| Abstract |
Smart cities are confronted with a variety of disaster threats. Among them, natural fires pose a serious threat to human lives, the environment, and asset security. In view of the fact that existing research mostly focuses on the analysis of accident precursors, this paper proposes a dynamic risk-modeling method based on Stochastic Petri Nets (SPN) and Bayesian theory to deeply explore the evolution mechanism of urban natural fires. The SPN model is constructed through natural language processing techniques, which discretize the accident evolution process. Then, the Bayesian theory is introduced to dynamically update the model parameters, enabling the accurate assessment of key event nodes. The research results show that this method can effectively identify high-risk nodes in the evolution of fires. Their dynamic probabilities increase significantly over time, and key transition nodes have a remarkable impact on the emergency response efficiency. This method can increase the fire prevention and control efficiency by approximately 30% and reduce potential losses by more than 20%. The dynamic update mechanism significantly improves the accuracy of risk prediction by integrating real-time observation data and provides quantitative support for emergency decision making. It is recommended that urban management departments focus on strengthening the maintenance of facilities in high-risk areas (such as fire alarm systems and emergency passages), optimize cross-departmental cooperation processes, and build an intelligent monitoring and early-warning system to shorten the emergency response time. This study provides a new theoretical tool for urban fire risk management. In the future, it can be extended to other types of disasters to enhance the universality of the model. © 2025 by the authors. |
