Precise quantitative evaluation of the risk level of coal mine water inrush accidents based on the CW-BN model

Water inrush accidents in coal mine pose a serious threat to production safety and personnel's life safety. In view of the difficulties in the risk analysis of coal mine flooding accidents, such as the difficulty in determining the key influencing factors and strong subjectivity of index weights, this study proposed an assessment method based on the combination-weighted Bayesian network model to achieve accurate quantitative assessment of risk levels. Typical water inrush accident cases are analyzed by data-driven method, and various factors affecting mine water inrush are systematically considered by integrating expert suggestions, and then the main factors are identified, the risk index system of coal mine water inrush accidents is constructed, and the Bayesian network topology is generated by mapping accordingly. Experts are invited to conduct a prior probability quantitative evaluation according to the questionnaire. AHP and entropy method are combined to determine the combined weights and calculate the conditional probabilities of the child nodes in the Bayesian network. The key risk factors of coal mine flooding accidents were accurately identified, and the risk level was assessed by using the forward causal reasoning, reverse diagnosis, and sensitivity analysis functions of the Bayesian network. Taking a coal mine in Shaanxi Province as an example, this research method is compared with the fuzzy comprehensive evaluation method, and the results show that both of them are consistent with the actual situation. This method is convenient, intuitive and reliable, and can be used as a more operational method for risk assessment of rich-water mines.