Fault Class Isolation in linear systems with unknown inputs

In fault isolation, it is often necessary to first make a decision on which class of fault has occurred. Such a decision requires isolating the occurred fault class amongst different possible fault classes. We shall refer to this problem as fault class isolation (FCI), and this problem is the subject of this paper. In order to address the FCI problem, a general methodology is proposed. As an application of the methodology, FCI schemes using unknown input observers are designed to achieve the isolation between actuator and sensor class of faults for certain linear systems with unknown inputs. The results demonstrate that the FCI between actuator and sensor class of faults can be achieved under certain conditions. Examples are provided to show the effectiveness of the proposed FCI schemes.