一种适用于PPP动力学模型异常的自适应Kalman滤波

An Adaptive Kalman Filter for Dynamics Model Abnormity of PPP

为了削弱PPP参数估计中动力学模型异常对Kalman滤波解的影响,针对PPP状态向量中各类参数不符值对动力学模型异常描述特性的不同,以位置相关为条件对参数进行分类,构建分类因子自适应Kalman滤波用于PPP参数估计。选取6个IGS站点3 d的数据,使用标准Kalman滤波与构建的自适应Kalman滤波进行PPP解算分析。结果表明,相较于标准Kalman滤波,自适应Kalman滤波能通过自适应因子调节状态预测协方差,加速PPP收敛。静态模式下,平均收敛时间从28.2 min缩减到19.4 min,N、E、U方向的平均精度为1.50 cm、3.34 cm、5.55 cm,分别提高7%、14%、19%;动态模式下,构建的自适应Kalman滤波解N、E、U方向偏差的RMS值为2.7 cm、3.6 cm、6.3 cm,较标准Kalman滤波分别提高13%、28%、43%。

In order to weaken the kinetic model anomaly when using Kalman filtering for PPP parameter estimation,we propose adaptively filtering with classified adaptive factors.According to the different reliability of PPP parameters to be evaluated,the state parameter vector is divided into two groups,namely position and others.Through three days observations from six IGS stations,the solutions of different Kalman filtering are obtained.The results show that the adaptive Kalman filter can accelerate the PPP convergence by adjusting the state prediction covariance by the adaptive factor.In the static modes,the average convergence time is reduced from 28.2 min to 19.4 min,the average accuracies are notably improved by 7%,14%and 19%,which are 1.50 cm,3.34 cm and 5.55 cm in the N,E and U directions.In the kinematic mode,the average accuracies are notably improved by 13%,28%,and 43%,which are 2.7 cm,3.6 cm,and 6.3 cm in the N,E and U directions.