In tunnel fire safety prevention and control, artificial intelligence (AI) has gradually become a key means of accurate monitoring and intelligent warning of tunnel fires because of its ability to solve problems of traditional monitoring, such as long response delays, long prediction times, and high false alarm rates. Prediction models based on empirical formulas take minutes to calculate, failing to meet the needs of early intervention. Therefore, using AI to accurately predict fire development trends (such as smoke spread and temperature distribution) is crucial for formulating emergency plans and ensuring safety. In particular, fire monitoring and intelligent early warning methods based on AI have become an important research direction in tunnel fire safety research.

Applications of AI to tunnel fires include the development of few-shot and self-supervised learning methods to enhance model generalization ability. They also involve promoting system integration and standardization to realize platform-based collaborative management. In multisource data collection, multi-sensor fusion adopts an improved hierarchical architecture based on D–S evidence theory. It integrates temperature, smoke, and gas data, thereby improving fire identification reliability by 45% in complex environments. Video monitoring relies on CNN (convolutional neural network) and YOLOv8 algorithms, combined with tunnel CCTV (closed-circuit television) systems, to analyze flame and smoke characteristics. It achieves 96% recognition accuracy and reduces the false alarm rate by 30%. Edge computing has achieved up to 96% accuracy and supports real-time alarms. At the platform level, AI-based disaster prevention and response systems (e.g., Shanghai's intelligent system) enable real-time visualization of fire locations and temperatures. They automatically trigger coordinated control of ventilation and sprinkler systems, reducing response delays by more than 50% compared with manual operation. In terms of intelligent early warning, generative AI, such as GANs (generative adversarial networks) and Transformers, can generate fire spread simulations within 5 s. LSTM–TCNN (long short-term memory-temporal convolutional neural network) reduces temperature field prediction from minute-level to second-level (with 90% accuracy), and digital twins construct 1∶1 virtual tunnels to generate synthetic data, thereby reducing the demand for training data by 50%.

AI can effectively improve detection accuracy and response efficiency in tunnel fire monitoring and early warning. However, several challenges remain, including the scarcity of real-world samples (applying highway models to railways reduces accuracy by 15%–20%), the limited ability of traditional algorithms to capture global features, a lack of standardization in system integration, and high deployment costs. Future research will focus on using generative diffusion models to generate high-fidelity data and alleviate the sample scarcity issue, while reinforcement learning will be employed to optimize the collaborative control of equipment. In addition, a three-dimensional visualization platform based on BIM (building information modeling) and digital twins will be developed to enable VR/AR-based simulations. Further improvements in multimodal fusion are expected to enhance data reliability and cross-scenario adaptability, thereby advancing the intelligence of tunnel fire prevention and control. This research will contribute to improve the intelligence level of tunnel fire early warning and emergency response and promote the practical application of AI in tunnel fire engineering.

Effective control of ventilation parameters is critical in metro tunnel safety research. Since cross-passage wind is driven by inter-tunnel pressure differentials, investigating the impact of tunnel air supply on wind speed is therefore vital. Furthermore, train-induced piston wind can damage cross-passage fire doors, compromising operational safety and highlighting the need to assess the feasibility of eliminating fire doors through an optimized air supply design. Thus, this study clarifies the influence of air supply parameters on cross-passage airflow, compares cross-passage wind speeds derived from full-scale cold smoke experiments and thermal simulations of real fire scenarios, evaluates the feasibility of eliminating fire doors, and supports the optimization of metro fire ventilation systems and fire prevention research.

Here, full-scale experiments were combined with numerical simulation to explore the study objectives. Specifically, multi-condition ventilation and smoke tests were conducted in a Zhengzhou metro tunnel, utilizing a portable large-section anemometer to ensure the accuracy of the experimental data. The test setup comprised the tunnel structure, cold smoke device, and corresponding measuring systems. As cold smoke could not replicate the thermal buoyancy of real fires, full-scale thermal smoke simulations were performed using a fire dynamics simulator (FDS) and computational fluid dynamics software developed by NIST, which was validated for fire dynamics studies through multi-scale tests. Building on this, the smoke flow was analyzed under various air supply and exhaust conditions, with cross-passage wind speeds compared between the cold smoke experiments and FDS-simulated real fire scenarios.

First, the simulation data obtained under fire-free conditions were consistent with the results of cold smoke tests, verifying the feasibility of the numerical simulation method. Second, real fires generate significant heat, causing hot smoke to rise owing to thermal buoyancy. This enhances vertical airflow in the tunnel and results in higher cross-passage wind speeds compared with cold smoke tests, though the velocity increase was limited by wall friction. Third, the distance (spacing) between air supply and exhaust ports, along with the status of platform screen doors, alters the inter-tunnel pressure differential; furthermore, the presence of a train can obstruct pressure-driven airflow, slightly reducing cross-passage wind speed. Fourth, cold smoke tests confirmed that a rational ventilation design can achieve the cross-passage wind speeds exceeding 2 m/s. Fifth, the ventilation modes corresponding to Conditions 1-3, 2-3, and 3-3 effectively increased wind speeds in both tunnels and cross-passages across different train positions and fire locations.

Based on the study results, the following conclusions are drawn: first, cold smoke experiments demonstrated that optimized ventilation can maintain a cross-passage wind speed exceeding 2 m/s during emergencies. This indicates the feasibility of eliminating fire doors, which could reduce construction and maintenance costs and enhance cross-passage evacuation efficiency. Second, ventilation Conditions 1-3, 2-3, and 3-3 optimize the emergency ventilation effect of metro tunnels, providing practical references for engineering applications. Third, the validity of the effective model is confirmed by the consistency between the fire-free simulation results and experimental data. In fire scenarios, cross-passage wind speed is influenced by thermal buoyancy, smoke viscosity, and smoke density. Among them, thermal buoyancy increases the speed, whereas wall friction suppresses it.

The continuous expansion and intensification of urban underground space utilization are leading to increasingly complex utility tunnel designs, including structures with significant longitudinal slopes and extended ventilation compartments. In contrast to cable fires in conventional utility tunnels, the interaction between slope-induced airflow and ventilation airflow path creates a unique underground environment that significantly affects fire spread and post-fire smoke exhaust. The development of fires in these complex utility tunnels must be investigated to provide robust, evidence-based guidelines for designing safer energy infrastructure and developing more effective fire protection strategies, thereby enhancing the resilience and safety of urban underground infrastructures. This study aims to quantify the impact of tunnel slope on fire spread behavior and to evaluate how daily operational ventilation strategies affect both sloped and long-ventilation-compartment tunnels.

This study employs the numerical simulation software Fire Dynamics Simulator using an actual utility tunnel project as the engineering basis to investigate the development patterns of cable fires in utility tunnels under varying longitudinal slopes (0%, 1%, 3%, 5%, and 10%), ventilation compartment lengths (200, 400, and 600 m), and daily air exchange rates (2, 4, and 6 h⁻¹).

The findings show that for utility tunnels with longitudinal slopes, the natural airflow generated by the stack effect of a cable fire interferes with combustion, flame spread, and smoke propagation, resulting in accelerated fire spread on the right side of the tunnel. At a 10% slope, the maximum flame spread rate on the right side is 0.183 m/s, a 22% increase over the zero-slope conditions. Increasing the daily air exchange rate has a greater effect on suppressing fire spread in areas with large slopes. At an air exchange rate of 6 h⁻¹, the right-side spread range of a cable fire reduced by 9 m under a 10% slope, while it reduced only by 7 m under a 3% slope. For utility tunnels with long ventilation compartments, extending the compartment length increases the airflow travel distance, reducing airflow pressure along the tunnel and weakening the influence of the airflow on fire development. Increasing the air exchange rate has a limited impact on the development of cable fires in long ventilation compartments of utility tunnels. Compared to no-ventilation conditions, air exchange rates of 2, 4, and 6 h⁻¹ suppress fire spread only by 0.55%, 0.55%, and 1.67%, respectively.

The aforementioned results indicate that increased air exchange rates can effectively constrain fire spread in steeply sloped tunnels by counteracting the stack effect. However, their impact is markedly diminished in long ventilation compartments due to airflow attenuation over distance. The research results provide data and technical support for fire safety design and protection requirements in actual underground utility tunnels. The quantitative data and mechanistic explanations presented in this study can inform the development of enhanced safety standards and operational protocols, ultimately mitigating the risks associated with cable fires in the increasingly complex underground lifelines of modern cities.