基于行程时间的车辆通行连续性分析

定点检测器采集的路段行程时间样本中通常包含连续通行和非连续通行两部分，针对连续通行样本的识别，传统的固定阈值方法无法提供合理的结论.将行程时间样本时间序列化后，利用其差值的平稳随机性质，采用基于统计学的离群点检测的参数方法分离样本比传统方法更加科学有效.通过实证分析，非连续通行样本的占比是一种能够分离样本的指标.关联性分析的结论表明，该指标在4 分钟车程长度以内的路段中取值平稳，在4 分钟车程长度以上的路段中取值则受到道路等级、路段长度、交通运行状态、区位用地性质等多种因素的综合影响.

Continuous Travel Characteristics Based on Travel Time Data

Travel time samples collected by fixed traffic detectors usually consist of two parts: continuous travel time and discontinuous travel time. In the previous study, the continuous travel time samples are recognized by the fixed threshold; however, the outcomes cannot be proved accurate and correct. Converting travel time samples to time series whose difference value is stationary stochastic process, recognition becomes more scientific and effective by using a statistical parametric method of outlier detection. The ratio of discontinuous travel data is a typical indicator of recognition based on empirical study. The results of the association analysis indicate that the ratio is stable under 4 minute ride whereas when it comes to beyond this range, the ratio is influenced by the road grade, road length, traffic condition and locational factors.