Article ID | Journal | Published Year | Pages | File Type |
---|---|---|---|---|
695692 | Automatica | 2013 | 10 Pages |
Abstract
The detection of changes in the eigenstructure of a linear time invariant system by means of a subspace-based residual function has been proposed previously. While enjoying some success in its applicability in particular in the context of vibration monitoring, the robustness of this framework against changes in the noise properties has not been properly addressed yet. In this paper, a new robust residual is proposed and the robustness of its statistics against changes in the noise covariances is shown. The complete theory for hypothesis testing for fault detection is derived and a numerical illustration is provided.
Related Topics
Physical Sciences and Engineering
Engineering
Control and Systems Engineering
Authors
Michael Döhler, Laurent Mevel,