Comparison of fuzzy bayesian network methods for maritime applications

Abstract

Fuzzy bayesian networks are commonly used to incorporate expert opinions in risk assessment approaches. Various methods exist to aggregate expert statements with fuzzy methods. Yet few efforts have been made to compare these methods and how they affect risk probabilities in fuzzy bayesian networks. This paper aims at filling this gap. We construct a bayesian network for the entry of a ship in an offshore wind farm and analysed it using pythagorean fuzzy weighted geometric, the linear opinion pool, and the similarity aggregation method. We compare results for every node and the network overall. For the calculation of the nodes the normalisation and calculation of failure probability are added or omitted to all three methods and for the similarity aggregation method the beta value is varied. Our paper shows that the choice of fuzzy methods can have a significant impact on the risk probabilities of bayesian networks depending on the combination of answers and weightings. Adding fuzzy failure probabilities has the highest impact on probability calculations in bayesian networks.