基于灰色关联分析的路段行程时间卡尔曼滤波预测算法

为改善卡尔曼滤波用于时间序列预测时的自适应性能,提出基于灰色关联分析的路段行程时间实时预测算法．首先,利用灰色理论对行程时间序列的各影响因素进行灰色关联分析,根据灰色关联度的大小来选取路段行程时间的主要影响因素,由此建立相应的动态方程．在此动态方程基础上,通过卡尔曼滤波递推进行路段行程时间预测．文中利用深圳某交通干道上的实测行程时间进行仿真实验,结果表明该算法的综合预测性能优于常规卡尔曼滤波方法,可应用于正常交通流状况下的路段行程时间预测．

Estimation Algorithm with Kalman Filtering for Road Travel Time Based on Grey Relation Analysis

In order to improve the self-adaptability of Kalman filtering applied to the estimation of time series, a real-time estimation algorithm for road travel time is put forwarded based on the grey relation analysis. The grey relation analysis of various factors to affect the travel time series is first carried out based on the gray theory, and the main factors to influence the road travel time are picked out according to the grey relevancy degree. Then, the corresponding dynamic equations are set up, and the estimated values are obtained using a set of reeursion formulas. The real data of the road travel time collected in an artery in Shenzhen City are finally used to perform simulated experiments. The results indicate that the proposed algorithm is suitable for the estimation of road travel time in natural traffic flow because it is of better integrated performance than the conventional Kalman filtering model.