摘要
为提高高速路短时交通流预测的准确度,建立了一种基于时空相关分析和BP神经网络的短时交通流预测方法,首先,通过分析高速路网上下游断面间的时空相关性和空间互相关性,选取与预测目标相关性较大的历史时段和相关断面。然后,将各相关断面交通流时间序列与其时间延迟序列进行重构,选取历史时段和重构后的相关断面作为BP神经网络预测模型的输入。利用四川省某高速路数据对该预测方法进行性能评价,实例证明该方法与只考虑高速路时间特性或空间特性的预测模型相比具有更高的预测精度,提高了交通流预测的实时性和可靠性。由此可见,该方法可作为高速路短时交通流预测的有效手段。
Aiming at the shortage of traffic flow prediction based on single cross-section, the interaction of the adjacent cross-sections in the high speed road network is studied. Then according on the analysis of the spatial-temporal characteristics,a short-time traffic flow forecasting model based on the multiple cross-sections was established .The model is extended to the prediction model based on a single cross-section,and temporal-spatial characteristics of high speed traffic flow and time delay characteristics of space interaction are considered and they determine the input dimension of the forecasting model.Finally, BP neural network is the forecasting tool to estimate the prediction results.The experimental results show that the prediction model has higher prediction accuracy compared with the traditional single cross-section prediction model and improved the real-time performance and reliability of traffic flow prediction.It is of great significance to improve the traffic efficiency of high speed road.
出处
《数字技术与应用》
2017年第3期46-50,共5页
Digital Technology & Application
基金
四川省交通科技项目(2013c7-1)
关键词
短时交通流预测
时空相关分析
BP神经网络
时间延迟
short-term traffic flow forecasting
spatio-temporal correlation analysis
BP neural network
time delay
