Robust Fault Estimator Design for Uncertain Networked Control Systems

In order to avoid production deteriorations or damages, system faults have to be identified and decisions that stop the propagation of their effects have to be made. This gives the rise to the research on fault detection and isolation (FDI) and in recent years, the problem has attracted lots of attention from researchers. Among them, the model-based approach is the common approach, see survey papers [88, 89, 90]. The prime importance [116, 117] in designing a model-based fault-detection system is the increasing robustness of residual to unknown inputs and modeling errors and enhancing the sensitivity to faults. Two approaches are mainly applied in FDI to address these two issues. One is to use the

$\mathcal H_\infty$

norm of transfer function matrix from fault to residual signal as a measure to estimate the sensitivity to the faults [119, 120]. Another method is to adopt the

$\mathcal H_\infty$

filtering formulation to make the the error between residual and fault as small as possible [118, 121]. Furthermore, the existence of time-delays is commonly encountered in dynamic systems and has to be dealt with in the realm of FDI. Some results have been obtained to address this issue, see [91]-[95]. However, these results are mostly obtained for systems with state delays.