Study of an adaptive and fault-tolerant federated filter algorithm based on RBF

With the information-sharing scheme′s influence on the performance of the federated filter, the filters with various schemes differ in fault-tolerance and estimation precision. Considering that the local filter with fault (separated) from the federated filter can not share the information, a fault-detecting method is given based on the RBF. Then an adaptive sharing scheme for the federated filter algorithm is first presented by introducing the fault (factor.) The key to the proposed method is that fault detection and information sharing are considered as a whole. The information (sharing) coefficients for the local filters without fault are adjusted adaptively, according to the change of the estimation quality. With the sharing coefficient reset zero, the local filter whose failure has been (detected) is isolated from the federated filter structure by the fault factor, and recovered to the structure after its (failure) is removed. Simulation shows that the ability of fault-tolerance and the precision of the filter are improved by this method.