Integrated Fault Tree Analysis and Bayesian Network for Fire Risk Assessment in University Laboratories

The rapid expansion of higher education has introduced new safety challenges in university laboratories, where fire incidents now represent a critical threat to campus safety. In this paper, we propose a method for analyzing the causes of fire accidents by combining Fault Tree Analysis (FTA) and Bayesian Network (BN), which breaks through the limitations of the static analysis of traditional risk assessment. Based on accident cases at home and abroad, the direct and indirect causes of fires in university laboratories were analyzed by using the accident tree. The fire accident tree of the university laboratory, containing 21 basic events, was constructed and mapped into a Bayesian model with dynamic reasoning ability according to the transformation principle. The model was then optimized according to the actual situation. GENIE5.0 software was used for quantitative analysis, forward reasoning of the occurrence probability of fire accidents, and reverse inference of the posterior probability of the root node. The key factors of fires were determined through probability importance analysis and sensitivity analysis. The analysis results show that the probability of laboratory fire is 2.39%, and the three key risk factors are unreasonable storage of combustibles, insufficient or ineffective laboratory fire extinguishing equipment, and chemical reaction. Prevention and control measures such as “five fixed management” and virtual reality emergency drills are proposed to provide a basis for fire analysis and prevention in university laboratories.