基于实时多模态时空数据的时空图卷积网络精准鲁棒交通流预测模型

A Spatial-temporal Graph Convolutional Network Model for Accurate and Robust Traffic Flow Prediction Based on Real-time Multimodal Spatial-temporal Data

随着我国城市规模的不断扩张,绕城高速公路被各大城市相继采用,在疏导城市过境交通、缓解市内拥堵、提高通行效率等方面起到积极作用。然而,少有学者针对绕城高速公路开展全面和系统的研究,在其交通流特性、态势评估预测、运营管理措施等方面有着较大的研究空间。传统基于单一或部分数据源进行交通预测受到数据源的限制,在精准度、鲁棒性和实用性上都有待提高。本研究选择成都绕城高速公路(国道G4202)作为试点路段,基于系统视频智能分析平台对车流速进行实时监测,并基于实时多模态时空数据的时空图卷积网络进行精准鲁棒交通预测模型的研究。该方法结合了时空图卷积网络以及卡尔曼滤波(Kalman Filtering)校正相点的未来演化规律,预测模型通过自学习对比实际交通监测数据进行优化,并提供交通预测模型的有效性与可信度评估。实际交通流数据测试结果表明,文中所提出的预测模型与传统预测模型相比具有更高的预测精度,是一种有效的交通流预测方法。

With the continuous expansion of city size in China,ring expressways have been adopted in most cities,playing a positive role in diverting urban transit traffic,relieving urban congestion,and improving traffic efficiency.However,few scholars have conducted comprehensive and systematic studies on ring expressways.Traffic flow characteristics,situation assessment and prediction,and operation management measures are still in research.Traditional traffic prediction based on single or partial data source is limited by data sources,and needs to be improved in terms of accuracy,robustness,and practicability.Selecting the Chengdu Ring Expressway(National Road G4202)as the pilot section,based on the system video intelligent analysis platform,the real-time vehicle flow speeds are monitored,and the research on the accurate and robust traffic prediction model is carried out based on the ST-GCN of real-time multi-modal spatial-temporal data.This method combines the ST-GCN and Kalman filtering to correct the future evolution of the phase point.The prediction model is optimized through self-learning and comparison with actual traffic monitoring data,and the evaluation of the effectiveness and credibility of the traffic prediction model is also provided.The test result of actual traffic flow data shows that the proposed prediction model has higher prediction accuracy than traditional prediction models,it is an effective traffic flow prediction method.