Federated strong tracking filtering for nonlinear systems with multiple sensors

Abstract

In this paper, the filtering problem is investigated for nonlinear systems with multiple sensors. A federated strong tracking filter is designed to track state mutations by making full use of limited sensor information. Several subsystems are composed of different sensor combinations, and their states are independently estimated by using local filters in parallel. The strong tracking filter, as local filters, adaptively adjusts gain matrices of filters by introducing fading factors to track state mutations timely. Based on the theory of boundedness and inequality technique, the fusion estimation error is proved to be exponentially bounded in mean square. Finally, the feasibility and effectiveness of the proposed method is demonstrated by an experiment concerning the rotary steering drilling tool system.