Nonlinear Analysis: Theory, Methods & Applications
Multirate sampling and input-to-state stable receding horizon control for nonlinear systems
Introduction
Stabilization of controlled systems is one of the central topics in control theory that has led to a wealth of different stabilization methods. Among these methods, receding horizon control (RHC) strategies, also known as model predictive control (MPC), have become quite popular (see for example [5], [16] for an overview). The receding horizon method obtains the feedback control by solving a finite horizon optimal control problem at each time instant using the current state of the system as the initial state for the optimization and applying “the first part” of the optimal control. The use of digital computers in the implementation of the controllers necessitated the investigation of sampled-data systems. A sampled-data control system consists of a continuous-time plant controlled by a discrete-time (digital) controller. An overview and analysis of existing approaches for the stabilization of sampled-data systems can be found in [19], [21], [22] (see also the references of these papers).
Two main approaches of sampled-data control design can be distinguished: the first one consists in the implementation of a continuous-time stabilizing control law at a sufficiently high sampling rate, while the second way is to discretize the continuous-time model and design a stabilizing controller on the basis of the approximate discrete-time model.
Stabilization of disturbance-free sampled-data nonlinear systems via their approximate discrete-time models was presented in the recent papers [21], [22]. In these papers, sufficient conditions are presented which guarantee that the same family of controllers as stabilizes the approximate discrete-time model also practically stabilizes the exact discrete-time model of the plant. These results yield a general framework for the control design. A significant amount of work is also devoted to the study of control design methods which results in controllers satisfying the above mentioned sufficient conditions (see e.g. [6], [10] and [11]). Input-to-state stabilization and integral versions of input-to-state stabilization of sampled-data nonlinear systems with disturbances are studied by [20], [18], respectively. All of these investigations deal with the case when the sampling rates of the control function and the state measurement coincide, i.e. a single-rate approach is presented. In some applications, the sampling rates for the control function and the state measurement are different (see e.g. [1], [2], [3] and [23]). In [3], we presented a multirate version of the receding horizon algorithm for the stabilization of sampled-data nonlinear systems under a “low measurement rate” in the presence of measurement and computational delays. In this paper, we do not consider the delays; instead we consider the presence of disturbances.
When the disturbances are present, several receding horizon control approaches can be proposed. Some of them are based on the nominal prediction [17], [24], [13], [4]; others are based on the technique [9] or on min–max MPC [14], [15]. The drawback of or the min–max technique is that it becomes much more difficult to solve the optimization problem needed to generate the feedback law. In the first approach, the Lipschitz (or at least) continuity of the value function is needed. An important fact is that, in the presence of an explicit final state constraint, the continuity of the value function is not generic. One way to overcome this problem is to make an additional assumption of Lipschitz (continuity) of the value function but it is difficult to verify this condition. Therefore, in the present paper, the proposed receding horizon control method is based on the nominal prediction with an implicit final state constraint. A great majority of works deal either with continuous-time systems with sampling [4] or without taking into account any sampling [7], [8] or with discrete-time systems considering the model given directly in discrete time [9]. In [4], a sampled-data control is applied to the continuous-time system without taking into account any approximation in the plant model.
The aim of the present paper is to derive a multirate version of the receding horizon algorithm based on the nominal approximate discrete-time model, and establish that the receding horizon controller which stabilizes the nominal approximate discrete-time model also practically input-to-state stabilizes the exact discrete-time system with disturbances. The basic idea of handling the disturbances is very similar to that of [4], but, in contrast to this work, the design of the controller is based on the approximate model of the nominal system in the present paper. The importance of taking this fact into account is supported by a series of counter-examples (see e.g. [22], [21] and [10]), which show that even for disturbance-free systems one can design a controller to stabilize the approximate model, but the exact discrete-time model is destabilized by the same controller. In what follows, the notation will be used both in and ; and , and denote the usual class-, class- and class- functions (see e.g. [22]).
Section snippets
Preliminaries and problem statement
Consider a continuous-time nonlinear control system with disturbances given by where , , are the state, control input and exogenous disturbance, respectively, , with , is closed and , is compact and . Let be such that . Let be a given compact set containing the origin and consisting of all initial states to be taken into account. The system is to be controlled digitally using piecewise constant
Nominal stability of -step RHC
Consider the nominal system Let be the solution of (4) with given , and .
The nominal exact discrete-time system is given by where . We note that . Remark 2 If Assumption A1 is valid, then is continuous in and and it is locally Lipschitz.
We note that, since is typically nonlinear, in (5) is not known in most cases, and therefore the controller design can be carried out by means
Input-to-state stability
The concept of input-to-state stability has been widely used in stability analysis and control synthesis of nonlinear systems. Recently, considerable attention has been paid to the practical input-to-state stabilization of the general parametrized family of discrete-time nonlinear systems, which commonly arises in sampled-data control design (see [12], [20]). Definition 2 System (2) is practically input-to-state stabilizable (P-ISS) in about the origin with the parametrized family if there exist a
Acknowledgment
The author is grateful to Prof. E. Gyurkovics (School of Mathematics, Budapest University of Technology and Economics) for her valuable ideas and comments, and to the anonymous reviewer for constructive suggestions, which helped to improve the presentation of the paper.
References (24)
- et al.
State and output feedback nonlinear model predictive control: An overview
European Journal of Control
(2003) Receding horizon control via Bolza-type optimization
Systems & Control Letters
(1998)- et al.
Stabilization of sampled-data nonlinear systems by receding horizon control via discrete-time approximations
Automatica
(2004) - et al.
Input-to-state stability for discrete-time time-varying systems with applications to robust stabilization of systems in power form
Automatica
(2005) - et al.
Input-to-state stability of min–max MPC controllers for nonlinear systems with bounded uncertainties
Automatica
(2006) - et al.
Constrained model predictive control: Stability and optimality
Automatica
(2000) - et al.
Sufficient conditions for stabilization of sampled-data nonlinear systems via discrete-time approximation
Systems & Control Letters
(1999) - et al.
Multirate versions of sampled-data stabilization of nonlinear systems
Automatica
(2004) - et al.
Discrete-time stability with perturbations: Application to model predictive control
Automatica
(1997) Stabilization of sampled-data nonlinear systems by an -step receding horizon control based on their approximate discrete-time models
Pure Mathematics and Applications
(2003)
Cited by (13)
Discretization of nonlinear input-driven dynamical systems using the Adomian Decomposition Method
2012, Applied Mathematical ModellingCitation Excerpt :Any attempt to locally linearize these processes may cause poor identification and ultimately alternate behavior [15]. Consequently, nonlinear controller synthesis’ and monitoring system design problems have been extensively explored and researched over a range of years [16–20]. Furthermore, in light of the above considerations, the viability and performance of digitally computer-based controls and monitoring supervision strategies critically depend on the selection and properties of the underlying time-discretization methods that are applied to the original continuous-time input-driven dynamical systems and processes.
Robustness of prediction based delay compensation for nonlinear systems
2011, IFAC Proceedings Volumes (IFAC-PapersOnline)HIV dynamics: Analysis and robust multirate MPC-based treatment schedules
2009, Journal of Mathematical Analysis and ApplicationsStabilizing output feedback receding horizon control of sampled-data nonlinear systems via their approximate discrete-time models
2007, IFAC Proceedings Volumes (IFAC-PapersOnline)Attitude Sampling Tracking Control for A Rigid Spacecraft Via Backstepping
2019, 9th International Conference on Information Science and Technology, ICIST 2019Hepatitis B virus dynamics: Modeling, analysis, and optimal treatment scheduling
2013, Discrete Dynamics in Nature and Society