基于EWM-TOPSIS的城市卡口地点车速数据诊断

Diagnosis for urban bayonet spot speed data based on EWM-TOPSIS method

为解决因交通信息监控设备年久失修导致时空数据不完备,对交通时空数据质量状况进行综合诊断,为后续数据质量恢复与应用研究奠定数据基底。首先利用地点车速数据的时序特性和箱线图判别准则(Boxplot)客观判别了问题数据类型,其次针对不同问题数据类型特点构建了多属性综合评价指标图谱,采用熵权法(EWM)为各评价指标赋予信息熵权值,利用逼近理想解(TOPSIS)评价法得到了各条路段地点车速数据的贴近度并给予排序,最后综合道路属性及贴近度排序结果评价了区域路网地点车速数据质量状况。以2021年乌鲁木齐市1周的原始卡口交通数据为例,对应不同问题数据类型提出了多个数据质量评价指标,利用多属性评价指标图谱及道路属性对EWM-TOPSIS综合评价方法进行验证。研究结果表明:问题数据判别算法能够精确判别出冗余数据、缺失数据及异常数据3类问题数据,对应的评价指标均值分别为5.66%、10.04%、5.04%,综合可用性指标均值为84.66%,各路段数据质量整体良好;利用EWM-TOPSIS综合评价方法得到的路段地点车速数据贴近度均值为0.25,区域路网地点车速数据质量指数均值为0.19,道路属性是影响区域路网数据质量的关键因素之一。所提数据质量诊断方法能够辨识原始卡口交通问题数据类型,综合道路属性客观评价区域路网的整体数据质量,为交通数据质量诊断及未来应用服务提供了新的解决思路。

In order to solve the incompleteness of spatiotemporal data caused by the disrepair of traffic information monitoring equipment,a comprehensive diagnosis for traffic spatiotemporal data quality was carried out,which lays a solid data base on subsequent data quality recovery and application research.First,the time series characteristics of spot speed data and Boxplot criterion were used to objectively identify the problem data types.Secondly,a multi-attribute comprehensive evaluation indexes map was constructed according to the characteristics of different problem data types,then the entropy weight method(EWM)was utilized to assign information entropy weights for each evaluation indexes.The technique for order preference by similarity to ideal solution method(TOPSIS)was applied to obtain the closeness degree of spot speed data on road segments giving the closeness degree ranking as well.Finally,the spot speed data quality on regional road network was comprehensively evaluated by combining the closeness degree ranking results with the road attributes.The original bayonet traffic data of Urumqi road network for one week in 2021was taken as an example,multiple data quality evaluation indicators were proposed by different problem data types.The EWM-TOPSIS method was verified through multi-attribute evaluation indexes map and road attribute.The results show that the problem data discrimination algorithm can accurately distinguish redundant data,missing data and abnormal data,meanwhile,the corresponding evaluation indexes are 5.66%,10.04%,5.04%respectively,and the average comprehensive availability value is 84.66%.The data quality for each road segment is better as a whole.The mean closeness degree of spot speed data on road segments is 0.25,and the mean quality index of spot speed data on regional road network is 0.19evaluated by EWMTOPSIS comprehensive evaluation method.The road attribute is also the one of key factors affecting the spot speed data quality on regional road network.The proposed method can identify the original traffic problem data types,objectively evaluate the spot speed data quality for regional road network by considering the road attributes,and provide a novel solution for traffic data quality diagnosis and future application services.1tab,6figs,27refs.