Fault detection and isolation for a class of nonlinear systems using an adaptive observer

In this paper we study a class of nonlinear systems with unknown inputs and uncertain parameters for which nonlinear observers with linearizable error dynamics in appropriate coordinates can be designed. Various conditions for accomplishing this task is stated. Furthermore our findings evidently encompass some results on the unknown input observers (UIO) for linear and bilinear systems. The observation scheme is then utilized as a mean for model based monitoring and failure diagnosis within the system. More specifically, a simple approach for fault detection and isolation (FDI) of actuator faults is presented. Selection of threshold value with reliability is discussed. Finally, the paper concludes with examples, illustrating applicability of the reported results in linear and nonlinear systems.