International Conference on Fire Safety Engineering Research and Practice

Big data and artificial intelligence (AI) based techniques have made smarter operation and management of urban infrastructure possible. The applications of AI in fire detection, risk assessment, and data-driven fire and emergency management are emerging. This chapter provides a framework to use the cutting-edge AI tool to enable real-time forecasting of the evolution of major urban infrastructure fire accidents. Choosing tunnel fires as an example, the guidelines for constructing a reliable fire database with both experimental and numerical data is introduced. AI algorithms with promising potential to forecast fire scenarios are discussed, including time-sequence prediction, multi-modal data fusion and fire pattern generation. Finally, the proposed AI model is demonstrated in three typical fire scenarios, including a reduced-scale chamber fire, a full-scale room fire, and a tunnel fire. Results show that AI can predict the fire development and the occurrence of critical events (e.g., flashover and backdraft) under various environments in real-time or even with a certain lead time. AI can act as a powerful tool to provide useful fire incident evolution information, thereby promoting urban resilience against the fire hazard.