为了更加精准地描述动态交通流演变过程,得到状态的最优无偏估计,本文基于多项式趋势模型和卡尔曼滤波理论,提出一种实时动态OD矩阵估计的多项式趋势滤波算法.该算法首先将状态变量定义为实际OD流量相对于其历史值的偏差,并将该偏差表述为一个具有滑动趋势的随机演变过程,然后通过建立一个多项式趋势滤波模型实现对动态OD矩阵的估计与预测.最后以一条接近实际路况的高速路网为研究对象进行仿真.大量仿真试验结果表明,本文提出的算法性能优于传统方法,能获得更准确的OD矩阵估计与预测信息.%To more accurately describe the evolution process of dynamic traffic flow, and get an unbiased estimation of state vectors, a novel polynomial trend filtering method is proposed for real dynamic OD matrix estimation on the basis of the polynomial trend model and the Kalman filtering theory in the paper. Firstly, the state variables are defined as the deviations of the actual OD flow from the historical values. These deviations are presented as a stochastic evolution process with a sliding trend. Furthermore, the dynamic OD matrix is estimated and predicated by establishing a polynomial trend filtering model. Finally, a simulation freeway is used as the research object, and a large number of simulation results prove that the performance of the algorithm proposed in this paper is better than the traditional method, and this algorithm can acquire more accurate estimation and prediction information for OD matrix.
展开▼
