Dynamic Risk Assessment of Roof Fall and Rib Spalling Accidents in Non-Coal Mines: Based on FFTA-DBN and the 24Model

Non-coal mines constitute a fundamental pillar of the global energy transition, providing essential raw materials across various sectors. Accidents in these facilities cause severe direct damage, including personal injuries and economic losses, while triggering broader systemic impacts, such as supply chain disruptions of essential energy materials. Taking China as an example, roof fall and rib spalling accidents account for 30% to 40% of all safety incidents in non-coal mines. Consequently, investigating the occurrence and evolutionary process of roof fall and rib spalling risks in these environments holds significant importance. To identify the fundamental factors of the accident, the 24Model and the fault tree are employed in this research. Expert elicitation and fuzzy set theory are used to determine the occurrence probability of each basic event. To overcome the limitations of traditional static risk assessment, a dynamic Bayesian network is introduced to capture the time-varying characteristics of risk factors, evaluating the overall accident probability over a 52-week period, equivalent to approximately one year. In this study, 30 basic events and 18 intermediate events were identified. Quantitative results show that effectively controlling the critical events (X26, X4, X20, X27, X2) reduces the accident probability by 56.39%. Furthermore, the dynamic Bayesian network analysis demonstrates that under the specific assumption of continuous risk accumulation across five dimensions (human, material, management, cultural, and environmental factors) without the prompt implementation of targeted interventions, the occurrence probability of the accident reaches 0.95023 after 52 weeks. The results demonstrate that this model surpasses static models by effectively identifying the critical causal factors of the accident and evaluating their occurrence probabilities through systematic causation analysis and dynamic accident evolution. This approach facilitates precise accident early warning, offering a practical reference for relevant enterprises and personnel involved in roof safety management.