一种新的惯性导航初始对准滤波方法

Unscented卡尔曼滤波(UKF)在算法实现和估计精度方面均优于传统的扩展卡尔曼滤波(EKF)。但是当系统状态的维数比较高时，非局部的采样导致估计误差变大，此时需要采用尺度变换模式的UKF(SUKF)方法。中在惯导系统静基座初始对准的非线出虑波问题中引入SUKF，并通过仿真对比了新方法和EKF的估计效果。实验表明，新方法的收敛速度和估计精度均好于EKF。

A New Filtering Method for Initial Alignment of INS

Unscented Kalman Filter (UKF) is a new filtering method, which is superior to the EKF in accomplishment and in estimation precision. But when the dimension of the state is high, estimation errors become larger due to non-local samples. In this circumstance, it is required to use the Scaled UKF (SUKF). In this paper, the SUKF is introduced to the initial alignment filtering problem of stationary base of INS, and a simulation is made to compare the performance of the new filter with that of the EKF. The results indicate that the new filter is better than the EKF in convergence rate and estimation precision.