Analytical Study on the Prediction of Fire Evacuation Time in Large Complex Buildings Using the Ensemble Learning Technique

Abstract

With the recent increase in the risk of fire in buildings, the number of casualties that occur in the event of a fire have increased. This emphasizes the importance of performance-based design. However, simulating a performance-based design requires a lot of manpower and time, and re-simulation with minor changes is a difficult task. Therefore, in this study, we attempt to develop a prediction model that can easily predict the ASET for each fire distance as a fire factor and spatial factor by applying ensemble learning. The prediction model developed using machine learning based on FDS data showed a high coefficient of determination of 0.91, and we believe that ASET for each distance can be derived in real time by applying this prediction model.