Upload
morteza2885
View
216
Download
0
Embed Size (px)
Citation preview
8/11/2019 21642583.2014.913510
1/7
This article was downloaded by: [85.185.144.149]On: 22 September 2014, At: 02:59Publisher: Taylor & FrancisInforma Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House37-41 Mortimer Street, London W1T 3JH, UK
Systems Science & Control Engineering: An Open
Access Journal
Publication details, including instructions for authors and subscription information:http://www.tandfonline.com/loi/tssc20
Observer-based adaptive stabilization of a class ofuncertain nonlinear systemsMohammad M. Arefi
a, Jafar Zarei
b& Hamid R. Karimi
c
aDepartment of Power and Control Engineering, School of Electrical and Computer
Engineering, Shiraz University, 71348-51154 Shiraz, IranbDepartment of Control Engineering, School of Electrical and Electronics Engineering, Shir
University of Technology, 7155713876 Shiraz, IrancDepartment of Engineering, Faculty of Engineering and Science, University of Agder,
N-4898 Grimstad, NorwayPublished online: 09 May 2014.
To cite this article:Mohammad M. Arefi, Jafar Zarei & Hamid R. Karimi (2014) Observer-based adaptive stabilization of aclass of uncertain nonlinear systems, Systems Science & Control Engineering: An Open Access Journal, 2:1, 362-367, DOI:
10.1080/21642583.2014.913510
To link to this article: http://dx.doi.org/10.1080/21642583.2014.913510
PLEASE SCROLL DOWN FOR ARTICLE
Taylor & Francis makes every effort to ensure the accuracy of all the information (the Content) contained inthe publications on our platform. Taylor & Francis, our agents, and our licensors make no representations orwarranties whatsoever as to the accuracy, completeness, or suitability for any purpose of the Content. Versionsof published Taylor & Francis and Routledge Open articles and Taylor & Francis and Routledge Open Selectarticles posted to institutional or subject repositories or any other third-party website are without warrantyfrom Taylor & Francis of any kind, either expressed or implied, including, but not limited to, warranties ofmerchantability, fitness for a particular purpose, or non-infringement. Any opinions and views expressed in thiarticle are the opinions and views of the authors, and are not the views of or endorsed by Taylor & Francis. Thaccuracy of the Content should not be relied upon and should be independently verified with primary sourcesof information. Taylor & Francis shall not be liable for any losses, actions, claims, proceedings, demands,costs, expenses, damages, and other liabilities whatsoever or howsoever caused arising directly or indirectly inconnection with, in relation to or arising out of the use of the Content.
This article may be used for research, teaching, and private study purposes. Terms & Conditions of access anduse can be found at http://www.tandfonline.com/page/terms-and-conditions
It is essential that you check the license status of any given Open and Open Select article to confirmconditions of access and use.
http://www.tandfonline.com/page/terms-and-conditionshttp://www.tandfonline.com/page/terms-and-conditionshttp://dx.doi.org/10.1080/21642583.2014.913510http://www.tandfonline.com/action/showCitFormats?doi=10.1080/21642583.2014.913510http://www.tandfonline.com/loi/tssc208/11/2019 21642583.2014.913510
2/7
Systems Science & Control Engineering: An Open Access Journal, 2014
Vol. 2, 362367, http://dx.doi.org/10.1080/21642583.2014.913510
Observer-based adaptive stabilization of a class of uncertain nonlinear systems
Mohammad M. Arefia
,Jafar Zareib
and Hamid R. Karimic
aDepartment of Power and Control Engineering, School of Electrical and Computer Engineering, Shiraz University, 71348-51154
Shiraz, Iran; bDepartment of Control Engineering, School of Electrical and Electronics Engineering, Shiraz University of Technology,7155713876 Shiraz, Iran; cDepartment of Engineering, Faculty of Engineering and Science, University of Agder, N-4898 Grimstad,
Norway
(Received 25 December 2013; final version received 7 April 2014 )
In this paper, an adaptive output feedback stabilization method for a class of uncertain nonlinear systems is presented. Sincethis approach does not require any information about the bound of uncertainties, this information is not neededa priorianda mechanism for its estimation is exploited. The adaptation law is obtained using the Lyapunov direct method. Since all thestates are not measurable, an observer is designed to estimate unmeasurable states for stabilization. Therefore, in the designprocedure, first an observer is designed and then the control signal is constructed based on the estimated states and adaptationlaw with the -modification algorithm. The uniformly ultimately boundedness of all signals in the closed-loop system isanalytically shown using the Lyapunov method. The effectiveness of the proposed scheme is shown by applying to a unified
chaotic system.
Keywords: output feedback; adaptive control; uniformly ultimately boundedness; Lyapunov-based design
1. Introduction
The problem of the robust output feedback regulation of
uncertain nonlinear systems is the design of a feedback con-
trol law for such systems in a way that the boundedness of
signals in the closed-loop is guaranteed (Chen & Huang,
2005a,2005b;Huang, & Chen, 2004). However, in many
practical applications, measurement of all the states is not
possible. Therefore, observer design is an essential step in
this approach.
The design of observers has received a great deal ofattention recently (Liu, 2009). However, there are two
main restrictive conditions in the design of observer-based
controllers. First, the nonlinearities are only functions of
measurable signals, which is a common assumption in the
literature. Moreover, it is assumed that the unknown non-
linearity is bounded by output injection terms and this
unknown nonlinearity satisfies a global Lipschitz condi-
tion, which is the second restriction (Liu, 2009). In practical
cases, these conditions are not held.
In order to make the design procedure more practical,
we should consider a mechanism to relax these condi-
tions. These conditions were further relaxed recently inLiu
(2009),Choi and Lim (2005),Alimhan and Inaba (2006)andHou, Wu, and Duan (2009). InLiu and Zheng (2009), a
Fuzzy logic systemis employed to estimate theupper bound
of nonlinear uncertainties. This procedure could be done
by using other approximation tools such as neural networks
(NNs) (Arefi & Jahed-Motlagh, 2011;Du & Chen, 2009).
Corresponding author. Email:[email protected]
However, the design of fuzzy logic system and NNs needs
to incorporate the knowledge of an expert. In this paper,
these limitations are thoroughly relaxed by the estimation
of upper bound of uncertainties using an adaptive control
strategy. Compared withLiu and Zheng (2009)andDu and
Chen (2009), this method needs neither any information
about the bound of uncertainties nor an experts or system
specialist. Moreover, in the presented method in Liu and
Zheng (2009)andDu and Chen (2009),semi-global results
are obtained, while our proposed approach provides glob-ally uniformly ultimately boundedness (UUB) for all the
signals in the closed-loop system.
This paper is organizedas follows: theproblem formula-
tion of output feedback stabilization of uncertain nonlinear
systems is presented in the second section. Adaptive out-
put feedback, observer design and the stability analysis
of the algorithm are presented in Section 3.In Section4,
an uncertain unified chaotic system is adopted to evalu-
ate the effectiveness of the proposed method. Finally, the
conclusions are given in Section5.
2. Problem formulation
Consider the following uncertain nonlinear system:
x(t)= Ax(t)+Bf(x)+Bu,
y= CTx,(1)
2014 The Author(s). Published by Taylor & Francis.
This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/3.0), which permits
unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. The moral rights of the named author(s) have been
asserted.
mailto:[email protected]:[email protected]8/11/2019 21642583.2014.913510
3/7
Systems Science & Control Engineering: An Open Access Journal 363
wherex = [x1, . . . ,xn]T n corresponds to the state vec-
tor of the system, y m is the system output, u=
[u1, . . . , um]T m is the input vector of the plant, f(x)=
[f1(x) . . . , fm(x)]T m is the unknown nonlinear function
vector, A nn, B nm, andC nm are constant
matrices with appropriate dimensions. In the system (1),
not allxiare assumed to be measurable and only the system
outputyis assumed to be available for measurement. In the
controller design, we need the following assumptions.
Assumption1 For the nonlinear function f(x), there
exists a positive constant such that
f(x) .
Remark 1 Although Assumption1is restrictive, one can
suppose that is large enough so that this assumption is
satisfied.
Assumption2 (A, B) is controllable and (A, CT) is
detectable.The main goal is to design an output feedback con-
troller such that the states of(1) in the presence of unknown
nonlinear function are bounded.
Iff(x)is known and the state vectorx is available, the
controller can be chosen as
u= f(x)kTcx, (2)
wherekc is the state feedback gain matrix. By substituting
Equation(2)into Equation(1)we have
x(t)= (ABkTc )x(t). (3)
Since the pair (A, B) is controllable, the gain matrix kc
in Equation(3)can be chosen such that the characteristicpolynomial of ABkTc is strictly Hurwitz. Then, it can
be shown thatlimtx(t)= 0. However, becausef(x)is
unknown and only the system outputy is measurable, this
controller cannot be implemented in practice. A solution is
to estimate the upper bound of the known function using the
adaptive control strategy and design a suitable observer to
estimate the state vectorx using the measurable outputy.
3. Adaptive output feedback controller and observer
design
In the previous section we assumed that only the system
output is measurable and other states cannot be used inthe controller design. So, we need to design an observer
to estimate the unmeasurable states. Suppose that x is the
estimation of state vectorx. Thefollowing observer is given
as
x(t)= (ABkTc )x(t)+ko(yCTx),
y= CTx.(4)
Since the pair(A, CT) is observable, the gain matrixko
nm in Equation (4)can be chosen so that AkoCT is
strictly Hurwitz. Let the estimation errore = x x. From
Equations(1) and(4), we have
e= (AkoCT)e+BkTcx+Bf(x)+Bu,
e= cTe.(5)
Assumption3 There exist positive-definite matrices P
andQ satisfying
(AkoCT)TP+P(AkoC
T)+Q = 0,
PB= C.(6)
Remark 2 Ifko can be chosen such that the triple (A
KoCT, B, C) is strictly positive real (SPR), one can use
the KalmanYakubovichPopov lemma (Slotine & Li,
1991), which guarantees the existence of positive-definite
symmetric matricesP andQ in Equation(6).
Theorem1 Consider the nonlinear system (1) and the
observer given in Equation (4). Under Assumptions 13,
construct the following adaptive controller:
u= kTcx e2
e + , (7)
and the adaptation law is as follows:
= e , (8)
where >0, >0, (t0) >0, and >0 are the design
parameters.
Then all the signals in the closed-loop system are
UUB. Furthermore, the estimation error can approach an
arbitrarily small value by choosing the design parameter
appropriately.
Proof Choose the following continuously differentiable
function as a Lyapunov candidate
V = 1
2
eTPe+
1
2
, (9)
where = and is the adaptation gain given in
Equation(8). Thederivativeof Equation (9), usingEquation
(5) is
V = 1
2eT(ATo P+PAo)e+e
TPBkTcx+eTPBf(x)
+eTPBu+ 1
, (10)
where Ao = A koCT. Accordingto Equation (6),wehave
eTPB= eTC= eT. (11)
By using Equations (6),(7), (10), and(11) we have
V = 1
2eTQe+ eTf
e2 2
e + +
1
. (12)
8/11/2019 21642583.2014.913510
4/7
364 M.M. Arefiet al.
Now, using the fact that eTQe min(Q) e 2,
where min(Q) is the minimum eigenvalue of Q and
regarding to Assumption1we have
V 1
2min (Q) e
2 + e e2 2
e + +
1
.
(13)
Furthermore, the following inequality is true for the thirdterm of the right-hand side of inequality (13)
e2 2
e + = e
1+
e +
e + . (14)
Considering inequality(14) we have
V 1
2min(Q) e
2 e + + 1
. (15)
Finally, by substituting the adaptation law (8) into
Equation(15), we obtain
V 1
2min (Q) e
2 +,
1
2min (Q) e
2 +2 +| |.
(16)
Using a2 +ab= a2/2 12
(ab)2 +b2/2 a2/2
+b2/2 for anya, bwe can write
V 1
2min(Q) e
2 +2
2+
||2
2. (17)
From Equations (9) and (17)andmin(P) e 2eTPe
max(P) e 2 we have
V cV+ +||2
2, (18)
where
c= min
min(Q)
max(P),
. (19)
Solving inequality (18)gives
0 V(t) +||2/2
c
+
V(t0)
+||2/2
c
ect t 0.
(20)
Thus, V(t) max{V(t0), (+ (||2/2)/c}, t 0.
From the definition ofV(t) in Equation(9), the error vector
e(t), is bounded by
e(t)
max{V(t0), (+ (||2/2))/c}
min (P). (21)
Equation (21) means that V(t) is eventually bounded by
(+ (||2/2))/c. Thus, all signals of the closed-loop sys-
tem, i.e. e(t), are UUB (Khalil, 2002). Besides, since
Equation (17) implies that V T. Since the characteristic polynomial ofABkTc is
strictly Hurwitz, it can be concluded from Equation(4) that
x is bounded. Then according to e= x x, we can also
conclude that x is also bounded. In addition, based on the
definition of u in Equation (7), u is also bounded. This
completes the proof.
4. Simulation results
To show theproficiency of thepresented algorithm, thesim-ulation results are presented in this section. A class of more
general nonlinear systems is studied in this section. For
example, the following unified chaotic system is considered
(Liu & Zheng, 2009):
x1 = (25+10)(x2x1),
x2 = (2835)x1 x1x3+ (291)x2+ u2,
x3 = x1x2+ 8
3x3+ u3,
(23)
where [0, 1]. When [0,0.8), it is a Lorenz chaotic
system, = 0.8 is the Lu chaotic system, and (0.8, 1]is Chens chaotic system. The system can be easily trans-
formed into the canonical form of Equation (1) with the
following parameters:
A=
2510 25+ 10 02835 291 0
0 0 +83
,
B=
0 01 0
0 1
,
f(x)=
x1x3x1x2
, C=
0 1 00 0 1
T.
It is worth mentioning that Assumption 1,which imposes
an upper bound for f(x), is simply satisfied in chaotic
systems due to the boundedness of trajectories in these
systems (Arefi & Jahed-Motalgh, 2012). The simulation is
carried out with initial conditions x0 = [2,1, 1]T, x0 =
[3,2,1]T. It is straightforward to verify that the triple
(AKoCT, B, C) can be made SPR by the choice of the
observer gain.
8/11/2019 21642583.2014.913510
5/7
Systems Science & Control Engineering: An Open Access Journal 365
Besides, the observer and state feedback gain are
selected as
kc =
0 200 50
50 10 100
T, ko =
60 10 15
30 100 15
T.
Furthermore, the parametersof the controller and adaptation
law are as follows:
= 0.01, =0.5, =0.001, (0)= 0.1.
Figure1 shows the chaotic behavior of system(23) with
u= 0 and = 0.5.
The state trajectories of the system by applying the con-
troller (7) with = 0.5 are shown in Figure 2.Itcanbeseen
from the simulations that the adaptive output feedback con-
troller (7)makes the state estimations tend the actual states
precisely. As this figure shows, the proposed controller sta-
bilizes the system in the presence of unknown nonlinear
uncertainties.
The time responses of the control input and adaptation
parameter are shown in Figures3and4,respectively.
We see that the control input is smooth and imple-
mentable. Furthermore, the value of is bounded.
0 5 10 15 20 25 3020
0
20
40
x1
0 5 10 15 20 25 3050
0
50
x2
0 5 10 15 20 25 3050
0
50
100
time [S]
x3
Figure 1. The chaotic behavior of system withu = 0 where is chosen as = 0.5.
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 52
0
2
4(a)
(b)
(c)
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 51
0
1
2
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 54
2
0
2
Time [S]
Figure 2. (a)x1 (solid line) andx1 (dashed line); (b) x2 (solid line) andx2 (dashed line); (c)x3 (solid line) andx3 (dashed line).
8/11/2019 21642583.2014.913510
6/7
366 M.M. Arefiet al.
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5600
400
200
0
200
u1
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5300
200
100
0
100
Time [S]
u2
Figure 3. Control inputs of the system (a)u1 and (b)u2.
Figure 4. Adaptation parameter.
5. Conclusion
In this paper, an adaptive output feedback stabilization
strategy for a class of uncertain nonlinear systems was pro-
posed. Since all the states are not measurable, an observerwas presented to estimate unmeasurable states. The design
method is based on the Lyapunov stability Theorem, and
it was shown that all signals in the closed-loop system are
UUB. Additionally, in this method, the mere knowledge of
boundedness of uncertain term is sufficient.
Simulation results for the stabilization of a unified
chaotic system show that the proposed approach has a fast
response in stabilization. Moreover, the norm of the estima-
tion errorsis bounded, while thecontrol signalis completely
smooth.
References
Alimhan, K., & Inaba, H. (2006). Output feedback controlfor a class of nonlinear systems. International Journal ofAutomation and Computation, 3(3), 215221.
Arefi, M. M., & Jahed-Motlagh, M. R. (2011). Observer-basedadaptive neural control for a class of nonlinear non-affinesystems with unknown gain sign, 18(4)(pp. 26442649). 18thIFAC world congress, Milano.
Arefi, M. M., & Jahed-Motalgh, M. R. (2012). Robust synchro-nization of Rossler systems with mismatched time-varyingparameters.Nonlinear Dynamics, 67(2), 12331245.
Chen, Z.,& Huang, J. (2005a). Global robustoutputregulation foroutput feedback systems. IEEE Transactions on AutomaticControl, 50, 117121.
Chen, Z., & Huang, J. (2005b). A general formulation and solv-ability of the global robust output regulation problem. IEEETransactions on Automatic Control, 50, 448462.
8/11/2019 21642583.2014.913510
7/7
Systems Science & Control Engineering: An Open Access Journal 367
Choi, H. L., & Lim, J. T. (2005). Global exponential stabi-lization of a class of nonlinear systems by output feed-back. IEEE Transaction on Automatic Control, 50(2),255257.
Du, H., & Chen, X. (2009). NN-based output feedback adaptivevariable structure control for a class of non-affine nonlinearsystems: A nonseparation principle design.Neurocomputing,72(79), 20092016.
Hou, M., Wu, A., & Duan, G. (2009). Adaptive control of aclass of nonlinear systems by output feedback(pp. 238243).Proceedings of the7th Asian control conference, Hong Kong,China.
Huang,J., & Chen, Z.(2004). A general frameworkfor tackling theoutput regulation problem.IEEE Transactions on AutomaticControl, 49, 22032218.
Khalil, H. K. (2002).Nonlinear systems (3rd ed.). New Jersey, NJ:Prentice-Hall.
Liu, Y. (2009). Robust adaptive observer for nonlinear systemswith unmodeled dynamics.Automatica, 45(8), 18911895.
Liu, Y. J., & Zheng, Y. Q. (2009). Adaptive robust fuzzy control
fora class of uncertainchaotic systems.Nonlinear Dynamics,57, 431439.
Slotine, J. J., & Li, W. (1991). Applied nonlinear control.Englewood Cliffs, NJ: Prentice-Hall.