An extended Bayesian network model for calculating dam failure probability based on fuzzy sets and dynamic evidential reasoning

Effective risk analysis is crucial for ensuring the safe operation of hydropower dams. Existing methods typically rely solely on subjective or objective data. The subjectivity of expert opinions and insufficient objective data affect the reliability of results. To reduce subjectivity in the risk analysis process and enhance the utilization of subjective and objective data, this study proposes an extended Bayesian network (BN) model for calculating the dam failure probability based on fuzzy sets (FSs) and dynamic evidential reasoning. To address the fuzziness and uncertainty of expert opinions, a Gaussian FS was introduced to transform expert opinions into fuzzy numbers. Additionally, a dynamic evidential reasoning approach was proposed to combine subjective and objective data, considering their differences, subjectivity, and potential conflicts. To maintain consistency in the fuzzification and defuzzification processes, an improved Onisawa formula was proposed. Finally, the BN was employed for probability prediction and sensitivity analyses. The model was applied to the Z hydropower dam in China. The results showed that the failure probability was 5.2 × 10−4, and the main controllable risk factors were identified. A comparative analysis was conducted and the results showed that the proposed method had a better ability to handle uncertainty, reduce subjectivity, and mitigate conflicts.