H∞ Dynamic observer design with application in fault diagnosis

Most observer-based methods applied in fault detection and diagnosis (FDD) schemes use the classical two-degrees of freedom observer structure in which a constant matrix is used to stabilize the error dynamics while a post filter helps to achieve some desired properties for the residual signal. In this paper, we consider the use of a more general framework which is the dynamic observer structure in which an observer gain is seen as a filter designed so that the error dynamics has some desirable frequency domain characteristics. This structure offers extra degrees of freedom and we show how it can be used for the sensor faults diagnosis problem achieving detection and estimation at the same time. The use of weightings to transform this problem into a standard H∞problem is also demonstrated.