In this paper, a new approach to estimation of unknown inputs and faults in a class of nonlinear systems is presented. The approach is based on the design of a cascade connection of two sliding mode observers. The first observer is used for the estimation of state and unknown inputs and the second is used for the fault detection and isolation. An important feature of the proposed approach is that the state trajectories do not leave the sliding manifold even in the presence of unknown inputs and faults. This allows for faults and unknown inputs to be completely reconstructed based on the information retrieved from the equivalent output injection signals. The proposed approach is tested on a nonlinear model of single link flexible joint robot system.
