An enhanced FMEA based risk analysis for ship fire safety incorporating the dynamic consensus reaching process with bounded confidence opinion dynamics

In high-risk scenarios like ship fires, achieving expert consensus in failure mode and effect analysis (FMEA) is vital. However, existing FMEA models often lack mechanisms to effectively reconcile divergent expert opinions while accounting for the dynamic influence of social trust networks, limiting their practical effectiveness. To overcome these limitations, this study introduces an enhanced FMEA that incorporates a dynamic consensus reaching process (CRP) based on a bounded confidence DeGroot hybrid model and a social trust network (STN). First, expert evaluations are converted into cloud models, and the directed weighted Louvain algorithm reduces data complexity and identifies subgroups through the STN. Next, expert and subgroup weights are determined, and a double-layer dynamic consensus convergence detection mechanism identifies non-consensus experts. Their opinions are then adjusted using the DeGroot hybrid model, thereby constructing a dynamic CRP. Subsequently, comprehensive risk factor weights are determined. Finally, risk priorities are determined using the TOPSIS method. Applied to a ship fire risk assessment case, the model demonstrates feasibility; sensitivity and comparative analyses further show that it achieves consensus more quickly and with higher efficiency. This work bridges STN theory with marine safety engineering, offering a robust tool for risk management in scenarios demanding coordinated expert judgment.