基于探测车数据的路段行程时间估计

Link Travel Time Estimation Based on Probe Vehicle Data

利用探测车数据进行路段行程时间估计面临着两类误差:采样误差和非采样误差,从而导致估计结果精度不高和可靠性差。在回顾已有估计方法的基础上,有针对性地引入了自适应式卡尔曼滤波,建立了相应的状态方程和观测方程,利用相似时间特征的历史数据标定了状态转移系数,并对滤波进行了求解。以实际数据对估计方法进行了验证,平均相对误差为13.13%。研究表明,自适应式卡尔曼滤波能够应用到基于探测车数据的路段行程时间估计中来,并具有估计精度高、收敛速度快、参数少、对初值不敏感等优点。

Two kinds of errors including sampling error and non-sampling error can be encountered by using probe vehicle data to estimate link travel time,leading to low accuracy and low reliability of the estimation.The paper introduces an adaptive Kalman filter,sets up state equation and observation equation,and calibrates state transition parameter by using data with similar time features to obtain the solution of the filter.The estimation method is validated by using actual data,where the mean relative error is 13.13%.This result shows that adaptive Kalman filter can be applied to estimating link travel time from probe vehicle data with relatively high level of accuracy.This method converges rapidly and involves fewer parameters,which are also insensitive to initial values.