基于UPF算法的车辆GPS/DR组合导航研究  

UPF Filter Algorithm for Vehicle Integrated Navigation

作　　者：李桂芳[1] 孙勇成[2] 林坚[1] 黄圣国[1] 

机构地区：[1]南京航空航天大学民航(飞行)学院,南京210016 [2]中国电子科技集团28所,南京210007

出　　处：《科学技术与工程》2012年第31期8143-8146,共4页Science Technology and Engineering

摘　　要：车辆GPS/DR组合导航系统是非线性系统。采用扩展卡尔曼滤波(EKF)对其进行状态估计时,系统线性化过程将导致较大的滤波误差。为了获得更好的估计性能,将一类改进的粒子滤波方法 (UPF),即以无位卡尔曼滤波(UKF)为建议密度的粒子滤波方法(PF)应用于车辆GPS/DR组合导航系统中,避免了EKF方法的线性化近似过程,提高载体的定位精度。为验证该方法的有效性,将其与EKF分别用于GPS/DR组合导航系统的滤波仿真。仿真结果表明:UPF能减小导航定位误差,滤波性能明显优于EKF。The model of GPS/DR integrated navigation system is nonlinear, and there are great estimation er- rors due to necessarily linearize the intrinsic nonlinear systems if the Extended Kalman Filter （EKF） is used. In or- der to acquire better performance, an advanced particle filter algorithm named UPF is applied in GPS/DR integrat- ed navigation system, which adopts Unscented Kalman Filter （UKF） as its＇ proposal distribution. Due to avoiding the approximate linearizing process, the method can improve the positioning accuracy of carrier. In order to test the validity of the UPF, the two methods are compared in the process of estimating state of the vehicle integrated GPS/ DR navigation systems. The simulation results show that the UPF can reduce the errors of navigation position, and outperform EKF in terms of accuracy.

关 键 词：GPS/DR系统 组合导航 状态估计 扩展卡尔曼滤波 粒子滤波 

分 类 号：P228.4[天文地球—大地测量学与测量工程] TP399[天文地球—测绘科学与技术]