Performing state estimation gives information about internal variables that are not measured, but are indicative of the health of a system. But does state estimation add any value to FDI? Is it possible to have faults such that they do not affect measurements, but do have a significant effect on internal states? In those cases does state estimation assist in detecting and isolating the fault?
Hello.
You should perhaps look up some linear systems theory. The ideas you talk about are formalised. Look at controllability, observability and their generalisations. Fault detection is deal with in a number of places I am aware of. You could try condition monitoring, manoeuvring target tracking and high integrity navigation systems.
My only concern is this:
All the estimator knows is the model and measurements. Since the model does not change, and the fault does not affect the measurements to a large extent, the estimated states will not be largely different from the ones obtained without any fault. How does estimation aid fault detection in such cases? Or does it, at all?
Dear Laya,
When a fault occur, sometimes the model change. in addition, a fault can affect a specific measurement (i.e. the current in electric machines - see the first attached paper).
Estimation could be considered as another way for fault detection if some measurements are not possible or induce an important cost.
You can have a look on one of our previous work I have attached for you.
Regards,
Mohamed
- Badges
- Science topic
- Similar topics
- Organizational Psychology
- Training