摘要
交通流的可预测时间间隔是实现交通控制和诱导的关键。为考察交通流时间序列在不同时间间隔下的可预测性,以北京市二环路2min、4min、6min、8min、10min、12min、14min和16min间隔的交通量时间序列为研究对象,应用R/S分析法计算不同观测尺度下的Hurst指数值,发现同一天内Hurst指数值随观测尺度的变大而增大,同一观测尺度下,Hurst指数随着样本量的增加而降低。最后,采用近似熵法计算序列的复杂度,发现复杂度指数有相似的变化规律,证明观测尺度大小与序列的随机性强弱存在负相关。
The predictable interval of traffic flow is the key to achieve real-time traffic control and guidance. In order to investigate the predictability of traffic flow time series in different intervals, this paper focuses on traffic volume time series with the intervals of 2min, 4min, 6min, 8min, 10min, 12min, 14min and 16min from Beijing 2nd-Ring Road. Using R/S analysis method to calculate the Hurst exponent of different observation scales, discovering that Hurst exponent increases as the observed scales get larger in the same day, and it increased as the sample size decreases in the same observation scales. Finally, using approximate entropy method (ApEn) to calculate the complexity of sequence, discovering the complexity index had a similar variation, which confirms that there is a negative correlation between the observation scales and the strength of randomness of series.
出处
《系统工程》
CSSCI
CSCD
北大核心
2010年第5期75-80,共6页
Systems Engineering
基金
教育部"新世纪优秀人才计划"项目(NCET-08-0718)
国家自然科学基金资助项目(60874078)
国家科技支撑计划课题(2006BAG01A01)
高等学校博士学科点专项科研基金资助项目(20070004020)