Federated Filter Based on Dynamic Information Allocation with Unscented Kalman Filter
Jue Huang, Bin Yan · Advances in computer science research · 2015
Because of the complex environment of the underwater noise and the difficulty for detecting the failure of the underwater sensors, a distributed filtering algorithm for the nonlinear target is proposed by applied the unscented kalman filter into the federated filter frame.Furthermore, the information allocation parameter is dynamically designed to be the ratio of the trace of the covariance matrix of the sub filter and the trace of the global covariance matrix.The soft failure sensitivity of the sub system is improved.The simulation result shows that the UKF federated filter can track the nonlinear target effectively and the soft failure sensitivity of the sub system can be identified easily.