小型无人机组合导航系统的平方根UKF技术研究

Square-root Unscented Kalman Filter of Small UAV Integrated Navigation System

由于小型无人机组合导航系统的误差模型与观测模型均为非线性,对模型进行线性化的扩展卡尔曼滤波会引入线性化过程中忽略高次项的误差,导致滤波器性能下降。提出将平方根UKF方法(SRUKF)用于小型无人机IMU/GPS组合导航,SRUKF方法利用协方差平方根代替协方差参加递推运算,保证滤波算法的数值稳定性,提高组合定位的精度和可靠性。仿真结果表明,SRUKF方法是组合导航中一种很好的非线性滤波方法,实现了低成本、高精度的实时定位。

The process models and measurement models of the integrated navigation system, which is applied to small UAV, are strongly nonlinear. When the effects of the higher order terms of the Taylor series expansion become significant, EKF can seriously affect the accuracy or even lead to divergence of any inference system. A new filter algorithm based on square root unscented Kaman filter （SRUKF） is presented and used to the integrated navigation system of small UAV. SRUKF takes covariance square root matrix instead of covariance one in,filter iteration, which effectively avoided filtering divergence, and meanwhile improved the convergence velocity and stability of the algorithm. The results of simulation prove that SRUKF is a more＇ideal nonlinear filtering method for the integrated navigation of small UAV and can obtain higher accuracy solutions.