Abstract: This essay utilizes Changsha city road transportation data and compares foreign classical queue length models. With the integration of local conditions, SIGNAL94 model is improved to construct a new queue length model consistent with Changsha transportation characters. When impose time smoothing to reduce the randomicity of transportation, 15 minutes is proved to be the appropriate smoothing parameters to lower the tracking error of the bodel.

Key words: queue length, signalized intersection, saturated, dynamic model

摘要： 使用长沙市道路交通状况的真实数据，对比分析了国外多个经典排队长度模型的计算结果，并在结合当地交通特性的基础上，对SIGNAL94进行改进，建立了符合本地实际交通状况的排队长度优化模型.同时，在利用时间平滑技术减少交通流随机性对计算精度影响的过程中，发现以15分钟作为平滑参数，能够进一步降低优化模型的跟踪误差.

关键词: 排队长度, 信号口, 饱和, 动态模型