为解决现有高速公路行程时间估计并未考虑交通流实时变化特性的问题，以宏观交通流中 流量—密度—速度之间的关系为基础，以速度与密度的线性模型和指数模型为例，假设车辆到达 服从泊松分布，构建了高速公路路段M/G(n)/C/C状态相关性排队模型，该排队模型服务时间与系 统内的车辆数有关，充分反映了交通流的随机性。在Vissim 中构建仿真路网，通过对比仿真实验 结果与模型计算结果可知，两者基本一致，误差在5%以内，且指数模型较线性模型适用范围更 广，更符合实际情况。

The existing estimation of expressway travel time can′t reflect the real time variation charac⁃ teristics of traffic flow adequately. To solve this problem, the relationship among flow, density and speed in the macroscopic traffic flow was considered. The linear model and the exponential model between speed and density were taken as examples, and the vehicle arrival was assumed to follow Poisson distri⁃ bution. The M/G(n)/C/C state dependent queuing model of expressway was developed. The service time of this model was related to the number of vehicles in the system, and the randomness of traffic flow was reflected adequately. At the same time, simulation road network was built in Vissim. Then the simulation experiment was compared with the model calculation and the two results were almost similar. The error between the two results is less than 5%. The exponential model applies wider than the linear mode, which is more in accordance with the actual conditions.