Iterative learning control for a class of singular distributed
parameter systems
Xingyu Zhou
1
, Xisheng Dai
1,3†
, Senping Tian
2
, Sange Mei
1
1. School of Electrical and Information Engineering, Guangxi University of Science and Technology, Liuzhou 545006, P. R. China
E-mail: staryuz@163.com, mathdxs@163.com, 15177720949@163.com
2. School of Automation Science and Engineering, South China University of Technology, Guangzhou 510640, P. R. China
E-mail: ausptian@scut.edu.cn
3. the Colleges and Universities Key Laboratory of Intelligent Integrated Automation, Guilin University of Electronic Technology, Guilin,
541004, P. R. China
Abstract:Thispaperaddressesiterativelearningcontrolproblemforsingulardistributedparametersystemswithparabolictype.
Owingtosingularvaluedecompositiontheory,thesingulardistributedparametersystemsaretransformedintoitsdynamicde-
compositionstandardform.Then,invirtueoftheBellman-Gronwallinequalityandcontractionmappingapproach,thelearning
convergenceofL
2
normofoutput errors hasbeenguaranteed through rigorousanalysis.Sufficient convergenDF conditionsare
providedundertwocases.Intheend,numericalsimulationsarepresentedtovalidatetheeffectivenessofP-typeILCscheme.
KeyWords: Iterativelearningcontrol;4ingulardistributedparametersystem;$onvergence
1 Introduction
Iterative learning control (ILC) is an effective intelligen-
t control strategy for repetitive operation systems in a fixed
time region, which was presented formally by Arimoto in the
midst of the 1980s. ILC has the characteristic with memory
and correction mechanism. The basic thought of ILC is to
iteratively regulate the input signal by utilizing the previous
error information, which was incorporated into the control
for the subsequent iterations, and ultimately obtain the de-
sired output target [1-3]. Nowadays, increasing attention has
been paid to applying this algorithm in industrial systems
with high tracking accuracy requirements, uncertainty and
repetitive executions [4], such as time-varying robotic sys-
tems [5], injection-molding process [6], micro electro me-
chanical systems [7], chemical batch process [8], etc.
Singular distributed parameter systems (SDPSs) are the
more generalized dynamic systems with broad engineering
background, such as singular perturbation model in power
systems [9], the temperature distribution of synthetic ther-
mal conductor [10], satellite positioning systems [11], etc.
This system is distinct from the finite-dimensional singular
system, but also different from the general distribution pa-
rameters system. On the one hand, SDPSs can regard as
an infinite-dimensional singular system. On the other hand,
the system will not only unstable, but the system structure
has changed dramatically under disturbance, such as caus-
ing the pulse behavior [12]. Since has widespread applica-
tions in practical engineering, SDPSs has attracted wide at-
tention in the past several decades and achieved some cer-
tain accomplishments. For example, the paper [13] dis-
cussed the feedback pole placement and stabilization of S-
DPSs based on operator semigroup theory. In paper [14-15],
the Fourier approach is applied in solving second order S-
DPSs. [16] structured special switching manifold to design
This work is supported by National Natural Science Foundation (NNS-
F) of China under Grant 61364006 and 61374104, the Key Laboratory
of Intelligent integrated automation of Department of Guangxi Education,
Project of Outstanding Young Teachers Training in Higher Education Insti-
tutions of Guangxi. † Corresponding author.
dynamic variable structure control in the light of separating
principle. For singular systems and distributed parameter
systems of ILC also appealed to many scholars. In [17],
the convergence conditions for linear time-invariant gener-
alized systems with P-type, PD-tyte and PID-type ILC laws
are proposed separately by using matrix inequality and con-
traction mapping approach. [18] studied D-type anticipatory
ILC law for inhomogeneous heat equations in terms of the
Banch fixed-point theorem. In paper [19-20], the design of
P-type ILC law for uncertain distributed parameter system-
s and MIMO second-order hyperbolic DPS is considered in
the light of Green formula and Gronwall-Bellman inequali-
ty. Galerkins method has been employed to divide DPS into
two subsystems which include slow part and fast part [21-
23]. Eigenspectrum-based technique is adopted for tackling
ILC problem for parabolic DPS in [24]. However, there is
no report about the ILC of the singular distributed parameter
systems to the best of my knowledge.
In this work, we extend the framework of ILC to singu-
lar distributed parameter systems with parabolic type in the
light of its dynamic decomposition standard form
[16]
. Un-
der some given assumed conditions, we divide into two sit-
uations about the direct transmission matrix N(t) whether
is the zero matrix in terms of P-type ILC algorithm and
Gronwall-Bellman inequality, as well as Green formula, the
convergence of the algorithm can be derived through rigor-
ous analysis. Finally, an illustrative numerical example is
given.
Notations: Let H =(h
1
,h
2
, ··· ,h
n
) is vector, then the
Euclidean norm of H is H =
n
i=1
h
2
i
. And if H is
matrix, H =
λ
max
(H
T
H) is its matrix norm, where
λ
max
(·) is the maximum eigenvalue of H.IfX(ξ)=
(X
1
(ξ),X
2
(ξ), ··· ,X
n
(ξ))
T
∈ R
n
∩L
2
(Ω), then X
L
2
=
{
Ω
(X
T
(ξ)X(ξ))dξ}
1
2
.Forg(ξ, t):Ω× [0,T] → R
n
,
g(·,t) ∈ R
n
∩ L
2
(Ω), ∀t ∈ [0,T], its (L
2
,λ) is defined as
g
(L
2
,λ)
=sup
0tT
{g(·,t)
2
L
2
e
−λt
}.
2017 IEEE 6th Data Driven Control and Learning Systems Conference
Ma
26-27, 2017, Chon
qin
, China
978-1-5090-5461-9/17/$31.00 ©2017 IEEE
DDCLS'17
42