基于因果关联的交通拥堵传播分析

Traffic congestion spreading analysis based on causal nexus

为揭示拥堵区域之间的内在联系,提出基于因果关联的交通拥堵传播分析方法。该方法通过对出租车GPS数据进行清洗、地图匹配等预处理后,计算蜂窝网格的速度并以蜂窝网格而非路段为单元检测拥堵事件,以便更精确地描述各区域的交通状态;基于拥堵的时空特征,提出拥堵传播事件提取方法,以确定拥堵的时空范围;提出网格关联置信度与拥堵传播源强度,推断蜂窝网格间拥堵传播的方向。研究结果表明:该方法能定位出易发生且易传播拥堵的重点传播源,并发现晚高峰相对早高峰更容易出现范围广、持续时间长的拥堵传播事件;该方法能用于揭示拥堵的演化规律,可为交通管理部门制定拥堵缓解策略提供科学依据。

In order to reveal the correlation between congestion areas,a causal nexus method was proposed to investigate congestion propagation in urban areas.Taxi GPS data were preprocessed via data cleaning and map matching.Hexagon speed was calculated and congestion events were detected on hexagon level instead of road segment level to receive more accurate results.Congestion spreading events were extracted using the spatiotemporal characteristics of hexagons.The confidence index and congestion intensity index were used to infer the congestion spreading diretion between hexagons.The results show that the proposed method can identify recurring congestion sources that spread congestion to other areas.The congestion spreading during evening rush hours was larger in scale and longer in duration compared to morning rush hours.The proposed method can uncover the evolutionary patterns of traffic congestion and provide significant intelligence for transportation authorities to mitigate traffic congestions.