Iterative learning based estimation of periodically occurring faults

In this study, the estimation problem of periodically occurring faults is investigated for a class of non-linear time-varying systems. Unlike most existing results, the system subject to bounded disturbance and measurement noise is considered. A novel non-linear iterative learning (IL) observer is constructed by using previous output estimation errors and inputs for the purpose of fault estimation (FE). The estimated fault is updated through a novel designed IL law. The recursive analysis and robust control technology are used to prove the condition under which the ultimate boundedness of the FE error can be guaranteed. Finally, a simulation study on a practical manipulator is provided to demonstrate the effectiveness and applicability of the proposed FE scheme.