베이지안 네트워크를 이용한 아차사고 평가 모델 개발 및 주요 원인 도출

• Published in: 한국안전학회지 Vol. 38; no. 4; pp. 54 - 59
• Main Authors: 하선영, Seon Yeong Ha, 이미정, Mi Jeong Lee, 백종배, Jong-bae Baek
• Format: Journal Article
• Language: Korean
• Published: 한국안전학회 31.08.2023
• Subjects: assessment; bayesian network; near miss; sensitivity analysis; 안전공학; assessment; bayesian network; sensitivity analysis; near miss
• ISSN: 1738-3803; 2383-9953

The relationship between near misses and major accidents can be confirmed using the ratios proposed by Heinrich and Bird. Systematic reviews of previous national and international studies did not reveal the assessment process used in near-miss management systems. In this study, a model was developed for assessing near misses and major factors were derived through case application. By reviewing national and international literature, 14 factors were selected for each dimension of the P2T (people, procedure, technology) model. To identify the causal relationship between accidents and these factors, a near-miss assessment model was developed using a Bayesian network. In addition, a sensitivity analysis was conducted to derive the major factors. To verify the validity of the model, near-miss data obtained from the ethylene production process were applied. As a result, “PE2 (education),” “PR1 (procedure),” and “TE1 (equipment and facility not installed)” were derived as the major factors causing near misses in this process. If actual workplace data are applied to the near-miss assessment model developed in this study, results that are unique to the workplace can be confirmed. In addition, scientific safety management is possible only when priority is given through sensitivity analysis.