9 February 2023 in Accident Analysis & Prevention
15 October 2024 in Ocean Engineering
22 January 2026 in IEEE Transactions on Intelligent Transportation Systems
With the rapid development of the marine economy and the escalating intensity of maritime traffic, compounded by the influence of geopolitical tensions, climate change and other factors, the risk of marine accidents has significantly increased. Therefore, proposing an appropriate analysis method to find the critical elements of marine accidents is of great practical significance. However, due to the numerous elements of marine accident causation and the complex coupling, manual identification and analysis based on traditional methods are inefficient and unable to tackle complex tasks. In this paper, a Structured Prompt-based Accident Causation Elements Extraction Model (SPACE) is constructed to extract and classify accident causal factors and their coupling links automaticially. Moreover, a Multi-elements Coupling Network of Accident Causation (MCNAC) for marine accidents is built to analyze the critical elements of marine accident by calculating network topology indicators. A comparative evaluation analysis is conducted on the extraction performance with mainstream language models. The study shows that the accuracy, recall, and F1 score indicators of the SPACE model have reached 91%, 90%, and 90%, respectively, which are superior to existing model indicators. The key system component “captain” has higher importance in the network, and the two most common factors in the causation of accidents are the “failure to take effective collision avoidance actions” and “negligence in lookout”. The results of this study can provide a rapid method for identifying accident causation elements and key factors for maritime departments and related research institutions, thereby improving decision-making efficiency and scientificity.