Practical stability of perturbed event-driven controlled linear systems

Abstract

Many plants are regulated by digital controllers that run at a constant sampling frequency, thereby requiring a high processor load for the computations. To guarantee a good control performance, such a high sampling frequency might be required at some periods of time, but not necessarily continuously. By using an event-driven control scheme that triggers the update of the control value only when the (tracking or stabilization) error is large, the average processor load can be reduced considerably. Although event-driven control can be effective from a CPU-load perspective, the analysis of such control schemes is much more involved than that of conventional schemes and is a widely open research area. This paper investigates the control performance of an event-driven controlled continuous-time linear system with additive disturbances in terms of practical stability (ultimate boundedness). By using the derived results, the event-driven controller can be tuned to get satisfactorily transient behavior and desirable ultimate bounds, while reducing the required average processor load for its implementation. Several examples illustrate the theory