火灾是城市轨道交通车站内影响最为严重的事故之一，科学合理的安全疏散方案是突发火灾时确保乘客出行安全的重要保障。然而，目前地铁车站火灾疏散方案中的疏散路线难以根据火场情况动态调整。针对地铁车站疏散路径固定单一的弊端，基于地铁车站内的监控系统，利用计算机视觉技术识别人员分布信息和火灾发生位置，建立空间拓扑模型，利用改进的蚁群算法规划出耗时最短且转弯次数较少的疏散路线，实现站内乘客更科学高效的疏散，最后通过 3 个场景的案例应用验证本文所提疏散方法的有效性。

Fire is one of the most serious accidents occurring at urban rail transit stations. A scientific and reasonable safetyevacuation plan is essential to ensure the safety of passengers in the event of a fire. However, it is difficult to dynamicallyadjust the evacuation routes in the current subway station fire evacuation plan according to the fire situation. This study usedcomputer vision technology to identify personnel distribution information and fire locations based on the monitoring system inthe subway station. A spatial topology model was developed, and an improved ant colony algorithm was used to plan an evacuationroute that takes the shortest time and has fewer turns to provide a more scientific and reasonable evacuation route for evacuatingpassengers. The effectiveness of the evacuation plan was verified by applying it to three scenarios.

U231.96

国家自然科学基金青年项目(52202385)