Periodic event-triggered control for congested networked control systems

Abstract

With the recent development of control systems, event-triggered control (ETC) has been introduced to prevent unnecessary usage of resources, which often happens under time-based control implementations. This thesis presents a novel approach to periodic event-triggered control (PETC) that aims at reducing the number of transmissions between the controller and the sensors even further. This goal is particularly important in networked control systems (NCSs), where communication and computation resources are scarce. In this report, a relaxed triggering condition is introduced that relies on bounding the Lyapunov function of the continuous-time closed-loop system with an exponentially decaying function, rather than requiring its monotone decrease. The relaxed PETC achieves significantly less transmissions compared to existing PETC implementations. The thesis pushes the limit of event-triggered
control even further, by introducing an algorithm for a scheduler of NCS that allows to skip some of the events. This can be seen as a ‘last resort’ approach, that postpones the transmission as much as possible. It is inspired by methods used in self-triggered control (STC) and scheduling event-based NCS. Reducing the communication between the plant and the controller introduces some trade-offs that are also discussed in this report. Finally, several modifications of presented ideas are given that can be applied depending on the main objectives
on the performance of the control loop.