Achieve asymptotic stability using Lyapunov's second method

IOSR Journal of Mathematics (IOSR-JM)
e-ISSN: 2278-5728, p-ISSN: 2319-765X. Volume 13, Issue 1 Ver. I (Jan. - Feb. 2017), PP 72-77
www.iosrjournals.org
DOI: 10.9790/5728-1301017277 www.iosrjournals.org 72 | Page
Achieve asymptotic stability using Lyapunov's second method
Runak Mohammed Saeed
(University of Kirkuk,Iraq)
Abstract: This paper discusses asymptotic stability for autonomous systems by means of the direct method of
Liapunov.Lyapunov stability theory of nonlinear systems is addressed .The paper focuses on the conditions
needed in order to guarantee asymptotic stability by Lyapunov'ssecond method in nonlinear dynamic
autonomous systems of continuous time and illustrated by examples.
Keywords:Lyapunov function, Lyapunov’s second method, asymptotic stability, autonomous nonlinear
differential system.
I. Introduction
The most useful and general approach for studying the stability of nonlinear systems is the theory
introduced in the late 19th
century by the Rusian Mathematician Alexander MikhailovichLyapunov [1,2].
Lyapunov stability is a fundamental topic in mathematics and engineering, it is a general and useful approach to
analyze the stability of nonlinear systems. Lyapunov stability concepts include two approaches: Lyapunov
indirect (first) method and Lyapunov direct(second) method. For Lyapunov indirect method the idea of system
linearization around a given point is used and one can achieve local stability with small stability regions. On the
other hand the Lyapunov direct method is the most important tool for design and analysis of nonlinear systems.
This method can be applieddirectly to a nonlinear system without the need to linearization and achieves global
stability. The fundamental concept of the Lyapunov direct method is that if the total energy of a system is
continuously dissipating, then the system will eventually reach an equilibrium point and remain at that point.
Hence, the Lyapunov direct method consists of two steps. Firstly, a suitable scalar function v(x) is chosen and
this function is referred as Lyapunov function [3,4]. Secondly, we have to evaluate its firstordertime derivative
along the trajectory of the system. If the derivative of a Lyapunov function is decreasing along the system
trajectory as time increase, then the system energy is dissipating and the system will finally settle down[5].
In this paper the tools of Lyapunov stability theory will be considered . Lyapunov's second method is
presented to achieve asymptotically stable of nonlinear systems . some examples illustrate the procedure for
studying the asymptotic stability of nonlinear system.The paper is organized as follows. In sec. 2 A brief review
of Lyapunov stability theory is presented . In sec.3 Lyapunov's methods (direct and indirect) methods is studied
. Lyapunov stability theory will desscuss to achive the main subject of the paper . Examples to illustrate The
above concept is presented in Sec .4. Concluding remarks are given in Sec. 5 .
II. A Brief Review of LyapunovStability Theory
Consider the autonomous systems
𝑥 = 𝑓 𝑥 (1)
𝑓: 𝐷 → 𝑅 𝑛
, 𝐷 =open connected subset 𝑅 𝑛
, 𝑓 locally Lipschitz , the system (1) has an equilibrium point 𝑥 ∈ 𝐷
i,e., 𝑓 𝑥 = 0 . For convenience , we state all definitions and theorems for case when the equilibrium point is at
the origin 𝑥 = 0 .
Definition1: The equilibrium point 𝑥 = 0 of (1) is
1- Stable , if for each 𝜀 > 0, there is 𝛿 = 𝛿(𝜀) > 0 such that
𝑥(0) < 𝛿 → 𝑥(𝑡) < 𝜀 for all 𝑡 ≥ 0. (2)
2- Asymptotically stable, if it is stable and 𝛿 can be chosen such that
𝑥(0) < 𝛿 → lim𝑡→∞ 𝑥 𝑡 = 0. (3)
3- Unstable, if not stable.
Definition 2 : Let 𝑉: 𝐷 → 𝑅 be a continuously differentiable function defined in a domain 𝐷 ⊂ 𝑅 𝑛
that
contains the origin , the derivative of 𝑉 along the trajectories of (1) , denoted by 𝑉(𝑥), is given by

𝑉 𝑥 =
𝑑
𝑑𝑡
𝑉 𝑥 =
𝜕𝑉
𝜕𝑥𝑖
𝑑
𝑑𝑡
𝑥𝑖
𝑛
𝑖=1
=
𝜕𝑉
𝜕𝑥1
𝜕𝑉
𝜕𝑥2
…
𝜕𝑉
𝜕𝑥 𝑛
𝑥 =
𝜕𝑉
𝜕𝑥
𝑓(𝑥). (4)
If 𝑉(𝑥) is negative , 𝑉 will decrease along the trajectory of (1) passing through 𝑥.
A function 𝑉(𝑥) is
1- Positive definite if
a- 𝑉 0 = 0 and
b- 𝑉(𝑥) > 0 for 𝑥 ≠ 0.
2- Positive semidefinite if
a- 𝑉 0 = 0 and
b- 𝑉(𝑥) ≥ 0 for 𝑥 ≠ 0.
3- Negative definite if
a- −𝑉 𝑥 is positive definite
4- Negative semidefinite if
a- −𝑉 𝑥 is positive semidefinite
Lyapunov's stability theorem states that the origin is stable if , in a domain 𝐷 that contains the origin, there is a
continuously differentiable positive definite function 𝑉(𝑥) so that 𝑉(𝑥) is negative semidefinite, and it is
asymptotically stable if 𝑉(𝑥) is negative definite, when the condition for stability is satisfied, the function
𝑉(𝑥) is called a Lyapunov function [3].
III. LyapunovDirect Method
Lyapunov’s direct method is a mathematical extension of the fundamental physical observation that an
energy dissipative system must eventually settle down to an equilibrium point. It states that if there is an energy-
like function 𝑉for a system, that is strictly decreasing along every trajectory of the system, then the trajectories
are asymptotically attracted to an equilibrium. The function 𝑉 is then said to be a Lyapunov function for the
system [9].
To prove the equilibrium is asymptotically stable we have to seek a scalar function of the states and
this function is positive definite in region around the equilibrium point :𝑉(𝑥) > 0 , except 𝑉 𝑥 = 0 . The
existence of a Lyapunov function is sufficient to prove stability in the region . If 𝑉(𝑥) is negative definite , the
equilibrium is asymptotically stable[9].
Theorem 1[10]: Let𝑥 = 0 be an equilibrium point for (1) where𝑓: 𝐷 → 𝑅 𝑛
is a locally Lipchitz and𝐷 ⊂ 𝑅 𝑛
a
domain that contains the origin. Let 𝑉: 𝐷 → 𝑅be a continuously differentiable, positivedefinite function in 𝐷
such that
𝑉 0 = 0 and 𝑉(𝑥) > 0, ∀𝑥 ∈ 𝐷{0} ,
Then 𝑥 = 0 is a stable equilibrium point, if
𝑉 𝑥 ≤ 0, ∀𝑥 ∈ 𝐷. (5)
Moreover, if
𝑉(𝑥) < 0 , ∀𝑥 ∈ 𝐷{0} , (6)
Then 𝑥 = 0 is an asymptotically stable equilibrium point.
In both cases above V is called a Lyapunov function. Moreover, if the conditions hold for all𝑥 ∈ 𝑅 𝑛
and
𝑥 → ∞implies that 𝑉(𝑥) → ∞, (7)
then 𝑥 = 0 is globally stable in (5) and globallyasymptotically stable in (6).

Proof[10]: Suppose 𝜖 > 0, choose 𝑟 ∈ 0, 𝜀 such that 𝐵𝑟 = 𝑥 ∈ 𝑅 𝑛
, 𝑥 ≤ 𝑟 ⊂ 𝐷. Let 𝛼 = min 𝑥 =𝑟 𝑉 𝑥 .
Choose 𝛽 = (0, 𝛼) and define Ω 𝛽 = 𝑥 ∈ 𝐵𝑟 , 𝑉 𝑥 ≤ 𝛽 . It holds that if 0 ∈ Ω 𝛽 → 𝑥 𝑡 ∈ Ω 𝛽 ∀𝑡 because
𝑉(𝑥 𝑡 ) ≤ 0 → 𝑉(𝑥 𝑡 ) ≤ 𝑉(𝑥 0 ) ≤ 𝛽
Further ∃ 𝛿 > 0 such that 𝑥 < 𝛿 → 𝑉 𝑥 < 𝛽. Therefore, we have that
𝛽𝛿 ⊂ Ω 𝛽 ⊂ 𝛽𝑟
And furthermore
𝑥(0) ∈ 𝛽𝛿 → 𝑥(0) ∈ Ω 𝛽 → 𝑥(𝑡) ∈ Ω 𝛽 → 𝑥(𝑡) ∈ 𝛽𝑟
Finally, it follows that
𝑥(0) < 𝛿 → 𝑥(𝑡) < 𝑟 ≤ 𝜀, ∀𝑡 > 0.
This means that the equilibrium point is stable at the point 0x .
In order to show asymptotic stability, we need to show that 𝑥(𝑡) → 0 as 𝑡 → ∞. In this case, it turns out that it is
sufficient to show that 𝑉(𝑥 𝑡 ) → 0 as 𝑡 → ∞. Since 𝑉 is monotonically decreasing and bounded from below by
0, then
𝑉 𝑥 → 𝑐 ≥ 0,as 𝑡 → ∞
Finally, it can be further shown by contradiction that the limit 𝑐 is actually equal to 0.
To prove globally stable and globally asymptotically stable
Proof[10]: Given the point 𝑥 ∈ 𝑅 𝑛
, let 𝑐 = 𝑉(𝑥). Condition (7) implies that for any > 0 , there is 𝑟 > 0 such
that 𝑉(𝑥) > 𝑐 whenever 𝑥 > 𝑟 . Thus Ω 𝛽 ⊂ 𝛽𝑟 , which implies that Ω 𝛽 bounded.
IV. Application And Illustrative Examples
Example 1 [10]:Consider the following system:
)(
)(
22
2
2
1212
2
22
2
2
111




xxxxx
xxxxx


(8)
Now we will choose Lyapunov function )(2/1)( 2
2
2
1 xxxV  we have
)(.)( xfxV 
 21, xx  T
xxxxxxxx )(,)( 22
2
2
121
2
2
22
2
2
11  
)()( 22
2
2
1
2
2
22
2
2
1
2
1   xxxxxx
))( 22
2
2
1
2
2
2
1  xxxx
Since, 0)( xV and 0)( xV , provided that
22
2
2
1 )( xx  , it follows that the origin is an asymptotically
stable equilibrium point .
Example 2 [9]:Consider the simple pendulum (pendulum with friction), (Figure 3.1), k is a coefficient of
friction, l denotes the length of the rod and mdenotes the mass of the bob. Let  denote the angle subtended
by the rod and the vertical axis through the pivot point. The gravitational force equal to mg , where g is the
acceleration due to gravity. Using Newton's second law of motion and take the state variables as 1x and
2x . Therefore the state equations are

212
21
sin x
m
k
x
l
g
x
xx




)(
)(
2
1
xf
xf


(9)
where 0x is an equilibrium point , to study the stability of the equilibrium at the origin we propose a
Lyapunov function candidate )(xV . In this case we use total energy )(xE which is a positive function as the
sum of its potential and kinetic energies and 0)0( E , we get
PKE  (Kinetic plus potential energy )
= mghwlm 2
)(
2
1
Where
2xw 
)cos1()cos1( 1xllh  
Finally,
E= )cos1(
2
1
1
2
2
2
xmglxml 
We know define )(xV E (positive definite) as
)cos1(
2
1
)( 1
2
2
2
xmglxmlxV  ,
and the energy is
 
1
0
2
21
2
2
2
1
)cos1(
2
1
sin)()(
x
xxaxyaxVxE
Thus the derivative of )(xV is
T
x
m
k
x
l
g
xxmlxmglxV ]sin,][,sin[)( 2122
2
1  2
2
2
xkl
As we saw )(xV is negative semi-definite (not negative definite) because at 0)( xV we find 02 x
regardless of the value of 1x (thus 0)( xV along the 1x axis), this means that the origin is stable by
(theorem 1), not asymptotic stability.
Here Lyapunov function candidate fails to indentify an asymptotically stable equilibrium point by having )(xV
negative semi definite.
In order to prove an asymptotical stability we need to define LaSalle’s Invariance Principle.
LaSalle’s Invariance Principle , developed in 1960 by J.P. LaSalle, the principle Basically verify that if there
is a Lyapunov function within the neighborhood of the origin , has a negative semi-definite time derivative
along the trajectories of the system which established that no trajectory can stay identically at point where
0)( xV except at the origin , then the origin is asymptotically stable .In order to understand that we present a
definition and theorems related to LaSalle’s Invariance Principle[7] .
Definition 3 [5]: A set M is said to be an invariant set with respect to the system (1) if Mx )0( 
Mtx )( 
 Rt .
Theorem 3 [10]:(LaSalle's theorem): Let RDV : be a continuously differentiable function and assume
that
i) DM  is a compact set, invariant with respect to the solutions of (1).
ii) 0V in M
iii)  0,::  VandMxxE  ; that is, E is the set of all points of M such that 0V
iv) :N is the largest invariant set in E .
Then every solution starting in M approaches N as t .

Proof [10]:Consider a solution )(tx in (1) starting in M . Since MxV  0)( , )(xV is a decreasing
function of t . Also, since (.)V is a continuous function, it is bounded from below in the compact set M . It
follows that ))(( txV has a limit as t . Let  be the limit set of this trajectory. It follows that M
since M is (an invariant) closed set. For any p  a sequence nt with nt and ptx n )( . By
continuity of )(xV , we have that
)(pV
n
lim atxV n ))(( ( a constant)
Hence, axV )( on  . Also, consider  is an invariant set, and moreover 0)( xV on  (since )(xV
is constant on  ). It follows that
MEN 
Since )(tx is bounded this implies that )(tx approaches  ( its positive limit set ) as t .Hence )(tx
approaches N as t .
Corollary [9]: Let 𝑥 = 0 ∈ 𝐷 be an equilibrium point of the system (1). Let 𝑉: 𝐷 → 𝑅 be a continuously
differentiable positive definite function on the domain 𝐷, such that 𝑉 ≤ 0, ∀𝑥 ∈ 𝐷. Let 𝑆 = {𝑥 ∈ 𝐷𝑉 𝑥 = 0}
and suppose that no solution can stay identically in 𝑆, other than the trivial solution 𝑥 𝑡 = 0. Then ,the origin is
asymptotically stable.
Theorem 4 [10]: The equilibrium point 0x of the autonomous system (1) is asymptotically stable if there
exists a function )(xV satisfying
i) )(xV positive definite Dx , where we assume that D0
ii) )(xV is negative semi definite in a bounded region DR  .
iii) )(xV does not vanish identically along any trajectory in R , other than the null solution 0x .
Proof [10]:By (Theorem 1), we know that for each 0 there exist 0
0x  )(tx
That is, any solution starting inside the closed ball B will remain within the closed ball B . Hence any
solution ),,( 00 txtx of (1) that starts in B is bounded and tends to its limit set N that is contained in B .
Also )(xV continuous on the compact set B and thus is bounded from below in B . It is also non increasing
by assumption and thus tends to a non-negative limit L as t . Notice also that )(xV is continuous and
thus, LxV )( x in the limit set N . If N is an invariant set with respect to (1), which means that any
solution that starts in N will remain there for all future time. But along that solution, 0)( xV since )(xV is
constant )( L in N . Thus, by assumption, N is the origin of the state space and we conclude that any
solution starting in BR  converges to 0x as t .
Example 4 [10]: In Example 2, when the origin of the nonlinear pendulum was stable by using Lyapunov direct
method however , asymptotic stability could not be obtained .
Return to the system
212
21
sin x
m
k
x
l
g
x
xx




(10)
And the candidate Lyapunov function is:
2
21
2
1
)cos1()( xxaxV  , (11)
2
2
2
)( xklxV  . (12)
which is negative semi definite since 0)( xV for )0,( 1xx  , if we apply (theorem 4, conditions (i),(ii)) we
satisfy in the origin








2
1
x
x
R .
With   1x , and axa  2 , for any

Ra . If we check condition (iii) which V can vanish
identically along the trajectories trapped in R , other than the null solution. Using (12) we get
0V  2
2
2
0 xkl 02 x t  02 x .
And using (11) we obtain:
21sin0 x
m
k
x
l
g
 since 02 x  0sin 1 x .
Restricting 1x to ),(1 x condition (iii) is satisfied if and only if 01 x . So 0)( xV does not
vanish identically along any trajectory other than 0x , therefore 0x is asymptotically stable by (Theorem
4).
V. Conclusion
When we use Lyapunov direct method, we can get from the system if it is stable or asymptotic or
unstable, but in some cases, like our example, this method fails to achieve the stability and this does not mean
that the system is not stable, just only means that such stability property cannot be established by using this
method. In this case we applied Lasalle's invariance principle to obtain the asymptotic stability . when we saw
the origin is stable, not asymptotic stable.
References
[1]. Debeljkovic D. L. J., BuzurovicI. M., (2011), “Lyapunov stability of Linear Continuous Singular”, International Journal of
Information and systems sciences, vol. 7,pp. 247-260.
[2]. Luciano M. L., Eduardo N. V., Samuel I. N.,(2014) "On the construction of Regions of Stability", Pure and applied Mathematics,
Journal, vol.3,pp. 87-91.
[3]. Pukdeboon c.,(2011)," A Review of Fundamentals of Lyapunov Theory",The Journal of Applied Science, vol. 10, pp. 55-60.
[4]. Mawhin J.,(2015), "AlexandrMikhailovichLiapunov, The general problem of the stability of motion (1892)", ResearchGate, pp. 1-
17.
[5]. Mitropolskii y. A., Borne P., Martynyuk A.A.,(2007), "Nonlinear Dynamics and systems Theory", vol. 2 pp.113-120.
[6]. Hedih K. S., (2007),"Nonlinear Dynamics and AleksandrMikhailovichLyapunov (1857-1918)", Mechanics, Automatic Control and
Robotics, vol. 6, pp.211-218.
[7]. Almog j.,(2009),"Stability Analysis of Dynamic Nonlinear Systems by means of Lyapunov Matrix-Valued Functions",thesis,
University of the Witwatersand, Johannesburg, pp. 27-59.
[8]. Bacciotti, A. and Rosier, L.,2005,"Liapunov function and stability in control theory",(Second Edition), Springer,vol., 267, p. 105-
108.
[9]. Khalil H. K., (2002),"Nonlinear Systems" (Third Edition), prentice Hall, Upper Saddle River, New Jersey, p. 111-160.
[10]. Horacio J. M., (2003), "Nonlinear Control System Analysis and Design", John Wiley & Sons, Inc., Hoboken, New Jersey, p. 65-
133.
[11]. Vidyasagar M., (1993),"Nonlinear System Analysis", (Second Edition), Prentice Hall, Englewood Cliffs, New Jersey, p. 135-268.
[12]. Lyapunov's theorem for stability and asymptotic stability of an equilibrium point of a nonlinear system

Achieve asymptotic stability using Lyapunov's second method

More Related Content

What's hot (20)

Similar to Achieve asymptotic stability using Lyapunov's second method (20)

More from IOSRJM (20)

Recently uploaded (20)

Achieve asymptotic stability using Lyapunov's second method