Stabilization of switched nonlinear parameter-varying systems
with application to multiple controllers switching design
Hao Yang
1
, Bin Jiang
1
, Gang Tao
21
, Yuehua Cheng
3
1. College of Automation Engineering, Nanjing University of Aeronautics and Astronautics, Nanjing, 210016, China
E-mail: haoyang@nuaa.edu.cn, binjiang@nuaa.edu.cn
2. Department of Electrical and Computer Engineering, University of Virginia, Charlottesville, VA 22904-4743, USA.
E-mail: gt9s@virginia.edu
3. Academy of Frontier Science, Nanjing University of Aeronautics and Astronautics, Nanjing, 210016, China
E-mail: chengyuehua@nuaa.edu.cn
Abstract: This paper addresses stabilization issue of switched nonlinear systems where each mode may undergo both stable and
unstable behaviors over the time due to the variation of uncertain parameters. Two stabilizability conditions as well as a state-
dependent stabilization switching law are proposed. The new result is applied to the control of nonlinear plant via switching
of multiple redundant controllers in the presence of uncertain time-varying parameters in each controller. A spacecraft attitude
control example verifies the practical motivation and the efficiency of the proposed approaches.
Key Words: Switched systems; Time-varying parameter; Unstable modes; Stabilization; Switching control.
1 Introduction
Stabilization of switched nonlinear systems with unstable
modes are of particular interest in some engineering applica-
tions, e.g., supervisory control where the plant may be unsta-
ble temporarily under non-appropriate controllers [1], asyn-
chronous networked control where both the stable and un-
stable operations through the network are needed [2], multi-
agent systems where each agent may either follow the group
or not under the switching connection topology [3], etc.
In the presence of partial unstable modes, Ref. [4] and [5]
show that if the activating period of unstable modes is smal-
l enough compared with that of stable modes, the switched
system is stable. A switched nonlinear system with all un-
stable modes may still be stabilized if each mode has some
stable states and all the stable states compose the whole s-
tate space [3, 6, 7]. In [8], sufficient and necessary stabili-
ty conditions are proposed that allow each mode to have a
Lyapunov-like function which may increase within a bound
over the “switched on” time sequence of the corresponding
mode. Most of the above results consider the determined
case, i.e. each mode is always stable or unstable.
On the other hand, the analysis of switched systems would
become more complicated if there exist some time-varying
parameters in each mode [9]. These parameters can be di-
vided into two classes: C1. physical variables which can be
regulated, e,g, the speed or the attack angle in the missile
autopilot system [10], aircraft systems [11], etc. In this case,
each mode can be stabilized under appropriate control law
and switching law; C2. uncertainties which are often han-
dled by adaptive control methods [12]-[14]. However, it is
difficult to estimate parameters that vary fast or even arbi-
trarily, it is also hard to adjust controller to compensated for
these parameters if they vary, e.g., in zero dynamics or in the
controller itself.
In this paper we are interested in parameters of C2, and
This work is supported by NSF of China (61104116, 61273171,
61203091), Doctoral Fund of Ministry of Education of China
(20113218110011), and Shanghai Aerospace Science and Technology
Innovation Fund (SAST201234).
focus on a class of switched nonlinear systems where each
mode’s dynamics contains uncertain time-varying parame-
ters, whose variations may make the mode stable in some
time periods and unstable in other ones. Such switched sys-
tem is quite different from those in previous works [3]-[7]
where stability/instability of each mode is definite and is
known a priori. We provide two stabilizability conditions
under which a state-dependent stabilization switching law is
proposed. It shows that if the period in which at least one
mode is stable is long enough compared with that when all
modes are unstable, then the origin of the switched system is
asymptotically stabilizable.
As an important application of the new results, we con-
sider the control problem of nonlinear plant with uncertain
parameters in the controller, this is different from general
supervisory control problem where the uncertain parameters
exist in the plant [1], [13]. To improve the reliability of the
system, a switching scheme among a family of redundant
controllers are proposed.
In the rest of the paper, stabilization of switched systems
are discussed in Section 2. Section 3 focuses on switching
control design, which is applied to spacecraft attitude con-
trol in Section 4, followed by some concluding remarks in
Section 5.
2 Stabilization of switched nonlinear systems
2.1 Preliminaries
Considered an autonomous switched system
˙x(t)=f
σ(t)
(x(t),θ
σ(t)
(t)) (1)
where x ∈
n
are the measurable states. Define M =
{1, 2,...,m}, where m is the number of modes. σ(t):
[0, ∞) →Mdenotes the switching function. f
i
, i ∈M
are smooth functions with f
i
(0, 0)=0. θ
i
:[0, ∞) → Ω
i
⊂
n
p
denote the unknown time varying parameters of mode
i where Ω
i
is a known compact set. The function θ
i
(t) has
no unique and determined form, the initial values θ
i
(0) are
random in Ω
i
, we also do not impose any bound on
˙
θ
i
(t).
3URFHHGLQJVRIWKHQG&KLQHVH&RQWURO&RQIHUHQFH
-XO\;LDQ&KLQD
2030