What is it about?

Fault estimation is an important topic, and sliding mode observers have been traditionally used to achieve it. For descriptor systems, it typically requires the system to be infinitely observable. Here, we present a method to perform fault estimation for descriptor systems that are not infinitely observable, and also present a method to design the observer so that the fault estimate will be least corrupted by other disturbances.

Featured Image

Why is it important?

This method enables fault estimation for descriptor systems that are not infinitely observable. A physical implication of infinite observability is that there needs to be a certain number of sensors available; this could increase system cost and complexity. By enabling fault estimation for non-infinitely observable systems, it could mean that we will require less sensors. In addition, our method of designing the observer to be robust against disturbances could also mean that we can reduce the number of fault channels, and model them as disturbances instead; this again could reduce the number of sensors that we need.


This is one of my first articles in contributing to the area of fault estimation of descriptor systems, and I look forward to many more in the near future.

Chee Pin Tan
Monash University

Writing this paper was an interesting journey, and the findings useful. The method used to re-express systems that are not infinitely observable as infinitely observable ones can lead to many new applications for observers.

Dr Joseph Chang Lun Chan
Chonbuk National University

Read the Original

This page is a summary of: Robust fault reconstruction for a class of infinitely unobservable descriptor systems, International Journal of Systems Science, January 2017, Taylor & Francis,
DOI: 10.1080/00207721.2017.1280552.
You can read the full text:



The following have contributed to this page