Leak Detection Research Articles

Enhanced Data Leakage Detection in Cloud and Enterprise Environments through NLP-Driven Classification, Metadata Tagging, and Anomaly-Based Learning

Data leakage occurs when information from outside a training dataset inadvertently influences a machine learning model, leading to overly optimistic performance estimates and reduced generalizability. Detecting data leakage is crucial to maintain model integrity, prevent overfitting, and ensure accurate deployment results in real-world applications. Traditional methods for leakage detection are limited by their inability to capture subtle, complex forms of leakage that arise in highdimensional data or intricate workflows. This study proposes an improved data leakage detection framework that leverages a combination of statistical testing, cross-validation anomaly checks, and interpretability techniques. Our approach systematically identifies suspicious patterns, assesses feature-target relationships across training and test sets, and flags inconsistent data flows that may signal leakage. By implementing these methods, we demonstrate enhanced sensitivity to various leakage types, including label, feature, and temporal leakage, across several case studies in healthcare, finance, and image processing. Our findings highlight the importance of robust leakage detection techniques in developing reliable machine learning models and suggest practical guidelines for integrating these methods into machine learning pipelines. This approach ultimately promotes the development of models with better generalizability, fairness, and trustworthiness.

Research and experiment of a novel leak detection robot

Purpose This paper addresses the issue of exterior wall leakage inspection in the construction industry. Currently, leakage inspections primarily rely on manual operation and empirical judgment, which often fail to meet industry standards and result in low efficiency, leading to frequent leakage issues during the operation and maintenance stages. This paper proposes the design of a new leakage inspection robot and a novel control method to enhance effectiveness and efficiency. Design/methodology/approach The robot integrates multiple sensors and maintains operational stability through a posture control strategy. Using technologies such as machine vision, the robot locates the window seam and navigates the sliding platform to the target. The pressure control module ensures that water-spraying factors at the seam meet relevant standards. Findings Compared to manual methods, the robot maintains stable water pressure that complies with relevant standards through collaborative control using IMU, encoders and other sensors, demonstrating superior performance over human labor. Additionally, using machine vision, the robot navigates the sliding platform to the target window seam for the water-spraying test. Originality/value This paper addresses the limitations of traditional leakage inspection methods by developing an innovative leakage inspection robot and control strategy. Significant results have been achieved through various applications, further validating its effectiveness and efficiency.

LPG Transport Tracking and Leakage Detection with Accident Prevention Alert System

The LPG Transport Tracking and Leakage Detection with Accident Prevention Alert System offers a comprehensive solution for enhancing the safety and monitoring of LPG transportation. Powered by an Arduino microcontroller, the system integrates multiple sensors, including a gas sensor to detect leaks, a load cell to monitor cylinder weight, and a DHT11 for environmental conditions. In the event of a leak, the system triggers a buzzer alert and automatically sends notifications via GSM to pre-defined contacts. Real-time vehicle tracking is enabled through a GPS module, while an LCD display provides continuous updates on gas levels, weight, and location. This system ensures timely detection and response to hazards, significantly reducing risks and improving transport safety. Keywords: LPG transport, leakage detection, accident prevention, Arduino, GPS tracking, gas sensor

Haemoconcentration in diagnosing dengue haemorrhagic fever: evidence from a rural Sri Lankan observational study.

One of the World Health Organization criteria for dengue haemorrhagic fever (DHF) is a 20% increase in haematocrit. This study on dengue hospitalised patients compares haematocrit levels at baseline, at the onset of the critical phase and the maximum during the critical phase in rural Sri Lanka. This observational study included patients with dengue in the febrile phase who progressed to the critical phase. Haematocrit was recorded and ultrasound scans were performed thrice daily when the platelet count dropped to <100000/μl. The onset of the critical phase was confirmed by ultrasound-detected plasma leakage. Forty-three patients were included in the final analysis. The mean haematocrit at baseline and at the onset of the critical phase was 40.6% (standard deviation [SD] 4.7) and 41.3% (SD 5.1), respectively, with no difference (p=0.14, paired t-test). None of the participants showed a 20% increase in haematocrit at the onset of the critical phase, with the maximum observed increase being 17.0%. A 20% increase in haematocrit was not observed during the critical phase, suggesting that haemoconcentration is unreliable for diagnosing DHF in hospitalised patients. This study recommends revisiting the WHO criterion of a 20% haematocrit increase for diagnosing plasma leakage in dengue patients. It suggests that frequent ultrasound scans may be a more reliable method for early detection of plasma leakage.

A proof-of-concept study towards developing digital twins for operational excellence in large-scale water distribution networks

ABSTRACT In the current face of water scarcity, water losses due to leakages in large-scale water distribution networks (WDN) and non-revenue water are challenging factors. In this direction, Digital Twins integrate concepts like IoT, ML (machine learning), and DL (deep learning) with a water pave path for smart urban water infrastructure. Herein, we propose a holistic digital twin systems framework and its application in leak detection, validated with field-data on Indian Institute of Technology Jodhpur (IIT-J) campus WDN. A detailed methodology developing monitoring digital twins supported on the python platform and using open-source WDN simulators; EPANET and WNTR for hydraulic simulations and a Graph-convolution Neural Network-based leak detection model is elucidated. Results are analysed and demonstrated for the highest water consumption zone of the campus, with the model giving an accuracy of 90% for leakage detection. Further, a test scenario is described where the proposed framework shows water savings of up to 58% which would have been otherwise lost due to leaks in WDN.

Combined L-Band Polarimetric SAR and GPR Data to Develop Models for Leak Detection in the Water Pipeline Networks

Water pipeline leak detection in a fast and accurate way is of much importance for water utility companies and the general public. At present, the rapid development of remote sensing and computer technologies makes it possible to detect water pipeline leaks on a large scale efficiently and timely. The leakage will cause an increase in the water content and dielectric constant of the soil around the pipeline, so it is feasible to determine the leakage site by measuring the subsurface soil relative dielectric constant (SSRDC). In this paper, we combine the SAOCOM-1A L-band synthetic-aperture radar (SAR) and the ground-penetrating radar (GPR) data to develop regression models that predict the SSRDC values. The model features are selected with the Boruta wrapper algorithm based on the SAOCOM-1A images after pre-processing, and the SSRDC values at sampling locations within the research area are calculated with the reflected wave method based on the GPR data. We evaluate multiple linear regression (MLR), random forest (RF), and multi-layer perceptron neural network (MLPNN) models for their ability to predict the SSRDC values using the selected features. The experimental results show that the MLPNN model (R2 = 0.705, RMSE = 1.936, MAE = 1.664) can better estimate the SSRDC values. Further, in the main urban area of Tianjin, China, which has a large water pipeline system, the SSDRC values of the area are obtained with the best model, and the locations where the predicted SSDRC values exceeded a certain threshold were considered potential leak locations. The empirical results indicate an encouraging potential of the proposed method to locate the pipeline leaks. This will provide a new avenue for the monitoring and treatment of water pipeline leaks.

An 8-Micrometer-Thick Film Strain Sensor with Conformal 3D Microstructure for Accurate Detection of Body Motion and Air Leakage.

Elastomer-based resistive super-thin film strain sensors show great application potentials in electronic skins, human-machine interaction systems, wearable devices for healthcare, and machine learning algorithms. However, it is challenging to accurately monitor the deformation of human body joints and organs with curved surfaces (e.g., knees, throats, finger joints) by only taking advantage of material thickness and elasticity of conventional2D film strain sensors. Herein, a simple strategy is developed to fabricate conformal elastomeric thin film sensors with periodic 3D microstructure inspired by the ridges and valleys of human skin for accurate signal acquisition. Specifically, an 8-micrometer-thick elastic film strain sensor with 3D microstructure is fabricated via thermoforming followed by in situ chemical growth of silver nanoparticles. The 3D film strain sensors exhibit excellent signal linearity (R2 = 0.99) and relatively high sensitivity (gauge factor = 14) over a relatively wide strain range (≈43%), with an ultra-low strain detection limit of 0.025%, enabling potential applications in human healthcare monitoring and air leakage detection. Thus, this study unveils a simple methodology to process microstructure-enabled conformable 3D film strain sensors, which show good conformability and multiple mechanical sensing functions for advancing the development of next-generation flexible strain sensors.

Research on Leak Detection and Localization Algorithm for Oil and Gas Pipelines Using Wavelet Denoising Integrated with Long Short-Term Memory (LSTM)–Transformer Models

Traditional leakage prediction models for long-distance pipelines have limitations in effectively synchronizing spatial and temporal features of leakage signals, leading to data processing that heavily relies on manual experience and exhibits insufficient generalization capabilities. This paper introduces a novel leakage detection and localization algorithm for oil and gas pipelines, integrating wavelet denoising with a Long Short-Term Memory (LSTM)-Transformer model. The proposed algorithm utilizes pressure sensors to collect real-time pipeline pressure data and applies wavelet denoising to eliminate noise from the pressure signals. By combining LSTM’s temporal feature extraction with the Transformer’s self-attention mechanism, we construct a short-term average pressure gradient-average instantaneous flow network model. This model continuously predicts pipeline flow based on real-time pressure gradient inputs, monitors deviations between actual and predicted flow, and employs a pressure curve distance algorithm to accurately determine the leakage location. Experimental results from the Jilin-Changchun long-distance oil pipeline demonstrate that the model possesses superior leakage warning and localization capabilities. Specifically, the leakage prediction accuracy reaches 99.995%, with a leakage location error margin below 2.5%. Additionally, the model can detect leaks exceeding 0.6% of the main pipeline flow without generating false alarms during operation.

Leakage detection in pipeline systems using machine learning

Leakage detection in pipeline systems using machine learning

Fine-Grained Leakage Detection for Water Supply Pipelines Based on CNN and Selective State-Space Models

The water supply pipeline system is responsible for providing clean drinking water to residents, but pipeline leaks can lead to water resource wastage, increased operational costs, and safety hazards. To effectively detect the leakage level in the water supply pipelines and address the difficulty of accurately distinguishing fine-grained leakage levels using traditional methods, this paper proposes a fine-grained leakage identification method based on Convolutional Neural Networks (CNN) and the Selective State Space Model (Mamba). An experimental platform was built to simulate different leakage conditions, and multi-axis sensors were used to collect data, resulting in the creation of a high-quality dataset. The signals were converted into frequency-domain images using Short-Time Fourier Transform (STFT), and CNN was employed to extract image features. Mamba was integrated to capture the one-dimensional time dynamic characteristics of the leakage signal, and the CosFace loss function was introduced to increase the inter-class distance, thereby improving the fine-grained classification ability. Experimental results show that the proposed method achieves optimal performance across various evaluation metrics. Compared to SVM, BP neural networks, and CNN methods, the accuracy was improved by 17.9%, 15.9%, and 3.0%, respectively. Compared to Support Vector Machine (SVM), Backpropagation neural network (BP), attention mechanism with the LSTM network (LSTM-AM), CNN, and inverted transformers network (iTransformer) methods, the accuracy improved by 17.9%, 15.9%, 7.8%, 3.0%, and 2.3%, respectively. Additionally, the method enhanced intra-class consistency and increased inter-class differences, showing outstanding performance at different leakage levels, which could contribute to improved intelligent management for water pipeline leakage detection.

Association Between Skull Base Fracture Location and Likelihood of Developing a Traumatic CSF Leak

Background: Skull base fractures (SBFs) pose a significant risk for traumatic cerebrospinal fluid (CSF) leaks, which can lead to serious complications if undetected. Despite anatomical correlations, the predictive value of fracture location on CSF leak development remains inadequately explored in contemporary clinical settings. Objective: To evaluate the association between specific skull base fracture locations (anterior, middle, posterior) and the likelihood of developing a traumatic CSF leak in adult head-injured patients. Methods: This cross-sectional observational study was conducted at the Neurosurgery Department of Shaheed Mohtarma Benazir Bhutto Institute of Trauma (SMBBIT), Karachi, from July to December 2024, including 200 patients with radiographically confirmed skull base fractures. Adults over 18 years were included, excluding those with pathological, surgical, or congenital causes of CSF leaks. Clinical diagnosis of CSF rhinorrhea or otorrhea was supported by bedside glucose testing, and fracture location was confirmed via CT/MRI. Ethical approval was obtained, and all participants provided informed consent per the Declaration of Helsinki. Statistical analyses were conducted using SPSS v25, employing chi-square tests, Pearson correlation, and linear regression to assess associations. Results: CSF leaks occurred in 10% of cases, predominantly in anterior fractures (12/75), but the chi-square test showed no statistically significant association (p = 0.118). Regression analysis revealed fracture location (β = -0.234, p = 0.005) and gender (β = -0.180, p = 0.030) as significant predictors, while age was not. Conclusion: Although anterior skull base fractures were more frequently associated with CSF leaks, fracture location alone did not predict leakage in univariate analysis. Gender and anatomical site, however, emerged as significant predictors in multivariate modeling. These findings support a multifactorial risk assessment approach to enhance early CSF leak detection and improve outcomes in head trauma management.

CFD Analysis on Double Leaks of Subsea Pipeline Leakage

The transportation of materials through industrial pipelines is essential in many sectors. However, pipeline leakage is a common and potentially devastating issue, particularly in subsea environments. Although extensive research has been conducted on pipeline leak detection, the specific dynamics of double leaks in subsea pipelines under varying fluid velocities have not been thoroughly investigated. This study used numerical simulations with ANSYS FLUENT, based on the standard k–ε model under steady-state conditions, to examine the effects of fluid velocity on the leak flow rate, pressure distribution, and turbulence kinetic energy in double-leak subsea pipeline models. The results indicate that changes in pipeline fluid velocity have a minimal impact on the flow behaviour in the leak regions. Notably, the leak flow rate in the first leak consistently exceeded that in the second leak. The pressure around the leaks exhibits fluctuations owing to turbulence, which diminishes further from the leak points towards the pipe centreline. These findings suggest that leak detection systems should be positioned near the pipe surface to enhance the detection accuracy and effectiveness. This study provides valuable insights into the fluid dynamics of double leaks in subsea pipelines and contributes to improved leak management strategies

The role of drain placement in post-bariatric surgery bleeding and leak detection: Palestinian main center experience (2017–2021)

IntroductionBariatric surgery, such as sleeve gastrectomy (SG) and gastric bypass, is a common option for weight loss in patients with obesity and metabolically ill individuals. However, complications like bleeding and leaks can occur. Surgeons often use intraoperative drains to detect these issues, but their effectiveness is debated due to conflicting evidence. Our study aims to evaluate the benefits of intra-abdominal drains in detecting postoperative bleeding and leaks.MethodsThis is a retrospective cross-sectional study, with data from 494 patients who underwent bariatric surgery at the primary bariatric center in Palestine, between 2017 and 2021. Patient demographics, comorbidities, surgical complications, and drain usage were collected, managed, and analyzed using SPSS.ResultsThe study included 494 patients, predominantly females (69.0%) with a median age of 39.5 years. Sleeve gastrectomy was the most common procedure (78.1%). Postoperative complications occurred in 3.0% of patients, with bleeding being the most prevalent (1.4%). Drain placement was routine (82.0%), but no significant correlation was found between drain output volume and signs of bleeding, such as heart rate and blood pressure. However, a weak inverse correlation was observed between volume for bloody drain character and hemoglobin levels on the first postoperative day.ConclusionDrains are commonly used in bariatric surgery; however, their effectiveness in detecting complications like bleeding and leaks remains uncertain. There was no association between drain output volume and signs of bleeding and leak complications, and the clinical assessment, especially the vital signs, is the most effective method in identifying postoperative issues.Clinical trial numberNot applicable.

Android Based Gas Booking System and Leakage Detection Using IOT

Liquefied Petroleum Gas (LPG) is widely used in households and industries, but gas leakage and the unavailability of gas cylinders pose significant safety and convenience challenges. This project presents an IoT-based Gas Leakage Detection and Automatic Gas Booking System that enhances safety and automation. The system integrates Arduino Uno and ESP32 microcontrollers to monitor gas levels and detect leakage in real-time. A Load Cell is employed to measure the weight of the gas cylinder, ensuring continuous monitoring of gas consumption. When the weight of the cylinder drops below a predefined threshold, the system utilizes the GSM module to automatically book a new gas cylinder by sending an SMS notification to the authorized distributor. For safety measures, an MQ-05 Gas Sensor is incorporated to detect any gas leakage. Upon detecting leakage, the system promptly sends an SMS alert to an authorized mobile number, allowing immediate action to prevent hazards. The proposed system enhances safety, reduces manual intervention, and ensures the timely availability of gas cylinders. It also provides a cost-effective and scalable solution for households and industries. By leveraging IoT-based automation, the system significantly improves efficiency in gas booking and safety management, contributing to smart home and industrial applications. KEYWORDS- MQ-05 gas sensor, GSM module, Gas weight monitoring module, SMS alert, Load Cell, IOT

Smart Waterflow and Pipeline Leakage Detection Using IOT

Water scarcity and pipeline inefficiencies are growing concerns globally, with substantial economic and environmental impacts. To address these issues, this project presents a solution using the Internet of Things (IoT) for smart waterflow monitoring and pipeline leakage detection. The system employs a network of sensors integrated into the pipeline infrastructure, allowing real-time monitoring of water flow, pressure, and potential leaks. Data collected by these sensors is transmitted to a central system for analysis, providing early detection of leakage, irregularities, or inefficiencies in the water distribution network. The IoT-enabled platform offers real-time alerts and enables proactive maintenance, minimizing water wastage and reducing operational costs. Key words: Pipeline Leakage Detection, Water Flow Sensors

Image-Based Underwater Liquid Leak Detection and Transfer Learning

This paper addresses the critical challenge of liquid leaks in the oil and gas industry by leveraging advanced computer vision and deep learning methodologies. The objective is to develop practical models for detecting underwater objects with low image quality in adverse conditions. We train and test CNN detectors using Facebook's Detectron2 Faster R-CNN. The model was evaluated on a custom dataset of underwater oil spill videos, focusing on detection accuracy and processing speed. The results demonstrated that even using images of smoke in the sky as training made it possible to detect the underwater oil leak accurately.

Transient hydraulic pressure sensor based on single-hole-dual-core fiber Bragg grating

Transient hydraulic pressure is crucial for hydraulic pressure and pipeline leak detection in the fields of petroleum pipelines, gas pipelines, and so on. Real-time, long-distance, online, high-pressure, and high-precision measurement requirements can be precisely achieved through fiber optic sensing technology. In response to the need for pipeline leak detection, we propose a fiber-optic transient hydraulic pressure sensor based on single-hole-dual-core fiber Bragg grating (SHDC-FBG). By inscribing FBG on SHDCF using a deep ultraviolet laser, we conduct modal field analysis on the SHDCF and theoretical simulation analysis of the resonance wavelength of the SHDC-FBG. The simulation results closely match experimental data, demonstrating that air holes induce significant pressure-driven refractive index changes in the center core. Furthermore, the static and dynamic pressure sensing characteristics have been discussed. Experimental results indicate that the SHDC-FBG sensor exhibits a sensitivity of -5.4 pm/MPa within a wide dynamic hydraulic range of 0-20 MPa. To measure the hydraulic leak process, we compare the SHDC-FBG sensor with a conventional resistive pressure sensor. The experimental results show that the performance of the two sensors is essentially the same. Nevertheless, the compact size, low cost, and electromagnetic immunity make our sensor more competitive in harsh, long-distance environments.

Unsupervised Deep Learning for Blood-Brain Barrier Leakage Detection in Diffuse Glioma Using Dynamic Contrast-enhanced MRI.

"Just Accepted" papers have undergone full peer review and have been accepted for publication in Radiology: Artificial Intelligence. This article will undergo copyediting, layout, and proof review before it is published in its final version. Please note that during production of the final copyedited article, errors may be discovered which could affect the content. Purpose To develop an unsupervised deep learning framework for generalizable blood-brain barrier (BBB) leakage detection using dynamic contrast-enhanced (DCE) MRI, without requiring pharmacokinetic (PK) models and arterial input function (AIF) estimation. Materials and Methods This retrospective study included data from patients who underwent DCE MRI between April 2010 and December 2020. An autoencoder-based anomaly detection (AEAD) identified 1D voxel-wise time-series abnormal signals through reconstruction residuals, separating them into residual leakage signals (RLS) and residual vascular signals (RVS). The RLS maps were evaluated and compared with the volume transfer constant (Ktrans) using the structural similarity index (SSIM) and correlation coefficient (r). Generalizability was tested on subsampled data, and IDH status classification performance was assessed using areas under the receiver operating characteristic curves (AUCs). Results A total of 274 patients were included (164 male; mean age 54.23 ± [SD] 14.66 years). RLS showed high structural similarity (SSIM = 0.91 ± 0.02) and correlation (r = 0.56, P < .001) with Ktrans. On subsampled data, RLS maps showed better correlation with RLS values from original data (0.89 versus 0.72, P < .001), higher PSNR (33.09 dB versus 28.94 dB, P < .001), and higher SSIM (0.92 versus 0.87, P < .001) compared with Ktrans maps. RLS maps also outperformed Ktrans maps in predicting IDH mutation status (AUC = 0.87 [95% CI: 0.83-0.91] versus 0.81 [95% CI: 0.76-0.85], P = .02). Conclusion The unsupervised framework effectively detected blood-brain barrier leakage without PK models and AIF. ©RSNA, 2025.

Interpretable deep learning for acoustic leak detection in water distribution systems.

Interpretable deep learning for acoustic leak detection in water distribution systems.

Photon-Counting CT Myelography for the Detection of Spinal CSF Leaks.

Photon-counting CT myelography is a relatively recently described technique for the detection of spinal CSF leaks in patients with spontaneous intracranial hypotension.1-3 In this Video, we outline the fundamental procedural techniques used in CT myelography for CSF leak detection, as well as specific advantages and reconstruction options available when using photon-counting CT.4-7 Additionally, we show examples demonstrating the typical appearance of common spinal CSF leaks detected on photon-counting CT myelography, which mainly include dural tears and CSF-venous fistulas.8 This Video draws on technical insights gleaned from a multi-institutional perspective, while specific techniques continue to vary at different centers. Future refinement of photon-counting CT myelographic techniques at multiple centers will be helpful to maximally leverage this nascent technology.