广义联邦滤波器的设计与实验

为了克服联邦滤波器理论中对滤波器维数的限制,基于信息滤波算法和信息分配原理,发展了广义联邦滤波器设计理论。提出了当主滤波器维数和局部滤波器维数不相同时,广义联邦滤波器达到全局最优的解析补偿方法。并将上述设计方法应用到双MIMS/GPS组合导航系统实验中。实验结果表明:经解析补偿的广义联邦滤波器的定位精度同集中Kalman滤波器的定位精度十分接近,具有全局滤波最优性。广义联邦滤波器的设计使联邦滤波理论更加完善,增强了其实用性。

Design and experiment for generalized federated filters

In order to overcome the limitation on filter dimensions in federated filter theory, this paper developed a design theory for federated filters based on the information filtering algorithm and the information sharing principle. Analytical compensation was used to optimize the global filtering when the master filter dimension does not equal the local filter dimension. The algorithm was used to design a filter for a double MIMS/GPS integrated navigation system. Tests demonstrated that the position precision of the generalized federated filter designed using analytic compensation is close to that of centralized Kalman filters, so the global filtering design is optimized.