基于Gamma分布的交通流时间序列分割模型

删除或更新信息，请邮件至freekaoyan#163.com(#换成@)

本站小编 Free考研考试/2024-01-16

-->

王本超,李丹,秦攀,顾宏.基于Gamma分布的交通流时间序列分割模型[J].,2020,60(3):293-299

基于Gamma分布的交通流时间序列分割模型

Gamma distribution based traffic flow time series segmenting model

DOI：10.7511/dllgxb202003010

中文关键词:交通流时间序列 Gamma分布时间序列分割非负主成分分析

英文关键词:traffic flow time series Gamma distribution time series segmentation nonnegative principal component analysis

基金项目:国家自然科学基金资助项目(61502074,61633006);中国博士后科学基金资助项目(2016M591430);大连理工大学基本科研业务费专项资金资助项目(DUT17RC(4)09).

作者	单位
王本超,李丹,秦攀,顾宏

摘要点击次数:328

全文下载次数:344

中文摘要:

准确获取交通流量变化点,对后续的交通流预测、分类及多时段控制具有重要意义.鉴于交通流时间序列的非负性及异方差性,采用Gamma分布拟合交通流时间序列,并对其进行有效分割.针对多元交通流时间序列,首先利用非负主成分分析方法实现降维并提取特征序列,之后利用最大似然估计得到Gamma分布参数,通过不同参数的Gamma分布拟合特征序列的不同片段,并由赤池信息准则(AIC)确定最优分割边界及分割阶数.实验结果表明,所建立的分割模型能够反映不同时段的交通流变化,与现有分割方法相比,取得了更好的分割结果.

英文摘要:

It is significant to accurately obtain the change points of traffic flow for the subsequent traffic flow prediction, classification and multi\|time control. Considering the nonnegative and heteroscedasticity, the traffic flow time series are fitted by Gamma distribution and segmented effectively. For multiple traffic flow time series, dimension reduction is carried out by the nonnegative principal component analysis (NPCA) for feature extraction. Then, the likelihood of the principal component is constructed to obtain the parameters of the Gamma distribution. Consequently, the change points are determined from degree of fitting using the different parameters of the Gamma distribution by maximizing the likelihood. The Akaike information criterion (AIC) is used to select the optimal segmentation order and boundary. The experimental results indicate that the proposed segmenting model can reflect the change of traffic flow at different time and has better segmentation results than other existing methods.

查看全文查看/发表评论下载PDF阅读器

关闭

基于Gamma分布的交通流时间序列分割模型

本站小编 Free考研考试/2024-01-16

相关话题/

领限时大额优惠券,享本站正版考研考试资料!