为更好地模拟城市轨道交通站台区域乘客在突发情况下的应急疏散过程，为城市轨道交通运营管理提供高效的疏散方案，建立一种基于元胞自动机（Cellular Automata, CA）的站台疏散仿真模型。通过研究乘客行为特征选择采用冯诺依曼型CA模型作为站台的疏散模型；结合对疏散空间特点的分析确定基于动态规划的疏散路径规划方法，并给出最优路径的计算方法；根据站台乘客的初始分布提出基于欧式距离和综合考虑路径长度与排队时间的两种出口选择方案。最后针对仿真案例，运用上述模型和出口选择方案进行了仿真分析。结果表明：模型通过对每个乘客元胞进行疏散路径规划，实现了对乘客排队行为的模拟，符合运营管理介入情况下乘客疏散的实际情形；基于欧氏距离的出口选择方案可用于站台乘客初始分布大致均匀情况下的疏散仿真；综合考虑路径长度与排队时间的出口选择方案由于排队权重参数的引入，可较好地模拟站台乘客初始分布不均时乘客选择路径远但排队人数少的出口这一行为。模型的演化规则可成为运营管理人员组织疏散的依据。

In order to better simulate the emergency evacuation process of passengers in the platform areas of urban rail transit and provide efficient evacuation scheme for urban rail transit operation and management, a platform evacuation simulation model based on CA (Cellular Automata) was established. By studying the behavior characteristics of passengers, von Neumann CA model was selected as the simulation model of platform. Combined with the analysis of the characteristics of the evacuation space, the evacuation path planning method based on dynamic programming was determined, and the calculation method of the optimal path was given. According to the initial distribution of passengers, two exit-selection schemes based on Euclidean distance and synthetically considering the path length and queuing time were proposed. Finally, a simulation case was analyzed by using the above model and exit-selection schemes. The results show that through the evacuation path planning for each passenger cell, the model realizes the simulation of passenger queuing behavior, which is in line with the actual situation of passenger evacuation under the intervention of operation management; the exit-selection scheme based on Euclidean distance can be used for evacuation simulation when the initial distribution of passengers on the platform is roughly uniform; due to the introduction of queuing weight parameter, the exit-selection scheme considering the path length and queuing time can better simulate the behavior of passengers choosing the exit with long route but less queuing number when the initial distribution of passengers is uneven. The evolution rules can be the basis for operation management personnel to organize evacuation.