A Bayesian network model integrating data and expert insights for fishing ship risk assessment

Abstract

Marine accidents can result in severe economic losses and casualties, underscoring the critical need for effective risk assessment.. In this study, quantitative marine accident reports from Korea that objectively describe accident variables were collected and classified to analyze marine accidents of fishing ships To analyze the causes of accidents involving different types of fishing ships, a survey with subject matter experts (SMEs) was conducted. A fishing ship accident Bayesian network (FABN) scenario was then developed by integrating fishing ship accident data with SME insights. The FABN was comprehensively modeled based on the scenario, with marine accidents being modeled based on causal variables each marine accident. Changes in the output value of the FABN were verified via a sensitivity analysis, and the independence and statistical significance of the model were confirmed using a statistical analysis of the collected data. FABN allows for the immediate assessment of the probability of marine accidents related to fishing ships by utilizing network structures, and provides the advantage of structurally assessing ship accident risks