摘要
天文导航系统是典型的非线性和噪声非高斯分布的系统.针对传统的扩展卡尔曼滤波不适于非线性和噪声非高斯分布的系统,和一般粒子滤波存在的粒子退化和采样枯竭问题,提出了一种基于遗传算法进行再采样的月球探测器自主天文导航粒子滤波新方法.计算机仿真结果显示了该方法可以有效的克服传统粒子滤波方法的缺点,提高天文导航系统的定位精度.
Autonomous celestial navigation system is a typical nonlinear, non-Gaussian dynamic system. Extended Kalman filter （EKF） is widely used in spacecraft navigation. It only uses the first order terms in the Taylor series expansion. To nonlinear and non-Gaussian system, EKF may introduce large estimation error. Particle filter（PF） is a computer-based method for implementing a recursive Bayesian filter by Monte Carlo simulations. PF is an effective solution at dealing with nonlinear and/or non-Gaussian problems. The performance of PF relies on the choice of importance sampling density and resampling scheme. To overcome the particle degeneration and sample impoverishment problems existing in traditional particle filter method, a new autonomous celestial navigation method for lunar explorer based on genetic algorithm particle filter method is presented. Simulation results demonstrat the validity and feasibility of this new method.
出处
《北京航空航天大学学报》
EI
CAS
CSCD
北大核心
2006年第11期1273-1276,共4页
Journal of Beijing University of Aeronautics and Astronautics
基金
国家自然科学基金资助项目(60574086)
新世纪优秀人才支持计划资助项目(NCET-04-0162)
关键词
月球探测
自主天文导航
粒子滤波
遗传算法
扩展卡尔曼滤波
lunar exploration
autonomous celestial navigation
particle filter
genetic algorithm
extended Kalman filter
