针对SINS/GPS紧耦合组合导航中的非线性融合问题,提出了一种基于高斯和粒子滤波(GSPF)的紧耦合导航方法。建立了SINS和GPS的紧耦合非线性模型,分析了GSPF算法对最优贝叶斯滤波的近似原理,并设计了基于GSPF的紧耦合融合步骤。对系统进行了仿真,结果表明在GPS可见卫星数目低于4颗时紧耦合仍可实现组合,并且GSPF在紧耦合导航中可达到较高的估计精度,当系统具有较大姿态误差时可获得比粒子滤波(PF)更好的精度和更快的收敛速度。

To resolve nonlinear fusion problem in closely-coupled SINS/GPS integrated navigation system, a data fusion method based on Gaussian sum particle filter (GSPF) was presented. The approximation principle of GSPF algorithm was analyzed from the perspective of approximation Bayesian filter. Nonlinear state models and measurement models were derived. Dynamic simulation was carried out, and the simulation results show that the closely-coupled can still navigate when the number of satellite is less than four. The GSPF approach offers excellent estimation performances in closely-coupled system, by which better estimation accuracy can be obtained over normal particle filter (PF) when system attitude errors are large.