用于短时交通流预测的多项式分布滞后模型

A PDL Model Used for Short-term Traffic Flow Forecasting

摘要:

提出了用于短时交通流预测的多项式分布滞后模型.其建模思想是交通状态时间序列同时受自身滞后项之外的多个因素影响,并且影响分布到了多个时段.通过与ARIMA模型(自回归整数移动平均模型)的预测精度对比分析,表明PDL（多项式分布滞后）模型具有与ARIMA相同的预测精度,而在模型可移植性、算法复杂性和实现方面更具优势.研究结果为短时交通流预测理论提供一种新的研究思路.

Abstract:

polynomial distributed lags (PDL) model is proposed for short-term traffic forecasting,the key idea of which is that the time series of traffic condition are affected not only by the own lagged terms,but also by other factors relevant to the traffic condition.The PDL model also assumes that the effects of other factors are distributed to a number of terms.Comparison with the autoregressive integrated moving average (ARIMA) model shows that with the same prediction accuracy as the ARIMA model,the PDL model is more portable and easier but less complicated to implement.This research result provides a new research idea for the short-term traffic flow forecasting theory.