Full metadata record
| DC Field | Value | Language |
|---|---|---|
| dc.contributor | Department of Land Surveying and Geo-Informatics | en_US |
| dc.contributor.advisor | Lai, Wallace (LSGI) | en_US |
| dc.contributor.advisor | Zhu, Xiaolin (LSGI) | en_US |
| dc.creator | Zhou, Yimin | - |
| dc.identifier.uri | https://theses.lib.polyu.edu.hk/handle/200/14446 | - |
| dc.language | English | en_US |
| dc.publisher | Hong Kong Polytechnic University | en_US |
| dc.rights | All rights reserved | en_US |
| dc.title | Characterizing and detecting urban underground hazards by temporal ground penetrating radar measurements | en_US |
| dcterms.abstract | The increasing urban population has intensified demands for buried water supplies, drainage systems, energy distribution, and telecommunication networks. Repetitive road openings for new installations and replacement of aging utilities have made underground infrastructure more complex and congested. Aging utility infrastructure often leads to issues like ground subsidence caused by voids and water main leakages. Effective imaging and diagnostics are critical for decision-making on rehabilitation or replacement strategies. Ground penetrating radar (GPR) has emerged as the most suitable non-destructive technology for underground imaging. However, traditional GPR surveys using single- or dual-channel systems are labor-intensive, time-consuming, and limited to small-scale applications, making them impractical for large-scale urban hazard detections of void and leakage. Multichannel GPR (MCGPR), as an evolving technology in recent years, enhances efficiency by enabling large-scale and temporal data acquisition because vehicle-towed antennae can be arranged side-by-side together. But is the GPR processing simply an exercise of channel adding up? Can the conventional GPR survey scheme be simply adopted? The answers to the above questions are all negative, and migration of the data processing approach from GPR to MCGPR is required but underexplored in the GPR community. | en_US |
| dcterms.abstract | This thesis bridges critical gaps in MCGPR-based hazard detection through three interconnected contributions: | en_US |
| dcterms.abstract | (1). Leakage Signature Database & Frequency-Domain Characterization | en_US |
| dcterms.abstract | A comprehensive database of 78 real-world leakage scenarios was created using a 2,400 sq.m well-designed experimental site 'Q-Leak'. This database reduces reliance on simulated data by capturing leakage patterns across different pavement types and pipe attributes. Each leakage scenario was designed and measured in a before-and-after emulation scheme, enabling the database to capture both static leakage patterns and dynamic leakage patterns through temporal GPR measurements. To facilitate dynamic pattern recognition, three sub-datasets were created. This database provides a solid foundation for advancing the recognition of temporal leakage patterns. | en_US |
| dcterms.abstract | Additionally, GPR-Instantaneous frequency slices (IFS) in the time-frequency domain were developed as a new method for characterizing water leakage patterns. The developed method addresses the limitations of traditional amplitude slice analysis, and the restrictions of signal aliasing and windowing in short-time Fourier transform (STFT) and wavelet transform methods. | en_US |
| dcterms.abstract | (2). MCGPR Error Mitigation Framework | en_US |
| dcterms.abstract | Two error mitigation approaches were developed to reduce geo-location misalignment of MCGPR trajectory (above-ground) and harmonize signal variations in MCGPR antennae channels (underground). The first method focuses on correcting geolocation misalignments, which are a common yet often neglected issue in MCGPR trajectory data. A novel post-processing method that, for the first time, leverages cross-channel MCGPR measurements was proposed to enhance positioning accuracy. This approach achieves centimeter-level accuracy and is compatible with various positioning systems, making it a versatile solution for improving geolocation reliability. For signal harmonization, A two-stage correction method was implemented to improve signal consistency, ensuring that data from multiple surveys and channels are comparable. These error mitigation techniques enhance the overall quality of MCGPR data, enabling more accurate analysis and hazard detection. | en_US |
| dcterms.abstract | (3). Change-detection Algorithms | en_US |
| dcterms.abstract | There are two foci of algorithm development for identifying underground hazards, i.e., voids and water leakages. For void detection, an unsupervised superpixel-based method combines fuzzy C-means (FCM) clustering with a Markov Random Field (MRF) model. This approach integrates 3D neighborhood information to iteratively refine change maps. For leakage detection, a hierarchical methodology leverages both amplitude and instantaneous frequency analyses. Potential leakage regions are identified using 3D FCM clustering, followed by IFS to extract frequency attributes associated with water leakage, further narrowing regions of interest due to suspected leakage. | en_US |
| dcterms.abstract | The individual research contributions in the above three perspectives were integrated to build a holistic forensic investigation in a real-world case study of leakage. The study began with an initial MCGPR survey over a 5.5 km road segment of known leakage, but it had no idea where it could be. By studying the anomaly of water pressure records, a 1.6 km subsection was identified as a preliminary area of suspected leakage, where three subsequent temporal MCGPR surveys were conducted. The developed error mitigation methods and change detection algorithms were applied to data collected during two distinct periods (Period 1 and Period 2), divided by the three temporal surveys. The suspected leakage regions identified during Period 2 aligned closely with the actual pipe burst location on 3rd November 2024, validating the effectiveness of the developed methods. | en_US |
| dcterms.abstract | MCGPR, as the most recent development in the GPR community, is not merely an antenna array stacked and towed behind a vehicle. In reality, its adoption and change-detection diagnosis of underground requires a series of fundamental understanding of antennae and signals, and algorithm research and development. This research advances MCGPR-based underground hazard characterization by developing and validating a series of scientific approaches. The contributions of this work are believed to benefit engineering, surveying, and infrastructure management authorities by offering innovative tools that pave the way for the non-invasive and data-driven detection, characterization, and prediction of underground hazards. | en_US |
| dcterms.extent | xxi, 255 pages : color illustrations | en_US |
| dcterms.isPartOf | PolyU Electronic Theses | en_US |
| dcterms.issued | 2026 | en_US |
| dcterms.educationalLevel | Ph.D. | en_US |
| dcterms.educationalLevel | All Doctorate | en_US |
| dcterms.accessRights | open access | en_US |
Copyright Undertaking
As a bona fide Library user, I declare that:
- I will abide by the rules and legal ordinances governing copyright regarding the use of the Database.
- I will use the Database for the purpose of my research or private study only and not for circulation or further reproduction or any other purpose.
- I agree to indemnify and hold the University harmless from and against any loss, damage, cost, liability or expenses arising from copyright infringement or unauthorized usage.
By downloading any item(s) listed above, you acknowledge that you have read and understood the copyright undertaking as stated above, and agree to be bound by all of its terms.
Please use this identifier to cite or link to this item:
https://theses.lib.polyu.edu.hk/handle/200/14446