基于矩阵摄动理论的联邦滤波信息分配方法  

Information-sharing approach to federated filtering based on matrix perturbation theory

作　　者：李金梁[1] 吴训忠[1] 张宗麟[1] 

机构地区：[1]空军工程大学工程学院三系,陕西西安710038

出　　处：《系统工程与电子技术》2007年第11期1940-1944,共5页Systems Engineering and Electronics

基　　金："十一五"惯性技术预研项目基金资助课题(51309060302)

摘　　要：针对软故障条件下系统估计精度和容错性问题,将矩阵摄动理论引入联邦滤波器,提出并证明了"全局污染上界定理"。在此基础上,提出一种引入权衡因子的自适应信息分配策略,能够根据具体应用场合灵活调整局部滤波精度和容错性,尤其适合于软故障情况下改善无故障子滤波器的抗污染能力。最后通过SINS/GPS/ADS组合导航仿真证明该方法的有效性和可行性。To analyze the system accuracy and fault tolerance under ＂soft＂ failures, the matrix perturbation theory is introduced in federated filter, and a globally contaminate maximum theorem is put forward and proved. Based on which, an adaptive information-sharing approach with a balance coefficient is presented. In terms of the occasion, the approach can neatly adjust local filters＇ accuracy and fault tolerance; especially, it is well suited to provide major improvements in good local filters＇ ability of resisting contamination under ＂soft＂ failures. Finally simulation results on SINS/GPS/ADS integrated navigation show that the validity and feasibility of the method.

关 键 词：组合导航 信息分配 矩阵摄动 软故障 联邦滤波器 

分 类 号：TP13[自动化与计算机技术—控制理论与控制工程]