Thesis

Machine learning and case-based reasoning for damage stability decision support

Downloadable Content

Download PDF
Creator
Rights statement
Awarding institution
  • University of Strathclyde
Date of award
  • 2025
Thesis identifier
  • T17328
Person Identifier (Local)
  • 201994186
Qualification Level
Qualification Name
Department, School or Faculty
Abstract
  • This thesis investigates the integration of Machine Learning (ML) and Case-Based Reasoning (CBR) to enhance decision support for real-time damage stability assessment in passenger ships under dynamic flooding scenarios. Despite advances in maritime safety regulations, flooding incidents remain a leading cause of catastrophic vessel loss, necessitating innovative approaches to assess and mitigate risks. The research presents a novel framework that combines probabilistic and case-based methodologies, leveraging simulations to predict outcomes such as time to capsize (TTC) and critical survival factors. Key contributions include the development of a probabilistic decision support system that utilises Bayesian and Dempster-Shafer approaches to fuse predictions from ML and CBR, effectively managing uncertainty and enhancing situational awareness during emergencies. This system incorporates real-time monitoring, feature engineering from historical incident data, and dynamic simulations to account for complex flooding patterns. The methodology is validated through case studies and a Monte Carlo-based uncertainty analysis, demonstrating its efficacy in improving accuracy and reliability. The research highlights the implications of these advancements for passenger ships, offering insights into their survivability under extreme conditions. It aligns with industry trends and regulatory objectives, contributing to safer ship designs and operational strategies. Recommendations for future work include exploring advanced computational models, integrating human factors, and expanding the framework to other ship types, emphasising the adaptability and scalability of the proposed solution.
Advisor / supervisor
  • Boulougouris, Evangelos
Resource Type
Note
  • Previously held under moratorium from 15 May 2025 until 15 May 2026.
DOI
Funder

Relations

Items

Thumbnail Title Date Uploaded Visibility Actions
file details PDF of thesis T17328 2025-06-10 Public Download