Event-Triggered Approach to Control of Nonlinear Systems

点击翻译按钮获取中文摘要

Control systems are typically implemented using a digital computer. In classical control, periodic control (i.e. sampling at regular intervals) is typically used. Recent research has raised important questions regarding the efficiency of periodic sampling in comparison with aperiodic solutions. Event-triggered systems have emerged as an attractive alternative to periodic sampling. In an event-driven system, information is exchanged between system components only when changes in the measured signals justify computing a new control action. The primary advantage of event-triggering over periodic sampling is better resource utilization by significantly reducing the rate of transfer of information between subsystems. This characteristic of event-triggered systems makes then particularly attractive in the context of Networked Control Systems (NCSs) in which the control loops are closed through a real-time (shared) network, thus limiting the amount of information that can be transmitted in real-time between system components. A second problem caused by the insertion of computer networks in control is the possibility of cyber threats which may include insider intentional threats and malicious actors. Today's threat landscape is, in fact, increasing at an alarming pace, and there is an increased urgency in protecting NCSs against cyber attacks. The long term objective of this research is to develop the foundations of a theory of control that can provide reliably solutions to the control of complex systems connected over a communication network. In such a theory: (i) Communication networks will be integrated seamlessly allowing communication between system components without risks (ii) Network congestion problems, delays, and packet dropouts will have minimal effect on system performance. (iii) Control systems will have integrity, i.e. will be able to detect cyber attacks and remain operational.Our approach consists of employing efficient solutions based on event-triggered control. Event-triggered control is in its infancy and much work is still required. We will focus on the solution of classical control problems with focus of robust solutions that can operate in the presence of modelling errors, and network effects (transmission delays, limited bandwidth, packet-dropouts), a problem that remains largely unsolved.We will also focus on secure control of nonlinear systems. Our research is aimed at exploiting the compatibility of measurements with the underlying physical process and control mechanism with the goal of providing protection against cyber attacks.