论文部分内容阅读
针对出行者出行时对交通信息预报以及动态路径规划的要求,对路段的历史交通流时间序列数据进行了研究,利用城市路段交通流的周期相似性特征提出了基于纵横序列相似性的短期交通流预测VHSSA模型,该模型克服了以往预测模型只考虑纵向时间序列周期性相似的缺陷,将全时间序列数据进行小波变换后分解为反映基本变化规律的基序列和反映波动变化情况的波动序列,既可只进行基序列预测,也可通过置信区间对波动序列进行修正,再与基序列叠加进行全序列预测。经试例验证,VHSSA模型和基于纵向序列相似性的VSSA模型分别与实测序列的基序列和全序列进行比对,VHSSA模型的预测效果总体优于VSSA模型,误差可满足实际要求。
Aiming at the requirements of traffic information forecast and dynamic path planning for travelers, this paper studies the historical traffic flow time series data of road sections, and proposes the short-term traffic flow based on the vertical and horizontal sequence similarity by using the cycle similarity of urban road traffic flows The VHSSA model is predicted. The model overcomes the shortcomings of the previous prediction model, which considers only the periodic similarity of the longitudinal time series. The whole time series data are decomposed into the basic sequence reflecting the basic variation and the fluctuation sequence reflecting the fluctuation. Only base sequence prediction can be carried out, and the fluctuation sequence can also be corrected through the confidence interval, and then the full sequence prediction can be performed by superimposing with the base sequence. The experimental results show that the VHSSA model and the VSSA model based on the similarity of longitudinal sequences are compared with the base sequence and the complete sequence respectively. The prediction results of VHSSA model are better than VSSA model, and the error can meet the actual requirements.