UKF方法在惯性导航系统初始对准中的应用研究

Research of unscented Kalman filter in initial alignment of inertial navigation systems

针对初始对准的精度直接影响系统工作性能问题,在初始对准中引入Unscented卡尔曼滤波算法(UKF)。Unscented卡尔曼滤波算法没有把非线性系统近似为线性系统,而是使用确定性样本的方法处理非线性问题,使采样点的均值和方差完全符合实际的非线性系统的均值和方差,解决了惯性导航系统正常工作的基本条件。仿真结果显示,使用Unscented卡尔曼滤波时惯性导航系统初始对准的精度和稳定性都好于普遍使用的广义卡尔曼滤波方法。

In view of the performance of inertial navigation systems is directly influenced by the accuracy of initial alignment, the unscented Kalman filter （UKF） algorithm is drawn into initial alignment. UKF does not regard nonlinear systems as linearization, but that it takes a deterministic sampling approach to the nonlinear systems. Thus the sample points completely capture the true mean and covariance of the nonlinear systems, and the postulate that can ensure a inertial navigation system running normally is solved. Simulation results show that UKF can yield the better performance of stability and higher accuracy than EKF.