摘要
针对广义卡尔曼滤波(extended Kalman filter,EKF)和无迹卡尔曼滤波(unscented Kalman filter,UKF)缺乏对系统异常的在线自适应调整能力、导致滤波器精度降低的问题,提出了一种将强跟踪滤波(strongtracking filter,STF)和UKF相结合的滤波算法,并进一步采用部分状态信息作为间接观测量,同时量测噪声方差阵实时调整,从而避免了对观测方程求取Jacobi矩阵的过程,使滤波器的设计得到简化。将该算法应用于航天器自主导航系统中,仿真结果表明,该算法在系统出现突变或缓变异常时,能够迅速检测出异常,在保证较高估计精度的同时,提高了系统的可靠性。
An improved filter algorithm combined strong tracking filter (STF) with unscented Kalman filter (UKF) is proposed to enhance poor performance of extended Kalman filter (EKF) and UKF in online adaptive adjustment ability and estimation accuracy when systems are abnormal. The process of solving Jacobi matrix in observer equation is avoided by deeming partial state information as indirect measurement and adjusting the measurement noise variance matrix online, which makes the filter design more simplified. The algorithm is applied to spacecraft autonomous navigation and simulation results show that when ahrupt or slow abnormalities of systems occur, the proposed algorithm can detect abnormalities rapidly, and guarantee high estimation accuracy and reliability of the system at the same time.
出处
《系统工程与电子技术》
EI
CSCD
北大核心
2011年第11期2485-2491,共7页
Systems Engineering and Electronics
基金
国家自然科学基金(60825302)资助课题
关键词
自主导航
强跟踪滤波
无迹卡尔曼滤波
间接量测
autonomous navigation rect measurement strong tracking filter (STF)
unscented Kalman filter (UKF)
indi