服务于智能交通系统的离群交通数据识别

为了提高交通建模的准确性和可靠性,或者提取重要的有价值的隐藏信息,将离群数据挖掘技术引入交通数据处理.首先分析了3种典型的离群数据挖掘算法:基于统计的方法、基于距离的方法以及基于密度的方法的原理、特点和时间复杂性;其次给出了2个实例分析,一是在建立交通流量预测模型前,将基于统计的方法和基于距离的离群检测方法分别用于交通量时间序列,寻找离群数据;二是将基于距离的方法和基于密度的方法用于路面平整度检测.实例研究表明,离群数据挖掘算法可有效识别异常交通数据,在交通工程领域具有较大的应用潜力.

Outlier detection in traffic data sets serving for intelligent transportation system

Outlier mining technique is introduced into traffic data process to detect and analyze the outliers of traffic data sets in order to improve the veracity and reliability of traffic modeling,or draw out important and valuable hidden information.First,the principle,characteristic and time complexity of three typical outlier mining approaches,that is statistical-based approach, distance-based approach,density-based approach,are analyzed. Two examples are then given.One of them is to detect outliers hidden in traffic volume before prediction model built based on statistical-based and distance-based outlier mining approaches,the other applies statistical-based and distance-based outlier mining approaches to pavement roughness detection.The experimental results illustrate that these outliers mining methods are feasible and valid to detect outliers in traffic data sets,and have a good potential of application in traffic engineering.