摘要
分析了平台式惯导系统 (PINS)静基座条件下的可观测性 ,采用系统可观测矩阵的条件数来定量计算PINS静基座可观测性能 ,找出该条件下的可观测矩阵条件数最小的 3个不可观测变量 .采用了自适应Kalman算法和常规Kalman算法对简化模型进行了仿真比较 .仿真结果表明前者比后者滤波收敛快 .进而提出了一种基于Elman神经网络的快速对准方法 .
The observability of stationary alignment for PINS (platform inertial navigation systems) was analyzed systemically, and condition number of observable matrix was adopted to compute quantificationally the observability of stationary alignment for PINS. Three unobservable states with the least condition number were chosen in this condition, then both adaptive Kalman filter and general Kalman filter were employed to compare for simplified model. Simulation results show that the former is faster. An alignment method based on Elman neural network was proposed.
出处
《北京航空航天大学学报》
EI
CAS
CSCD
北大核心
2002年第6期617-620,共4页
Journal of Beijing University of Aeronautics and Astronautics
关键词
对准
惯性导航
可观测性
神经网络
姿态角误差
Algorithms
Computer simulation
Kalman filtering
Mathematical models
Neural networks
