交通流的时间序列建模及预测

采用时间序列ARMA模型进行交通流量趋势预测，建模时利用现场测得的交通流非平稳时间序列进行差分变换及标准化处理，从而转化成标准正态平稳时间序列；模型参数估计采用极大似然估计法，并根据AIC准则为模型定阶；最后通过实测数据进行验证．结果表明该ARMA模型能够较好地拟合交通流时间序列并可获得较高的中短期预测精度，因而可用于动态交通信号控制．

Time Series Modeling and Predicting for Urban Traffic Flow

This paper proposes a trend prediction method for traffic volume by setting up time series ARMA model. In modeling the non-stationary time series obtained on the spot is transformed into stationary time series normally distributed by difference transformation and standardization, then the MLE method is used to estimate parameters and the AIC rule is used to estimate orders respectively. The simulation of field data shows that this ARMA model can fit traffic flow time series quite well. Moreover, the model can achieve better prediction accuracy in the medium and short-time term and can be applied to the dynamic traffic signal control.