基于时空图神经网络的交通拥堵预测与疏导策略研究

Research on Traffic Congestion Prediction and Relief Strategy Based on Spatiotemporal Graph Neural Network

摘要: 针对城市交通拥堵日益严峻的现实问题，通过对交通拥堵时空演化特征所进行的深入分析，揭示了传统预测方法在处理复杂时空关联性方面所存在的局限性。研究成功构建出基于时空图神经网络的预测模型，将交通路网抽象为动态图结构，并设计出多尺度时空特征提取与融合机制，有效捕捉了交通流的时空依赖关系。在多个真实数据集上所开展的实验验证表明，该模型在预测精度以及稳定性方面显著优于传统方法。基于所获得的预测结果，开发出动态路径诱导机制以及信号灯协同控制策略，实现了从预测到疏导的闭环管理。原型系统在实际城市场景中所进行的部署应用，证明了该研究对于缓解交通拥堵以及提升道路利用效率具备重要实践价值。

Abstract: Addressing the increasingly severe issue of urban traffic congestion, this study analyzes the temporal-spatial evolution characteristics of traffic congestion and reveals the limitations of traditional prediction methods in handling complex spatiotemporal correlations. A prediction model based on Spatiotemporal Graph Neural Networks (STGNNs) is constructed, which abstracts the traffic road network into a dynamic graph structure. A multi-scale spatiotemporal feature extraction and fusion mechanism is designed to effectively capture the spatiotemporal dependencies of traffic flow. Experimental validation on multiple real-world datasets demonstrates that the proposed model significantly outperforms traditional methods in both prediction accuracy and stability. Based on the prediction results, a dynamic route guidance mechanism and a coordinated traffic signal control strategy are developed, achieving closed-loop management from prediction to congestion mitigation. The deployment of a prototype system in a real-world urban scenario proves the significant practical value of this research in alleviating traffic congestion and enhancing road utilization efficiency.