Fault Isolability Evaluation Based on Orthographic Projection for Linear System With Bounded Noise

The fault isolability evaluation of a system is the prerequisite for the realization of fault diagnosis process. To evaluate fault isolability better, a novel quantitative evaluation criterion based on orthographic projection is proposed for linear discrete systems with unknown distribution of fault and interference inputs with bounded amplitude. The output zonotopes are orthographicly projected to the subspace, transforming the zonotopes into the subspace sets. Then, this brief quantifies the intersection ratio of system output under the two fault modes, to achieve the fault isolability evaluation. The computational complexity of the
$r$
-zonotope in
$\mathbb {R}^{n}$
projected orthogonally onto
$\mathbb {R}^{m}$
and converted into polyhedron as
${\mathcal{O}}(C_{n}^{m} \cdot C_{r}^{r - m + 1} \cdot m^{2} \cdot {(m - 1)^{3} \cdot r})$
. Finally, the effectiveness and accuracy of this proposed algorithm are demonstrated through the simulation example.