Underground commercial building fire incidents, Risk of fire casualty, Fuzzy fault tree analysis, Bayesian network, Expert evaluation, Risk in industry. Risk management, HD61

With the development of urbanization, underground commercial buildings (UCB) are facing severe challenges in fire safety management due to their unique structure and environmental characteristics. This study constructed a fire casualty risk assessment model that combines fuzzy fault tree analysis (FFTA) and Bayesian network (BN), aiming to quantitatively analyze the dynamic risk of casualties caused by fires in UCB. Fault tree analysis (FTA) is used to comprehensively identify the key risk factors leading to fire casualties in UCB, involving 55 basic events, and the occurrence probability of basic events was calculated via a fuzzy set. The FTA model was transformed into a BN structure via conversion rules and was optimized. The optimized BN model can dynamically analyze the specific fire evolution process and quantify the impacts of different emergency response measures on fire control, evacuation, and casualties. Innovatively, from the post-incident (a historical case study) and pre-incident (two potentially different fire scenarios) perspectives, various emergency plans were scientifically evaluated, providing reasonable suggestions and decision support for emergency management. The results indicate that the model can effectively guide the formulation of fire prevention and control strategies and emergency response work of UCB and provide an innovative tool for improving the safety of UCB and reducing fire accidents and casualties.