摘要
道路交通状态预测是交通流诱导和交通信息发布系统的重要依据.本文提出了一种基于最大熵模型的城市快速路交通状态预测方法,该方法通过提取影响交通状态的时间、空间等各种特征,运用最大熵模型训练得到各特征权重,直接预测道路交通状态等级.最大熵模型能够有效融合时间、空间等多种特征,并且不需要考虑各特征之间的相关性,具有很强的适应性.实验结果表明,基于最大熵模型直接预测交通状态等级同样具有较高的准确性.最大熵模型的成功使用,也证实了将交通状态预测作为一种模式分类问题来解决的可行性,进一步扩展了交通状态预测的思路.
The traffic state forecasting is an important basis for the traffic-induced system and traffic information dissemination system. The paper proposes a new forecasting method based on the maximum entropy model towards the urban freeway. The maximum entropy model can get the weights for the different features by the data training, and predict the road traffic state level directly. This method can fuse the temporal and spatial features efficiently, regardless of the relativity between the different features, which make this model more adaptable. The experiment results show that this traffic state prediction method can achieve a high accuracy. The success of this model prove that the traffic state prediction problem can be resolved through the pattern classification method, which enhances the methods of traffic state forecasting.
出处
《交通运输系统工程与信息》
EI
CSCD
2010年第2期112-116,共5页
Journal of Transportation Systems Engineering and Information Technology
基金
国家863研究计划项目(2006AA11Z231)
北京市科技计划重点项目(D07020601400707)
国家自然科学基金项目(60674002)
国家科技支撑计划项目(2007BAK12B04-15)
关键词
城市交通
交通状态
城市快速路
最大熵模型
预测
模式分类
urban traffic
traffic state
urban freeway
maximum entropy model
forecast
pattern classification
