State and fault (unknown input) estimation for descriptor systems that are not infinitely observable

What is it about?

Existing methods on state and fault estimation for non-infinitely observable descriptor systems have traditionally required the removal of some states (and treating them as unknown inputs or faults); this increases the number of unknown inputs/faults which could more easily render those methods infeasible. This paper proposes a scheme to circumvent that condition.

Featured Image

Why is it important?

Through our proposed 2-observer scheme, we successfully eliminated the requirement of removing certain states (and treating them as faults/unknown inputs), and therefore avoided the increase in the number of unknown inputs (or faults) and increases the chances of the scheme being feasible. In addition, we showed that both observers can be designed independently and it will not affect the stability of the overall scheme; this is a very useful feature that eases the load of the designer.

Perspectives