提出了数据驱动与中观交通仿真融合的交通流预测框架.该框架将数据驱动方法在路网局部断面和路段的高精度预测能力与中观交通仿真的路网范围预测能力结合起来,通过可信度高的路网局部断面和路段预测值,在线修正中观交通仿真模型的参数,使得中观交通仿真模型能够逼近、反映交通流运动趋势,提高路网范围交通状态预测精度.通过结合路段旅行时间预测与中观交通仿真的实例分析证明,断面和路段预测和中观交通仿真结合发挥了两者各自的优势,预测结果优于单一的中观交通仿真方法.%A framework that combines data-driven predictors and mesoscopic traffic simulators is proposed for traffic network state prediction. In the framework, the mesoscopic simulation model with network prediction ability is calibrated on-line by the highly reliable road segment traffic flow prediction results from the data-driven predictors. Therefore, the calibrated mesoscopic simulation system is able to capture the evolution of traffic flow and improve the network traffic prediction accuracy. A case study, which combines the road segment travel time predictor and the mesoscopic traffic simulator, is conducted to verify the effectiveness of the framework. The empirical results show that the framework fully utilizes the advantages of the two prediction approaches and achieves better traffic network state prediction performances than the method that applies the mesoscopic traffic simulation only.
展开▼
