基于季节模型及Kalman滤波的道路行程时间

发布时间：2018-09-14 08:07

【摘要】：道路行程时间是影响城市交通出行行为的重要因素。当前大多数出行时间研究基于路段进行,假设驾驶人沿着理想最短路径或最快路径行驶,难以对交叉口排队延误等相关时间参数进行精确估计。针对城市任意OD间的出行时间进行分析,采用Kalman滤波方法,利用历史数据对总行程时间进行有效预测。鉴于总行程时间分布存在比较明显的周期性特点,单一Kalman滤波算法难以反映出这种周期性,引入基于季节模型的Kalman滤波算法进行建模和优化。最后,利用深圳浮动车2011年12月连续3d的数据进行实证。研究结果表明:相对于传统的SARIMA模型及普通Kalman滤波算法,优化模型同时考虑总行程时间分布的周期性和时变性,具有较小误差及更好的拟合度;所得预测时间的平均绝对误差(MAE)分别在传统SARIMA模型及普通Kalman滤波算法结果基础上降低了37%和52%,其余误差指标,如均方根误差(RMSE)及最大相对误差(MRE)均有较大下降,从而证明了研究模型的有效性。
[Abstract]:Road travel time is an important factor affecting urban traffic travel behavior. At present, most of the travel time research is based on the road section, assuming that the driver goes along the ideal shortest path or the fastest path, it is difficult to accurately estimate the relevant time parameters such as queue delay at the intersection. Based on the analysis of travel time between any OD in city, the Kalman filtering method is used to predict the total travel time effectively by using historical data. In view of the obvious periodicity of the total travel time distribution, it is difficult to reflect the periodicity of the single Kalman filtering algorithm, so the seasonal model-based Kalman filtering algorithm is introduced to model and optimize. Finally, using the Shenzhen floating car in December 2011 consecutive 3 d data for empirical analysis. The results show that compared with the traditional SARIMA model and the ordinary Kalman filtering algorithm, the optimization model takes into account the periodicity and time-varying of the total travel time distribution, and has smaller error and better fitting degree. The average absolute error (MAE) of the predicted time is reduced by 37% and 52% respectively on the basis of the traditional SARIMA model and the ordinary Kalman filtering algorithm. The other error indexes, such as the root mean square error (RMSE) and the maximum relative error (MRE), have been greatly reduced. The validity of the model is proved.
【作者单位】：上海交通大学船舶海洋与建筑工程学院海洋工程国家重点实验室;上海交通大学船舶海洋与建筑工程学院交通研究中心;佛罗里达大学城市与区域规划系;
【基金】：国家自然科学基金青年科学基金项目(71101109) 长沙理工大学公路工程教育部重点实验室开放基金项目(kfj120108)
【分类号】：U491.1

【参考文献】