Abstract: In order to clarify the influence of facade wing walls on spilled flame, an experimental study on the spilled flame under different facade wing wall conditions was conducted and the real- time prediction of the fire temperature field of the facade was predicted. The facade fire prediction database was constructed through facade fire experiments under different wingwall spacing and depth conditions, and applied to the training and validation of convolutional neural network (CNN), so as to establish a real-time prediction model for the flame temperature of spilled flame under different facade wing- wall conditions, and provide support for the design of facade construction forms and facade fire protection.

Key words: spilled flame, facade, temperature field, CNN