Navigation risk assessment of intelligent ships based on DS-Fuzzy weighted distance Bayesian network

To address the challenges on navigation risk assessment of intelligent ships, including lack of historical accident data, strong uncertainties, complicated influencing factors, and difficulties in clarifying logical relationships among these factors, a novel navigation risk assessment scheme based on DS-Fuzzy weighted distance Bayesian network is innovatively proposed. Specifically, by combining the extracted risk influence factors with the DS evidence theory-based expert knowledge fusion method, a Bayesian network topology is first constructed. Then, by incorporating the fuzzy weighted distance into the calculation of the influence weight of the parent node, an automatic conditional probability assignment algorithm is further proposed to effectively reduce the computational complexity of the intermediate node parameters. By applying the fuzzy comprehensive evaluation strategy to the constructed evaluation model, the cascade risk quantitative analysis mechanism is further designed to facilitate the quantification and visualization of risk evaluation. Finally, based on practical navigation cases, experiments are conducted to verify the effectiveness and superiority of the proposed method.