Study of a robust federated filtering algorithm based on fault factor function

This paper firstly proposes an improved moving residual test with high fault detection sensitivity based on the state propagator for the federated filter. Then the fault factor function is defined to share the system information of the federated filter. With an analysis of the sensor failure's influence on the good subsystems with different information coefficients, an adaptive information sharing algorithm is presented to improve the recovery capability of the federated filter by heightening the robustness of the good subsystems. By this algorithm, the information sharing coefficients are adaptively adjusted according to the fault factors, and the faulty sensor's contamination on the good sensors via the global fusion reset is reduced. So after fault isolation, the fault information still in good subsystems will only last a very short time in the regenerated federated filter. The simulation shows this method is effective.