GPS/INS组合导航系统中的高斯粒子滤波混和算法

Gaussian particle filtering hybrid algorithm for GPS/INS integrated navigation system

采用非线性滤波器的惯性组合导航系统中,非线性滤波器的精度和实时性直接决定了惯性组合导航系统的性能。计算量和精度之间的矛盾是制约粒子滤波在GPS/INS组合导航系统中应用的主要因素。在分析高斯粒子滤波算法原理的基础上,提出了一种高斯粒子滤波混和算法,对系统线性部分采用线性递推方式,对系统非线性部分采用非线性递推方式,从而提高高斯粒子滤波精度和实时性。针对GPS/INS组合导航系统,混和算法利用卡尔曼滤波的线性递推方式进行量测更新,仿真结果表明混和算法在较少粒子条件下,相对高斯粒子滤波算法精度提高20%,滤波时间降低40%。

When a nonlinear filter is applied to inertial integrated navigation system, the precision and real-time performance of the nonlinear filter will directly determine the performance of the system. The incompatibility between computation and precision is a big obstacle for the application of particle filtering to GPS/INS integrated navigation system. In this research, a hybrid algorithm founded on Gaussian particle filter is proposed. In the new algorithm, a linear deduction will be employed for the linear part and a non-linear method will be used for the non-linear part to improve the precision and real-time performance of Gaussian particle filter. In response to the GPS/INS integrated navigation system, the hybrid algorithm measurement update is implemented by linear Kalman filter. Results from the simulation show that under the condition of fewer particles, the hybrid algorithm can improve the precision by 20% and reduce filtering time by 40%, compared with those by Gaussian particle filter algorithm.