基于路网空间拓扑结构的交通流数据填补模型研究

A Study on Improved Imputation Methods for Traffic Flow Data Based on Spatial Topology of Road Network

由于传感器故障、状况不稳定或数据传输等问题,交通流数据存在大量丢失的情况,对依靠这些数据进行交通科学决策产生了很大的影响。基于以上问题,提出一种基于空间拓扑结构的交通流数据填补方法,模型通过选取区域的典型道路训练上下游之间相关参数,根据道路类型匹配典型道路,应用典型道路的参数进行交通流数据填补,并利用武汉市实际交通检测器数据进行模型验证。选取武汉市实际道路上2个典型地磁检测器进行数据填补误差分析,不同填补方式下的平均相对误差为52.88%和51.93%,相对于传统的交通流数据填补建模来说,基于空间拓扑结构的填补模型在实际的应用过程中,能够在更大的范围内对丢失的交通流数据进行填补。

Due to breakdown or technical issues of sensors and/or unstable data transmission,a large amount of traffic flow data is missing,which has significant impacts on scientific traffic decisions rely on the data.Based on the prob⁃lems,an imputation model of traffic flow data based on spatial topology of the road network is proposed.The model trains the relevant parameters by using the data from typical upstream and downstream road segments in one region.The selected model with trained parameters is used to impute the missing data from the candidate road segment based on the traffic flow data from upstream or downstream.The traffic detector data from Wuhan are used to verify these models.Two typical geomagnetic detectors are selected for accuracy analysis of the proposed imputation models.The APE of improved methods is found to be 52.88%and 51.93%,respectively.The results show that compared to tradi⁃tional imputation models,the models proposed in this paper can impute the missing data of traffic counts at a much larger scale,even for those detectors without sufficient historical,local data.