베이지안 네트워크를 이용한 어선 해양사고 분석에 관한 연구

Abstract

Fishing vessels have different fishing methods and operate in different areas depending on the type of fisheries. In addition, as the causes of accidents and work environment factors are complex, a research on various analytical methods is necessary. In this study, data of marine safety judgment decisions and the transmission of marine accident situations were collected and identified to classified marine accident variables. To investigate the causes of accidents, Subject Matter Experts(SMEs) opinions were used to investigate each type of accident. Based on the results of the SMEs’ opinion survey, scenario modeling was conducted for analyzing the seven types of accident causes. In addition, a fishing vessel marine accident scenario was modeled by combining classified data of situational transmission and SMEs’ opinions. Posterior probability changes were inferred by adding a probability case from the results of Bayesian Network scenario modeling. Reducing the probability of occurrence of a node in a fire explosion scenario confirmed that the probability of an accident occurring decreased. Sensitivity analysis was performed to verify the modeled Bayesian Network scenario.