分车型的高速公路短时交通流量预测方法研究

发布时间：2019-01-17 08:25

【摘要】：针对混合交通流中车辆类型的不确定性和随机性,导致了直接对总车流量进行预测时难以获得较高的预测精度问题,提出一种分车型的流量预测方法。依据各种车型的车流量变化规律不同的特点,选用改进的时间序列算法对大型车和拖挂车的流量进行预测,选用二次指数平滑法对小客车和中型车的流量进行预测;然后通过车辆折算系数将各车型的流量预测值进行加权求和,从而得到总车流量预测值;最后利用渝武高速公路上微波车检器的实测数据对提出的预测方法进行了实验验证,并与非参数回归预测方法和卡尔曼滤波预测方法进行了对比。实验结果表明,无论在工作日还是节假日,分车型的流量预测方法均具有更高的预测精度,该结果为进一步提高高速公路管控能力建立了基础。
[Abstract]:In view of the uncertainty and randomness of vehicle type in mixed traffic flow, it is difficult to obtain high prediction precision when forecasting the total traffic flow directly. According to the different characteristics of vehicle flow, the improved time series algorithm is used to predict the flow of large vehicles and towing vehicles, and the quadratic exponential smoothing method is used to predict the flow of passenger cars and midsize cars. Then, the traffic forecast value of each vehicle model is weighted summing up through the vehicle conversion coefficient, and the total vehicle flow forecast value is obtained. Finally, the proposed prediction method is verified by using the measured data of microwave vehicle detector on Yu-Wu expressway, and compared with non-parametric regression prediction method and Kalman filter prediction method. The experimental results show that the forecasting method of traffic flow by type of vehicle has a higher accuracy both on weekdays and on holidays. The result establishes the foundation for further improving the ability of expressway control and control.
【作者单位】：重庆工程职业技术学院信息工程学院;重庆大学自动化学院;
【基金】：中国工程院重点咨询项目(2012-ZX-22) 重庆市自然科学基金重点资助项目(cstc2012jj B40002) 国家教育部博士点基金资助项目(20120191110047) 重庆市教委科学技术研究项目(KJ1403208,KJ1403209)
【分类号】：U491.1

【参考文献】