Adaptive optimisation of road safety strategic management

Jiao, Bosong ORCID: 0009-0000-7476-3930 (2025). Adaptive optimisation of road safety strategic management. University of Birmingham. Ph.D.

Preview

Jiao2025PhD.pdf
Text - Accepted Version
Available under License All rights reserved.
Download (17MB) | Preview

Abstract

Road crashes cause significant injuries and fatalities, hindering social and economic progress despite the essential role of a well-developed road network. In 2021, road crashes resulted in 1.19 million fatalities globally, underscoring this critical health issue. Various initiatives by the UN, WHO, Global Road Safety Facility (GRSF), and national road entities focus on reducing traffic crashes through strategic road infrastructure safety management. However, current tools and methods used for road safety management suffer from fragmented data analysis, focusing on specific segments instead of the whole network, limiting comprehensive crash insights. There is also poor integration across decision-making levels and reliance on subjective, localised countermeasures rather than adaptive, data-driven investment strategies. This creates a need for a system utilising Data Science and Artificial Intelligence (AI) to enhance road safety management.
To achieve this, the thesis begins with a systematic literature review of cutting-edge technologies and best practices in road safety. It emphasises the role of information technology, particularly big data, in crash data analysis, identifying a gap in the comprehensive understanding of crash databases and the need for advanced techniques from AI and multi-objective optimisation. These insights guide subsequent research aimed at developing an adaptive, data-driven system for strategic management, encompassing both data analysis and resource allocation.
Existing methods for data analysis often suffer from subjectivity and bias, as they rely on simplistic predefined parameters for frequent pattern searches without critical and logical thinking. Furthermore, they face a significant limitation when applied to the infrequent patterns search in road safety analysis, particularly in cases where fatalities are minimal. An analytical framework using association rule mining and K-means clustering analyses the UK STATS 19 database (2018) to identify associations with fatalities and overview crash data characteristics. This mitigates bias by integrating these methods to explore infrequent patterns, offering a comprehensive perspective to inform strategy development.
Additionally, a novel two-stage model for optimising road safety strategies is introduced, integrating multi-objective optimisation and association rule mining. This model enhances decision-making and resource allocation, validated through a case study in Utrecht. Improvements across objectives demonstrate its efficiency, allowing local authorities to tailor investment plans to their road network characteristics.
In conclusion, this thesis presents a novel conceptual framework for road safety strategic management using advanced data information technology. The adaptable analytical tools developed provide valuable insights for decision-makers at various administrative levels, significantly contributing to road safety. This research establishes a foundation for future studies and practical applications aimed at reducing road traffic crashes.

Type of Work:

Thesis (Doctorates > Ph.D.)

Award Type:

Doctorates > Ph.D.

Supervisor(s):