Fault tolerant tracking control for nonlinear systems based on derivative estimation

This paper presents a fault tolerant control (FTC)-architecture for trajectory tracking, where generalized actuator faults are online diagnosed and compensated for by the control system. Employing least-squares derivative estimators for identifying the faults inserts time-delays into the control loop. When a reference trajectory is to be tracked, the closed FTC loop represents a nonlinear time-varying time-delay system. Linearizing its dynamics around the reference trajectory makes it possible to determine tolerable delay times, which allows to deduce admissible intervals for the values of the FTC parameters. The FTC scheme is illustrated by simulations of an underactuated satellite with faulty actuators.