NONLINEAR OBSERVER DESIGN FOR L-V SYSTEM

Health Informatics - An International Journal (HIIJ) Vol.1, No.1, August 2012
1
NONLINEAR OBSERVER DESIGN FOR L-V SYSTEM
Sundarapandian Vaidyanathan1
1
Research and Development Centre, Vel Tech Dr. RR & Dr. SR Technical University
Avadi, Chennai-600 062, Tamil Nadu, INDIA
sundarvtu@gmail.com
ABSTRACT
This paper investigates the exponential observer problem for the Lotka-Volterra (L-V) systems. We have
applied Sundarapandian’s theorem (2002) for nonlinear observer design to solve the problem of observer
design problem for the L-V systems. The results obtained in this paper are applicable to solve the
problem of nonlinear observer design problem for the L-V models of population ecology in the food webs.
Two species L-V predator-prey models are studied and nonlinear observers are constructed by applying
the design technique prescribed in Sundarapandian’s theorem (2002). Numerical simulations are
provided to describe the nonlinear observers for the L-V systems.
KEYWORDS
Lotka-Volterra Models, Nonlinear Observers, Exponential Observers, Observability, Ecosystems.
1. INTRODUCTION
Nonlinear observer design is one of the central problems in the control systems literature. An
observer for a control system is an estimator of the state of the system, which is very useful for
implementing feedback stabilization or feedback regulation of nonlinear control systems.
The problem of designing observers for linear control systems was first introduced and solved
by Luenberger ([1], 1966). The problem of designing observers for nonlinear control systems
was first proposed by Thau ([2], 1973). There has been significant research done on the
nonlinear observer design problem over the past three decades [2-12].
A necessary condition for the existence of an exponential observer for nonlinear control systems
was obtained by Xia and Gao ([3], 1988). On the other hand, sufficient conditions for nonlinear
observer design have been obtained in the control systems literature from an impressive variety
of points of view. Kou, Elliott and Tarn ([4], 1975) obtained sufficient conditions for the
existence of exponential observers using Lyapunov-like method.
In [5-8], suitable coordinates transformations were found under which a nonlinear control
system is transformed into a canonical form, where the observer design is carried out. In [9],
Tsinias derived sufficient Lyapunov-like conditions for the existence of asymptotic observers
for nonlinear systems. A harmonic analysis approach was proposed by Celle et. al. ([10], 1989)
for the synthesis of nonlinear observers.
A complete characterization of local exponential observers for nonlinear control systems was
first obtained by Sundarapandian ([11], 2002). Sundarapandian’s theorem (2002) for nonlinear
observer design was proved using Lyapunov stability theory. In [11], necessary and sufficient
conditions were obtained for exponential observers for Lyapunov stable nonlinear systems and
an exponential observer design was provided which generalizes the linear observer design of
Luenberger [1]. Krener and Kang ([12], 2003) introduced a new method for the design of
nonlinear observers for nonlinear control systems using backstepping.

2
Lotka-Volterra (L-V) system is an important interactive model of nonlinear systems, which was
discovered independently by the Italian mathematician Vito Volterra ([13], 1926) and the
American biophysicist Alfred Lotka ([14], 1925). Recently, there has been significant interest in
the application of mathematical systems theory to population biology systems [15-16]. A survey
paper by Varga ([15], 2008) reviews the research done in this area.
This paper is organized as follows. In Section 2, we review the concept and results of
exponential observers for nonlinear systems. In Section 3, we detail the design of nonlinear
observers for two species L-V systems. In Section 4, we provide numerical examples for the
nonlinear observer design for L-V systems.
2. NONLINEAR OBSERVER DESIGN
By the concept of a state observer for a nonlinear system, it is meant that from the observation
of certain states of the system considered as outputs, it is desired to estimate the state of the
whole system as a function of time. Mathematically, we can define nonlinear observers as
follows.
Consider the nonlinear system described by
( )
x f x
=
 (1a)
( )
y h x
= (1b)
where n
x∈R is the state and p
y ∈R the output.
It is assumed that : ,
n n
f →
R R : n p
h →
R R are 1
C mappings and for some ,
n
x∗
∈R the
following hold:
( ) 0, ( ) 0.
f x h x
∗ ∗
= =
Note that the solutions x∗
of ( ) 0
f x = are called the equilibrium points of the dynamics (1a).
The linearization of the nonlinear system (1) at the equilibrium x∗
is given by
x Ax
=
 (2a)
y Cx
= (2b)
where
x x
f
A
x ∗
=
∂
 
=  
∂
 
and
x x
h
C
x ∗
=
∂
 
=  
∂
 
.
Next, the definition of nonlinear observers for the nonlinear system (1) is given as stated in [15].

3
Definition 1. (Sundarapandian, [11], 2002)
A 1
C dynamical system described by
( , ), ( )
n
z g z y z
= ∈
 R (3)
is called a local asymptotic (respectively, exponential) observer for the nonlinear system (1) if
the composite system (1) and (3) satisfies the following two requirements:
(O1) If (0) (0),
z x
= then ( ) ( ), 0.
z t x t t
= ∀ ≥
(O2) There exists a neighbourhood V of the equilibrium x∗
of n
R such that for all
(0), (0) ,
z x V
∈ the estimation error ( ) ( ) ( )
e t z t x t
= − decays asymptotically
(resp. exponentially) to zero.
Theorem 1. (Sundarapandian, [11], 2002)
Suppose that the nonlinear system (1) is Lyapunov stable at the equilibrium x∗
and that there
exists a matrix K such that A KC
− is Hurwitz. Then the dynamical system defined by
( ) [ ( )]
z f z K y h z
= + −
 (4)
is a local exponential observer for the nonlinear system (1).
Remark 1. If the estimation error is defined as ,
e z x
= − then the estimation error is
governed by
( ) ( ) [ ( ) ( )]
e f e x f x K h x e h x
= + − − + −
 (5)
Linearizing the dynamics (5) at x∗
yields the system
,
e Ee
=
 where .
E A KC
= − (6)
If ( , )
C A is observable, then the eigenvalues of E A KC
= − can be arbitrarily placed in the
complex plane and thus, a local exponential observer of the form (4) can be always found so
that the transient response of the error decays fast with any desired speed of convergence.
3. NONLINEAR OBSERVER DESIGN FOR TWO-SPECIES L-V SYSTEMS
The two species Lotka-Volterra (L-V) system consists of the following system of differential
equations
1 1 1 2
2 1 2 2 2
( )
x ax bx x
r
x cx x x M x
M
= − +
= − + −


(7)

4
In the system (7), 1( )
x t and 2 ( )
x t represent the predator and prey population densities
respectively as a function of time.
The parameters , , ,
a b c r and M are all positive, where a represents the natural decay rate of
the predator population in the absence of prey, r represents the natural growth rate of prey
population in the absence of predators, b represents the efficiency and propagation rate of the
predator in the presence of prey, c represents the effect of predation on the prey and M is the
carrying capacity.
The equilibrium points of the L-V system (7) are obtained by setting 1 2
0, 0
x x
= =
  and solving
the resulting nonlinear equations for 1
x and 2.
x
A simple calculation shows that the system (7) has three equilibrium points, viz.
(0,0), (0, )
M and ( )
1 2
, , .
r a a
x x x M
Mc b b
∗ ∗ ∗  
 
= = −
 
 
 
 
Using Lyapunov stability theory, it can be easily shown that the equilibrium points (0,0) and
(0, )
M are unstable, while the equilibrium point x x∗
= is asymptotically stable.
Figure 1 depicts the state orbits of the L-V system (7) when 1
a b c r
= = = = and 400.
M =
This figure shows that the equilibrium x x∗
= is locally asymptotically stable.
Figure1. State Orbits of the L-V System
Since only the stable equilibrium 1 2
( , )
x x x
∗ ∗ ∗
= pertains to problems of practical interest, we
consider only the problem of nonlinear observer design for the two species LV system (7) near
.
x x∗
= Since the carrying capacity M is large, it is evident that the coordinates of the
equilibrium 1 2
( , )
x x x
∗ ∗ ∗
= are positive.

5
Next, we suppose that the prey population is given as the output function of the LV system (7).
Since we know the value of 2 ,
x∗
it is convenient to take the output function as
2 2
y x x∗
= − (8)
The linearization of the plant dynamics (7) at x x∗
= is given by
1
2 2
0 bx
A r
cx x
M
∗
∗ ∗
 
 
=
 
− −
 
 
(9)
The linearization of the output function (8) at x x∗
= is given by
[ ]
0 1
C = (10)
From (9) and (10), the observability matrix for the L-V system (7) with output (8) is found as
2 2
0 1
( , )
C
C A r
CA cx x
M
∗ ∗
 
   
= =
   
− −
 
 
O
which has full rank. This shows that the system pair ( , )
C A is completely observable.
Also, we have shown that the equilibrium x x∗
= of the L-V system (7) is locally asymptotically
stable.
Thus, by Sundarapandian’s result (Theorem 1), we obtain the following result, which gives an
useful formula for the construction of nonlinear observer for the two species L-V system.
Theorem 2. The two species L-V system (7) with the output function (8) has a local exponential
observer of the form
[ ]
( ) ( ) ,
z f z K y h z
= + −
 (11)2
where K is a matrix chosen such that A KC
− is Hurwitz. Since ( , )
C A is observable, a gain
matrix K can be chosen so that the error matrix E A KC
= − has arbitrarily assigned set of
eigenvalues with negative real parts.
Example 1. Consider the two-species L-V system (7), where 0.1, 0.3, 0.7, 10
a b c r
= = = =
and 100.
M = Thus, we consider the L-V system as
1 1 1 2
2
2 2 1 2 2
0.1 0.3
10 0.7 0.1
x x x x
x x x x x
= − +
= − −


(12)
The system (12) has the stable equilibrium given by
1 2
( , ) (14.2281,0.3333)
x x x
∗ ∗ ∗
= =
We consider the output function as
2 2 2 0.3333
y x x x
∗
= − = − (13)
The linearization of the system (12)-(13) at x x∗
= is obtained as

6
0 4.2714
0.2333 0.0333
A
 
=  
− −
 
and [ ]
0 1
C =
As we have already shown for the general case, the system pair ( , )
C A is observable. Thus, the
eigenvalues of the error matrix E A KC
= − can be arbitrarily placed.
Using the Ackermann’s formula for the observability gain matrix [24], we can choose K so
that the error matrix E A KC
= − has the eigenvalues{ }
4, 4 .
− − A simple calculation gives
1
2
64.3000
7.9667
k
K
k
−
   
= =
   
 
 
By Theorem 2, a local exponential observer for the given two species L-V model (12)-(13)
around the positive equilibrium 1 2
( , ) (14.2281,0.3333)
x x
∗ ∗
= is given by
1 1 1 2 1 2
2
2 2 1 2 2 2 2
0.1 0.3 ( 0.3333)
10 0.7 0.1 ( 0.3333)
z z z z k y z
z z z z z k y z
= − + + − +
= − − + − +


(14)
Figure 2. Time History of the Estimation Error

7
Figure 3. Local Exponential Observer for the L-V System
For numerical simulation, we take the initial conditions as
15
(0)
2
x
 
=  
 
and
10
(0) .
5
z
 
=  
 
Figure 2 depicts the time history of the error estimates 1 1 1
e z x
= − and 2 2 2
e z x
= − .
Figure 3 depicts the synchronization of the states 1
x and 2
x of the L-V system (12) with the
states 1
z and 2
z of the observer system (14).
3. CONCLUSIONS
For many real problems of population ecosystems, an efficient monitoring system is of great
importance. In this paper, the methodology based on Sundarapandian’s theorem (2002) for
nonlinear observer design is suggested for the monitoring of two-species Lotka-Volterra (L-V)
population ecology systems. The results have been proved using Lyapunov stability theory.
Under the condition of stable coexistence of all the species, an exponential observer is
constructed near a non-trivial equilibrium of the Lotka-Volterra population ecology system
using Sundarapandian’s theorem (2002). A numerical example has been worked out in detail
for the local exponential observer design for the 2-species Lotka-Volterra population ecology
systems.
REFERENCES
[1] Luenberger, D.G. (1966) “Observers for multivariable systems”, IEEE Transactions on
Automatic Control, Vol. 2, pp 190-197.
[2] Thau, F.E. (1973) “Observing the states of nonlinear dynamical systems”, International J.
Control, Vol. 18, pp 471-479.
[3] Xia, X.H. & Gao, W.B. (1988) “On exponential observers for nonlinear systems”, Systems and
Control Letters, Vol. 11, pp 319-325.

8
[4] Kou, S.R., Elliott, D.L. & Tarn, T.J. (1975) “Exponential observers for nonlinear dynamical
systems”, Inform. Control, Vol. 29, pp 204-216.
[5] Krener, A.J. & Isidori, A. (1983) “Linearization by output injection and nonlinear observers”,
Systems and Control Letters, Vol. 3, pp 47-522.
[6] Krener, A.J. & Isidori, A. (1985) “Nonlinear observers with linearizable error dynamics”, SIAM
J. Control and Optimization, Vol. 23, pp 197-216.
[7] Xia, X.H. & Gao, W.B. (1988) “Nonlinear observer design by canonical form”, International J.
Control, Vol. 47, pp 1081-1100.
[8] Gauthir, J.P., Hammouri, H. & Othman, S. (1992) “A simple observer for nonlinear systems –
Applications to bioreactors”, IEEE Transactions on Automatic Control, Vol. 37, pp 875-880.
[9] Tsinias, J. (1989) “Observer design for nonlinear systems”, Systems and Control Letters, Vol.
13, pp 135-142.
[10] Celle, F., Gauthir, J.P., Kazakos, D. & Salle, G. (1989) “Synthesis of nonlinear observers: A
harmonic analysis approach,” Mathematical Systems Theory, Vol. 22, pp 291-322.
[11] Sundarapandian, V. (2002) “Local observer design for nonlinear systems”, Mathematical and
Computer Modelling, Vol. 35, pp 25-36.
[12] Krener, A.J. & Kang, W. (2003) “Locally convergent nonlinear observers”, SIAM J. Control and
Optimization, Vol. 42, pp 155-177.
[13] Volterra, V. (1926) “Fluctuations in the abundance of species considered mathematically”,
Nature, Vol. 118, pp 558-560.
[14] Lotka, A.J. (1925) Elements of Physical Biology, Williams and Wilkins, Baltimore, U.S.A.
[15] Varga, Z. (2008) “Applications of mathematical system theory in population biology”,
Periodica Mathematica Hungarica, Vol. 56, pp 157-168.
[16] Lopez, I., Gamez, M. & Molnar, S. (2007) “Observability and observers in a food web”, Applied
Mathematics Letters, Vol. 20, pp 951-957.
[17] Ogata, K. (1997) Modern Control Engineering, 3rd
edition, Prentice Hall, New Jersey, U.S.A.
Author
Dr. V. Sundarapandian obtained his Doctor of Science degree in
Electrical and Systems Engineering from Washington University, St.
Louis, USA in May 1996. He is a Professor at the R & D Centre at Vel
Tech Dr. RR & Dr. SR Technical University, Chennai, Tamil Nadu,
India. He has published over 270 papers in refereed international journals.
He has published over 170 papers in National and International
Conferences. He is the Editor-in-Chief of the AIRCC Journals -
International Journal of Instrumentation and Control Systems,
International Journal of Control Systems and Computer Modelling, and
International Journal of Information Technology, Control and
Automation. His research interests are Linear and Nonlinear Control
Systems, Chaos Theory and Control, Soft Computing, Optimal Control,
Operations Research, Mathematical Modelling and Scientific Computing.
He has delivered several Key Note Lectures on Control Systems, Chaos
Theory, Scientific Computing, Mathematical Modelling, MATLAB and
SCILAB.

NONLINEAR OBSERVER DESIGN FOR L-V SYSTEM

More Related Content

Similar to NONLINEAR OBSERVER DESIGN FOR L-V SYSTEM (20)

More from hiij (20)

Recently uploaded (20)

NONLINEAR OBSERVER DESIGN FOR L-V SYSTEM