基于有限记忆在线预测的卡尔曼滤波突变型故障容错设计  

A Fault Tolerance Designing of Kalman-Filter for Abrupt Changing Fault Based on Finite Memorial Online Forecasting Method

作　　者：曾静[1] 周永斌 张国良[1] 杨春[1] 

机构地区：[1]第二炮兵工程学院,西安710025 [2]第二炮兵驻7102厂军代室,成都610100

出　　处：《弹箭与制导学报》2004年第S4期272-275,共4页Journal of Projectiles,Rockets,Missiles and Guidance

摘　　要：具有良好的容错性是组合导航系统的基本要求。针对弹道导弹 INS/GNSS/CNS 组合导航系统中辅助导航系统的突变型故障,提出了有限记忆在线预测滤波器残差校验的突变型故障检测方法,采用四阶自回归模型进行滤波器残差序列的预测与故障检测。这种方法可以避免采用均值进行故障检测的方法在有渐变型漂移时出现误判的情况。基于提出的故障检测方法设计了卡尔曼容错滤波算法并完成了系统仿真。仿真结果证明了文中所提出的故障检测与滤波算法的有效性。Fine fault-tolerance is fundamental for the integrated navigation system.To deal with the abrupt changing fault of the assistant navigation system of the INS/CPS/CNS integrated navigation system for ballistic missile,a detec- ting method of abrupt changing fault via filter residual error based on finite memorial online forecasting is presented, and four order AR model was' adapted for residual error forecasting of filter and fault detecting.Thus can avoid fault a- lert of mean value detecting method while the gradual changing shift.Then the algorithm of Kalman fault-tolerance fil- tering was designed based on the fault detecting method presented before.Finally,the system simulation was finished, and the results show that the fault detecting and filtering algorithm presented in this paper is valid.

关 键 词：故障检测 时间序列 卡尔曼滤波 组合导航 

分 类 号：V249.32[航空宇航科学与技术—飞行器设计]