Abstract

Prior studies have noted the importance of ceiling temperature distribution prediction of tunnel fires. To overcome the limitations of precise modeling difficulties for model-driven methods and biased toward the convergence for the data-driven methods due to loss of physical interpretations, a hybrid model-driven and data-driven fusion algorithm is established to predict ceiling temperature distribution in tunnel fires based on dimensional analysis method and ant colony optimization algorithm. Thus, full-scale tunnel fire experiments with different ventilation speeds are conducted to support the algorithm. In contrast to the experimental data as well as the known theoretical model, data-driven back propagation neural network algorithm and hybrid ant colony optimization and back propagation neural network algorithm, the effectiveness and ability of the algorithm are verified, which can be used for the tunnel fire prediction.

Full Text
Published version (Free)

Talk to us

Join us for a 30 min session where you can share your feedback and ask us any queries you have

Schedule a call