Physical model and multiple moth-flame optimization fusion temperature field prediction in large-space building fires

Abstract

To achieve accurate and fast temperature field prediction in large-space building fires, a fusion method is developed, which combines a developed physical model based on heat transfer as well as physical characteristics and an improved multiple moth-flame optimization. The artificial intelligence-based method has advantages like real-time prediction, physical explanations, and no prior data training. According to these advantages, the method can meet the real firefighting application requirements. Supported by two numerical cases of temperature predictions in a large underground parking fire and a large logistics warehouse fire, the results support that the developed method is effective and superior to the traditional moth-flame optimization algorithm and its variant. The developed fusion method of the physical model and multiple Moth-flame optimization can support an effective and useful tool to achieve quick temperature field prediction in large-space building fires for better fire rescue.