Designing Resilient Linear Systems

Critical systems must be designed resilient to malfunctions and especially to a loss of control authority over actuators. This malfunction considers actuators producing uncontrolled and possibly undesirable outputs. We investigate the design of resilient linear systems capable of reaching their target even after such a malfunction. In contrast with the settings considered by robust control and fault-tolerant control, we consider undesirable but observable inputs of the same magnitude as controls since they are produced by a faulty actuator belonging to the system. The control inputs can then depend on these undesirable inputs. Building on our previous work, we focus on designing resilient systems able to withstand the loss of one or multiple actuators. Since resilience refers to the existence of a control law driving the state to the target, we naturally continue with the synthesis of such a control law. We conclude with the application of our theory to the ADMIRE-fighter-jet model.