基于多源数据的交叉口分流向交通需求预测

对交通流进行科学预判是实施精细化智能管控的基础,为了解决目前方法对于过饱和状态下需求预测精度较低的问题,利用交叉口地磁和上游路段微波数据,结合Markov转移矩阵及加权移动平均法对交叉口内分方向的流向比例依照时间序列进行动态预测,由上游路段的车辆通过率获取交叉口内的交通需求,进而构建交叉口分方向流量动态预测模型。最后通过实测数据对模型进行验证,结果显示总平均误差为13.46%,比使用传统预测模型的预测误差减少了4.13%,尤其是过饱和状态下的预测误差减少5.81%,有效提升了过饱和状态下的交通需求预测精度,这对于城市交叉口过饱和状态下的分流向交通组织及控制具有重要意义。

Traffic Demand Forecasting for Flow Direction at Intersection Based on Multi-source Data

The scientific prejudgment of traffic flow is the basis for the implementation of sophisticated intelligent control. In order to solve the problem of low precision of the current demand forecasting method in over-saturation, this paper uses the geomagnetic date at intersection and the microwave data of upstream section, combines with Markov transfer matrix and weighted moving average method, the directional flow ratio at intersection is predicted dynamically according to the time series. The traffic demand at intersection is obtained by vehicle passing rate of the upstream section, and then a dynamic prediction model of flow direction at intersection is constructed. Finally, the model is validated by the measured data. The results show that the overall mean error is 13.46%, and the prediction error is reduced by 4.13% than using the traditional prediction model. Especially in over-saturation, the prediction error is reduced by 5.81%, which effectively improves the accuracy of traffic demand prediction in over-saturation. This is important for the flow direction to organize and control under the over-saturation of urban intersections.