Abstract:Early detection of fuel leakage at service stations with underground petroleum storage systems is a crucial task to prevent catastrophic hazards. Current data-driven fuel leakage detection methods employ offline statistical inventory reconciliation, leading to significant detection delays. Consequently, this can result in substantial financial loss and environmental impact on the surrounding community. In this paper, we propose a novel framework called Memory-based Online Change Point Detection (MOCPD) which operates in near real-time, enabling early detection of fuel leakage. MOCPD maintains a collection of representative historical data within a size-constrained memory, along with an adaptively computed threshold. Leaks are detected when the dissimilarity between the latest data and historical memory exceeds the current threshold. An update phase is incorporated in MOCPD to ensure diversity among historical samples in the memory. With this design, MOCPD is more robust and achieves a better recall rate while maintaining a reasonable precision score. We have conducted a variety of experiments comparing MOCPD to commonly used online change point detection (CPD) baselines on real-world fuel variance data with induced leakages, actual fuel leakage data and benchmark CPD datasets. Overall, MOCPD consistently outperforms the baseline methods in terms of detection accuracy, demonstrating its applicability to fuel leakage detection and CPD problems.

What problem does this paper attempt to address?

The paper attempts to address the problem of early detection of fuel leaks in underground petroleum storage systems at gas stations. Existing data-driven fuel leak detection methods mainly rely on offline statistical inventory reconciliation, which leads to significant detection delays, potentially causing severe economic losses and environmental impacts. Therefore, the paper proposes a Memory-based Online Change Point Detection (MOCPD) framework to achieve real-time or near-real-time fuel leak detection. ### Main Issues 1. **Detection Delay**: Existing methods typically require weeks of daily data for static, offline experiments, resulting in delayed detection outcomes. 2. **Economic Losses and Environmental Impact**: Prolonged leak detection delays can lead to severe soil and groundwater contamination, threatening drinking water resources, and the cleanup process may take years and cost millions of dollars. 3. **Low-Granularity Data**: Existing methods rely on low-granularity data, whereas high-granularity data provided by Automatic Tank Gauges (ATGs) can enable earlier detection. 4. **Detection Accuracy**: Existing methods apply a uniform approach to all tanks without considering the specific characteristics of each tank, which may affect detection accuracy. 5. **Data Noise**: Fuel deviations are not only caused by leaks but may also result from measurement errors and evaporation losses, which increase data noise and affect detection performance. ### Solution The MOCPD framework proposed in the paper addresses the above issues in the following ways: 1. **Real-Time Detection**: MOCPD can continuously process incoming fuel inventory data and detect state changes (i.e., from normal to leak state or vice versa) in real-time or near-real-time. 2. **Memory Management**: MOCPD maintains a finite-sized memory that stores representative historical data and periodically updates the samples in memory to adapt to the latest trends of each tank. 3. **Adaptive Threshold Mechanism**: MOCPD dynamically calculates thresholds based on the samples in memory, ensuring detection accuracy and timeliness. 4. **Multiple Detection Methods**: MOCPD can be combined with statistical techniques and machine learning methods for dissimilarity measurement, making it suitable for engineering problems. ### Experimental Validation The paper validates the effectiveness of MOCPD through experiments on real-world fuel deviation data, simulated leak data, actual fuel leak data, and benchmark change point detection (CPD) datasets. The experimental results show that MOCPD outperforms other baseline online CPD methods in terms of detection accuracy, especially in fuel leak datasets containing multiple sequence instances, slight state transitions, and long-duration states. ### Main Contributions 1. **First Real-Time Fuel Leak Detection**: Compared to existing offline methods, MOCPD can continuously monitor each tank's data in a real-time environment, achieving early detection. 2. **Proposed New Memory Management Framework**: MOCPD designs an update phase to determine when and how to update the memory, making detection more accurate and timely. 3. **Adaptive Threshold Mechanism**: MOCPD improves detection precision and recall by dynamically calculating thresholds. 4. **Wide Applicability**: MOCPD is not only applicable to fuel leak detection but can also be extended to other types of change point detection problems.

Real-time Fuel Leakage Detection via Online Change Point Detection

A Real-Time Method to Detect the Leakage Location in Urban Water Distribution Networks

Time-Series-Based Leakage Detection Using Multiple Pressure Sensors in Water Distribution Systems.

An Angle-Based Leak Detection Method Using Pressure Sensors in Water Distribution Networks

Real-Time Burst Detection in District Metering Areas in Water Distribution System Based on Patterns of Water Demand with Supervised Learning

An Integrated Bottom-Up Approach for Leak Detection in Water Distribution Networks Based on Assessing Parameters of Water Balance Model

Leakage Identification in Water Distribution Networks with Error Tolerance Capability

Leak Detection in Water Distribution Systems by Classifying Vibration Signals

Towards a more realistic approach to the problem of detecting fuel leaks in filling stations: Mixed time windows

Real-time Detection of Urban Gas Pipeline Leakage Based on Machine Learning of IoT Time-Series Data

A Novel Hybrid Internal Pipeline Leak Detection and Location System Based on Modified Real-Time Transient Modelling

Pipeline leak localization using matched-field processing incorporating prior information of modeling error

Memory-free Online Change-point Detection: A Novel Neural Network Approach

A Highly Precise and Lightweight Detection Model for Water and Oil Leakage Detection

Adaptive Dynamic Thresholding Method for Fault Detection in Diesel Engine Lubrication Systems

Hierarchical Leak Detection and Localization Method in Natural Gas Pipeline Monitoring Sensor Networks

Burst Detection in District Metering Areas Using Deep Learning Method

Pipeline leakage localization method based on transient signal detection and data fusion from multi-sensor

Simultaneous quantification and changepoint detection of point source gas emissions using recursive Bayesian inference

TagLeak: Non-Intrusive and Battery-Free Liquid Leakage Detection with Backscattered Signals

Change-Point Detection in Industrial Data Streams based on Online Dynamic Mode Decomposition with Control