Fuzzy dynamic Bayesian network based on a discrete aggregation method for risk assessment of marine nuclear power platform hinge joints accidents

Abstract

Hinge joints play a critical role in the positioning system of Marine Nuclear Power Platforms (MNPP) by facilitating the transmission of mooring forces and accommodating rotational motions of the hull. Failure of hinge joints during operation can result in severe consequences for both the environment and human safety. Therefore, this study aims to address the limitations associated with the dynamic characteristics and uncertain information pertaining to the risk systems of hinge joint structures in MNPP. To tackle the challenge of conflicting expert opinions in group decision-making, a novel method called the discrete aggregation method (DAM) is proposed. DAM utilizes fuzzy opinion aggregation, taking into account the degree of consensus among experts. This approach effectively overcomes the problem of conflicting opinions. Furthermore, building a hinge joint accidents model using Bow-tie (BT) for hazard identification, and then mapping the BT model to a dynamic Bayesian networks (DBN) model. Fuzzy set theory (FST) and DBN are employed to handle cognitive uncertainty caused by data limitations and incomplete knowledge. These techniques enable causal inference for time-varying dynamic risk systems. To validate the proposed method, a case study of the positioning system of the first MNPP in China is conducted. The dynamic fatigue failure probability of hinge joints is calculated using predictive technology based on Bayesian theory. Additionally, diagnostic analysis techniques are employed to accurately identify key events and their underlying causes. The analysis results provide valuable data support for on-site operations and management personnel, aiding in the development of effective risk mitigation measures.