摘要：AbstractReal-time prediction of congestion propagation is essential for urban traffic control and guidance. In this paper, a traffic pattern knowledge graph and its associated reasoning framework are proposed to approach this issue. By training a reinforcement learning-based agent to learn relation reasoning paths, it is possible to predict congestion propagation on real-time traffic data in an efficient manner.
关键词：KeywordsTraffic ControlKnowledge GraphCongestion PropagationTime Series Data
