随着智能交通的发展,实时动态交通分配成为当前研究热门问题。短时交通流预测是实时动态交通分配的关键技术之一,在当今交通控制以及车辆导航中具有不可替代的地位。通过对交通流数据进行分析,得出交通系统具有耗散系统特性,并且存在混沌。在此基础上,运用混沌理论对交通流数据进行相空间重构,并用多元局域预测法对时间序列进行预测。通过分析预测数据,得出基于混沌理论的短时交通流量预测在2~5min内具有较高的预测精度。
With the development of intelligent traffic, real-time dynamic traffic assignment (DTA) has become one of the most popular researches. Short-time traffic forecast is the key technology of real-time dynamic traffic assignment, and plays an important role in traffic control and navigation. After analyzing data of traffic flow, the paper addresses that transportation system is dissipative and chaos exists in traffic status. Therefore, phase space reconstruction and multivariate local method is employed to generate and predict time series. The proposed method has high accuracy in two to five minutes.