Welcome to Scribd, the world's digital library. Read, publish, and share books and documents. See more
Download
Standard view
Full view
of .
Look up keyword
Like this
1Activity
0 of .
Results for:
No results containing your search query
P. 1
Ubiquitous Computing and Communication Journal_104

Ubiquitous Computing and Communication Journal_104

Ratings: (0)|Views: 4|Likes:
Published by Usman Tariq

More info:

Published by: Usman Tariq on Oct 31, 2008
Copyright:Attribution Non-commercial

Availability:

Read on Scribd mobile: iPhone, iPad and Android.
download as PDF, TXT or read online from Scribd
See more
See less

05/09/2014

pdf

text

original

 
 
1 
A NOVEL APPROACH TO ADAPTIVE CONTROLOF NETWORKED SYSTEMS
A. H. Tahoun
1
, Fang Hua-Jing
11
School of Information Technology and Engineering,Huazhong University of Science and TechnologyWuhan 430074, Chinaalitahoun@yahoo.com 
ABSTRACT
The insertion of communication network in the feedback adaptive control loopsmakes the analysis and design of networked control systems more and morecomplex. This paper addresses the stability problem of linear time-invariantadaptive networked control systems. Our approach is novel in that the knowledgeof the exact values of all system parameters is not required. The case of statefeedback is treated in which an upper bound on the norm of matrix
 A
is required tobe known. The priori knowledge of upper bound on norm
 A
is not required inconstructing the controller but it is required only to determine an upper bound onthe transmission period
h
that guarantees the stability of the overall adaptivenetworked control system under an ideal transmission process, i.e. no transmissiondelay or packet dropout Rigorous mathematical proofs are established reliesheavily on Lyapunov's stability criterion. Simulation results are given to illustratethe efficacy of our design approach.
Keywords
: Networked control systems, Transmission period
 ,
Adaptive control,Lyapunov's stability
.
 
1
 
INTRODUCTION
Networked control systems (NCSs) are feedback control systems with network communicationchannels used for the communications betweenspatially distributed system components like sensors,actuators and controllers. In recent years, thediscipline of networked control systems has becomea highly active research field. The use of networks asmedia to interconnect the different components in anindustrial control system is rapidly increasing. Forexample in large scale plants and in geographicallydistributed systems, where the number and/orlocation of different subsystems to control make theuse of single wires to interconnect the control systemprohibitively expensive [1]. The primary advantagesof an NCS are reduced system wiring, ease of systemdiagnosisand maintenance, and increase system agility[2]. The insertion of the data network in the feedback control loop makes the analysis and design of anNCS more and more complex, especially foradaptive systems in which systems parameters notcompletely known. Conventional control theorieswith many ideal assumptions, such as synchronizedcontrol and non-delayed sensing and actuation, mustbe reevaluated before they can be applied to NCSs.Specifically; the following issues need to beaddressed. The first issue is the network induceddelay (sensor-to-controller delay and controller-to-actuator delay) that occurs while exchanging dataamong devices connected to the shared medium.This delay, either constant (up to jitter) or timevarying, can degrade the performance of controlsystems designed without considering the delay andcan even destabilize the system. Next, the network can be viewed as a web of unreliable transmissionpaths. Some packets not only suffer transmissiondelay but, even worse, can be lost duringtransmission [3].The main challenge to be addressed whenconsidering a networked control system is thestability of the overall NCSs. In this paper, we treatthe stability analysis of networked control adaptivesystems, when the network is inserted only betweensensors and the controller. Under an idealtransmission process, i.e. no transmission delay orpacket dropout, we have derived a sufficientcondition on the transmission period that guaranteesthe NCS will be stable. This case is treated in [4] and[5], with completely known systems.This paper is organized as follows; the problemis formulated in Section 2. The main result is givenin Section 3. Section 4 presents an example andsimulation results, finally we present our conclusionsin Section 5.
 
 
2
2
 
FORMULATION OF THE PROBLEM
Consider an NCS shown in Fig. 1, in whichsensor is clock-driven and both controller andactuator are event driven.Figure 1 The block diagram of NCSIn Fig. 1, a class of linear time-invariant plantsis described as
)()()(
bu Ax x
+=
&
 
[
 , t 
k+1
) ,
= 0,1, 2, ...(1)where
 x
(
)
 R
n
is a state vector,
u
(
)
 R
is a controlinput vector, (
 A
,
b
) is controllable,
 A
is a constantmatrix with
unknown elements
,
b
is a knownconstant vector. We assume that the control isupdated at the instant
and kept constant until nextcontrol update is received at time
k+1
. Let
h
be thetransmission period between successivetransmissions, that is,
h
=
k+1
. For this paper, weassume that the transmission process is ideal, thereare no delays, no data losses (packet losses) duringthe transmission. In future work, we will relax theseassumptions.Our objective is to design an adaptive stabilizerfor the networked system and to find an upper boundon the time transmission period (sampling period)
h
 such that the NCS is still stable.The control input is of the form
)()()(
 xu
=
(2)where
(
) is an
n
-dimensional
control parameter 
 vector,
denotes transpose. From Eqs. (1) and (2),we get
)()()()(
 xbk  Ax x
+=
&
 
)()()()(
*
 xbk  xbk  x A
+=
(3)where
bk  A A
*
+=
is Hurwitz matrix satisfyingthat
Q APP A
=+
,
P
and
Q
are symmetric andpositive–definite matrices, and
*
is the true value of 
(
). Define
*
)()(
=
φ 
as the control parametererror vector, and
e
(
) =
 x
(
) –
 x
(
) as the transmissionerror, Eq. (3) can be rewritten as
)()()()()(
*
ebk  xb x A x
+=
φ 
&
(4)
3
 
MAIN RESULT
The main result of this paper will be treated inthe following theorem.
Theorem 1:
Let an NCS with linear time-invariantplant (1), an adaptive stabilizer with control input (2)is globally stable if the adaptive control law takes theform [6]
Pb x x
)()()(
α φ 
=
&
(5)and the transmission period satisfies
 
{ }
321
,,,1min
hhhh
<
, where,
α 
is an
n
×
n
 symmetric positive-definite adaptation gain matrix,and 
 ⎠ ⎞⎝ ⎛ +=
11
1ln1
ζ 
uppupp
 A Ah
 
 ⎠ ⎞⎝ ⎛ +=
2min2
)(1ln1
ζ  βλ 
uppupp
 AQ Ah
 
 ⎠ ⎞⎝ ⎛  ⎠ ⎞⎝ ⎛  +=
123
)1() 41()41(1ln1
ζ  β  β  β 
uppupp
 A Ah
 where
 
α ζ 
221
)(21)(
upp
 x pbbk  A
++=
 and
 
( )
)()(14
min12
bk  A AQP
upp
++++=
λ ζ ζ 
.
 
To prove the stability of the NCS, firstly, wewill find an upper bound on the transmission error
e
(
), a lower and an upper bound on the state
 x
(
), andfinally, we will use these bounds in Lyapunovfunction to prove Theorem 1.
 Lemma 1:
 
(Transmission Error Upper Bound) Thetransmission error e
(
)
is bounded between twosuccessive transmissions by
)()(
 xe
γ 
(6)
where
)1()(21)(
)(22
++=
upp
 Auppupp
e A x pbbk  A
α γ 
 ,
PlantControllerSensorActuatorNetwork 
 
 
3
 A
upp
is an upper bound on A such that;
upp
 A A
.
Proof 
: From the definition of 
e
(
), it can be foundthat
)()()()()(
 xbk  Ax xe
+==
&&
 
)()()()(
 xbk  Ax Ae
++=
 Taking the integral on both sides, and taking intoaccount that
e
(
) = 0, we have
( )
ds xbk  Axs Aeee
∫ 
+++=
)()()()()()(
 
dss Ae x x x pbb  xbk  Ax
∫ 
++=
)())(()()( 21))](()()([
2
α 
 If we choose
-
 
<
1, Therefore,
dsse A  x pb  xbk  x Ae
∫ 
+++
)()]()( 21)()()([)(
32
α 
 If we know an upper bound of 
 A
that is;
upp
 A A
,and applying Bellman-Gronwall Lemma [2], yields
dsdw A x  x pbbk  Ae
suppupp
 ⎠ ⎞⎝ ⎛ ×++
∫ ∫ 
exp)(])(21)([)(
22
α 
 Then
)()(
 xe
γ 
 
 Lemma 2: The state of the NCS, x(t), betweensuccessive transmissions is bounded by
)()1()()()1(
 x x x
γ γ 
+
(7)
Proof 
: As
e
(
) =
 x
(
) –
 x
(
), then
)()()()()(
 xe xe x
+
 
Using Eq. (6), it can be concluded that
)()1()()()1(
 x x x
γ γ 
+
 
Now we turn our attention to proof of Theorem1
.
Consider a positive-definite Lyapunov function
(
) of the form
 
)()()()()(
1
Px x
φ α φ 
+=
(8)Differentiating
(
)
 
with respect to
, we have
)()( )()()()()()()(
11
 xP xPx x
φ α φ φ α φ 
&&&&&
+++=
(9)Substituting for
)(
 x
&
and
)(
φ 
&
from Eqs. (4) and (5),there results
Pb x x  xPxb ePbk  x  xPb x  x AP xPxbe Pxb xPx A x
)()()( )()()( )()( )()()( )()()()( )()()()()()(
**
φ φ φ φ 
+++=
&
(10)Rearranging Eq. (10), yields
)()()(2 )()(2 )()()(2)()()(
*
 xPb x ePbk  x  xPb xQx x
φ φ 
+=
&
(11)
)(
&
becomes bounded from above as
)()(2 )()()(2 )()()(
*2min
e xbk P  xebP  xQ
++
φ λ 
&
(12)From (6) and (7), where we choose
h
< 1, then
)()(2 )()()( )1(2)()()(
*2min
 x xbk P  x xbP  xQ
γ φ γ γ λ 
++
&
(13)Using (7), and rearranging,

You're Reading a Free Preview

Download
scribd
/*********** DO NOT ALTER ANYTHING BELOW THIS LINE ! ************/ var s_code=s.t();if(s_code)document.write(s_code)//-->