Section: New Results

Signal processing

Active failure detection

Participants : R. Nikoukhah, S.L. Campbell, A. Esna Ashari.

We develop a novel theory of robust active failure detection based on multi-model formulation of failures. The results of years of research have been published in a book in 2004.

We have continued to work on the extension of our approach to more general situations. Since 2008, we have mainly worked on considering the effects of feedback on our approach. This work is carried out as part of the thesis work of A. Esna Ashari. The multi-model approach is still used to model the normal and the failed systems; however the possible advantages of using linear dynamic feedback in construction of the auxiliary signal for robust fault detection is considered and the results are compared to previously developed open-loop set-ups.

In our formulation of the active fault detection problem using feedback, we cannot use as cost criterion the norm of the auxiliary signal as it was done previously because in the feedback case the auxiliary signal depends on the noise through the feedback. So, we have formulated a more general cost function by considering the worst case scenario. This type of formulation is often used in robust control theory. We have given a complete solution to the problem.

We continue the work on active failure detection by developing more efficient algorithms to be used for very large problems.

Modal analysis and diagnosis

Participant : M. Goursat.

The new results concern mainly modifications of the identification procedure to improve the robustness and the quality of the results. The domain where theoretical results are still under consideration are for more general models with influence of exogenous phenomena (noise, temperature, fluids...).

For the diagnosis the studies are focused on fast on-line detection (e.g. for flutter) and improvement of the physical diagnosis.