基于卡尔曼滤波的交叉口动态O-D反推模型与算法

交叉口各进出口道之间的实时转向交通量是信号控制系统重要的输入数据,也是难以获得的数据.针对已有模型收敛速度较慢、无法满足实际系统应用需要的问题,提出了基于卡尔曼滤波的状态空间模型,设计了顺序卡尔曼滤波进行求解,并采用裁切和标准化对反推结果进行了修正.实例研究表明,模型和算法具有较高的效率和准确性,能够为实时自适应信号控制系统的开发提供支持.

Model and Algorithm of Dynamic Origin-Destination Flows Estimation for Intersections Based on Kalman Filtering

Real-time traffic counts between entries and exits of intersections are important input data for traffic signal control system, and also difficult to get. Existing models are mainly represented using optimization methods, which are rather slow to converge, and can not be used in practical systems. This paper put forward a state-space model based on Kalman filtering, designed a sequential Kalman filtering algorithm to solve it, and revised the estimated results using truncation and normalization. Case study shows that the model and algorithm are efficient and accurate, and can support the design of adaptive traffic control systems.