Distributed Fault Detection and Isolation in Formation Flight of Satellites using Extended Kalman Filters

In this paper, a distributed fault detection and isolation (FDI) scheme for formation flight of satellites is developed. In our proposed approach each satellite attempts to detect actuator faults of its neighboring agents in addition to its own faults. The main challenges are on integrating and fusing the FDI information generated by all agents in order to reduce and minimize the effects of false positive and false negative alarms. The attitude dynamics of a rigid body satellite is described by a system of nonlinear equations. Therefore, an Extended Kalman Filter (EKF) scheme is utilized to generate residuals for the nonlinear formation system in presence of additive noise representing the dominant environmental disturbances in space. Every satellite in the formation estimates its neighbor’s states by using the EKF and based on the neighboring satellites dynamic model. The local or individual agent level FDI system produces residuals by using the state information of the neighbors that have been exchanged. Simulation results presented illustrate the effectiveness and capabilities of the proposed FDI design methodology for a network of formation flight satellites. Keywords-Formation flight of satellites; Extended Kalman Filter; Distributed fault diagnosis.