平方根球形无味卡尔曼滤波机载无源定位算法

摘要

As the unscented Kalman filtering (UKF) in airborne passive location has greater impact on positioning accuracy and stability because of the initial value and numerical calculation error, an improved smoothing algorithm based on square root spherical unscented Kalman filter (Sqrt-UKFST) is presented. To guarantee that all sampling points on a unit hypersphere algorithm to improve the robustness of the initial value, the algorithm uses the unit hypersphere sphere tasteless converted by reducing the number of sampling points and the spherical radius .And the algorithm utilizes the square root matrix in the process of estimation to improve the stability of the filter. After 100 times Monte-Carlo experiments, simulation results show that the Sqrt-UKFST algorithm has better performance in the filter's stability, convergence velocity and the robustness of the initial value. When the initial state estimation reaches big error, the positioning precision maintained less than 30%.%针对机载无源定位系统中,初始值误差和数值的舍入计算对无味卡尔曼滤波(un-scented Kalman filtering,UKF)算法的定位精度和滤波稳定性影响较大的问题,提出一种基于平方根球形无味的卡尔曼滤波算法(square root spherical unscented Kalman filter,Sqrt-UKFST).该方法以单位超球体球面无味变换为基础,通过减少采样点数目和球面半径,保证所有采样点在一个单位超球体上,从而提高算法对初始值的鲁棒性,并采用平方根滤波提高算法的数值稳定性.对该算法进行100次Monte-Carlo实验,仿真结果表明,Sqrt-UKFST算法收敛速度快,滤波性能稳定;当初始状态估计误差较大时,Sqrt-UKFST算法的定位精度保持在30%以内,提高系统对初始值的鲁棒性.