Fire location detection method based on deep learning

The enclosed environment of tunnels poses significant challenges to the effective implementation of firefighting strategies in the event of a fire. Due to the complex internal structure and narrow space of tunnels, traditional emergency response methods are often inefficient under such conditions. Therefore, it is particularly crucial to quickly and accurately determine the location of a fire, develop the optimal firefighting plan, and continuously monitor the effectiveness of firefighting. Currently, many methods rely on simple deep learning models to locate fire sources, but in complex fire scenarios, these models often exhibit insufficient lightweight and flexibility. To address this issue, we propose a deep learning based fire localization model that can achieve automated fire detection in complex and ever-changing tunnel fire environments. Compared with traditional deep learning methods, this model not only improves accuracy, but also significantly accelerates response speed.To verify the effectiveness of the model, we compared seven existing fire detection models, including backpropagation neural network (BPNN), convolutional neural network (CNN), long short-term memory network (LSTM), CNN-LSTM, bidirectional long short-term memory network (BiLSTM), CNN BiLSTM, and CNN BiGRU. The final results indicate that our proposed method based on CNN BiGRU performs well in fire location detection tasks, with excellent detection performance, significant robustness, and lightweight advantages.