中文摘要：

为了更有效地预测城市路网交通流量,本文提出了一种城市道路交通预测模型.该模型基于网络层析成像（Network Tomography,NT）技术建立生成树,采用期望最大（Expectation Maximization,EM）算法得到路网子网车流概率分布,再结合路网子网中流量守恒原则,对待预测路段流量进行推测.实验结果表明,该模型优于现常用的人工智能模型,对城市交通流量预测更为有效,且提高了预测精度.

英文摘要：

To forecast the urban traffic flow more effectively, this paper proposes a Network Tomo-graphy based traffic flow prediction model. The model builds a spanning tree based on Network Tomo-graphy, estimates traffic probability distribution in road network subnet by Expectation-Maximization（EM） algorithm, forecasts the traffic flow according to the flow conservation in road network. Experi-mental results show that the new model has higher estimation accuracy compared to the Artificial intelli-gence model.