摘要
考虑到实际交通流的时变性与复杂性,本文在分布未知的情形下主要从非参数方法角度对交通流个数及位置进行检测估计。首先提出在具体交通领域的多变点问题;然后介绍基于SBC(BIC)信息准则的非参数极大似然变点个数及位置检测方法;最后运用动态算法通过R编程对贵阳市中心城区宝山北路与延安东路交叉口一周(周一至周日)实际车流量数据进行变点个数和位置的检测估计,以对实际交通状态划分建模及有效缓解交通拥堵提供依据及参考意见。
Considering the time-variability and complexity of traffic, the paper mainly estimates the number and location of traffic flow change-point from the perspective of non-parametric method in the case of unknown distribution. Firstly, multiple change-point problem in the specific traffic field is put forward. Then non-parametric maximum likelihood method based on SBC(BIC) information criterion to estimate the number and location of change-point are introduced. Finally, the dynamic algorithm is used to detect the number and location of change-point of the actual traffic flow data at the intersection of Baoshan North Road and Yan 'an East Road during one week(Monday to Sunday) with R, so as to provide basis and reference for the modeling of the actual traffic state and the effective alleviation of traffic congestion.
引文
[1]陈希孺.变点统计分析简介[J].数理统计与管理,1991,20(2):55-58.
[2]王晓原,张敬磊,马立云.基于加速遗传的交通流变点分析方法[J].理论新探,2013(7):32-35.
[3]王晓原,隽志才,贾洪飞,等.交通流突变分析的变点统计方法研究[J].中国公路学报,2002(4):69-74.
[4]杨新月,王晓原,赵胜川,等.基于Γ分布变点检验的交通流演化辨识方法[J].中国安全科学学报,2013(6):86-91.
[5]朱广宇,王雨晨,张彭,等.基于变点发掘的城市轨道交通客流预测模型[J].中南大学学报(自然科学版),2016(6):2153-2159.
[6]王红玉,施三支.多变点检验及在城市轨道交通中的应用[J].数学的实践与认识,2017(17):222-228.
[7]Zou C.L.,Yin G.S.,Feng.L.,Wang Z.J.Nonparametric maximum likelihood approachto multiple change-point problems [J].The Annals of Statistics,2014,42(3):970-1002.
[8]Tahmasbi R,Hashemi SM:Modeling and forecasting the urban volume using stochastic differential equations.IEEE Transactions on Intelligent Transportation Systems,2014(15):250-259.