Detection and isolation of actuator and sensor faults for networked control systems

Abstract

SummaryDetection and isolation of actuator and sensor faults in presence of disturbance for a class of linear networked control systems is considered, while unknown network‐induced delay is taken into account. The network‐induced delay effect is modeled by time‐varying polytopic uncertainties. Using eigen‐structure assignment approach, a fault detection filter is designed to decouple the fault and plant disturbance, while minimizing the effect of the induced delays in the network, using H∞ and H− index theories and benefiting from the free parameters in the eigen‐structure assignment approach. If the full disturbance de‐coupling is not feasible, another filter is designed to achieve partial de‐coupling and maximum robustness against the disturbance and network‐induced delay, while sensitivity to the fault is optimized. The actuator fault and disturbance vectors are augmented to define a new disturbance vector for isolation of the sensor and actuator faults. Numerical simulations are performed to evaluate the feasibility and applicability of the proposed approach. Copyright © 2016 John Wiley & Sons, Ltd.