Data-driven modelling and risk analysis of human evacuation from passenger ships

Wang, X (2026) Data-driven modelling and risk analysis of human evacuation from passenger ships. Doctoral thesis, Liverpool John Moores University.

[thumbnail of 2026xinjianwangphd.pdf]
Preview
Text
2026xinjianwangphd.pdf - Published Version
Available under License Creative Commons Attribution Non-commercial.

Download (1MB) | Preview
[thumbnail of 2026xinjianwangphdinternal.pdf] Text
2026xinjianwangphdinternal.pdf - Accepted Version
Access Restricted

Download (47MB)

Abstract

The safety and efficiency management of Human Evacuation from Passenger Ships (HEPS) is moving towards to a new scheme of data-driven risk analysis and evidence-based safety management. This PhD study by Published Works, presented as a compilation of six peer-reviewed articles, systematically establishes a holistic, risk-based analytical framework through integrating multidimensional risk analysis, causal-structural modelling, causal inference analysis and simulation-based prediction. It aims to advance the scientific foundation for HEPS safety by developing a coherent identification-analysis-simulation-management workflow that progresses from fundamental risk understanding to operational decision-support.

This PhD study by Published Works makes five fundamental contributions to maritime evacuation safety. The first contribution is pioneering the investigation of risk heterogeneity across operational contexts through an integrated Enhanced Multilevel Association Rule Mining (EMARM), the Weighted Influence Non-linear Gauge System (WINGS), the Total Adversarial Hasse Diagram Technology (TAHDT) and the Matrices Impacts Croises-Multiplication Appliance Classement (MICMAC) analytical model. It provides the first scientific evidence that critical risk influencing factors can be dynamically reconfigured across different watchkeeping periods, establishing a new scheme for developing temporally adaptive evacuation strategies. The second contribution establishes an integrated Association Rule Mining (ARM), the WINGS and the Adversarial Interpretive Structure Model (AISM) analytical framework for assessing maritime occupational accidents, systematically identifying risk influencing factors, quantifying their causal interactions, and constructing hierarchical topologies to reveal intrinsic risk transmission pathways. The third contribution pioneers the causal inference exploration of exit choice behaviour during emergency evacuation through an optimised Double Machine Learning (DML) analytical method. By optimising the nuisance function and adopting large-scale valid questionnaire data for quantitative causal analysis, it distinguishes causal effects from spurious correlations in evacuation decision scenarios, identifying core factors governing passenger exit choice and establishing a reliable paradigm for formulating targeted human evacuation intervention strategies onboard passenger ships. The fourth contribution proposes a comprehensive risk assessment framework for the entire HEPS process based on a Human–Ship–Environment–Organisation (HSEO) system. By integrating hybrid modelling approaches including Failure Mode and Effects Analysis (FMEA), Bayesian Belief Networks (BBNs), Evidential Reasoning (ER), Complex Network (CN) theory, Decision-Making Trial and Evaluation Laboratory (DEMATEL) and Interpretive Structural Modelling (ISM), this framework enables systematic identification, quantification, and prioritization of risks across all evacuation phases. The fifth contribution is the development of an integrated simulation-machine learning methodology for HEPS time prediction, combining agent-based simulation with state-of-the-art machine learning algorithms to achieve rapid and accurate evacuation forecasting while balancing predictive accuracy with computational timeliness for emergency decision-support.

By synthesizing these contributions, this PhD study by Published Works establishes a multi-layered analytical framework that spans from microscopic risk factor analysis to macroscopic system-level safety assessment. It transforms conventional experience-based safety management into a quantitatively-grounded, scientifically-driven paradigm, providing maritime stakeholders with advanced methodological tools for evidence-based safety planning. The demonstrated impact through high citation rates, industry awards for equipment development, and successful grant applications underscores this research's significant academic and practical value in enhancing evacuation safety for next-generation passenger ship operations.

Item Type: Thesis (Doctoral)
Uncontrolled Keywords: Emergency Evacuation; Maritime Safety; Maritime Transport; Passenger Ships; Risk Analysis
Subjects: H Social Sciences > HD Industries. Land use. Labor > HD61 Risk Management
H Social Sciences > HE Transportation and Communications
T Technology > TA Engineering (General). Civil engineering (General)
Divisions: Engineering and Built Environment
Date of acceptance: 22 October 2026
Date of first compliant Open Access: 18 November 2026
Date Deposited: 18 Nov 2026 13:40
Last Modified: 18 Nov 2026 13:45
Supervisors: Yang, Z, Matthews, C and Wang, J
URI: https://researchonline.ljmu.ac.uk/id/eprint/29857
View Item View Item