Data-driven Safety Incident Analysis and Risk Perception in Urban Railway Construction
數據驅動的城市軌道交通工程施工安全事件分析與風險感知研究
Student thesis: Doctoral Thesis
Author(s)
Related Research Unit(s)
Detail(s)
| Awarding Institution | |
|---|---|
| Supervisors/Advisors |
|
| Award date | 18 Dec 2023 |
Link(s)
| Permanent Link | https://scholars.cityu.edu.hk/en/theses/theses(767c504b-9c45-4596-9462-332b824d6058).html |
|---|---|
| Other link(s) | Links |
Abstract
This thesis investigates the high-risk environment of urban railway construction and proposes several data-driven, multi-technology strategies for accident prevention. The research encompassed four components, each targeting a vital issue in incident analysis and accident prevention.
Firstly, the study developed a unique database, the Urban Railway Construction Incident Database (UCRID), concentrating on past incidents in urban railway constructions. UCRID collected and managed historical accident data and near-miss situations from scattered sources to construct a relevant and effective database. The data analysis allowed the exploration of patterns and trends related to the time, region, and type of incidents, ultimately creating a solid foundation for further accident analysis and prevention protocols.
The study’s second part involved understanding the critical causal factors leading to accidents and near misses within urban railway construction. This effort employed a complex network approach to create an Urban Railway Construction Safety Risk Network (URCSRN). By studying and comparing the overall and local features of the URCSRN, the research identified unequal roles of different causations in triggering accidents or near misses. This helped in deciphering its spatial-temporal and worker-related characteristics leading to accidents, as well as near misses, thus adding to an effective prevention measure.
Thirdly, the research sought to identify the critical groups of causal factors that could lead to urban railway construction accidents. Using a Bayesian network (BN) based risk analysis model, the study explored the sequence and extent of causal factors leading to accidents. This model also proved relevant in predicting the probability of different types of accidents, enabling more accurate prevention measures.
The last segment aimed to construct a Collaborative Safety Risk Perception Framework (CSRPF) to address on-site safety risks in real-time. Taking up the case of mobile crane lifts as an example, CSRPF could recognize hand signals, mobile crane motion states, and unauthorized intrusion of workers on construction sites. The data from these various sources were collected and compared to detect potential safety risks. The CSRPF proved efficient in its field test, supporting real-time monitoring and alarms for safety threats.
In conclusion, this thesis advocates for the application of data-driven, multi-technology approaches to prevent urban railway construction accidents efficiently. The creation of the UCRID database, essential causal factors identification, the use of the BN model, and the development of CSRPF contribute towards better safety practices in urban railway construction.
Firstly, the study developed a unique database, the Urban Railway Construction Incident Database (UCRID), concentrating on past incidents in urban railway constructions. UCRID collected and managed historical accident data and near-miss situations from scattered sources to construct a relevant and effective database. The data analysis allowed the exploration of patterns and trends related to the time, region, and type of incidents, ultimately creating a solid foundation for further accident analysis and prevention protocols.
The study’s second part involved understanding the critical causal factors leading to accidents and near misses within urban railway construction. This effort employed a complex network approach to create an Urban Railway Construction Safety Risk Network (URCSRN). By studying and comparing the overall and local features of the URCSRN, the research identified unequal roles of different causations in triggering accidents or near misses. This helped in deciphering its spatial-temporal and worker-related characteristics leading to accidents, as well as near misses, thus adding to an effective prevention measure.
Thirdly, the research sought to identify the critical groups of causal factors that could lead to urban railway construction accidents. Using a Bayesian network (BN) based risk analysis model, the study explored the sequence and extent of causal factors leading to accidents. This model also proved relevant in predicting the probability of different types of accidents, enabling more accurate prevention measures.
The last segment aimed to construct a Collaborative Safety Risk Perception Framework (CSRPF) to address on-site safety risks in real-time. Taking up the case of mobile crane lifts as an example, CSRPF could recognize hand signals, mobile crane motion states, and unauthorized intrusion of workers on construction sites. The data from these various sources were collected and compared to detect potential safety risks. The CSRPF proved efficient in its field test, supporting real-time monitoring and alarms for safety threats.
In conclusion, this thesis advocates for the application of data-driven, multi-technology approaches to prevent urban railway construction accidents efficiently. The creation of the UCRID database, essential causal factors identification, the use of the BN model, and the development of CSRPF contribute towards better safety practices in urban railway construction.
