Model-based event-triggered control with time-varying network delays

Abstract

In this paper two approaches for reducing communication traffic in a control network, namely, ModelBased Networked Control Systems (MB-NCS) and eventtriggered control, are unified under a single framework. The use of a model of the plant in the controller node not only generalizes the Zero-Order-Hold (ZOH) implementation in traditional event-triggered control schemes but it also provides stability thresholds that are robust to model uncertainties. With respect to MB-NCS, the stability conditions presented here do not need explicit knowledge of the plant parameters as in previous work but are given only in terms of the parameters of the nominal model and some bounds in the model uncertainties. The resulting framework is capable of increasing the update time intervals compared with the individual approaches considered in this paper. I. INTRODUCTIONN Networked Control Systems (NCS) a digital communication network is used to transfer information among the components of a control system. NCS can also help to improve efficiency, flexibility, and reliability of the network interconnected system reducing reconfiguration and maintenance costs [1]. In contrast, the protocols used to establish an ordered communication between nodes and the number of control systems and different applications that share the communication network introduce time delays and loss of information. These situations force us to revaluate the tools that are commonly used in control design in order to account for limited feedback information in the analysis and design of NCS compared to traditional point-to-point control systems. Extensive research has been done in the area of NCS as described in [2] and references therein. Reducing the amount of communication between sensor and controller nodes without compromising the stability of the control system has been a topic of many papers. In particular, Walsh, et al.[3] introduced a network control protocol TryOnce-Discard (TOD) to allocate network resources to the different nodes in a Networked Control System, all of them may access the network at any time assuming each access occurs before the Maximum Allowable Transfer Time (MATI). The work in [4]-[5] uses more efficiently the packet structure, that is, reduction on communication is obtained by sending packets of information using all data bits available (excluding overhead) in the structure of the packet.Both authors are with the

Model-based event-triggered robust MPC/ISM

“…This block, relying on the measured state x, computes the state error e(t) =x(t) − x(t), and verifies the so-called triggering condition. In the present paper, following the suggestion in [11], we adopt a triggering condition with relative threshold. The threshold is progressively reduced as a function of the measured state, i.e.…”

“…Another valid solution is the so-called Model-Based Networked Control (see, for instance, [10] and [11]). In a Model-Based Networked Control System (MB-NCS), an explicit model of the plant is added to the controller in order to determine the control law, whenever possible, on the basis of the state of the model, rather than on the actual measurement.…”

Asynchronous Networked MPC With ISM for Uncertain Nonlinear Systems

“…In the present paper, following the suggestion in [32], we adopt a triggering condition with relative threshold. The threshold is progressively reduced as a function of the measured state, i.e.,…”

An improved self-triggered implementation for linear controllers

“…A shift in perspective was brought by the introduction of a new control paradigm known as event-triggered control byÅström and Bernhardsson [1999], Årzén [1999]. Although these references pointed to the benefits of event-triggered control in specific examples, the first systematic approach to obtain event-triggered implementations of control laws only appeared in Tabuada [2007] and was then improved by many researchers Heemels et al [2008], Cervin and Henningsson [2008], Lemmon [2011], Lunze andLehmann [2010], Mazo Jr. andTabuada [2011], Garcia and Antsaklis [2011], Dimarogonas et al [2012]. While in a traditional sampled-data paradigm (Åström and Wittenmark [1990]) new controller updates are performed periodically, regardless of the state of the system, event-triggered control is based on events triggered when stability or a pre-specified control performance is about to be lost.…”