作者投稿和查稿 主编审稿 专家审稿 编委审稿 远程编辑

计算机工程

• 人工智能及识别技术 • 上一篇    下一篇

无检测器道路交通流数据质量检测方法

王 方,李 华,杜金玲   

  1. (西安电子科技大学经济与管理学院,西安 710071)
  • 收稿日期:2013-01-15 出版日期:2014-03-15 发布日期:2014-03-13
  • 作者简介:王 方(1987-),男,博士研究生,主研方向:智能交通系统,决策分析;李 华,教授、博士生导师;杜金玲,博士研究生。
  • 基金资助:
    2012年西安市科技计划基金资助项目“道路交通与应急指挥系统研发”(CX1240)。

Quality Detection Method for Non-detector Road Traffic Flow Data

WANG Fang, LI Hua, DU Jin-ling   

  1. (School of Economics & Management, Xidian University, Xi’an 710071, China)
  • Received:2013-01-15 Online:2014-03-15 Published:2014-03-13

摘要: 一般交通流数据质量检测方法要求的原始数据量较大,而无检测器道路可获得的交通流数据又非常有限。为此,提出一种基于灰色系统理论的无检测器道路交通流数据质量检测方法。该方法将不同检测点获得的原始交通流数据处理成一组数据序列,通过对数据序列的灰生成、灰色关联度计算及标准化处理,求得不同数据序列相互间关系的密切程度参数λi,根据需求选出阈值λ,比较λi与λ之间的大小,实现无检测器道路交通流异常数据检测的目的。运用杭州市某一局部路网的浮动车交通流原始数据,将该方法与基于相似系数和的检测方法进行对比实验,结果证明,该方法的检测效果优于基于相似系数和的检测方法,平均错检率降低了21.00%,平均准确率提高了28.64%。

关键词: 智能交通, 交通流, 脏数据, 数据清洗, 数据质量, 灰色系统理论

Abstract: Conventional data quality detection method requires large number of initial data while traffic flow data at non-detector road is very limited. A new non-detector road traffic flow data quality detection method based on grey system theory is put forward to deal with the contradiction. The raw traffic flow data obtained by different detection points is processed into a set of sequence data. Through grey generating, calculating and standardizing of the set of sequence data, the closeness of the parameters λi which reflect the mutual relations between different data sequence is obtained. The purpose of detecting outliers is realized through the comparison of the size of λi and λ which is the selected threshold based on demand. Using the probe car traffic flow data which covers a local road network of Hangzhou, the efficiency of the proposed method is verified by comparing with the detection method based on similarity coefficient. The proposed method is better than the method based on similarity coefficient. For example, the average false detection rate of this method is lower than the method based on similarity coefficient by 21.00%, and the average accuracy rate is 28.64% higher than the latter one.

Key words: intelligent traffic, traffic flow, dirty data, data cleaning, data quality, grey system theory

中图分类号: