摘要
道路网交通流短时预测是实现智能交通控制和诱导系统的基础,本文以实时采集的交通流数据为基础,运用状态空间重构和G-P算法进行交通流短期特性研究.具体而言,通过分析交通流的混沌特性确定交通流率时间序列的嵌入维数,重构相空间,并以重构的交通流率时间序列作为输入,采用Elman神经网络实现道路网多断面交通流率同时预测.研究结果表明,在道路网交通流短时预测方面Elman网络优于经典的BP网络,基于混沌时间序列分析的Elman网络不仅能够实现道路网多断面同时预测,且预测效果优于无重构的神经网络.
Short-term traffic flow prediction of road network based on chaos analysis is presented in this paper.The chaos characteristics of the short-term traffic flow time series are studied in the method according to analysis the traffic flow dates,which are processed by the phase space reconstruction technology and G-P algorithm.First,the method confirms embedding dimension and constructs phase space.Second,the Elman Neural Network is adapted to realize short-term traffic flow prediction of road network and traffic flow time series with constructs phase space is inputted.The method is tested by using the traffic flow date on the road network and the experiment results are analyzed.The prediction result of Elman network based on chaos theory is super to common Elman network and better than the BP network,which can realize short-term traffic flow prediction of road network too.
出处
《系统工程学报》
CSCD
北大核心
2011年第3期340-345,共6页
Journal of Systems Engineering
基金
国家自然科学基金项目(50578009)
"973"国家重点基础研究发展规划项目(2006CB705500)
中国发展研究基金会2009年度"通用汽车中国发展研究青年奖学金"项目
北京交通大学优秀博士生科技创新基金资助项目(14108-2522)
