Nonlinear adaptive robust observer design for a class of nonlinear systems

To be practical, observers have to function not only in the ideal situation of having a perfect assumed model structure, but also in the presence of some degrees of model uncertainties. The adaptive observer designs in the literature have a major drawback in that they may destabilize in the presence of uncertain nonlinearities, which could occur in real systems. In this paper, based on the recently proposed adaptive robust control (ARC), robust filter structures and controlled parameter adaptation are effectively integrated to deal with some typical model uncertainties in a physical system to improve the state estimation properties of an observer. Theoretically, the proposed adaptive robust observer (ARO) is shown to possess the input-to-state-practically stable (ISpS) property. Practically, explicit on-line monitoring of certain persistence of excitation conditions is used as well to obtain better and robust parameter estimates to further improve the state estimation accuracy in implementation. Experimental results are obtained to verify the effectiveness of the proposed ARO design.