The maritime transport system has contributed a lot to the development of the global economy. In 2019, the volume of goods increased by 0.5% due to the COVID-19 pandemic, down from 2.8% in 2018, but in 2019, more than 11.8 billion tons of cargo were transported by ships. With the development of such maritime transportation, the marine transportation environment becomes complicated along with the enlargement, high speed, and automation of ships, and as a result, the risk of marine accidents is increasing. Marine accidents cause enormous human casualties and environmental and economic losses due to marine pollution.
According to the Ministry of Maritime Affairs and Fisheries' marine accident statistics, the number of marine accidents in the last five years (2016-2020) has been steadily increasing. According to the EMSA's 2020 annual report, ship crashes account for 22.6% of all maritime accidents, and 969 (54%) of a total of 1,801 accidents analyzed during the investigation period were reported as "human factors." In order to prevent collision accidents, it will be necessary to analyze the cause of the collision and take improvement measures according to the cause.
In order to propose an accident analysis model for human negligence, this study analyzed the causal relationship and prepared countermeasures for various factors caused by collision through the application of the Bayesian network to overcome the limitations of the existing ETA and FTA methods. VTA analysis techniques for human factor extraction were introduced, and causal relationships were analyzed through the derivation of accident influencing factors for generalized human negligence.
Through this study, by applying VTA to marine accident data and applying it to ETA based on extracted human factors, an RCO that enables quantitative countermeasures and evaluation under certain conditions in the event of a collision was established. The RIFs extracted from the data were applied to BN and the interaction and clear causal relationship between factors related to the collision accident could be identified.