Nguyen, Thinh Tan (2025) Accident data analytics and AI-based auditory enhancement for highway construction safety. PhD thesis, Clemson University, USA.
Abstract
This research aims to develop data-driven and AI-enhanced methodologies to improve worker safety in highway construction environments. The study integrates advanced analytical techniques and signal enhancement models to identify high-risk worker behaviors and enhance auditory safety warning systems in real time. The research comprises three core studies. The first study utilizes Sequential Pattern Mining (SPM) and Social Network Analysis (SNA) to explore over 1,000 construction accident reports, identifying high-risk worker actions and uncovering their hidden sequential relationships with accident types and consequences. This analysis offers critical insight into sector-specific risk patterns and supports the development of targeted, data-informed safety measures. The second study presents a novel AI-based auditory enhancement model using the Conformer-based Metric Generative Adversarial Network (CMGAN) architecture. The model is designed to improve the audibility and clarity of safety-critical signals, such as backup alarms and intrusion alerts, in noisy construction environments. By capturing both local acoustic details and long-range temporal dependencies, and using adversarial training with perceptual loss functions, the model effectively suppresses background noise while preserving essential signal characteristics, even under extreme noise conditions. The final study focuses on developing a real-time auditory enhancement prototype using the DeepFilterNet algorithm. This lightweight system is designed for on-site deployment and can be integrated into portable hearing protection devices (HPDs), enabling low-latency enhancement of safety-critical signals such as backup alarms and work zone intrusion alerts. The prototype was implemented on a low-cost Raspberry Pi 5 platform and evaluated through functionality testing, field deployment at active construction sites, and subjective listener surveys. Results demonstrate that the system effectively improves signal audibility while preserving essential auditory cues, confirming its feasibility for real-time operation in noisy highway construction environments. The proposed research is expected to make a significant contribution to the understanding of high-risk worker actions and the practical application of AI-driven auditory safety technologies. The findings hold promising implications for reducing accident severity, enhancing real-time communication, and improving overall safety practices in the highway construction industry.
| Item Type: | Thesis (Doctoral) |
|---|---|
| Thesis advisor: | Le, Tuyen Robert |
| Index terms: | platform, safety measures, testing, highway construction, real time, mining, survey, real-time operation, prototype, social network analysis, construction site, functionality, construction accident, methodology |
| Subjects: | design features, professional practice, research methods, digital design, health safety and environment, modelling and simulation, work location, data collection methods, geotechnical engineering, occupational health and safety management, project controls, civil engineering, management |
| Topics: | Design Practice, Digital Applications, Time Control, Site Management, Business Strategy, Research Practice, Health and Safety, Engineering Principles |
| Descriptive scope: | 4 PCTE |
N.B. Descriptive scope is a count of how many of the five facets of empirical research are indicated by the words used in title, abstract and keywords. It is not intended as a judgement on the research; merely a count of the kind of word we would expect to indicate Phenomenon, Concepts, Theoretical framing, Empirical techniques, Analytical techniques. If all five are present, then a code of “5 PCTEA” will indicate this. If you feel the coding for this record is questionable, we welcome discussion around the terms we matched or the way we categorized them. The facet you would expect may not be coded, or a facet may be coded inappropriately. This can also bear on a larger question, of which facets should be treated as defining in construction management research. Please get in touch, and we will look at it. More details here