak signals in WDNs. • The performance of LSTM GAN is evaluated through a series of validation approaches. • The method enriches the leak signals and enhances the water leak detection accuracy. Water distribution networks (WDNs) experience significant water loss due to leaks, necessitating advanced water leak detection methods. However, machine learning-based acoustic method heavily relies on signal information and is limited by data scarcity and the limited diversity of available data. To address this challenge and enhance water leak detection in WDNs, this study proposes an LSTM-GAN approach. Acoustic signals are collected from WDNs to train the LSTM-GAN model, which generates synthetic leak signals to enhance the dataset. The validity of the generative method is evaluated through t-SNE and acoustic characteristics analysis. LSTM-based water leak detection models are established and compared using the original and the generated datasets to confirm the efficacy of generated samples in improving water leak detection performances. The capability of LSTM-GAN has been evaluated through different perspectives, including sensitivity analysis and model comparison. The results validate the quality and consistency of the generated acoustic signals under leak conditions. Besides, the optimal number of generated samples should be determined according to the requirements and characteristics of the leak detection task. Furthermore, the comparison between the proposed method and other acoustic generative methods demonstrates the superiority of LSTM-GAN-generated signals in enhancing the performance of leak detection models. The proposed generative method offers an innovative approach to facilitate machine learning-based leak detection models with limited data, thereby enhancing robustness. [Display omitted] [ABSTRACT FROM AUTHOR]

[Display omitted] • Application of machine intelligence to water leak detection. • Identification of controlling parameters governing the leakage phenomenon. • The derived GEP expressions achieved a reasonable accuracy in leakage prediction. Globally, the loss of potable water through buried and service water distribution networks (WDNs) is one of the persistent challenges confronted by water utilities. Most of the WDNs inspection methods are ad-hoc and typically provide the current status of the pipes. Despite the significant research efforts, a reliable prediction method of the water leak of buried and service water distribution pipes is rare. To overcome this challenge, gene expression programming (GEP) models – that uses extracted features of acoustic signals collected from noise loggers attached to pipeline valves at the chamber – are developed and validated in the current study. To develop these models, extensive fieldwork hitherto unavailable was undertaken in this study to record the flow-induced acoustic signal of WDNs. The GEP models are developed using a step function – that returns a binary output – which is leveraged to exhibit the proposed detection and prediction methods using the acoustic signal collected from the buried WDNs. The models only consider the highly correlated features with the leakage. The models are found to be able to predict leakages on metal and nonmetal pipes with about 95% accuracy. These models were developed to reduce the time and cost of deployment of equipment for leakage detection of pipes. The proposed method presented in this study highlights the prospect and advantage of using the GEP in infrastructural management given the amazing capability of the machine intelligence technique to recognize multi-dimensional circumstances with ease, high prediction accuracy, and potential for unceasing improvement. [ABSTRACT FROM AUTHOR]

A reliable leak detection approach is necessary for reducing water loss caused by leakages. Our research reviews the existing acoustic leak detection approaches, including devices for collecting sound, signal processing methods, and ways to detect the presence of leakages. Firstly, a scientometric analysis of relevant bibliographic data from the Scopus database in the last thirty years is carried out. Secondly, based on the results of the scientometric review, the technologies of acoustic leak detection have been systematically reviewed. Acoustic leak detection approaches are classified into non-real time and real-time methods and are thoroughly discussed. Thirdly, acoustic signals collection through noise loggers, wireless sensor networks, accelerometers, hydrophones, or fiber optic sensors is elaborated. Fourthly, signal features extraction using different methods, including time-domain analysis, frequency domain analysis, time-frequency analysis, correlation analysis, and other pre-processing methods are explained. Fifthly, the determination of the presence of leaks using various detection methods is detailed. This paper provides a useful reference for academia and industry for the further development of advanced acoustic leak detection methods. • Acoustic signals are collected by noise loggers, wireless sensor networks, and so on. • Features are extracted by different analysis and signal processing techniques. • Leaks detected by a threshold or classification model. • Unresolved challenges: avoiding false alarms, detecting weak leaks, and so on [ABSTRACT FROM AUTHOR]

Leakages from water pipelines cause economic losses and environmental hazards. Despite the damages, it is challenging to avoid leaks throughout the lifetime. However, leak detection and localization, especially in real-time, minimize the damage. Owing to the recent advances, the micro-electromechanical systems (MEMS) based technologies have started to gaining recognition for water network monitoring in real-time, however, a systematic literature review to analyze the existing research trends, technological advances, and future research opportunities are largely missing. This study has based its investigation on three main MEMS-based technologies for real-time monitoring: MEMS sensors wireless networks, MEMS accelerometers, and MEMS hydrophones. Firstly, a scientometric analysis is conducted to 1) retrieve relevant research articles through Scopus, Web of Science, and Google Scholar, 2) visualize the publication trends, and 3) analyze the science mapping of influential authors, countries, organization, and top keywords occurrences. Secondly, qualitative discussions are made on research themes and sub-themes within three technologies: 1) MEMS WSNs are classified into static and mobile sensor-based wireless sensor networks. Seven sub-themes are categorized under static sensor-based wireless sensor networks such as PIPETECT, whereas three sub-themes are categorized under mobile sensor-based WSNs such as TriopusNet; 2) MEMS accelerometers are categorized into accelerometers based machine learning models and wireless systems; and 3) MEMS hydrophones are represented under one category. Thirdly, nine research opportunities including automated models, on-field real network-based experimental studies, optimal placement of sensor nodes for energy savings in wireless sensor networks, and a comparative analysis of real-time technologies are revealed. This study enhances the familiarity of early researchers with the application of MEMS-based technologies for leak detection and localization and provides seasoned researchers with a platform for future research development. • MEMS-based technologies create opportunities to enhance real-time leak detection and localization in water supply networks. • MEMS-based wireless sensor networks, accelerometers, and hydrophones are classified into different themes and sub-themes. • The bibliometric search process, scientometric analysis, and qualitative analyses are discussed in detail. • Future research directions are provided. [ABSTRACT FROM AUTHOR]