A data-driven dynamic Bayesian network for health monitoring and maintenance decision support of marine diesel engines

Usman, M, Celik, C orcid iconORCID: 0000-0002-2335-0455, Li, H orcid iconORCID: 0000-0002-4293-4763, Qu, Z, Chang, CH orcid iconORCID: 0000-0002-7351-8471, Wang, J orcid iconORCID: 0000-0003-4646-9106 and Yang, Z orcid iconORCID: 0000-0003-1385-493X (2026) A data-driven dynamic Bayesian network for health monitoring and maintenance decision support of marine diesel engines. Ocean Engineering, 369 (3). ISSN 0029-8018

[thumbnail of A data-driven dynamic Bayesian network for health monitoring and maintenance decision support of marine diesel engines.pdf]
Preview
Text
A data-driven dynamic Bayesian network for health monitoring and maintenance decision support of marine diesel engines.pdf - Published Version
Available under License Creative Commons Attribution.

Download (16MB) | Preview

Abstract

Performance monitoring and maintenance decision-making are critical components of machinery management for ensuring the safety, reliability, and operational efficiency of maritime systems. Recent advances in sensor technologies, data transmission capabilities, machine learning, and knowledge management have accelerated the transition from traditional fixed-interval maintenance towards condition-based and predictive maintenance strategies in the maritime industry. However, the health assessment of large two-stroke marine diesel engines remains challenging due to the presence of multiple interdependent operational parameters and their temporal uncertainties under varying sea and operational conditions. To address these challenges, this paper proposes a novel data-driven Dynamic Bayesian Network (DBN) framework for the health monitoring and maintenance decision support of large two-stroke marine diesel engines. The proposed framework integrates 15 operational parameters within a unified probabilistic model to capture temporal degradation and quantify health uncertainty through dynamic probabilistic reasoning. The framework was validated using operational data from a six-cylinder MAN B&W 6S50MC-C main engine. A five-day case study showed that cylinders 4 and 6 remained in the AMBER health state, while cylinder 5 was predicted to deteriorate from GREEN to AMBER after two days of operation, demonstrating the model's capability to predict health degradation trends. Sensitivity analysis identified Total Base Number (TBN) and iron content as the most influential indicators affecting engine health. Furthermore, water mist catcher performance, fuel viscosity, and exhaust temperature deviation exhibited the largest temporal variations, with average deviations of 4.31%, 3.91%, and 3.37%, respectively. The results demonstrate that the proposed DBN framework effectively captures temporal dependencies and operational uncertainty, enabling dynamic health assessment and risk-informed maintenance decision-making. By integrating multiple interdependent degradation factors within a unified probabilistic framework, the proposed approach provides an effective tool for condition-based maintenance and establishes a practical foundation for future maritime digital twin and intelligent maintenance applications.

Item Type: Article
Uncontrolled Keywords: Marine diesel engine; Health monitoring; Condition-based maintenance; Preventive maintenance; Dynamic Bayesian network (DBN); Maritime safety; 4012 Fluid Mechanics and Thermal Engineering; 4005 Civil Engineering; 4015 Maritime Engineering; 40 Engineering; Networking and Information Technology R&D (NITRD); Generic health relevance; 0405 Oceanography; 0905 Civil Engineering; 0911 Maritime Engineering; Civil Engineering; 4005 Civil engineering; 4012 Fluid mechanics and thermal engineering; 4015 Maritime engineering
Subjects: T Technology > T Technology (General)
T Technology > TA Engineering (General). Civil engineering (General)
Divisions: Engineering and Built Environment
Publisher: Elsevier
Date of acceptance: 14 October 2026
Date of first compliant Open Access: 20 November 2026
Date Deposited: 20 Nov 2026 15:28
Last Modified: 20 Nov 2026 15:28
DOI or ID number: 10.1016/j.oceaneng.2026.128885
URI: https://researchonline.ljmu.ac.uk/id/eprint/29908
View Item View Item