一种新的信号控制干道行程时间实时估计模型

A New Model for Estimating Real-time Travel Time on a Signalized Arterial

给出了一种新的信号控制干道行程时间实时估计模型.其建模思路是:将分析时段分为若干个较短的时间窗,然后进一步把一个时间窗离散为多个时间间隔.将干道各交叉口停车线前的车辆是否处于排队定义为干道系统的状态.在一个时间窗内,确定每个时间间隔上的干道系统状态,由此构造出一个无记忆特性的随机过程,根据离散马尔可夫决策过程理论,实现了单个时间窗的干道行程时间估计.在每个时间窗上应用该过程,实现了干道行程时间的实时估计.与现有模型相比较,文中模型的优势体现在:模型输入是通用的流量和信号配时数据,模型参数少且容易标定,模型应用方便、成本低和可移植性强.最后,该模型在广州市的某条实际干道上进行了检验.

A new model for estimating the real-time travel time on a signalized arterial is developed in this paper. The basic idea behind the proposed model is that the time horizon analyzed is first divided into multiple time windows, and these time windows are then further divided into shorter time intervals. In the paper, the arterial system states are defined as whether the vehicles at the stopping line of an intersection queue or not. After the determination of the arterial system states at each time interval within each time window, a stochastic process with memoryless property is then established. Consequently, the travel time on a signalized arterial within a time window can be estimated by using the discrete-time Markovian decision process （DTMDP）. This process is applied repeatedly to each time window, and the real-time travel time on the signalized arterial over the time horizon is then obtained. Compared to the previous related models, the model＇s inputs only include the data of traffic flow and signal setting, the proposed modelts parameters are less and easier to calibrate, the cost of model application is low, and the model has a good transplant property as well. Finally, the model proposed in this paper is validated on an actual signalized arterial in Guangzhou.