Robust detection of intermittent faults for linear discrete-time stochastic systems with parametric perturbations

In this paper, the robust detection problem of intermittent faults (IFs) for a class of linear discrete-time stochastic systems subject to time-varying parametric perturbations and unknown disturbances, is studied. Based on the reduced-order unknown input observer (UIO), a primary residual decoupled from perturbations and disturbances is designed and a sufficient solvability condition for the observer is provided. In order to detect all the appearing time and the disappearing time of the IF, a novel residual is constructed by introducing a sliding time window on the basis of the primary residual. Then two hypothesis tests are proposed to detect the appearing time and the disappearing time of the IF, respectively. Finally, a simulation example on a satellite is given to illustrate the effectiveness of the proposed robust detection scheme.