Real-time prediction of fire propagation in tunnels with discontinuous combustible materials based on bidirectional long short-term memory networks

Key Points

• Fire propagation dynamics are effectively predicted using a bidirectional long short-term memory network.
• The model demonstrates a significant accuracy improvement, achieving over 90% predictive performance during simulations.
• Assessment involved real-time predictive modeling of fire situations in tunnels with discontinuous combustible materials.
• The findings suggest enhanced safety measures can be developed for tunnel designs based on predictive outcomes.