INTERNATIONAL JOURNAL OF MATHEMATICS AND SCIENTIFIC COMPUTING, VOL. 1, NO.

1, 2011
1

AbstractThe robust guaranteed cost control for uncertain
switched systems with sampled-data state feedback and time delay
is investigated in this paper. Time-varying delay technique is
provided to solve the system with sampled-data state feedback
control. The upper bound for the sampled time of state is
estimated from the proposed LMI optimization approach. A
numerical example is illustrated to show the use of the main
results.

Index TermsSwitched time-delay system, sampled-data state
feedback; guaranteed cost control, LMI optimization approach;
time-varying delay, linear fractional perturbation.

MSC 2010 Codes 93D15, 93C30, 93C57.

I. INTRODUCTION
Switched systems are often encountered in automated highway
systems, automotive engine control system, chemical process,
constrained robotics, power systems and power electronics,
robot manufacture, and stepper motors [1]-[7]. Switched
system is composed of a class of subsystems and the switching
signal is used to specify which subsystem is activated in each
instant of time. Hence many complicated phenomena are
studied and proposed in recent years [4]. Time delay is often
encountered in various practical systems; such as aircraft
stabilization, manual control, models of lasers, neural networks,
nuclear reactors, ship stabilization, and systems with lossless
transmission lines. Sampled-data input is a practical and useful
tool to implement some complicate control schemes; such as
parallel distributed control in T-S fuzzy system [8]. Suppose
that the states of system are measured by some feasible sensors,
then the state values will be held until next measured instant to
renew the state [8]-[12]. There are many researchers to consider

This work was supported by the National Science Council of Taiwan,
R.O.C. under grant no. NSC 99-2221-E-022-003.
1
Department of Marine Engineering, National Kaohsiung Marine University,
Taiwan 811, R.O.C. (Tel: 886-7-8100888 ext. 5223, email:
chlien@mail.nkmu.edu.tw)
2
Department of Applied Geoinformation, Chia Nan University of Pharmacy &
Science, Tainan, Taiwan 717, R.O.C.
3
Department of Electronic Engineering, National Quemoy University,
Kinmen, Taiwan 892, R.O.C.

this important issue for estimating sampled time 0 > T to
stabilize the systems. In the past, time-varying delay technique
had been used to represent the sampled-data input [8]-[12].
This approach provides a useful analytic tool to estimate the
upper bound of sampled time 0 > T and attain the system
performance.

H control and guaranteed cost control are two stabilization

schemes with some respective performance indices [8], [12].

H control concept was proposed to reduce the effect of the

disturbance input on the regulated output within a prescribed
level and guarantee that the closed-loop system is stable. In our
past results in [8],

H control problem of fuzzy time-delay

system with sampled-data input had been considered.
Guaranteed cost control which not only makes the closed-loop
system asymptotically stable but also guarantees an adequate
level of performance. In [12], Lyapunov-Krasovskii functional
with Leibniz-Newton formula had been used to find guaranteed
control for T-S fuzzy systems by using time-varying delay input
approach. In [13], some linear fractional perturbations are
considered for T-S fuzzy time-delay systems. In this paper, the
guaranteed cost control for switched time-delay systems with
linear fractional perturbations is considered. Some additional
nonnegative inequalities are introduced to improve the
conservativeness of the proposed results in this paper.
Notations. For a matrix A , we denote symmetric nagative
definite by 0 < A . I means the identity matrix. B A means
that the matrix A B is symmetric positive semi-definite. N
is defined by { } N , , 2 , 1 .
II. GUARANTEED COST CONTROL FOR SWITCHED SYSTEMS
WITH NONLINEAR PERTURBATION AND SAMPLED-DATA
STATE FEEDBACK
Consider a continuous-time switched system with time delay:
( ) ( ) ( ) ( ) t u B t x f t x A t x A t x

+ + + =
2 1
) ( ) ( , 0 t , (1a)
( ) ( ), t t x = [ ] 0 , t , (1b)
where ( )
n
t x is the system state, ( )
p
t u is the control
input, time delay is a nonnegative constant, is a switching
signal which is a piecewise constant function and depends on t ,
takes its values in the finite set N , and the initial vector
Robust Guaranteed Cost Control for Uncertain
Switched Time-Delay Systems with
Sampled-Data State Feedback and Linear
Fractional Perturbations
Chang-Hua Lien
1*
, Ker-Wei Yu
1
, Hao-Chin Chang
1
, Long-Yeu Chung
2
, and Jenq-Der Chen
3

INTERNATIONAL JOURNAL OF MATHEMATICS AND SCIENTIFIC COMPUTING, VOL. 1, NO. 1, 2011
2
0
C , where
0
C is the set of continuous functions from
] 0 , [ to
n
. Matrices
i
A
1
,
i
A
2
, and
i
B , are given. ( ) ( ) t x f
i

is a perturbed nonlinear function satisfying
( ) ( ) ( ) t x F t x f
i i
, { } m i , , 2 , 1 , (2)
where
i
F is a given constant matrix. Define the following
functions ( )
i
, N i , that will be used to represent our
system:
( )

=
=
otherwise, , 0
, , 1 i
i

N i , (3)
where is defined in (1). The state equation of switched
system is rewritten as
( ) ( ) ( ) ( ) ( ) [ ]

=
+ + + =
m
i
i i i i i
t u B t x f t x A t x A t x
1
2 1
) ( ) ( , 0 t ,
(4a)
( ) ( ) t t x = , [ ] 0 , t , (4b)
where ( ) ( )
i i
=
2
, ( ) ( ) j i
j i
= , 0 , and ( ) 1
1
=

=
m
i
i
.
Define the cost function of system (1) with (2) as follows:
( ) ( ) ( ) ( ) [ ]ds s u S s u s x S s x J
T T

+ =
0
2 1
, (5)
where
n n
S


1
and
p p
S


2
are two given positive definite
symmetric matrices. We wish to design a sampled-data state
feedback control in (3) and find a positive constant
*
J , such
that the system (1) with (2) is asymptotically stable and
*
J J ,
where
*
J is the guaranteed cost for this sampled-data state
feedback control in (3) of switched system (1) with (2).
The following state feedback control is used to stabilize the
switched in this paper:
( ) ( ) , , i kT x K t u
i
= = ( )T k t kT 1 + < ,
where
n p
i
K

is designed in this paper. The final feedback
control is inferred as
( ) ( ) ( ),
1

=
=
m
i
i i
kT x K t u ( )T k t kT 1 + < . (6)
In this paper, we will provide a concept to treat the original
system with sampled-data input by time-varying delay. With
(6), the sampled-data control input can be described as follows:
( ) ( ) ( ) ( )

=
=
m
i
i i
t h t x K t u
1
, ( )T k t kT 1 + < , (7a)
where ) (t h is specified by
( ) kT t t h = , ( )T k t kT 1 + < . (7b)
The system (4) can be rewritten as follows:
( ) t x ( ) ( )

= =
=
m
i
m
j
j i
1 1

( ) ( ) ( ) ( ) [ ] t h t x K B t x f t x A t x A
j i i i i
+ + ) ( ) (
2 1

( ) ( ) ( ) ( ) ( ) [ ]

= =
+ + =
m
i
m
j
i i i i i i
t h t x K B t x f t x A t x A
1 1
2 1
) ( ) ( ,
0 t , (8a)
( ) ( ) t t x = , [ ] 0 , t , m i , , 2 , 1 = . (8b)
From (6b), we have ( ) T t h < 0 , 0 t . The following lemma
will be used to design the state feedback control.
Lemma 1. (Schur complement of [14]). For a given matrix
(

=
22
12 11
* S
S S
S with
T
S S
11 11
= ,
T
S S
22 22
= , then the following
conditions are equivalent:
(1) 0 < S ,
(2) 0
22
< S , . 0
12
1
22 12 11
<
T
S S S S

Now we present a result to design the sampled-data
guaranteed cost control (7) for system (8) with (2).
Theorem 1. Suppose for a given constant 0 > , the following
LMIs :
0

*

22
12 11
>
(
(

Q
Q Q
, 0

*

22
12 11
>
(
(

R
R R
,
11 11

Q R > ,
22 2

Q P > ,
22 2

R P > , (9a)
0
1
2

P P T <

, (9b)
(
(
(
(
(
(
(
(
(
(
(
(
(

=
i
i
i
i
i i
i i
i
i i
i i i i i i i
i
99
88
77
66
57 55
47 44
33
27 26
19 18 17 15 14 12 11

* * * * * * * *
0

* * * * * * *
0 0

* * * * * *
0 0 0

* * * * *
0 0

* * * *
0 0

0 0

* * *
0 0 0 0 0 0

* *
0 0

0 0 0 0 *

0

0

0
0 0 0 0 0 0 0
0 0 0 0 0 0 0
0 0 0 0 0 0 0
0 0 0 0 0 0 0
0 0 0 0 0 0 0
0 0 0 0 0 0 0
0 0 0 0 0 0

<
(
(
(
(
(
(
(
(
(

+ , for all { } m i , , 1 , (9c)

have a solution with positive constants
i
, positive definite
symmetric matrices
0

P ,
1

P ,
2

P ,
3

P ,
22

Q ,
n n
R


22

,
n n
Q
3 3
11

,
n n
R
3 3
11

, matrices
n n
Q


3
12

,
n n
R


3
12

,
n p
i
K

where * represents the symmetric form in the matrix

and
3 1 1 0 0 1 11

P P A P P A
T
i i i
+ + + = ,
i i i
K B

12
= ,
0 2 14

P A
i i
= ,
I
i i
=
15

,
T
i i
A P
1 0 17

= ,
T
i i
F P
0 18

= ,
0 19

P
i
= ,
T
i i
K

26
= ,
T
i
T
i i
B K

27
= ,
1 33

P
i
= ,
3 44

P
i
= ,
T
i i
A P
2 0 47

= , I
i i
=
55

, I
i i
=
57
,
1
2 66

= S
i
,
0 77

P
i
= , I
i i
=
88

,
1
1 99

= S
i
,
INTERNATIONAL JOURNAL OF MATHEMATICS AND SCIENTIFIC COMPUTING, VOL. 1, NO. 1, 2011
3
[ ] [ ]
T T
Q I I I I Q R T
12 12 11

0 0

+ + =
[ ] [ ]
T T
R I I I I R
12 12

0 0

+ + . (9d)
Then the system (4) with (2) is stabilizable by sampled-data
input (6) and (7) with
1
0


= P K K
i i
and the guaranteed cost is
given by
( ) ( )

+ =
0
1 0
*
) ( ) ( 0 0
T
T T
ds s x P s x x P x J
( ) ( ) ( )


+ + +
0
3
0
2
) ( ) (

ds s x P s x ds s x P s x T s
T
T
T
, (9e)
where
1
0 0

= P P ,
1
0
1
0


= P P P P
i i
, { } 3 , 2 , 1 i .
Proof. Define the Lyapunov functional
( ) ( )


+ =
t
T t
T T
t
ds s x P s x t x P t x x V ) ( ) ( ) (
1 0
( ) ( ) ( ) ( )

+
t
T t
T
ds s x P s x T t s
2

+
t
t
T
ds s x P s x

) ( ) (
3
, (10a)
where
1
0 0

= P P ,
1
0
1
0


= P P P P
i i
, { } 3 , 2 , 1 i , are positive
definite symmetric matrices. The time derivatives of ( )
t
x V ,
along the trajectories of system (8) with (2) satisfy
) (
t
x V

( ) ( ) { ( ) ( ) ( ) ( ) [ ]} t x f t x A P t x t x P A A P t x
i i
T
m
i
T
i i
T
i
+ + + =

=

2 0
1
0 1 1 0
2 ) ( ) (
( ) ( ) ( ) ( ) ( ) { } t h t x K B P t x
j i
T
m
i
m
j
j i

= =
0
1 1
2
( ) ( ) T t x P T t x t x P t x
T T
+
1 1
) ( ) ( ( ) ( ) t x P t x T
T

2
+
( ) ( )
( )
( ) ( )
( ) (

t
t h t
T
t h t
T t
T
ds s x P s x ds s x P s x
2 2

( ) ( ) ( ) ( ) + t x P t x t x P t x
T T
3 3
. (10b)
From the definition of ( )
i
with the above equation, we have
( ) ( ) ( ) ( ) ( ) { } t h t x K B P t x
j i
T
m
i
m
j
j i

= =
0
1 1
2
( ) ( ) ( ) ( ) { }

=
=
m
i
i i
T
i
t h t x K B P t x
1
0
2 . (10c)
Now we define a vector by
( ) ( ) ( ) ( ) ( ) [ ] T t x t h t x t x t X
T T T T
= .
By Leibniz-Newton formula and LMIs (9a), the following
additional nonnegative inequalities can be introduced:
( )
( )
( )
( )
( )

t
t h t
T
ds
s x
t X
Q
Q Q
s x
t X

22
12 11
*

( ) ( ) ( ) ( ) ( ) ( ) ( ) [ ] t h t x t x Q t X t X Q t X t h
T T
+ =
12 11
2
( ) ( )
( )
0
22
+

t
t h t
T
ds s x Q s x , (10d)
( ) ( )
( )

t h t
T t
T
ds
s x
X
R
R R
s x
X

22
12 11
*

( ) [ ] ( ) ( ) ( ) ( ) ( ) ( ) [ ] T t x t h t x R t X t X R t X t h T
T T
+ =
12 11
2
( ) ( )
( )
0
22
+

t h t
T t
T
ds s x R s x , (10e)
where
, 0

0 0 0
0

0 0
0 0

0
0 0 0

0 0 0
0

0 0
0 0

0
0 0 0

*
1
0
1
0
1
0
1
0
22
12 11
1
0
1
0
1
0
1
0
22
12 11
>
(
(
(
(
(

(
(

(
(
(
(
(

=
(

P
P
P
P
Q
Q Q
P
P
P
P
Q
Q Q

. 0

0 0 0
0

0 0
0 0

0
0 0 0

0 0 0
0

0 0
0 0

0
0 0 0

*
1
0
1
0
1
0
1
0
22
12 11
1
0
1
0
1
0
1
0
22
12 11
>
(
(
(
(
(

(
(

(
(
(
(
(

=
(

P
P
P
P
R
R R
P
P
P
P
R
R R


From condition (2), we have
( ) ( ) ( ) ( ) ( ) ( ) 0 t x f t x f t x F F t x
i
T
i i
T
i
T
, { } m i , , 2 , 1 . (11)
From the input in (7), we have
( ) ( ) ( ) ( ) ( ) ( ) [ ] ( ) ( ) [ ] t h t x K S t h t x K t u S t u
j
T
i
m
i
m
j
j i
T
=

= =
2
1 1
2

( ) ( ) ( ) [ ] ( ) ( ) [ ] t h t x K S t h t x K
i
T
i
m
i
i
=

=
2
1
. (12a)
By the similar derivation of (12), condition (9b), and system (8)
with
1
0 2
1
0 2


= P P P P , we have
( ) ( ) t x P t x T
T

2
( )

=
=
m
i
i
1

( ) ( ) ( ) ( ) [ ] ( ) + +
1
0 2 1

) ( ) ( P t h t x K B t x f t x A t x A
T
i i i i i

( ) ( ) ( ) ( ) [ ] t h t x K B t x f t x A t x A
i i i i i
+ + ) ( ) (
2 1
. (12b)
From (9a) and (10)-(12), we have
( ) ( ) ( ) ( ) ( ) t u S t u t x S t x x V
T T
t 2 1
+ +

( ) ( ) ( ) ( ) ( ) ( ) ( ) [ ]

+
m
i
i
T
i i
T
i
T
i i
t x f t x f t x F F t x
1
1

( ) ( ) ( ) ( ) ( )( ) ( ) [ ] {

=

m
i
T
i ij
T
i i
t X Q R t X t h t Y t Y
1
11 11

( )( ) ( )
( )
( )( ) ( )
( )
)
`

t
t h t
T
t h t
T t
T
ds s x Q P s x ds s x R P s x
22 2 22 2

( ) ( ) ( ) [ ]

=

m
i
i i
T
i i
t Y t Y
1
, (13a)
where
( ) ( ) ( ) ( ) ( ) [ ] t x f t x t X t Y
T
i
T T T
i
= ,
INTERNATIONAL JOURNAL OF MATHEMATICS AND SCIENTIFIC COMPUTING, VOL. 1, NO. 1, 2011
4
(
(
(
(
(
(

=
i
i
i
i i i i
i
55
44
33
15 14 12 11
* * * *
0 * * *
0 0 * *
0 0 0 0 *
0

( ) ( )
T
i
i
i
i
i
i
i
i
T
i i
P S
(
(
(
(
(
(

(
(
(
(
(
(

+
(
(
(
(
(
(

(
(
(
(
(
(

57
47
27
17
1
0
57
47
27
17
26
1
1
2
26
0

0
0
0
0
0
0
0
0
0

(
(
(

+
0 0 0
0 0 0
0 0
, (13b)
1
1
3 1 0 1 1 0 11
S F F P P P A A P
i
T
i i
T
i i i
+ + + + + =

,
i i i
K B P
0 12
= ,
i i
A P
2 0 14
= ,
0 15
P
i
= ,
T
i i
A
1 17
= ,
T
i i
K =
26
,
T
i
T
i i
B K =
27
,
1 33
P
i
= ,
3 44
P
i
= ,
T
i i
A
2 47
= ,
I
i i
=
1
55
, I
i
=
57
,
[ ] [ ]
T T
Q I I I I Q R T
12 12 11
0 0 + + =
[ ] [ ]
T T
R I I I I R
12 12
0 0 + + .
Pre- and post-multiplying the matrix
ij
in (13b) by
[ ] I P P P P diag
i

1
0
1
0
1
0
1
0

[ ] 0

0 0 0 0
> = I P P P P diag
i
,
with
1
0 0

= P P ,
1
0
1
0

= P P P P
i i
,
1
0

= P K K
i i
,
we have
(
(
(
(
(
(
(

=
i
i
i
i i i i
i
55
44
33
15 14 12 11

* * * *
0

* * *
0 0

* *
0 0 0 0 *

0

( ) ( )
T
i
i
i
i
i
i
T
i i
P S
(
(
(
(
(
(

(
(
(
(
(
(

+
(
(
(
(
(
(

(
(
(
(
(
(

0
0
0

0
0
0
0

0
47
27
17
1
0
47
27
17
26
1
1
2
26

(
(
(

+
0 0 0
0 0 0
0 0

, (14)
where
0 1 0 0 0
1
11 11

P S P P F F P
i
T
i i i i
+ + =

,
kli

,
{ } 7 , , 2 , 1 , l k are defined in (9d). By Lemma 1, LMI 0

<
i

in (9c) is equivalent to 0

<
i
in (14). The condition 0

<
i
in
(14) is also equivalent to 0 <
i
in (13b) for all { } m i , , 1 .
From (11) and (13a) with the condition 0 <
i
, there exists a
0 > such that
( ) ( )
2
t x x V
t

.
We conclude that the switched system (4) with (2) is
asymptotically stabilizable by sampled-data state input in (6)
with (7). Integrating the equation in (13a) from 0 to with
0 <
i
, we have
( ) ( ) ( ) ( ) ( ) ( ) [ ] 0
0
2 1
+ +

dt t u S t u t x S t x V x V
T T
.
With ( ) 0

x V , we have
( ) ( ) ( ) ( ) [ ] ( )
*
0
2 1
J V dt t u S t u t x S t x
T T
= +

,
where
*
J is the guaranteed cost and given in (9e). The system
(4) with (2) is stabilizable by sampled-data input (6) with (7)
and
1
0


= P K K
i i
.
In the next results, the optimal guaranteed cost control for
system (4) with (2) is provided. The minimization for the cost
function in (9e) is given in the following result.
Corollary 1.
Suppose for a given constant 0 > , the following
optimization problem:
( )
3 3 3 2 2 2 1 1 1
Minimize W W W W W W trace
T T T
+ + + , (15a)
subject to
(i) (9a)-(9c),
(ii)
( )
( )
0

0
0
0
<
(
(

P x
x
T

, 0

2
1
1 0
<
(
(

+
I
I P P
,
0

2
2
2 0
<
(
(

+
I
I P P
, 0

2
3
3 0
<
(
(

+
I
I P P
, (15b)
has a solution with constants 0 > , 0 >
i
, positive definite
matrices
0

P ,
1

P ,
2

P ,
3

P ,
22

Q ,
22

R ,
1
,
2
,
n n

3
,
n n
Q
3 3
11

,
n n
R
3 3
11

, matrices
n n
Q


3
12

,
n n
R


3
12

,
n p
i
K

, where
( ) ( )
T
T
T
W W ds s x s x
1 1
0
=

, ( ) ( ) ( )
T
T
T
W W ds s x s x T s
2 2
0
= +

,
T T
W W ds s x s x
3 3
0
) ( ) ( =

. (15c)
Then the control (6) with (7) and the gain
1
0


= P K K
i i
is the
guaranteed cost control of system (4) with (2), and the
guaranteed cost is given in (9e) with
1
0 0

= P P ,
1
0
1
0


= P P P P
i i
.
Proof. By lemma 1, LMIs (15b) are equivalent to
( ) ( ) < 0 0
0
x P x
T
, 0

2
1
0
< + +

k k
P P , { } 3 , 2 , 1 k . (16)
INTERNATIONAL JOURNAL OF MATHEMATICS AND SCIENTIFIC COMPUTING, VOL. 1, NO. 1, 2011
5
Note that
[ ] [ ] 0

2

1
0 0 0
1
0
1
0
+ =

k k k k k
P P P P P , { } 3 , 2 , 1 k .
The following results are obtained from condition (16):
0

0 0
< +
k k
P P P , { } 3 , 2 , 1 k . (17)
Conditions in (17) are equivalent to

k k k
P P P P < =
1
0
1
0

, { } 3 , 2 , 1 k .
Hence we have
( ) ( ) ( ) ( ) ( ) ( ) |

\
|
= |

\
|
=


0
1
0
1
0
1
T
T
T
T
T
T
ds s x s x P trace ds s x P s x trace s x P s x

( ) ( ) ( )
1 1 1 1 1 1 1 1 1
W W trace W P W trace W W P trace
T T T
= = ,
( ) ( ) ( ) ( ) ( )
2 2 2 2 2 2
0
2
W W trace W Q W trace ds s x P s x T s
T T
T
T
= +

,
( ) ( ) ( ) ( )
3 3 3 3 3 3
0
3
W W trace W P W trace ds s x P s x
T T T
=

.
By the similar formulation of [15], we can complete this proof.
Remark 1. For a given constant 0 > , the LMI
optimization problem in (15) can be solved by the LMI Toolbox
of Matlab. Simple for loop can be used to find the
minimization of the guaranteed cost.

III. GUARANTEED COST CONTROL FOR UNCERTAIN
SWITCHED SYSTEMS WITH LINEAR FRACTIONAL
PERTURBATIONS
Consider the system (1) with linear fractional perturbations
in the following form:
( ) ( ) ( ) ( ) ( ) ( ) ( ) t u t B t x f t x t A t x t A t x

+ + + =
2 1
) ( ) ( ,
0 t , (18a)
( ) ( ), t t x = [ ] 0 , t , (18b)
where ( ) t A
i 1
, ( ) t A
i 2
, and ( ) t B
i
are some system matrices with
perturbations and satisfying
( ) ( ) ( ) [ ] t B t A t A
i i i 2 1

( ) ( ) ( ) [ ] t B B t A A t A A
i i i i i i
+ + + =
2 2 1 1
, { } m i , , 2 , 1 ,
(19a)
where constant matrices
i
A
1
,
i
A
2
, and
i
B are given, ( ) t A
i 1
,
( ) t A
i 2
, and ( ) t B
i
are some perturbed matrices and satisfying
( ) ( ) ( ) [ ] ( ) [ ]
i i i i i i i i
N N N t M t B t A t A
3 2 1 2 1
= ,
{ } m i , , 2 , 1 , (19b)
( ) ( ) [ ] ( ) t t I t
i i i i
=
1
, I
T
i i
< , (19c)
where
i
M ,
ki
N , { } 3 , 2 , 1 k , and
i
are some given constant
matrices with appropriate dimensions. ( ) t
i
is an unknown
matrix representing the parameter perturbations which satisfies
( ) ( ) I t t
i
T
i
. (19d)

Remark 2. The perturbations in (19) are the generalization
form of the parametric perturbations in [12].
Lemma 2. [13] Suppose that ( ) t
i
is defined in (19c) and
satisfying (19d), then for real matrices
i
U ,
i
W and
i
X with
T
i i
X X = , the following statements are equivalent:
(I) The inequality is satisfied
( ) ( ) 0 < + +
T
i
T
i
T
i i i i i
U t W W t U X ,
(II) There exists a scalar 0 >
i
, such that
0
* *
* <
(
(
(

I
I
W U X
i
T
i i i
T
i i i i

, (20)
where the matrix
i
is defined in (19c).
From Corollary 1 with the switched system in (18) with (2),
(19), and Lemma 2, we can obtain the following results.
Theorem 2.
Suppose for a given constant 0 > , the following
optimization problem:
( )
3 3 3 2 2 2 1 1 1
Minimize W W W W W W trace
T T T
+ + + , (21a)
subject to
(i) (9a), (9c), (15b),
(ii)
(
(

=
i
i i
i
22
12 11
~
~
*
~
~
~
~
~
~

0
0 0 0 0 0 0 0 0 0
0 0 0 0 0 0 0 0 0
0 0 0 0 0 0 0 0 0
0 0 0 0 0 0 0 0 0
0 0 0 0 0 0 0 0 0
0 0 0 0 0 0 0 0 0
0 0 0 0 0 0 0 0 0
0 0 0 0 0 0 0 0 0
0 0 0 0 0 0 0 0

<
(
(
(
(
(
(
(
(
(
(
(
(
(

+ ,
for all { } m i , , 1 , (22b)
where
(
(
(
(
(
(
(
(
(
(
(
(
(

=
i
i
i
i
i i
i i
i
i i
i i i i i i i
i
99
88
77
66
57 55
47 44
33
27 26
19 18 17 15 14 12 11
11

* * * * * * * *
0

* * * * * * *
0 0

* * * * * *
0 0 0

* * * * *
0 0

* * * *
0 0

0 0

* * *
0 0 0 0 0 0

* *
0 0

0 0 0 0 *

0

0

~
~ ,
,
*
~
~
,
0 0 0 0 0

0

0 0

0 0 0 0 0
~
~
22
411 211 111
710 110
12
(

=
(
(

=
I
I
i
T
i i i
i
T
i
T
i
T
i
T
i
T
i T
i

has a solution with constants 0 > , 0 >
i
, 0 >
i
, positive
INTERNATIONAL JOURNAL OF MATHEMATICS AND SCIENTIFIC COMPUTING, VOL. 1, NO. 1, 2011
6
definite matrices
0

P ,
1

P ,
2

P ,
3

P ,
22

Q ,
22

R ,
1
,
2
,
n n

3
,
n n
Q
3 3
11

,
n n
R
3 3
11

, matrices
n n
Q


3
12

,
n n
R


3
12

,
n p
i
K

, where matrices
1
W ,
2
W , and
3
W are
given in (15c),
kli

, { } 9 , , 2 , 1 , l k , and

are defined in (9d)

and
i i i i
M = =
710 110

,
T
i i
N P
1 0 111

= ,
T
i
T
i i
N K
3 211

= ,
T
i i
N P
2 0 411

= . Then the control (6) with (7) and the gain
1
0


= P K K
i i
is the guaranteed cost control of the switched system
in (18) with (2) and (19), and the guaranteed cost is given in (9e)
with
1
0 0

= P P ,
1
0
1
0


= P P P P
i i
.
Proof. Consider the switched system (18) with (2) and (19), the
stability LMI condition in (9b) should be rewritten as follows:
(
(
(
(
(
(
(
(
(
(
(
(
(

=
i
i
i
i
i i
i i
i
i i
i i i i i i i
i
99
88
77
66
57 55
47 44
33
27 26
19 18 17 15 14 12 11

* * * * * * * *
0

* * * * * * *
0 0

* * * * * *
0 0 0

* * * * *
0 0

* * * *
0 0
~
0 0

* * *
0 0 0 0 0 0

* *
0 0
~

0 0 0 0 *

~
0

~
0
~ ~
~
(
(
(
(
(
(
(
(
(

+
0 0 0 0 0 0 0
0 0 0 0 0 0 0
0 0 0 0 0 0 0
0 0 0 0 0 0 0
0 0 0 0 0 0 0
0 0 0 0 0 0 0
0 0 0 0 0 0

(
(
(
(
(
(
(
(
(
(
(
(
(

=
i
i
i
i
i i
i i
i
i i
i i i i i i i
99
88
77
66
57 55
47 44
33
27 26
19 18 17 15 14 12 11

* * * * * * * *
0

* * * * * * *
0 0

* * * * * *
0 0 0

* * * * *
0 0

* * * *
0 0

0 0

* * *
0 0 0 0 0 0

* *
0 0

0 0 0 0 *

0

0


( ) ( )
T
i
i
T
i
T
i
T
i
T
i
T
i
T
T
i
T
i
T
i
T
i
i
i
i
M
M
t
N P
N K
N P
N P
N K
N P
t
M
M
(
(
(
(
(
(
(
(
(
(
(
(

(
(
(
(
(
(
(
(
(
(
(
(
(

+
(
(
(
(
(
(
(
(
(
(
(
(
(

(
(
(
(
(
(
(
(
(
(
(
(

+
0
0
0
0
0
0
0
0
0
0
0
0

0
0
0
0
0

0
0
0
0
0
0
0
2 0
3
1 0
2 0
3
1 0
0
0 0 0 0 0 0 0
0 0 0 0 0 0 0
0 0 0 0 0 0 0
0 0 0 0 0 0 0
0 0 0 0 0 0 0
0 0 0 0 0 0 0
0 0 0 0 0 0

<
(
(
(
(
(
(
(
(
(

+ , (22)
where
kli

, { } 9 , , 2 , 1 , l k , and

are defined in (9d), and

( ) ( )
3 1 1 0 0 1 11

~
P P t A P P t A
T
i i i
+ + + = , ( )
i i i
K t B

~
12
= ,
( )
0 2 14

~
P t A
i i
= , ( ) t A P
T
i i 1 0 17

~
= , ( ) t B K
T
i
T
i i

~
27
= ,
( ) t A P
T
i i 2 0 47

= .
This proof can be completed in the similar formulation of
Theorem 1 and Corollary 1.
IV. NUMERICAL EXAMPLE
Consider the switched system in (18) with (2), (19), and the
following parameters:
(

=
1 . 1 4 . 0
1 . 0 2
11
A ,
(

=
1 . 1 2 . 0
0 9 . 1
12
A ,
(

=
1 0
2 . 0 1
21
A ,
(

=
1 1
0 8 . 0
22
A ,
(

=
3 . 0
1
1
B ,
(

=
4 . 0
2 . 1
2
B ,
I F F = = 1 . 0
2 1
, 1 . 0
2 1
= = , 3 = ,
(

= =
2 . 0
1 . 0
2 1
M M , [ ] 1 . 0 2 . 0
12 11
= = N N ,
[ ] 05 . 0 1 . 0
22 21
= = N N , 05 . 0
32 31
= = N N ,
(

=
1 0
0 1
1
S , 1
2
= S , ( ) [ ] 0 , 3 ,
1
1

(

= t t x . (23)
(A) If the upper bound for state sampled time is given by
1 . 0 = T , the optimization problem in Theorem 2 with 80 =
has a feasible solution
[ ] 0.019 0.004

1
= K , [ ] 0.0288 0.0049

2
= K ,
INTERNATIONAL JOURNAL OF MATHEMATICS AND SCIENTIFIC COMPUTING, VOL. 1, NO. 1, 2011
7

(

=
0.0811 0.0366 -
0.0366 - 0.0298

0
P .
The switched system in (18) with (2), (19), and (23) is
stabilizable by sampled-data state input in (6) and (7) with
[ ] 0.6015 0.8711

1
0 1 1
= =

P K K ,
[ ] 0.9604 1.3435

1
0 2 2
= =

P K K .
The guaranteed cost is given by 138.1399
*
= J . In this case,
the guaranteed cost performance can be provided via the
stabilizing robust control in (6) and (7) with the sampled time
less than 0.1 second.
(B) If the upper bound of state sampled time is given by
2 . 0 = T , the optimization problem in Theorem 2 with 28 =
has a feasible solution
[ ] 0.0175 0.0016

1
= K , [ ] 0.0308 0.0003

2
= K ,

(

=
0.0809 0.0364 -
0.0364 - 0.0296

0
P
.
The switched system in (18) with (2), (19), and (23) is
stabilizable by sampled-data state input in (6) and (7) with
[ ] 0.5391 0.7161

1
0 1 1
= =

P K K ,
[ ] 0.8445 1.0285

1
0 2 2
= =

P K K .
The guaranteed cost is given by 141.486
*
= J . In this case,
the guaranteed cost performance can be provided via the
stabilizing robust control in (6) and (7) with the sampled time
less than 0.2 second.

V. CONCLUSION
In this paper, the guaranteed cost control problem for a class
of uncertain switched time-delay system with sampled-data state
feedback has been studied. Based on the LMI optimization
approach and time-varying delay transformation technique,
some delay-dependent criteria have been proposed to minimize
the upper bound of the guaranteed cost for the system with
nonlinear and linear fractional perturbations.
ACKNOWLEDGMENT
The research reported here was supported by the National
Science Council of Taiwan, R.O.C. under grant no. NSC
99-2221-E-022-003.
REFERENCES
[1] X. M. Sun, W. Wang, G. P. Liu, and J. Zhao, Stability analysis for linear
switched systems with time-varying delay, IEEE Trans. Syst. Man,
Cybernetics, Part B, vol. 38, pp. 528533, 2008.
[2] Y. G. Sun, L. Wang, and G. Xie, Delay-dependent robust stability and
stabilization for discrete-time switched systems with mode-dependent
time-varying delays, Appl. Math. Comput., vol. 180, pp. 428-435, 2006.
[3] Y. G. Sun, L. Wang, and G. Xie, Delay-dependent robust stability and

H
control for uncertain discrete-time switched systems with
mode-dependent time delays, Appl. Math. Comput., vol. 187, pp.
1228-1237, 2007.
[4] Z. Sun and S. S. Ge, Switched linear systems control and design.
London: Spring-Verlag, 2005.
[5] D. Xie, N. Xu, and X. Chen, Stabilisability and observer-based switched
control design for switched linear systems, IET Control Theory and
Applications, vol. 2, pp. 192-199, 2008.
[6] Zhai, G., Liu, D., Lmae, J. & Kobayashi, T. (2006) Lie algebraic stability
analysis for switched systems with continuous-time and discrete-time
subsystems. IEEE Trans. Circuits Syst., 53, 152156.
[7] L. Zhang, P. Shi, and M. Basin, Robust stability and stabilisation of
uncertain switched linear discrete time-delay systems, IET Control
Theory & Applications, vol. 2, pp. 606-614, 2008.
[8] C. H. Lien, K. W. Yu, C. T. Huang, P. Y. Chou, L. Y. Chung, J. D. Chen,
Robust

H control for uncertain T-S fuzzy time-delay systems with

sampled-data input and nonlinear perturbations, Nonlinear Analysis:
Hybrid Systems, vol. 4, pp. 550-556, 2010.
[9] H. Gao and T. Chen, Stabilization of nonlinear systems under variable
sampling: a fuzzy control approach, IEEE Trans. Fuzzy Systems, vol. 15,
pp. 972-983, 2007.
[10] H. K. Lam, F. H. F. Leung, Sampled-data fuzzy controller for time-delay
nonlinear system: LMI-based and fuzzy-model-based approaches, IEEE
Trans. Syst. Man Cybern. Part B: Cybernetics, vol. 37, vol. 617-629,
2007.
[11] H. K. Lam, Stability analysis of sampled-data fuzzy controller for
nonlinear systems based on switching TS fuzzy model, Nonlinear
Analysis: Hybrid Systems, vol. 3, pp. 418-432, 2009.
[12] J. Yoneyama, Robust guaranteed cost control of uncertain fuzzy systems
under time-varying sampling, Applied Soft Computing, vol. 11, pp.
249-255, 2011.
[13] J. Yang, W. Luo, G. Li, and S. Zhong, Reliable guaranteed cost control
for uncertain fuzzy neutral systems, Nonlinear Analysis: Hybrid Systems,
vol. 4, pp. 644-658, 2010.
[14] S. P. Boyd, L. E. Ghaoui, E. Feron, V. Balakrishnan, Linear Matrix
Inequalities in System and Control Theory. Philadelphia: SIAM, 1994.
[15] C. H. Lien, Delay-dependent and delay-independent guaranteed cost
control for uncertain neutral systems with time-varying delays via LMI
approach, Chaos, Solitons & Fractals, vol. 33, pp. 1017-1027, 2007.