Abstract—An repetitive learning control strategy is presented
for a class of bilinear parametric systems with mixed unknown
time-varying and time-invariant parameters and unknown time
varying delay. By reconstructing the system equation, the
time-varying delay is combined into an unknown periodic
time-varying vector which is estimated by a periodic adaptive
mechanism. The proposed control scheme includes a
differential-difference mixed-type adaptive law used to estimate
the unknown time-varying parameters and an adaptive
repetitive learning control. Based on the Lyapunov-Krasovskii
energy function synthesis, we show that all signals in the
closed-loop system remain bounded and the asymptotic
convergence of the tracking error. The simulation results show
the feasibility and effectiveness of the approach.
I. INTRODUCTION
Learning control is considered to be the tracking of given
periodic output reference of known period in the presence of
uncertainties in systems. In terms of the nature of a system,
learning control can be classified into iterative learning control
(ILC)
[1-5]
and repetitive control (RC)
[6-9]
. Although there exist
some differences between ILC and RC, they are identical in
the basic idea, i.e., using the information obtained from
previous trial (or period) to improve the control for current
trial (or period.)
In the last three decades, both learning control approaches
have made much progress
[1-10]
. Recently, by combining
adaptive control techniques into learning control, a kind of
new adaptive iterative learning control (AILC) schemes
[2-5]
and adaptive repetitive control (ARC)
[7-10]
, have been
developed, where the uncertain parameters are estimated by
various ways to generate the current control input. For
example, the unknown parameters are estimated in the time
domain
[2]
, or iteration domain
[4-5],
Although so many
approaches have been developed in the field of learning
control, To the best of our knowledge, only a few of them are
concerning time-delay systems
[11-12]
, and they are investigated
only in the framework of classical ILC. Delay often exists in
various engineering systems, such as electrical networks,
microwave oscillators, etc. The existence of delay is a source
of instability and poor performance
[13]
. Therefore, the
controller design and stability analysis for time-delay systems,
especially for uncertain nonlinear time-delay systems, have
attracted a number of researchers over the last years, and some
interesting results have been obtained
[14]
.
* Resrach supported by NSFC Foundation under grant #61164017 and
NSFYC under grant #2010ZC069
F. Yunping Sun. Author is with the School of Information, Yunnan
Normal University, Kunming, 650092 China (corresponding author:
86-871-6594-1300; fax: 86-871-6594-1300; e-mail:sunypxd@163.com).
S. Nayu Jia. Author, is with the School of Information Technology and
Engineering, Yuxi Normal University, Yuxi, 653100 China (e-mail:
yxjny@yxnu.net).
Motivated by the aforementioned discussion, in this paper,
we will focus on solving the repetitive learning control
problem for bilinear parametric systems
with unknown time-
varying delays and time-varying parameters. From the
viewpoint of technical approach, we overcome two obstacles:
one is how to handle the unknown time-varying delays, which
is successfully solved by reconstructing the system equation
and then combining time-varying delays into an unknown
time-varying vector that is estimated by a periodic adaptive
mechanism, and the other is how to deal with the bilinear
parametric terms dependent on the unknown time-varying
delays and the unknown time-varying parameters, which is
also overcome by using an inequality and constructing
Lyapunov-Krasovskii function.
The paper is organized as follows. The problem statement,
that includes the class of bilinear parametric time-varying
systems to be considered, is presented in Section 2. In Section
3 the shows designing the control law and the periodic
adaptive law. In Section 4 exhibits the rigorous analysis of
learning convergence. a simulation example is provided to
verify the validity of control approaches In Section 5. Finally,
the work of this paper is concluded in Section 6.
Throughout this paper,
[0, )C +∞ and [0, )
n
C +∞ denote the
continuous and continuously
n -order differentiable function
space.
For a signal ( ) :
P
tR R→ , we say that
()
t
is
2
T
convergent
if lim ()() 0
t
T
tT
t
ffd
σσσ
−
→∞
=
³
.
II.
SYSTEMS DESCRIPTION AND PROBLEM FORMULATION
To clarify clearly demonstrate the underlying idea
proposed in this paper, we mainly consider the first-order
nonlinear systems with the following nonlinear plant. (for
notational convenience, arguments are sometimes omitted
when no confusion arises)
max
() ( ,) () () ( (), ( ()), )
() (), [ ,0]
txt btut fxtxt tt
xt t t
θξ θ τ
ντ
=++−
®
=∈−
¯
(1)
where ( )xt R∈ and ( )ut R∈ are the system state and the
control input, respectively; ( ) [0, )t
θ
∈+∞ is unknown
time-varying parameters with known period
0,T >
f
(,,)f is
an unknown differentiable function, ( , )xt
ξ
is a known
continuous function;
1
() [0, )tC
τ
∈+∞denotes an unknown
time- varying delay with the same period
T as ( )t
θ
䯸 which
implies that there exists a constant
max 0
τ
> such
that
max()t
ττ
≤ 䯸 [0, )t∀∈ +∞, 0
θ
> is unknown time-
Repetitive Learning Control for Bilinear Parametric Systems with
Unknown Time-Varying Delays and Time-Varying Parameters *
Yunping Sun and Nayu Jia
2392
978-1-4799-4699-0/14/
31.00©2014 IEEE
Proceedings of 2014 IEEE Chinese Guidance, Navigation and Control Conference August 8-10, 2014 Yantai, China