International Journal of Engineering Sciences, 2(8) August 2013, Pages: 320-326

TI Journals

International Journal of Engineering Sciences

ISSN
2306-6474

www.tijournals.com

Parameterization of Time-Delayed Systems

Based on Passivity and LMI Approach
Hatem A. Elaydi 1, Mohamed Abu Riziq 2
1,2

Islamic University of Gaza, Gaza, Palestine.

AR TIC LE INF O

AB STR AC T

Keywords:

This paper investigates the delay dependent stability conditions of time delay systems. The method
exploited here is based on choosing a suitable Lyapunov-Krasovskii functional containingtriple
integral term in combination with passivity conditions, and to our knowledge no one have used
triple integral term with combination of the passivity conditions; thus constitute the main
contribution of this paper. Thus, constitute moreover, Jensens inequality was utilized to deal with
cross product terms that appeared when we derive the derivation of Lyapunov-Krasovskii function.
New delay dependent stability bound for particular time delay systems is derived. This is clear
through numerical example solved by convex optimization algorithm specifically by CVX toolbox
under MATLAB package.

Time delay systems

Lyapunov-Krasovskii functional
Passivity
LMI approach
Positive realness

2013 Int. j. eng. sci. All rights reserved for TI Journals.

1.

Introduction

Time delay systems (TDS), in many signals are known as dead time processes (DTP).Delays are encountered them in different branches in
control fields, such as chemical engineering systems, lag transportations, product manufactories, robotics, telecommunications, biosystems,
underwater vehicles and so on [1, 2, 3]. Time delay systems are difficult to deal with, because the presence of delays may cause the system
to be unstable or at least it degrades the performance of the control system. Delays in the systems may produce a decrease in the systems
phase and it imposes more restrictions and constraints on the system analysis and controller's design [2]. For these reasons and others the
control issues of the time delay systems are one of the most important fields that attracted the attention of many engineers and researches.
In 1957, the Smith predictor was developed to handle systems delays; however, it could not handle unstable systems. In spite of these
efforts, several problems still remain open and every year many papers are written to deal with different aspect of time delay process
control [3], and it motivates us to exploit different methods for studying the behavior of the time delay systems. In this work, state space
representation is used, in addition to linear matrix inequality and passivity theory.
In literature, there are not much surveys for time delay systems based on passivity notions and LMI approach despite the importance for
these concepts and the direct relation between passivity properties and stability criteria, and this is in turn motivates us to work on this
topic. It is true that the notion of passivity and generalization of this notion (dissipativity) date back to early 1960. The first one who
studied the concept of passivity was Popov [3] and he related this to the electrical networks which contain passive elements and does not
generate energy. A key concept of dissipative and in turn the passive systems are that of storage functions and supply rate functions [2,3,4],
and these concepts can be understood under certain conditions as a Lyapunov functions. On the other hand, express these notions in terms
of convex optimization settings such as Linear Matrix Inequality (LMI) method.
The main idea behind studying the dissipativity and passivity properties of the system is that many important physical systems have certain
input-output properties that are related to conservation, dissipation, and transport of energy [5]. This leads to so called energy based control
theory that is strongly deals with Lyapunov function which is known and may be the first example for an LMI. The most commonly used
representation for describing TDSs is functional differential equations [6].
Different approaches related to positive realness properties of time delay systems are discussed, since the positive real theory has played a
major role in stability and system theory[7,8,9]. It is well known fact that there is one to one relationship between the positive realness and
passivity theory, hence motivated us to study the passivity analysis for time delay systems. Some related works for a class of passive
systems are reported in [7,10-16].
The main contribution of this work is the further development of passive analysis of a class of time delay systems using a suitable
Lyapunov-Krasovskii functional containing triple integral term in combination with passivity conditions as mentioned above. Such systems
arise in the modeling effort of studying water quality constitutes in fresh stream [7]. We restrict attention on linear time invariant model
systems. One main result in this work that the amount of delay the system can tolerate without affecting the passivity; hence, the stability
properties of the closed loop system. The condition is given in terms of linear matrix inequalities (LMIs) and this in turn may be solved

* Corresponding author.
Email address: helaydi@iugaza.edu.ps

Parameterization of Time-Delayed Systems Based on Passivity and LMI Approach

321

Internat ional Jour nal of Engineeri ng Science s, 2(8) Au gust 2013

efficiently and readily by an efficient software packages. In our work, we proposed improved delay dependent stability criteria for linear
time delay system considered in [7].This paper is organized as follows:
Section 2, the problem is formulated, and related some lemmas are reviewed. Section 3, deals with new stability criteria for time delay
system.Section 4, gives the results of this work and compare the results with other works. Finally, section 5 concludes this paper.
Notations& Terminology: the matrix is said to be positive definite (positive semi definite) if > 0 ( 0) respectively. Similarly, it
is said to be negative (negative semi definite) if < 0 ( 0). The matrix is said to be symmetric if =
when
is the transpose

of
matrix. is the n-dimensional Euclidean space,
denotes the set of real matrices. , = ([, 0, ])denotes the
Banach space of continuous functions mapping the interval [, 0] into
symmetric matrix.

2.

represents the elements below the main diagonal of the

Problem Statement

Consider the following linear system with delay:

x (t ) A0 x (t ) A1x (t 1 ) B 1w (t )
z (t ) C 1x (t ) D11w (t )

(2.1)

y (t ) C 2 x (t ) D 2w (t )
x (t ) (t ), t 0.
Where

x (t ) R nx is the state; y (t ) R

ny

is the output measurment; w(t) R p ;exogenous input

z (t ) R nz is the controlled output;

(t) are continuous functions defined on (-,0]. A0 , A1, B1 ,C1 ,C 2 , D11,D2
given exogenous constant matrices with appropriate dimensions,
are the state delay and the control delay respectively. The system
(2.1) is approximate model for real system, namely for water quality system [7] and this is one of water quality studies on the River Nile. In
a typical model, the state variables are the concentrations of pollutants
(represented a mixture of the low-levels in the bio-strata) and
pollutant
(represented the mixture of the other levels in bio-strata). The control variables are signals proportional to the water speed and
the amount of effluent discharged into the reach at pre-selected points,for more detail see [7]. This system will be used as case study for
analysis problem. The task is to investigate delay dependent stability analysis of the system (2.1).
Definition 1:
The time delay control system (2.1) is said to be passive if

( ) ( ) > , (0, )
(2.2)

for some constant , which depends on the initial condition of the system. In addition, the systems is said to be strictly passive (SP) if it is
passive and
+
> 0.
Lemma 1 (Schur Complement) [

Where ( ) =

( )
( )

( )
> 0,
( )

( ), ( ) =

( ),

( ) depends affinely on

( ) > 0, ( ) ( ) ( )

Lemma 2 for any constant matrix

concerned are well defined, then

( )

Lemma 3[18] for any scalar

h 0

is equivalent to

( ) > 0, Or ( ) > 0, ( ) ( ) ( )

> 0, and a scalar

( )

( )

and any constant matrix

( ) > 0.

> 0, vector function : [0, ]

( )

M M T 0 the following inequality holds

T

t t
t t

t t

h2
T
x
(
u
)
Mx
(
u
)
duds

x
(
u
)
duds
M

x (u )duds

2 t h s
t h s

t h s

such that the integrations

Hatem A. Elaydi and Mohamed Abu Riziq

322

Int ernational Journal of Engi ne ering Sci ences, 2(8) Au gust 2013

3.

Main Results

In this section we proposed new stability criteria for time delay system (2.1) based on choosing for a suitable Lyapunov-Krasovskii
functional containing triple integral term.

Theorem 1: For a given positive scalar , the system (2.1) with time invariant delay is asymptotically stable and strictly passive if there
T

exist R R 0 , S S

0, P P 0 and Q Q T 0, such that the following LMI holds:

T
2 T
AT P PA Q R PA R
PB1 C1
0
0
0
A0 R
0
1
0

2 T

Q R
0
0
0
0
A1 R

T
2 T

(D11 D11 )
0
0
0
B1 R

S S
0
0

<0

S
0
0

0
2 S

3.1

Proof: consider the following Lyapunov-Krasovskii functional candidate containing a triple integral term:

V V 1 V 2 V 3 V 4
Where

V 1 x T (t )Px (t )
t

V 2 x T ( )Qx ( )d
t

V3

( )Rx ( )d ds

(3.2)

s
t

V4

t t

( )S x( )d d ds ,

t s

First, the time derivative of

can be calculated as

Second, the time derivative of

V1 x T (t ) Px (t ) x T (t ) Px (t )

or equivalently 2

can be obtained as

V2 x T (t )Qx (t ) - x T (t - 1 )Qx (t - 1 )
Third, for

the time derivative is

V3 12 xT (t )Rx (t ) 1

x T ( )Rx ( )d

t 1

Finally, by taking the derivative of

2

we get:

t t
V4 xT (t )Sx(t ) xT ( )S x( )d ds
2
2 t s
2

By using lemma 2 the upper bound of double integral term of can be calculated as shown below:
T

1 x ( ) Rx ( )d x ( )d R x ( )d
t

t 1
1

x (t ) R R x (t )

x (t 1 ) R R x (t 1 )
x T (t ) Rx (t ) 2x T (t )R (t 1 ) x T (t 1 ) Rx (t 1 ).

(3.3)

( )

( ).

Parameterization of Time-Delayed Systems Based on Passivity and LMI Approach

323

Internat ional Jour nal of Engineeri ng Science s, 2(8) Au gust 2013

Similarly, using lemma 3 an upper bound for double integral term of can be obtained as[18]:
T

t t
t t

2 t t
xT ( )S x( )d ds x( )d ds S x( )d ds
2 t s
t s
t s

x (t )
2

S
t

x (t )ds
t

x (t )
S t

S x (t )ds
t

(3.4)

and we can rewrite the previous quantity as :

t
2 T
T

x (t )Sx (t ) 2 Sx (t ) x (t )ds
t

x
(
t
)
ds

S x (t )ds
t

(3.5)

Now combine all the derivatives mentioned above we get the following:

V (x (t )) x T (t )A 0T Px (t ) x T (t - )A1T Px (t ) W T (t )B 1T Px (t )
x T (t )PA 0x (t ) x T (t ) PA1x (t - ) x T (t )PB 1W (t )
x T (t )Qx (t ) - x T (t - )Qx (t - ) x T (t ) Rx (t )
2

2
2 x Rx (t 1 ) x (t - 1 ) Rx (t - 1 ) xT (t )Sx(t )
2
2

(3.6)

t
t

2 Sx (t ) x (t )ds x (t )ds S x (t )ds

t
t
t

x T (t ) A0T P PA0 Q R x (t ) x T (t - )A1T Px (t )

+ x T (t )PA1x (t - ) W T (t )B1T Px (t ) x T (t )PBW
(t )
1
+ 2 x T (t -1 )Rx (t -1 ) 2x T Rx (t 1 ) - x T (t - )Qx (t - )
t

(3.7)

t
t
2
2 Sx (t ) x (t )ds x (t )ds S x (t )ds xT (t )Sx(t )
t
t
t
2
T

Where

A 0T RA 0 A 0T RA1 A 0T RB 1
A1T RA 0 A1T RA1 A1T RB 1
B 1T RA 0 B 1T RA1 B 1T RB 1
Now for passivity analysis and to show that the system is asymptotically stable and strictly passive (SP) we apply the passivity condition:

V (t ) 2z T (t )w (t ) x T (t ) A 0T P PA 0 Q R x (t )
x T (t - ) A1T Px (t ) 2 x T (t ) Rx (t )
x T (t ) PA1x (t - ) W T (t )(B 1T P C ) x (t )
x T (t )(PB 1 C 1T )W (t )+ 2 x T (t )Sx (t )
- x T (t - )Qx (t - ) - x T (t - ) Rx (t - )
t
T
W T (t )( D11
D11 )W (t ) 2 x T (t )S

x (t )ds

Hatem A. Elaydi and Mohamed Abu Riziq

324

Int ernational Journal of Engi ne ering Sci ences, 2(8) Au gust 2013

t
t
2
x (t )ds S x (t )ds xT (t )Sx(t )
t
t
2
x T (t )Rx (t - ) x (t - ) Rx (t )

(3.8)

Let us now write Eq.(3.8) in the compact form as follows:

x(t)

x(t )

w(t)
= x (t) (3.9)

x (s)ds
x (s)

A + PA

0
0
0

+ Q + R PA + R PB
(Q + R)

(D

0
0 0
0
C
0
0
0
0
0
0
0
+D )
0
S S

S
2

0 0 0]

(3.10)

Applying Schur complement for (3.10) we conclude that the LMI (3.1) holds.

4.

Practical Example

In this section we will present a numerical example for the water quality model for the Nile River to demonstrate the less conservativeness
of the proposed method in this paper:
Consider the system (3.1) with the following parameters:

0.3
3 2
0
A0
, A1

1 0
0.3 0.2
0.5
B 1 , C1 2 0 , D11 [2].
0.4
Using the CVX toolbox, and solving LMI (3.1) for the system we found:

A 0T P PA 0 Q R

13.5166 7.8249
7.8249 9.9649

1.1927 0.3883
0.3883 1.7824

PA1 R
Q R

PB 1 C 1T
0
0
0
T
( D11 D11 )
0

2S

0
0
0
S
S

0
0
0
0
0
2

2
S
2

=
= 10

13.1563 7.3203
7.3203 5.4082

0.8590 0.0094
0.0094 0.8652

2 A 0T R

2 A1T R
2 B 1T R

0
0

2
R

Parameterization of Time-Delayed Systems Based on Passivity and LMI Approach

325

Internat ional Journal of Engineeri ng Scie nc es, 2(8) Au gust 2013

Since all matrices are positive definite this means that the water equality model under consideration is asymptotically stable and strictly
passive for any satisfying
0 2.1898
This is confirmed by simulating the system under the influence of the initial conditions. The states of the system converge to equilibrium as
the time goes to infinity as shown in Fig. (1). It is concluded that the system is strictly passive; hence, asymptotically stable and tolerates
delay up to 2.1898 seconds.

Figure 1. State trajectories

It is obvious from Fig.1 that the delay is affected the state trajectories of the system, but the system is still passive; hence, asymptotically
stable. The example shows that Theorem (1) obtains less conservative than [7] and it relieves the constraints of stability caused by timedelay.

5.

Conclusion

In this paper, a new delay-dependent stability criterion for linear time delay systems is proposed. To obtain a less conservative result, an
appropriate Lyapunov-Krasovskii functional including triple integral term and second derivative of the original system is used with
combination with passivity conditions to improve the feasibility solution of an optimization problem in term of LMI constraints. Numerical
example has been given to demonstrate the effectiveness of proposed method and its improvement over the existing results.

References
[1]
[2]
[3]
[4]
[5]
[6]
[7]
[8]
[9]
[10]
[11]

Qing-Chang Zhong, Robust Control of Time Delay Systems,Springer, 2006.

A.S. Kloev V.S. Karpov, Synpaper of high speed regulators for time delay systems, Energoatomoizdat, Moscow, 1990.
Christian Ebenbauer, Tobias Raff and Frank Allgower, Dissipation Inequality in System Theory: An Introduction and Recent Results, Mathematics
Subject Classification, 2000.
J.E. Normey-Rico and E.F. Camacho. Control of Dead Time Processes, Springer, 2007.
JieBao and Peter L. Lee, Process Control, The passive Systems Approach, Springer, 2007.
Briat, C. Control and Observation of LPV Time-Delay Systems. Grenoble-INP. Accessed Oct 28, 2011, [online] Available:
http://www.briat.info/paper/PhDPaper.pdf
Magdi S. Mhmoud and LihuaXie, Passivity analysis and Synpaper For Uncertain Time Delay Systems, Mathematical Problems in Engineering,
Volume 7, pp. 445-484, 2001.
Anderson, B. D. O. A system Theory Criterion for Positive Real Matrices, SIAM .J. Control & Optimization, Vol. 5, pp. 171-182, 1967
Safonov, M.G. Jonckheere, E. A. Verma, M. and Limebeer, D. J. N. Synpaper for positive real multivariable feedback systems, Int. J. Control, Vol.
45, pp.817-842, 1987.
Fridman and U.Shaked Delay-Dependent Stability &H_Control: Constantand Time-Varying Delays, International Journal of Control, Vol. 76, pp.
48-60, 2003.
DimitriPeaucelle, AleksanderFradkov and Boris Andrievesky,.Robust Passification via Output Static feedback., Proc.16-th IFAC World Congress
on Automatic Control, Research supported by CNRS-RAS cooperation program No. 16394, Prague, 2005.

326

Hatem A. Elaydi and Mohamed Abu Riziq

Int ernational Journal of Engi ne ering Sci ences, 2(8) Au gust 2013

[12]
[13]
[14]
[15]
[16]
[17]
[18]

Min Gang Hua et.al. Dynamic Output Feedback Passive Control for Neutral Systems with Delay in Control Input, DCDIS Proceeding 3, Impulsive
Dynamical Systems & Applications, pp. 1687-1694, 2005.
Zho Bao Yan et. al. ThePassivity Control for a Kind of Uncertain T-S Fuzzy Descriptor Systems With Time-Delay, International Journal of
Information & Systems Sciences, Vol. 7, No. 1, pp.114-123, 2011.
Magdi S. Mahmoud. et. al. Dissipative Analysis &Synpaper of Time delay Systems, Mediterrian Journal of Measurements and Control, Vol.1, No.
3, pp-97-108, 2005.
Nichil Chopra, Passivity of Feedback Interconnected of Two Passive Systems When There are Time Varying Delays in the Communication,
Decision & Control, ISSN-0191-2216, pp. 4620-4625, Cancun, 2008.
Baozhu Du, Stability Analysis and Controller Synpaper of Continuous Time Linear Time Delay Systems, Hong Kong University, May 2010.
Accessed Nov. 2, 2011 [online], http://hdl.handle.net/10722/130913,
Boyd, S.P., Ghaoei, L.E., E.Feron, E., et al, Linear Matrix Inequalities in System and Control Theory, Philadephia, SIAM Studies on Applied
Mathematics, 1994.
Myeongjin Park, Ohmin Kwon, Juhyun Park and Sangmoon Lee, Delay Dependent Stability Criteria For Linear Time Delay System of Neutral
Type, Work Academy of Science, Engineering and Technology, 2010.