베이지안 네트워크를 이용한 원양어선 위험요인 분석에 관한 연구

Abstract

This study aims to analyze the risk factors contributing to marine accidents involving Korean distant water fishing vessels using a Bayesian network approach. As marine accidents in this sector often result in severe casualties and significant economic losses, understanding their underlying causes is critical. Based on official investigation records from the Korea Maritime Safety Tribunal (2000~2023), a dataset of 46 accident cases involving longliners, trawlers, and other fishing vessels was constructed. The analysis categorized accidents by vessel type, gross tonnage, vessel age, location, operating status, and specific causes, including poor lookout and inadequate maintenance. Following the Formal Safety Assessment (FSA) framework recommended by the International Maritime Organization (IMO), the study applied Bayesian networks to quantify the probabilistic relationships among risk factors. The results revealed that the most hazardous conditions for different accident types included: vessels with 300~500 GT, aged 20~40 years, operating outside harbor limits during navigation or fishing. Specifically, collision and grounding incidents were primarily associated with poor lookout, while sinking and fire/explosion incidents were linked to inadequate maintenance. The findings underscore the necessity of tailored safety control measures for each accident type and vessel category. This research provides empirical evidence to support decision-making for improving safety policies under the Act on the Punishment of Serious Accidents and the Distant Water Fisheries Development Act. Keywords: Bayesian network, marine accidents, distant water fishing vessels, risk analysis, Formal Safety Assessment (FSA)