用于无陀螺卫星姿态确定的预测UKF算法  

Predictive Unscented Kalman Filter for Gyroless Satellite Attitude Determination

作　　者：廖鹤[1] 王本利[1] 曹正礼[2] 

机构地区：[1]哈尔滨工业大学卫星技术研究所,黑龙江哈尔滨150080 [2]沈阳飞机设计研究所,辽宁沈阳110035

出　　处：《南京理工大学学报》2011年第5期687-692,共6页Journal of Nanjing University of Science and Technology

基　　金：长江学者和创新团队发展计划(IRT0520);国家自然科学基金(60904051)

摘　　要：为了降低卫星姿态动力学模型误差和初始姿态误差对无陀螺卫星姿态确定的影响,提出了一种预测无迹卡尔曼滤波(UKF)算法。该算法利用预测滤波中的泰勒级数展开方法对未知的卫星姿态动力学模型误差进行估计,然后将其带入UKF中实现无陀螺卫星姿态确定。用三维偏差四元数代替四元数作为状态变量,避免了估计过程四元数单位约束性被破坏的现象。仿真结果表明,当较大的卫星姿态动力学模型误差和初始姿态误差同时存在时,该算法仍能精确地实现卫星姿态确定。In order to reduce the reliance of gyroless satellite attitude determination on satellite attitude dynamic model and initial attitude information,a predictive unscented Kalman filter（UKF） is presented.The Taylor series expansion approach in predictive filter is used to estimate the model uncertainty of the unknown satellite attitude dynamics.It is propagated into UKF to estimate satellite attitude.Instead of the quaternion,the generalized three-dimensional attitude representation is used as the dynamic state to maintain the quaternion normalization in the filter.The simulation results indicate that the presented filter can estimate the satellite attitude accurately even when the big model uncertainty of satellite attitude dynamics and initial attitude-error exist in the meantime.

关 键 词：模型误差 初始姿态误差 无陀螺卫星姿态确定 预测无迹卡尔曼滤波 单位约束性 

分 类 号：V448.22[航空宇航科学与技术—飞行器设计]