Predicting Flashover Occurrence using Surrogate Temperature Data

Fire fighter fatalities and injuries in the U.S. remain too high fire fighting hazardous. Until now, fighters rely only on their experience to avoid life-threatening events, such as flashover. In this paper, we describe development of a flashover prediction model which can be used warn before occurs. Specifically, consider use simulation program generate set synthetic data an attention-based bidirectional long short-term memory learn complex relationships between temperature signals conditions. We first validate with measurements obtained from full-scale experiments. Then, account for realis-tic vent opening conditions multi-compartment structure. Results show that our proposed method achieves promising performance even when is completely lost room origin. It believed facilitate transformation tactics traditional experience-based decision marking data-driven reduce deaths injuries.