Federated Adaptive Kalman Filtering and its application

In order to deal with the problem in which the Federated Kalman Filtering (FKF) may be instable or divergent when noise statistics is unknown, a new federated filtering is presented, which is defined as Federated Adaptive Kalman Filtering (FAKF). A factor of modified the measurement noise covariance was built by using the ratio between filter residual and actual residual in FAKF. The adaptive estimation of FKF was realized by online modifying the measurement noise covariance. FAKF and FKF were compared using practical measuring data in inertial navigation system/global positioning system/double-star system (INS/GPS/DS) integrated navigation system. Simulation results show that FAKF has adaptability and has better estimation accuracy than the FKF when noise statistics information is unknown.