Deep Reinforcement Learning-Based Resilient Restoration of Ship Cyber–Physical Systems
Journal of Marine Science and Engineering · 22.04.2026 · 14 (9) · 765
DOI: 10.3390/jmse14090765 ↗Kurz gefasst
Fachartikel in der Zeitschrift Journal of Marine Science and Engineering, online erschienen am 22.04.2026. Verfasst von Yahui Liu und 4 weiteren Autor:innen. Die Arbeit nennt 41 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: National Natural Science Foundation of China.
Abstract (Originalsprache)
The rapid development of cyber–physical technologies has led to enhanced observability and controllability of shipboard power systems. However, the reliance of shipboard power systems on information networks undermines the traditional security provided by physical isolation; under malicious attacks, faults in the information domain can propagate rapidly, causing physical power outages and reducing the resilience of shipboard power systems. To address this issue, this paper investigates the cascading failure reconstruction and resilience enhancement in shipboard cyber–physical systems (SCPSs) under uncertain network attacks. First, a cascading failure propagation model is established to capture the interaction between attack paths and system vulnerabilities, revealing how cyberattacks spread through communication links and infiltrate the power topology. Then, a reinforcement learning-based load recovery strategy is developed, in which a masked proximal policy optimization (masked-PPO) algorithm is employed to optimize reconfiguration decisions under operational constraints. The proposed approach enables adaptive and efficient recovery actions in complex cross-domain environments. Case studies based on representative SCPS scenarios demonstrate that the proposed method improves cascading-failure reconfiguration capability by 13.21% and reduces the average decision time by 18.6%, validating its effectiveness, real-time performance, and scalability.
Förderung
- National Natural Science Foundation of China
Zitieren
APA
Liu, Y., Wen, S., Zhao, Q., Zhang, B., & Lu, Z. (2026). Deep Reinforcement Learning-Based Resilient Restoration of Ship Cyber–Physical Systems. Journal of Marine Science and Engineering, 14(9), 765. https://doi.org/10.3390/jmse14090765
Vancouver
Liu Y, Wen S, Zhao Q, Zhang B, Lu Z. Deep Reinforcement Learning-Based Resilient Restoration of Ship Cyber–Physical Systems. Journal of Marine Science and Engineering. 2026;14(9):765. doi:10.3390/jmse14090765
BibTeX
@article{Liu2026,
author = {Liu, Yahui and Wen, Shuli and Zhao, Qiang and Zhang, Bing and Lu, Zhangchao},
title = {Deep Reinforcement Learning-Based Resilient Restoration of Ship Cyber–Physical Systems},
journal = {Journal of Marine Science and Engineering},
year = {2026},
volume = {14},
number = {9},
pages = {765},
doi = {10.3390/jmse14090765}
}
EUVIMED