一种基于火灾图像和人工智能的热释放速率计算方法

The heat release rate(HRR)of a fire is one of the critical parameters in describing the fire behavior and hazard. However，due to the complexity of fire scenarios，traditional methods of measuring HRR cannot be applied. This study aims to explore a HRR prediction method based on artificial intelligence in fire scenarios，which will facilitate the identification of fire intensity and provide support for fire emergency rescue. The deep learning model(Swin Transformer)for HRR calculation is trained on the publicly available fire calorimetric database from the National Institute of Standards and Technology，which contains 89 fire tests and over 50 000 fire images taken at different times. The obtained model can identify the size of the fire in real time based on the fire images taken in the fire scenario. The results show that even in different fire scenarios，the deep learning model can effectively identify HRR based on the fire images，which will provide effective support for the development of future intelligent fire systems.