When inside a burning building, firefighters sometimes struggle to notice the signs of impending flashover, when most combustible items in a room ignite suddenly. Flashover is a leading cause of firefighter deaths. Research shows that artificial intelligence (AI) could provide the necessary warning.
Researchers at the National Institute of Standards and Technology (NIST), the Hong Kong Polytechnic University, and others developed a Flashover Prediction Neural Network (FlashNet) model to forecast lethal events. Published in Engineering Applications of Artificial Intelligence, FlashNet has an accuracy of up to 92.1% across more than a dozen typical residential floorplans. The best results compared to other AI-based flashover predicting programs.
Researchers beefed up their approach with graph neural networks (GNN) to address the variability of real fires. They digitally simulated more than 41,000 fires in 17 kinds of buildings. Factors such as the fire’s origin, furniture types, and whether doors and windows were open or closed varied throughout the simulation. This provided the GNN model with nearly 25,000 fire cases to use as study material and 16,000 for fine-tuning and final testing. Accuracy depended on the amount of data it had and the lead time it sought to provide firefighters. The tool produced the least false negatives, dangerous cases where the models failed to predict an imminent flashover.