基于时间占有率的短时交通预测模型

针对现有的交通预测模型大多以交通量为变量,而交通量作为二值函数,无法有效地判断道路交通是否处于交通顺畅或拥挤状态,采用时间占有率这一单值变量,建立了基于ARIMA(p,d,q)时间序列结构的短时交通预测模型。在建模过程中,以非参数检验法判定序列平稳性,AIC准则确定模型结构,最小二乘算法(LS)估计参数。以城市中心商务区交叉口为实例,对一个信号控制周期的占有率实测值进行预测分析。结果表明,时间占有率比仅仅用交通量更能描述交叉口的实际情况,且算法简单,精度高,适合于交通控制和信息诱导系统的实时在线预测。

Short-term Traffic Prediction Model Based on Occupancy

Many existing models for traffic prediction are based on traffic flows. As a dual-value function, flow data could not be used to find effectively whether the traffic on the road is good or in congestion. Time-occupancy, which is a single-valued function, is taken to establish a feasible analysis process and traffic prediction model based on ARIMA of urban intersection. In modeling a non-parametric inspection is used to validate the stationary of time series, the AIC rule to estimate orders and the LS method to estimate parameters. In the end, an updating prediction in a CBD intersection proves that occupancy is a better parameter and fit for online prediction in traffic control and information guidance systems quite well.