一种车载组合导航系统的联邦滤波新算法

针对车载GPS/DR组合导航系统的特点,提出了一种新的联邦滤波算法.子滤波器由两个滤波器构成:一个是卡尔曼滤波器,另一个是Unscented卡尔曼滤波器(UKF),它们分别处理GPS子系统和DR子系统的信息.UKF是一种较新的非线性滤波方法,相对传统的非线性滤波器EKF具有更高的滤波精度和更高的计算效率.联邦滤波器的信息分配因子采用一种基于观测误差的最小均方误差形式,最佳地融合了传感器的输出信息.这种结构降低了联邦滤波器的计算难度和计算量,提高了组合导航系统的可靠性和容错性.仿真结果表明,提出算法的定位精度很高,与UKF算法的效果相当.

A new federated filtering algorithm for vehicles' integrated navigation

In view of the characteristic of GPS/DR vehicle navigation system,a new federated filter was proposed.This filter was composed of two sub-filters: a Kalman filter and an unscented Kalman filter(UKF),and they were respectively used to process the information that came from GPS sub-system and the DR sub-system.In which,the information allocation factor was taken as a minimum mean square error(MMSE) based on observation error.In this way,the output information was fused optimally.With this federated filter,the computation load and difficulty decrease,but the reliability and fault tolerance were improved.Simulation results show that the position precision is high and is comparative with that of the UKF.