The Development of a Bayesian Network Framework with Model Validation for Maritime Accident Risk Factor Assessment

An integrative approach to maritime accident risk factor assessment in accordance with formal safety is proposed, which exploits the multifaceted capabilities of Bayesian networks (BNs) by consolidation modelling, verification, and validation. The methodology for probabilistic modelling BNs well known its application based on model verified though sensitivity analysis only, while validation often omitted due a lack established evaluation measures applicable scarce real-world data. For this reason, work, modified Lyapunov divergence measure proposed as novel quantitative assessor that can be efficiently exploited an individual scenario contributing causal identification, thus serve developed expert elicited BN. framework are showcased grounding small passenger ships Adriatic, complete disclosed, performed, utilization identification ranking presented. data from two cases demonstrate explanatory approach.