To reduce the loss of tunnel fire accidents, a rapid detection and 3D temperature field prediction model was constructed for tunnel vehicle fires based on deep learning to realize advanced tunnel fire prediction. An improved YOLOv8 network model combined with the OpenCV module was employed to achieve vehicle information detection and position prediction. A novel triangle rotation convolutional neural network (TRCNN) model and Triangle framework were proposed, which were integrated with the high-order polynomial regression (HOPR) model and Informer-long short-term memory (Informer-LSTM) network to form the HOPR-Informer-LSTM hybrid model for fire smoke recognition and precise 3D fire temperature prediction. The research results demonstrate that the vehicle classification and license plate recognition achieve mean average precision at 50% intersection over union (mAP@50) of 98% and 99% respectively, vehicle speed prediction reaches a coefficient of determination (R2) of 94%, and fire smoke recognition accuracy reaches 82%. Validation of the temperature prediction model using small-scale test bench measurements shows that the 3D temperature field prediction error remains within ±5 ℃. The research results provide an advanced tunnel fire prediction method, offering technical support for tunnel fire prevention and control.
LAI Peng-hao, HUANG You-bo, LUO Cheng-jia, DONG Bing-yan.
Three-dimensional Temperature Field Prediction of Tunnel Fire Based on Machine Vision and Temporal Forecasting Forecasting[J].
Science Technology and Engineering, 2026, 26(12): 5130-5143 DOI:10.12404/j.issn.1671-1815.2502857
现有隧道火灾温度场预测研究存在两大问题:传统线性回归与低阶多项式模型难以捕捉非线性温度演化特征,而多参数高阶模型虽能提升拟合精度,但计算复杂度与时间消耗呈指数级增长,难以满足轻量化实时预测需求。为此,提出基于长短期时序记忆网络(Informer long short-term-memory,Informer-LSTM)与高阶多项式回归(high-order polynomial regression,HOPR)的混合模型,以解决火灾三维温度场预测难题。Informer-LSTM基于先验火灾温度数据,建立时间维度与三维温度场分布及动态演化规律的长期依赖关系,有效平滑FDS模拟数据中的固有数值噪声,并抑制小尺度隧道火灾实验中火焰脉动与湍流卷吸效应引发的温度高频振荡问题。相比传统Transformer模型[16],Informer模型[17]采用稀疏自注意力与信息聚合机制,在处理长序列时计算效率更高,能准确捕捉时间序列的全局模式,提升模型精度与泛化性,适用于鲁棒性要求较高的火灾温度预测场景。HOPR模型凭借高效计算性能、紧凑参数架构与自适应维度控制能力,在低复杂度短期回归任务中具备较高拟合精度,且轻量化架构适用于资源受限的隧道火灾场景。融合两种算法实现隧道车辆火灾初期与蔓延阶段的时序预测,显著降低预测误差并提升评估稳定性。
AnH, JiangY, JiangJ. A survey of fire accidents during the process of highway tunnel operation in China from 2010 to 2021: characteristics and countermeasures[J]. Tunnelling and Underground Space Technology, 2023, 139: 105237.
[2]
LiuF, ZhaoZ, YaoH, et al. Application of aspirating smoke detectors at the fire earliest stage[J]. Procedia Engineering, 2013, 52: 671-675.
[3]
ChengG, ChenX, WangC, et al. Visual fire detection using deep learning: a survey[J]. Neurocomputing, 2024, 596: 127975.
DengShiqiang, DingHao, JiangShuping, et al. Intelligent detection algorithm for road tunnel fire smoke based on improved YOLOv5s[J]. Journal of Highway and Transportation Research and Development, 2024, 37(11): 194-209.
JiangLi, HeTingquan, GuoXin, et al. Inversion of fire source location and heat release rate intunnel fires based on deep learning[J]. Fire Science and Technology, 2024, 43(9): 1292-1297, 1315.
WangNian Nian, TianYuan, YuLi, et al. Research on predicting vehicle ignitiontime based on tunnel fire spread[J]. Chinese Journal of Safety Science, 2018, 28(12): 46-51.
[12]
ZhangX, JiangY, WuX, et al. AIoT-enabled digital twin system for smart tunnel fire safety management[J]. Developments in the Built Environment, 2024, 18: 100381.
ZhaiLihua, NongXingzhong, HeGuanhong, et al. Numerical si-mulation for the smoke spreading features in subway tunnels under the condition of blocking effect[J]. Journal of Safety and Environment, 2020, 20(3): 930-937.
BaiZhenpeng, LiYanfeng, ZhangPeng, et al. Multivariate regression prediction analysis of temperature distribution in tunnel fires[J]. Fire Science and Technology, 2018, 37 (10): 1304-1306, 1323.
HongYao, FuCeji. Rapid simulation and prediction of tunnel fires based on HPC and machine learning[C]// Proceedings of the 27th Annual Conference of Beijing Mechanics Association. Beijing: Beijing Society of Theoretical and Applied Mechanics, 2021: 119-121.
[19]
HochreiterS, SchmidhuberJ. Long short-term memory[J]. Neural Computation, 1997, 9(8): 1735-1780.
[20]
林跃江. 基于CFAST的火灾烟气流动与温度分布预测[D]. 重庆: 重庆大学, 2020.
[21]
LinYuejiang. Prediction of fire smoke flow and temperature distribution based on CFAST[D]. Chongqing: Chongqing University, 2020.
[22]
LuX, HeM, WangZ, et al. Prediction of flashover time in a compartment fire by CNN-LSTM based deep neural network considering wearable data collection equipment[J]. Journal of Building Engineering, 2024, 97: 110719.
JiaJinzhang, ChenJiaqi, ChenYinuo. Research on the application of long short term memory networks in real time disaster situation prediction of tunnel fires[J]. Journal of Safety and Environment, 2025, 25(4): 1298-1309.
[25]
VaswaniA, ShazeerN, ParmarN, et al. Attention is all you need[J]. Advances in Neural Information Processing Systems, 2017, 30: 3762.
[26]
ZhouH, ZhangS, PengJ, et al. Informer: Beyond efficient transformer for long sequence time-series forecasting[C]// Proceedings of the AIAA Conference on Artificial Intelligence. Reston: AIAA, 2021: 11106-11115.
[27]
DongZ, WuY, PeiM, et al. Vehicle type classification using a semisupervised convolutional neural network[J]. IEEE Transactions on Intelligent Transportation Systems, 2015, 16(4): 2247-2256.
[28]
ZhangX, ChenX, DingY, et al. Smart real-time evaluation of tunnel fire risk and evacuation safety via computer vision[J]. Safety Science, 2024, 177: 106563.
[29]
VargheseR, SambathM. YOLOv8: a novel object detection algorithm with enhanced performance and robustness[C]//2024 International Conference on Advances in Data Engineering and Intelligent Computing Systems (ADICS). New York: IEEE, 2024: 1-6.
[30]
XuZ, YangW, MengA, et al. Towards end-to-end license plate detection and recognition: a large dataset and baseline[C]// Proceedings of the European Conference on Computer Vision (ECCV). Munich: European Computer Vision Association, 2018: 255-271.
[31]
GongY, DengL, TaoS, et al. Unified Chinese license plate detection and recognition with high efficiency[J]. Journal of Visual Communication and Image Representation, 2022, 86: 103541.
[32]
ZhangY, SunP, JiangY, et al. Bytetrack: multi-object tracking by associating every detection box[C]// European Conference on Computer Vision. Cham: Springer, 2022: 1-21.
[33]
de Venancio P V A B, LisboaA C, BarbosaA V. An automatic fire detection system based on deep convolutional neural networks for low-power, resource-constrained devices[J]. Neural Computing and Applications, 2022, 34(18): 15349-15368.
China Highway and Transportation Society (CHTS). Highway video monitoring technology requirements: T/CHTS 10001[S]. Beijing: China Highway and Transportation Society (CHTS), 2021.
General Administration of Quality Supervision, Inspection and Quarantine of the People's Republic of China, China National Standardization Administration. Communication rules of the devices of surveillance and controlsystem for expressway—part 2: vehicle detector: GB/T 34428.2—2017[S]. Beijing: National Technical Committee for Standardization of Transportation Engineering Facilities (Highways), 2017.
[38]
WangQ, WuB, ZhuP, et al. ECA-Net: efficient channel attention for deep convolutional neural networks[C]// Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. New York: IEEE, 2020: 11534-11542.
[39]
WooS, ParkJ, LeeJ Y, et al. Cbam: convolutional block attention module[C]// Proceedings of the European Conference on Computer Vision (ECCV). Berlin: Springer, 2018: 3-19.