基于自调整粒子滤波的组合导航方法研究

在非线性模型非高斯噪声条件下，标准粒子滤波在组合系统的观测精度较低时能取得较好的滤波效果，但在高观测精度情况下会导致滤波发散。该文针对这一问题，提出一种自调整粒子滤波方法，根据观测噪声的统计大小，自适应调整似然分布的形状，使之与先验分布重叠的区域更大，有效提高滤波稳定性。将自调整粒子滤波算法应用到组合导航系统中，并在非高斯噪声、观测信息由低观测精度跳变到高观测精度条件下进行了仿真研究，结果表明，该自调整粒子滤波算法在组合导航系统具有高观测精度的情况下依然保持了滤波精度和稳定性。

Research on Integrated Navigation System Based on Self-adjust Particle Filter

In nonlinear and non-Gaussian integrated navigation system, the standard particle filter is effective when the observations are not accurate, but it is ineffective and diverges more greatly when the measurements are accurate. In this paper, a self-adjust particle filter is proposed to improve the stability of the filter. The self-adjust particle filter changes the likelihood distribution adaptively according to the statistical characteristic of the observation noises to increase the overlap area between the likelihood and prior distribution. The self-adjust particle filter is applied in non-linear and non-Gaussian integrated navigation system when the observation precision changes drastically from low to high. Simulation results show that the self-adjust particle filter is steady and accurate when the observation precision is high.