LUND UNIVERSITY PO Box 117 221 00 Lund
A Comparative Study and Performance Assessment of H °° Control Design
O'Young, Siu D.; Hope, J.; Åström, Karl Johan; Postlethwaite, Ian
1988
Document Version:
Publisher's PDF, also known as Version of record Link to publication
Citation for published version (APA):
O'Young, S. D., Hope, J., Åström, K. J., & Postlethwaite, I. (1988). A Comparative Study and Performance Assessment of H °° Control Design. (Technical Reports TFRT-7403). Department of Automatic Control, Lund Institute of Technology (LTH).
Total number of authors:
4
General rights
Unless other specific re-use rights are stated the following general rights apply:
Copyright and moral rights for the publications made accessible in the public portal are retained by the authors and/or other copyright owners and it is a condition of accessing publications that users recognise and abide by the legal requirements associated with these rights.
• Users may download and print one copy of any publication from the public portal for the purpose of private study or research.
• You may not further distribute the material or use it for any profit-making activity or commercial gain • You may freely distribute the URL identifying the publication in the public portal
Read more about Creative commons licenses: https://creativecommons.org/licenses/
Take down policy
If you believe that this document breaches copyright please contact us providing details, and we will remove access to the work immediately and investigate your claim.
CODEN: LUTFD2 / (TFRT-7403) / L-32l (1988)
A Comparative Study and Performance As ses sment
of H* Control Design
S. D. OYoung J. Hope K. J. ^Å"rtrti- I. Postlethwaite
Department of Automatic Control
Lund Institute of Technology
October 1988
Department of Automatic Control Lund fnstitute of Technology
P.O. Box
118S-22L 00
Lund
SwedenDocumcnt n¿tne
Report
D¿üc of i¡suc
October 1988 Document Numbe¡
CODEN: IUTFD2/(TFRT-7403)/ 1,-32 / (Ig88) Author(s)
S.D. O'Young,
J.
Hope, K.J. Åström,I.
Postleth- waiteSupervisor
S po nsoÅng organisatíon
Titlc ¿nd subtítlc
A comparative study and Performamce Assessment of .E[æ control Design.
Abstræt
This paper assesses the robust stability and robuet performance properties of different
.E-
methods, and reports the use ofa generalized two-input (vector) and two-output (vector) plant configuration in multivariableIl-
design. Two industrial design examples are used: a scalar robot arm and a multivariable generation station and grid model. ThefIæ
designs attempted tendto
be conservative dueto
the representation of plant uncertainty as(-t-)
norm bounded perturbations neglecting the phaee inform¿tion, and the merging of multiple design objectives into onef[-
norm. Despite the conservatism, thef[æ
approach is still systematic and useful for multivariable designs.Kcy wotdla
Classiñcatìon system and. /or índcx terms (íf any)
S upplcmcnt ary bìblìo graphìcal informatíon
ISSN and kcy títle ISBN
Languagc English
Numbcr of pages 32
Rncìpìent's notcs
,Sccurity clas sifrcat io n
Thc report nay bc orde¡ed Îrom thc Departtnent of Automatic Control or bo¡¡owcd úhrougå thc lJniversity Lìbrary 2, Box lolo,
3-227 03 Lund, Swedcn, Tclcx: 33248 fiubbis lund..
A
Comparative Study and Perfo¡mance Assessmentof
rYæ Control Designs$
by
S.D. O'Young,* J. Hope,$
K.J. Åströmt
andI.
postlethwaiteÏAbstract
This paper assesses the robust
stability
and robust performance properties of different -17- methods, and reports the useofa
generalized two-input (vector) and two-output (vector)plant
configurationin
multiy¿riable-E*
design.Two industrial
design examples are used:a
scalarrobot arm
an¿ a multiv¿riable generation station and grid model. The ,Eæ designs attempted tend to be conserv¿tive due to the representation of plant uncertainty as(,8*)
norm bounded perturbations neglecting the phase information, and the merging ofmultiple
design objectivesinto
oneJl* norm.
Despite the conseryatism, the 11æ approach isstill
systematic and useful for multiy¿riable designs.Keywords
fIæ
Design, Robustness,Multivariable,
Computer-Aided-Design,Industrial
Applications$
The research is supportedby the Science and Engineering Research Council,
U.K.
* Dept.
of Electrical Engineering, theUniv.
of Torãnto, To-ronto, Ontario, M5S144
Canada$ central Electricity
Generating Board., Barnwood, GL4 zRS, England,u.K.
t Dept.
ofAutomatic
Control,Lund Institute
of Technolog¡ Box 118, 5-221 00Lund
42, SwedenI Dept.
of Engineering,univ.
of Leicester, Leicester,LEl
zRH, England,u.K.
1. INTRODUCTION
The
objective ofthis
paperis to report
and assess the applicationof
J7æoptimization in
control system design,with two industrial
examples:a robot arm
andan electrical
power generationstation.
The robot arm is a scalar systemwith
large v¿riationsin
plant dynamics over its operatingrange. It is
chosento
assessthe robust stability
and robust performance propertiesof the lY-
method. The classical Horowitz and Sidi (1972) method is then compared against
-E*.
The powerstation
has 2 actuator inputs, 2 sensor signals, and 2 disturbance sources (one of which enters theplant
neither via the actuator nor the sensor).It
is chosen to demonstrate the useofa
generalizedtwo-input
(vector) and two-output (vector) plant configurationin
multiv¿riable,I1-
optimi zation- based design. For implementation, the dominant dynamics of the multiv¿riablelY*
controller areidentified,
andthe
state-spacel/-
controlleris
subsequently reducedto
classicalPI
controllerswith
constant cross-coupling terms.In -tlo"
designs, robust performance and robuststability
requirements are specified as bounds onthe
weighted-E-
norms ofindividual
closed-loop transfer functionmatrices.
These matrices are augmented(".S. by
stacking)into
one single transfer functionmatrix,
anda
stabilizing contoller is found to minimize the.E-
norm ofthis
augmentedmatrix.
The combination of multiple design objectives into one often gives conservative results because the ¡?æ norm on the augmentedmatrix
oniy imposes an upper bound onits
elements. A sample of designs usingthis
approach can be foundin
Postlethwaite (1-986, 1987a) and Safonov (1986). Doyle (1983) proposes the use of p analysis and synthesis methods to alleviate this conserv¿tism essentially by solving a multiobjective optimization problem. Designs using the ¡r approach have been reported by Doyle (1986, 1987) and Fan (198?).This paper deals only
with
¡loo designsbut
not p designs because theJ?-
optimization techniques are now well understood (Francis,1987). All the .Iy'*
computations can becaried out in
state- space (Doyle, 1988 and Chu 1986) and they have been implementedin.E-
Computer-Aided Design(CAD)
packages such asStable-Hl
(Postlethwaite, 1987b) andLINF
(Chiang, 1987). Forp,
theret 4lt.
the .design gxapn_leqin this
paper have been carriedout
usingthis CAD
package,at
OxfordUniversity
England, U.K.does not appear
to
be an efficient algorithmic implementation of the synthesis procedure and there are noCAD
packages available.This
paper differsfrom
other application paperson -ã-
designin that it
offersa tutorial intro-
duction as well as an assesment of the methodology. Thetutorial
introduction is made possible by abstracting the algorithmic implementation of,E-
optimization as a 'black box' OPT
procedure 2.Design rules
for
weights selection are introduced, and thel7-
design results are then assessed by comparisonwith the
classical Horowitz method (Horowitz andSidi,
1972).A
completemuitivari-
able design cycle from engineering specifications to implementation considerations is also reported.We begin by introducing the use of -Eæ optimization
in
controller designin
the next section.2. flæ
DESIGN PROCEDUREThe
plants usedin this
assessment haveall
been modelled aslinear time inr¿riant,
continuous time and lumped parameter systems which have both state-space realizations and transfer functionmatrixrepresentationsintheLaplacianv¿riabre
", " =lâ 3] r"u
G(s)= c(sI- A)-ra+n.
The system G is said
to
be stableif
the statematrix
.4 has no eigenvaluesin
the closed right-haif- plane. Suppose G is stable; then the .I1æ norm of G can be defined viaits
transfer functionmatrix
representation asllCll- = tlp a1G(iùJ,
wherea1c(jr))
denotesthe
largest singular valueof
Gat
frequency c.r. WhenG is
scalar,its
11æ normis simply the
highest gain onits
Bodeplot.
We also usethe
symbolRH* to
denote matriceswith
stable real-rational entriesin s, that is,
the real-rational subspace of .Eæ.2.L Formulation of
control
system Designs asllæ optimizations
The
general compensation configuration usedin this
assessmentis
shownin Fig. 1. It will,
inthe
sequel,be
referredto
asthe
Standard Compensation Configuration (SCC).The
objective is2
The interested reader is referredto
Chu (1986), Francis 11987) and Dovle 11988) forits
imnlemen-tation details. An
understandingof the'OPÍ
'procedurèis íot ur*"í"¿ ìor;Ad;á t; ;ã;l-ihi;
paper.
3
to
design a compensator .K(in
state-space form), usually known as the controller,for
theplant P
(represented alsoin
state-space form) suchthat
theinput/output
transfer characteristicsfrom
the externalinput
vector dto
the externaloutput
vector e is desirable, according to some engineering specifications. The internal compensation signal flow paths are represented by vectors 3r and u, and correspondto
the sensor signals and actuator demands, respectively.The
compensated system(that is, with the controller K in the internal
signalpath)
shownin Fig.
1is
saidto be internally
stableif the
augmentedA matrix of the
compensated system is stable.In
other words, when the external signal o=
0,the
states of bothP
andK wiit
goto
zero asymptoticallyfor
anyinitial
conditions. Such a controller is saidto be
stabi,Iizing.Let
M
denote the closed-loop transfer functionmatrix
mapping externalinput
o to externaloutput e,
andlet
W¿ and Wobe
weightsin -RIl*,
chosento
emphasize(or
de-emphasize)the
relative importance of the externalinput
andoutput
signal. The.ã-
approach isto
design a controller1l
such
that
theI/@
norm ofM
is minimized.In
other words, the objective isto
solve the following optimization problemOPT
where the
minimization
is over the whole seú of stabilizing controllers.3. AN INDUSTRIAL ROBOT ARM
A
simple model(Åström et
al.r 1987) ofa robot arm is
usedin
the assessment ofthe
robustness propertiesof the -E- method. The
transferfunction from the
controlinput (motor current) to
meariurementoutput (motor
angular velocity) isP7o(s)
= kmlJas2+ds+k
(s
+
pt)lJo,J *d(JaIJm)s*k(Ja+Jm)l
(1) I{ etabilizingmlnllw"Mwill*,
where
Jae1.0002,
0.002],Jm=
0.002,d= 0.0001,å =
100,lcm= 0.5andpr =
0.013.
Themoment of
inertia
"Iø of the robot arm varieswith
the arm angle. Bode piots of the plant gain forthe
extreme values of the arminertia Ja in (1),
wherePo:=
pJø=o.ooz andp" !=
pJo=0.0062, âr€given
in
Fig. 2.Since
this robot arm plant
haslarge
variationsin
dynamics overits required
operating rangeof
angular positions,the
objectiveis to
designa robust
controllerwhich is
insensitiveto
these variations and exhibits good tracking and disturbance rejection propertiesat all
angular positions.The ability to maintain stability
overthe entire
rangeof plant dynamic
v¿riationsis
referredto
as robuststability
(RS) and theability to sta¡ at
the sametime, within
certain performance requirements is referred to as robusú performance (RP). We also use the term nominal
performance(NP)
to referto
the performance of the closed-loop system pertainingto
a nominal plant.The most general fixed gain compensation configuration for satisfying both tracking and disturbance rejection requirements is a two-degree-of-freedom controller, consisting of an inner loop
for
robuststability
and disturbance rejection, and an outer loopfor
tracking. Wewill
concentrate mainly on the inner loop design.A
pre-compensator consists of a simple first-order lag, acting as a set-point scheduler, is adequate.Three approaches are undertaken
in
this comparative study: twof/æ
d.esigns, one aiming at achiev- ing robust performance and robuststability
simultaneousl¡ the secondat
only simultaneous nomi- nal performance and robuststabilit¡
and thethird
a classical design via the Horowitz method. Theplant
uncertaintyfor
thel?-
designsis
modelled as(ll*)
norm-boundedadditive
perturbations on a nominal plant, and for the Horowitz method as real parameter(,/ø)
perturbations.3
The small constan!ft
is added Lo- avoid the Model Matching Trans-fo-rm_ation zero (O'Young, 1g8g) calsed by.3, go_le at-theorigin. This
smallperturbatið" F;-.t;¿¿A;d iaü;* p;ò;reáì.'åolîã;
fully
specifiedbut
more complexII*
design problem.,5
3.1 Robust Performance and Robust
stabilization
via-E* optimization
Our
treatment of the robust performance and robuststability
requirements follows Doyle's (1984)method of
representing these design requirements asan unstructured additive perturbation
tothe
nominalplant.
Consider the inner-loop compensation configurationin Fig. 3
where Pois
the nominal plant andK
is the inner-loop feedback compensator, A1 and A2 are additive perturbations, representingplant
uncertaintyand
performance requirementsrespectivel¡ and W1
andW2
ate weights' We assumethat A1
andA2
are scaledviaWt
andW2 suchthat
they are closed unit-balls(:= {ó
eREØ' lláll* < 1}) in
-B.Eæ.For robust performance, suppose
that
the variationsin
dynamicsfrom the
nominal plant are con- tainedwithin
a filteredunit,ball in -&r?-:
{Wz6:6e A}f {P- Po:PeP},
(2)where
P :- {P¡o
;Ja €
[0.0002, 0.002]], and Po is the nominalplant.
Then a sufficient condition for robuststabilityis that If
stabilizes the set of plants{po*Wzó
: ó€ A2}
which, by (2), containsP.
Define
the
sensitivityfunction
,9o as Sp:=
(1+ Pf)-l
anda robust
performance requirement can be defined asVP €.p
and Vc.r€ ft, lsp(jùllwtjùl ( 1. In other
words,the
disturbance rejectionratio (:-
dfe,Fig.3)
is guaranteed, over all possibleP eP, to
be biggerthan lI4{l
atall
frequencies.To
representthe
requirrnentsfor robust
performance androbust stability
simultaneouslyin
the oPT
setting, the closed-loop transferfunction matrix
is defined asM
Wt Sp"W2I(
Sp" (3)The following robust performance and robust
stability
sufficiency resultis
then obtained.RPRS
ContrcIlerK
satisfres the robust performance and robuststability
rcquirementsif (i) K
stabilizes Po and
(ii) llMll* < tlø.
Proof: For robust stability,
we have V6r e A1 and y
6ze A2,
sinceA1
andA2
are unit_balls
in RE*,ll[ót, ár]ll- S \Æ. By (i), M is
stable, andvór e a1
and Vóze 42,
we havell[ór, óz]Mll- < ll[ár,
óz]11""llMll- < 1.
The small gain theroem then impliesthat
the intercon- nected systemin Fig. 3 is stable.
For robust performance, supposef
6ze A2
andf
c¿€ ft
suchthat
l(.9aao,W)(jr)l ) 1.
Then, there existsa
real-rationalfunction ór € Ar with the
appro_priate
gainand
phaseshift
suchthat
(^gp..,.6rW16)(j.l) = t. The unity-gain
positive feedback would destabilize the interconnected system, and hence contradicts the robuststability
condition.Q.E.D.
To
achieve RPRSfor the robot arm
example,the
SCCplant
correspondingto the
closed-loop transfer functionmatrix M in
theOP?
probiem is constructed by interconnecting the state-space realizations ofPo,W1 andW2
accordingto
Fig.3.
Weight W2 is chosento
satisfy Inequality 2;in
particular,it
is constructed as a stable real-rational function suchthat yp ep
and Vc.re
$1,lwz(j,)l> l(P - P,)(j./):.
(4)\Meight W1
is a
high-gain low-passfllter with
the highest possible cross-over frequencya. (:= u
;lw{ir)l =
1) choseniteratively
suchthat
the solutionto
theOPT
problem is achiev¿bleat
a cost lessthan Ll.'n.
The sensitivity functions corresponding
to
the nominal plantsP,
andP"
are shownin
Fig.4.
Notethat
the cross-over frequency c.r"for
the final design isat
6 rad/sec, andthat both
l,Sp"l and l^gal are both below 0dB at
6 rad/sec. The robust performance requirement is thus achieved.3.2 Nominal Performance and Robust Stabilization
via .ãæ
OptimizationThe NPRS design requirements 'u:ary slightly from those of RPSP
in
the sensethat
robuststability
is retained as an obvious hard design constraint,
but optimal
nominal performance is only requiredfor
the nominalplant.
The rationale behind such a strategyis that if
l.9p"lis
made small enough over the operating band, the closed-loop dynamicswithin
the inner ioop should also be relativelyI
insensitive
to plant perturbation,
hence achievingrobust
performanceindirectly (O'Young
and Francis, 1986).Consider again the compensated system
in
Fig. 3with A1
removed from the block diagram and letM in
theOPT
problem be defined as beforein
the RPRS case.NPRS
ContrcIlerK
satisfres the nominal peúormance and robuststability
requirementsif (i)
K
stabilizes Po and(ii) ll}fll- <
1.Proof: It
follows, by(ii), llMll." <
1 which impliesthat llw2KS¿ ll"" <
t.
Sincevóz e 42,
ó2 is stable and lló2ll<
1,it
follows by(i)
and the small gain thereomthat
the interconnected systemin
Fig. 3 is stable. Nominal Performance follows immediately from(ii)
sincellWrSp"ll." <
f. Q.E.D.
The NPRS condition differs from the RPRS case only by raising the cost of the
OPT
problem by a factor of{2. In
general, this number increasesat
the rate of1fr
wherez
is the number of blocks of additive unstructured perturbations, representingboth
robust performance and robuststability
requirements.The design procedure follows exactly as
in
the RPRS case where Wz is the same as before, and the highest achievable cross-ovet frequency c.r"for l7r
satisfyingInequality (ii) in the
NPRS design is 1-5 rad/sec; hence 2.5 times higher than the achiev¿ble cross-over frequencyof the
RPRS design, although the actual bandwidths of5p.
are similarin
both case. This showsthat
a NPRS design can sometimes satisfy RPRS requirements because robust performance depends on nominal performance in most feedback designs. The sensitivity functions corresponding toP,
andP"
are shownin
Fig. 4.3.3 Robust Performance and Robust
stability via
the Horowitz methodThe
I/*
approach is often criticized for being conservative, and to demonstrate this fact, we present the result ofÅström
eúal.
(1987), on the same designvia the
Horowitzmethod.
Horowitz dealswith
the RPRS requirementsby
characterizing the so-called Horowitz bounds on a setof
discretefrequency
points
{c,.r} overa
frequencyband, delimiting the
feasible compensator complex gain regions-8,
whereB.:= {a(ø)
:Ir
Ir + a@¡eç¡..,¡ (
c(c.r), YPeP
(5.1)and
a(u)P(ja)
I ó(r) - a(a),
YPe P\
(5.2)I + a(a)P(jc.')
on a Nichols' chart. Condition 5.1 is a direct characterization of the complex inner-loop compensator
gain
neededto
guaranteea
disturbance rejectionratio ) L/c
overall
possibleplant
dynamicsP
€.P.
Condition 5.2is
neededto
guarantee the existenceof
an outer-loop controller suchthat the
compensated frequency responsefor
command signal tracking staysat the
sametime within the
tolerancelimits b(r) -
a(cu). These tolerancelimits
are sometimes derivedfrom time
(step) response requirements for designs involving minimum phase plants. Notethat
the actuai perturbed set of plantsP
is usedin
the characterization ofB*
instead of a norm-bounded set(2)
asin
l?æ design.The
inner-loop compensatorIf is
synthesizedby
(pasting' together,for
example,by
real-rational function approximations, so that its frequency response lies within the Horowitz bounds and satisfiesthe
usual Nyquiststability condition. The
sensitivity functions correspondingto
Po andp"
areshown in Fig. 4. Note that the disturbance rejection bandwidths are in the region of 50
-
100 rad/sec,much higher than those achieved
via
the.Ifæ
methods.3.4 Conservatism
in
the11*
DesignsThe
conservatismin the .E*
designs stemmainly from the
representationof plant
uncertainty(2) and the
formulation ofa multi-objective
optimization problem(3)
asa
single-ob jectiveOpT
problem.
In
the Horowitz method, both the gain and phase information on the plant dynamics variations are usedin
the characterization offeasible regions for the compensator frequency response.In
theI/-
I
design, we
only
use the gaininformation via Inquality
2but
ignorethe
phaseinformation.
Phase information can only be ignoredat
high frequencies where the sensor signal is often dominated and corrupted by noise which contains no deterministic phase information. Phase information is howeverimportant at
low frequenciesor in the
cross-over(cut-off)
frequency range, where perturbations aretypically
structured.Because
of the
low-damping resonant peaksand
troughs occurringat
frequencies around 250- 300 rad/sec and higher (Fig. 2), a sufficient condition for robuststability
isto limit
the inner-loop disturbance rejectionbandwidth to
be lessthan
250 rad/sec as demonstratedby the
frequency responsesof the
-Eæ controllers as shownin Fig. 5. In fact, the optimal H*
controllers can be replacedby
4th-order low-passñlters with the
respective cut-off frequencieswith no
appreciable changein
closed-looptime
responses.In the
caseof the Horowitz
design,the controller
gaincan be kept high at
frequencies beyond 250 rad/sec because phaseinformation is
usedin
the characterization of plant uncertainty.Although the low-frequency gain of the Horowitz controller is also significantly higher than the 11æ
controllers,
the
high gainis not
neededto
rejectoutput
disturbance.The
low-frequency gainsof the -f/æ
controllers can be increased,if
necessargby
choosing higher low-frequencygain
for W1without
affecting the performance of the resultant design.Since 172 has
to
be synthesizedby
a real-rational function approximationto
satisfy the Inequality 3, allowance must be madefor
approximationelror.
Fig. 6 showsthe error
marginof
an eighth- order real-rational function approximation lP"-
Pol,with
the most pronounced error occulring at frequenciesjust
below the resonant frequency of Poat
around 100-300rad/sec. This
error forces theIf
æ controllersto
have lower gain bandwith than is actually constrained byInequality
2.In
the RPRS design, the obvious conserv¿tism comes from the requirementthat llMll-
be< IlØ
which implies
that llW2I{
Sp"lloo must also be< Llrfz. It
has however been shownin
the NPRS casethat llwrx S¿ll* (
1 is already sufficientto
guarantee robust stability.In the
NPRS design,the
objectiveslllfi.Íall- < L and llw2Kspjl- < l must be
satisfiedsimultaneously. These
two
objectives are imbeddedinto the
single-obje ctive OPT
problem by stacking them asM (3).The
two objectives are satisfied independently onlyif
they are prescribed over disjoint frequency bands. This seldom happens for practical problems since they usually have neariy equal gain aroundthe
cross-over frequencyband. The
design requirementfor
theoptimal
NPRS design isthat lS¿l <
1 at the cross-overfrequencybut
the actual gain is-3
dB because therobust
stability
condition constraint contributes also to the gain ofM.
In fact, the conservatismof
the stacking andthe
error marginin
the synthesis of W2 is takeninto
account, the actual costof
the NPRS design can be pushed up to about 1.8without violating
either the nominal performance or the robuststability
constraints.3.5 The Outer-Loop Design
The
designof the
pre-compensatorwill be
discussedbriefly to
completethe robot arm
design.Tracking is an open-loop property when there is no
plant uncertaint¡
and this is especially trueif
the inner loop has high enough gain such
that
the closed loop dynamics are sufficiently insensitiveto
variationsin plant
dynamics.This
appliesto our robot arm
example, and the require¿ outer- loop compensator is simply a first-orderlag: fr. With r
chosento
be)
0.03 second, the rateof
change of the command signal fed to the inner loop is slow enough not to excite the lowest resonant frequency (around 300 rad/sec)
of
therobot arm at all
anguiarpositions. Fig.
Z showsthe
step responsesof the
various inner-loop designsfor the
extreme valuesof Ja.
Notethat the
tracking response for the NPRS designwith
cost =l-.8 acheives nearly the same speed as the Horowitz design for the nominal design,but
the performance degrades significantlyfor
the perturbed plantp".
4. A
POWERGENERATION EXAMPTE
The power station shown
in
Fig. 8 is usually operated atfull
load and is tiedto
the loadgrid.
The load frequencytrÍ (in
%pu)
is affectedby
the electrical powerinput
fromthis
station andby
the external perturbation .E¿(in
%p")
on the load demand. The thermal powerinput to
the boiier is11
controlled by the fl.ow rate of
hot
carbon dioxide gas circulated through the reactor.It
is assumedthat
thehot
gas temperature is kept constantby
a relativelytight
regulation of thereactivity
via the reactor control rods. For this example, the hot gas feed rate is taken as the heatinput Q (h%
pu)
to
theboiler.
The boiler steam plessure P5(in
bars) is influenced by Q and thethrottle
valve opening .4'"(in
%p")
which acts a speed governorfor the turbine.
Thetotal
perturbationto
the boiler is modelled as additive steam perturbation P¿at
theoutput.
The measured outputsare.lf
and Ps, and the actuator inputs are Q and
.4".
The disturbance inputs arc E¿ and P¿. The design objectivefor the
station control system isto
suppressthe
disturbance ofI{
from -t¿ and P¿, andto
keep the variations ofPs within limits.
This
exampleis
usedto
demonstrate the useof J7- optimization for
designinga
controllerin
ageneralized
two-input
(vector) and two-output (vector) SCC plant and the resultingIl*
controllerwill
then besimplified.
Here, neither robust performance nor robuststability
is a design concern, since wewill
consider operationat full
power onlywith little
v¿riationsin
plant dynamics. Coor- dinated control of actuators Q andÁ"
for the best possible regulation ofIf
is ofprimary
interest.In
other words, we are dealingwith
a nominal performance optimization problem.4.1
A Multiv¿riableJl*
DesignThe
internal
configurationof the
SCCplant
(seethe
Appendixfor its
state-space realization) is shownin
Fig.9.
The externalinput
vector d:= lE¿
PalT represents the disturbancesto
the power station and the externaloutput
vector e:= [¡f
Ps QA"]T
represents the responses to be minimized.The actuator demands Q and
A"
are included as constraintsto
prevent saturationsin
the caseof
large andabrupt
disturbances. The internal feedback signal vector g:= IN Ps]"
and u,=
lQA"lT
are signals providedfor the control
systemfrom plant instrumentation. Let M be the
transfer functionmatrix
mapping the closed-loop external signalsfrom d to e.
Theã-
design problem is formulated as anoPT
problemto
minimize the weighted -Eæ norm ofM.
4.2 Weights Selection and Controller Design
Input
weight W¿ in' OPT
is used to scale the magnitudes of the worst-c aseE¿
and. P¿ disturbances, and has been chosen to be a constant diagonalmatrix
of thear- m =
l3 1] . O*n"t
weightIZ,
is chosen
to
be diagonalwith
entries u)rt'u)2¡u3
and.wain RHæ.
Weights ?o1 and u)2 arechosento
be low-pass ûlterswith
appropriatecut-off
frequenciesto
representthe
required disturbance rejection bandwidths from dto e.
\Meightsu3
and u)4 ã,re high-passfllters
representing the actual useful bandwidths of the actuators. The frequency responses of the weightsfor
the final design are shownin Fig.
10.The
-rYæ-optimal cont¡olleris
obtainedby solving
theOPT
problemiterativel¡
andtrading
off the relative bandwidths of tu1 and w2to
achieve a satisfactory compromise between the regulation of.lf
andP5.
The closed-loop response of e as the result of2%
pu dropin
power demand (E¿) is shownin
Fig. 11. These results compare favourablywith
the existing station control system which consists of scala¡ proportional plus integral(PI)
loops and constant feedforward terms.4.3 Controller Simplication
The 'full ordet'
controllerfrom Stable-H
has L5states. It
can be reducedto 6
states (see theAppendix) by the
minimal realization procedure proposedby
Tombs (1985)without
appreciable changesin
closed-loop ïesponses. The resultant 2x 2
6-state controllerstill
has 64 parameters and isstill
consideredto
be too complexfor
implementation.It
is desirableto simplify this
controllerby identifying its
dominant modesand
algebraic couplings sothat it
can be implemented using conventionalPI
control loops.By
using elementaryrow and column
operationson the B, C and D matrix of a
state-space realizationof
the 6-state controller/f
and observingthe Nyquist plots
and step responses of its scalar elements, the controlleris
diagonalizedat
low frequencies.The dominant
dynamics of the low-frequency modelof
the controller consists,in fact, of two PI
terms as shownin Fig.
12. By adding a further second-order term whose resonant frequency and dampingratio
correspond to the13
complex conjugate pair of eigenvalues
(at
s= -.49 + j0.83)
of the/. matrix of
the original state- space controllerK,
the frequency and steps responsesof the
simplifiedcontroller match
closelywith
thoseof the
state-space model exceptfor the
very fast transientmodes. It is felt that
the high-frequency dynamicsof the
J?æ controller should be dropped becauseit is
neither desirable nor usefulto
excite the powerstation with
fast control actions.The closed-loop responses to the same 2 %
pt
dropin
-E¿ correspondingto
the simplified controller are almost identical to the 6th-order state-space model. The second order term can also be replaced by a constant gain 1, resultingin
slightly faster transient responses,at
the risk ofreaching actuator ratelimits for
large (andabrupt)
disturbancesin
,0¿.5.
CONCTUSIONSThe
optimization of robust performance and robuststability
can be formulated as an-tæ
design problembut with
a conservativeresult. Alternativel¡
one can optimize nominal performance and can check posteúori whetheror not
the nominal design meets the robust performance as well.With the robot
arm example, we have identified and demonstratedtwo major
sourcesof
conser- vatism in11*
design: the representation of (real-parameter) plant perturbations by norm-bounded uncertainty and the inadequacy of a single-objectiveH*
OPT
problem to represent multiple design objectives. Doyle's ¡r approach addresses these problems, and should become a powerfulCAD
tool when a numerically reliable and efficientp
synthesis procedure is available.At
present, numerical optimization basedCAD
packages such as Boyd's (1988) qdes and Polak's (1982)DELIGHT
can also handle muitiple design objectives, specifiedboth in
time and frequency domains.For scalar and minimum phase systems, an experienced designer can usually do better
with
classicai techniquesthan
resortingto
the arsenal of a norm-based optimization methodlike 11-.
For mul- tivariable systems, despiteits
conservatism,the
systematic11- optimization is
howevera
viabie andattractive
design tool, as demonstrated by the power-station example.An
optimization basedtechnique can give a reasonable preliminary design
on
which refinements can be addedto
satisfy additional design objectives such as time domain performance which cannot be includeddirectly in l/*.
It
should be emphasizedthat
we have only used an additive perturbationfor
modellingplant
un- certaintyin the robot arm example.
Other representations such as stablefactor
perturbations (Vidyasagar, 1985)might
model the actualplant
uncertainty more accurately and makethe
re- sulting11*
design less conserv¿tive. Futhermore,-t1-
designs are highly sensitiveto
the choiceof
weights'It
isnot
claimedthat
therY-
design is the best achievable.A
challenging exercise could beto find other
weights and modelsof plant
uncertaintyfor
improvingthe Jl-
designsfor
the robot arm.The two-input two-output
SCC approachis a natural structure in
multivariable designs where disturbances, control variables, actuator inputs and sensor outputs originateat
different locationsof the plant. The
most general settingis to
considerthe
controllerto
be alsoin the
SCC form(Nett,
1986 and Desoer, 1987) thus allowing a distu¡bance signalto
be addedto
thecontroller. A
controller synthesis procedurein
this setting needsto
be considered.The approximation of a state-space controller obtained from a norm-based optimization procedure such as
l1- by
a simple and structured one asin
Section 4.3 is neededto
economize on hardware and software implementation costs. Also and perhaps moreimportantly
gain scheduling (e.g. forstart-up and
shut-down) nonlinear characteristics(e.g.
resetwind-up )
andintegrity
behaviour(e'g'
loss of sensor signais) may be moreintuitive
and hence more manageable. The simplification procedure is usually done on an ad-hoc trial-and-error basis, and this need.s to be performedwithin
a user-friendlyCAn
environment.15
ACKNOWTEDGEMENT
The authors would
like to
thank the CentralElectricity
Generating Boardin
the United Kingdomfor
permissionto
publishthis
paper.REFERENCES
Åström,
K,
T.,. Neumann and P. Gutman (1986). A comparison of robust and adaptivecontrol.
P¡oc.2nd
IFAC
Workshop on Adaptive Sysú.Contr.
and Signal Ptocessing, Lund, Sweden, 37.Boyd,
S.P.,V.
Balakrishnan,C.H. Barratt, N.M. Khraishi, X. Li, D.G.
Meyerand S.A.
Norman (1988).A
newCAD
method and associated architectures for linear controlleis.IEEE Trans.
Auto.Cont.,,
AC-33(3),268.
Chiang R.Y. and M.G. Safonov (1987). The
LINF
computer programfor.t-
controller design.[/niv.
of
SouthernCalif.
Reporú,EECG-0785-1.
Chu, C.C., J.C-
loyle
andE.B.
Lee (1986). The general distance problemin ¡Y* optimal
controltheory. Int. J. ContrcL,44,565.
Desoer
C.A.
andA.N.
Gündes (1987),with two-input two-output
plant anducB/ER,L
}d87/L.
Algebraic theory
of linear
time-invariant feedback systems compensator. Electronic ResearchLaboratory
UC Berkeley,Do_yle J.,
K.
Gloveqr P. Khargonekar and B. Francis (1988). State-space solution to standardH2
arrdIl-
control problems.Proc.
American Cont. Conf.,Atianta,
GA.Doyle,.M..Morari, R.S- Smith,
A.
Skjellum, S. Skogested and G.J. Ballas (1987). Case study session:applications of multiv¿riable robust control techñiques.
IFAC
World Congreós, Munich, FRG.D9{1", J.C.
-(1984). Lecture notes
in
advancesin
multiy¿riablecontrol.
ONL/Honeywell Workshop, Minneapolis,MN.
Doyle, J.C. (1983). Synthesis of robust controllers and filters
with
structured plant uncertainty. Proc.IEEE
Conf.Dec.
Control, ).09.Francis,
B.
(1987).A
Coursein ll-
Control Theory. Springer.Horowitz,
I.
andM. Sidi (L972).
Synthesis of feedback systemswith
largeplant
ignorancefor
pre- scribed time-domain tolerances. -[nú.J.
Control, LB(Z),, 2BT.Nett, C.A. (1986).
Algebraic aspects of linear control systemstability. IEEE
Trans.Auto.
Cont.AC-31(10),
94L.O'Young,
S.?., I.
Postlethwaite andD.-W.
Gu (1989).A
treatmentof
model matching zeroin
the optimalH2 and.&æ
control design. To appearin IEEE
Trans.Auto.
Cont..O'You1S¡ S.D. and
B.A.
Francis (1986). Optimal performance and robust stabilization.Automatica 22, I71.
Polak
q.' P.
Seigel,T. Wuu, W.
Nyeand D.
Mayne(1982). DELIGHT.MIMO: an
interactive optimization-based multiv¿riable control system design pacliâge.IEEE Contr.
Syst.Mag.,4.
Postlethwaite assessment based
I',
S' o'Y.or1ng, on induJtrialP,-w. appricatio"r.
Gu and J.-Hopepr"ã.-Èãö (lg87a)- worta
r?@c;ú;;;;;fiunich, control
system design: a FRG.critical
Postl-ethwaite
ouEL L687/87. I'r S' o'Young
andD.-w. Gu (1982b). stable-H
user,sGuide. univ. of ofloñ
Postlethwaite
.lræ optimization. I', D'-w'
proc.' Gu,IEEE
s. o'YoungConi
andTi;r:¿;;;;;;
M. Tombs (1g86).ìi:"'
Industrial control system design usingsafonov'
Proc. IEEE M'G'
Conf. on andR'Y' g$||g-(1996).
CACS"D: Waskngton, CACS? DC.using the state-spacezæ
theory-
a design example.Tombs' state-space
M'S'
system. andI'
PostlethwaiteInt. J.
Contrò|.(1987).
Truncated balanced realizationof a
stable non-minimalvidyasagar, Ma.
M'
(1985). Control System Synthesis, a factorization approach.MIT
press, cambridge,T7
CAPTIONS
FOR FIGURESFig.
L: The Standard Compensation ConfigurationFig. 2:
BodePlots
of. Po and P"Fig. 3:
RPRS Design ConfigurationFig. 4:
Bode Plots of .9p, and ,9p"for
RobotArm
DesignsFig. 5:
Bode Plots of the Controllersfor
RobotArm
DesignsFig. 6:
Approximation ofPlant
Uncertainty by WzFig. 7:
Step Responses for RobotArn
DesignsFig. 8:
Power Generation StationFig. 9:
SCCPlant for
the Power Generation ExampleFig.
10: Bode Plots of Weightsfor
the Power Generation Station ExampleFig.
LL: Step Responseto
2 %pt
Increasein
E¿Fig.
12: SimplifiedJlæ
Controller D.1_ -(1.1*
12s)t tL -
-l
s P12
=
0'016IM
where @n=
0.96 and( =
0.5.Kt =
-80,K2= -0.5, Kr =0.8 and
[1] = lri, ;H] l1;]
The power statlon SCC plant.:
-0.1000d-03 o.39ood-01 0.5635d_O4 o.OOOOd+Oo
0.0000d+00 -0.t.¡¡2Sd+00 O.?318d_03 O.OOOOd+Oo
0.0000d+00 0.o0ood+oo _0.?049d_01 o.4oood+ot 0.o000d+00 0.oo00d+oo o.oooodf0o _o.toood+oo
-0.2535d-02 0.1235d-04 -0.1063d+00 o. 1 997d-02 0.1?96d-02 0.2031d-02
B-
0.18?8d+01 -0.1102d-01 0.4019d+02 -0.7¿159d+00 -0.6629d+00 -0.7485d+00
0.21 96d-03 -0 .1 04 0d-03 -0.1887d-02
0.7849d-04 -0.3645d-03 -0.1883d-03
-0.9558d-01 0.3421d-01 -0.5071d+02 0.4 653d+01 0.1911d+01 o.238?d+01
0.88?8d-03 -0.1053d-02 -O.2007d_o2
-0.1367d-02 0.7294d-O2 0.t284d-O2 -0.3560d+01 -0.83Aod+OO -O.2l2Od+01
-0.2251d-01 0.6284d-01 0.4437d_Ol -0.6312d-01 -0.48??d-01 o.4043d_01 -0.5835d-01 -0.1422d+OO -0.748ld_Ol B=
0.5000d-03 0.0000d+00 0.0000d+00
0 .0000d+00
0.1000d+03 0.0000d+00 0.0000d+00 0.0000d+00
0 . I 000d+03 0. 0000d+00
0 - 0000d+00 0.0000d+00 0.0000d+00 0.0000d+00 0.1000d-02 0.0000d+00
0.0000d+00 0. 0000d+00 0.0000d+00 0.0000d+00
0.0000d+00 0. ?748d-04 0. 0000d+00 0. 1006d-02 0.0000d+00 -0. 281?d-01 0.1000d-02 0,0000d+00
0 ,0000d+00 0.0000d+00
0 .0000d+00 0.0000d+00 0 .0 000d+00 0 .0 000d+00
0 .0000d+00 0 .81 97d+00 0 .0000d+00 0.0000d+00 0.0000d+00 0.8197d+00
0.0000d+00 0.0000d+00 0.0000d+00 0.0000d+00
0 .0000d+00
0 .0000d+00
-o.2924d-O2 -0.1052d+00
0 .241 5d-02 -0.4287d-01 -0.6089d-01 0.2?93d-01
0.9528d-01 0.4041d+00
0.4592d-01 -0.4019d+02 0.7{65d+00 -0.6543d+OO -0.417sd+OO 0.3174d-01 -0.1226d+OO -0.6219d+00
c-
-0.7865d+00 -0 .1 705d+01 0.0000d+00
0.1 000d+0t 0.0000d+00 0.0000d+00 0 .0000d+00 0. 1 000d+0 1
0.0000d+00 0.0000d+00 0.0000d+00 -0.¡l?3od+oo 0.1000d+01 0.0000d+00 0.0000d+00 0.1000d+01 0.0000d+00 0.0000d+00 0.0000d+00 -0.¡l?30d+OO
D-
-0.1500d+0L -0.8739d-04
-0.8110d+01 -0.4?23d-03
t
rc ô
/ =
Open-l,oop Poles:
Real fmag.
No
I
2 3 4 5 6
I
2 3 4 5 No
-50.25263 -0.510088?
-0.4935145e-01 -0.4935145e-01 -0 .247 9007 e-02 -0. 9987098e-0¡t
0.8278035€-01 -0.82?8035e-01
0.6432413e-01 -0.6432¡113e-01 Flnlte Transmlsslon Zeros:
Real Inag.
-186.255{
-1.085320 -0.1379847 -0.1379847 -0.1665605e-01 The reduced Control.l.er:
e
v
d u
F'i
t
L^
i
^2
VJ2 !r
I
P
o e
K
d
trig. 3
Bode Plots of Po o nd pe
1
.5ØØ 2 2-5ØØ
Rod/Sec ( Powers of 1Ø)
StobLe-H V2
29- 7-1q89
1øØ
8Ø
6ø
4ø
m 2ø
!
Ø
-2Ø
-4ø
1
3
Po
-6ø
triq.
À.Bode Plots of S wlth Po
1 1
.5ØØ
Rod /Sec ( Powe rs of 1Ø)
StobLe-H V2
2- 8- 1989 1Ø
-1ø
-2Ø
-3Ø
Ø
m c
3 2
l-lorowìt:,'s ne
t h6l
NfRS W1
ßPA3
L
RPßg
PR3
-4Ø
Ø
Ø.5øØ 2.5ØØ
Bode Plots of S wlth pe
1 1
.5ØØ
Rod/Sec ( Powers of 1Ø)
StobLe-H V2
2- 8- 1989 1Ø
-1Ø
-2Ø
-3Ø
Ø
m T)
2 3
N Pn"s ra/1
Horowitz's
rre tLo¿{
Wr
ß,PRS
RPRs
PrrS
-4Ø
Ø
ø.5ØØ
F¡.1.
* ¡
2.5ØØ
Bode PLot of Inner ControLLers
1
2 3
StobLe-H V2
29- 7-19A9 4Ø
2Ø
Ø
n0 !
-2Ø
-4ø
-6Ø
-8ø
4
Rod /Sec ( Powers of 1Ø)
Ho'o-ìt:.'r
r¡g+hodr'V
NPnS,co+L=l'S
NPR,S
R.P RS
-1øØ
Ø
W2 vs (Po-Pe) stobLe-H v2
29- 7-19A9
m !
1øØ
6Ø
2Ø
-2ø
-4Ø
aØ
4Ø
ø
1.sØØ 2 2.5ØØ 3
Rod/Sec (Power of 1Ø)
?o Pe
hJ.
-6ø
1
l-ia
61.2ØØ
ø.8øØ
Ø.6ØØ
ø.4ØØ
ø.2øØ
1
Step Res ponses for Robot Arm
Ø.2ØØ Ø.3ØØ
Seco nd
stobLe-n v2
29- 7-1989
NPR'S- Pe
NP ÈS -Po
nPns -
Pn( fost-t.
g)
aggs -
Po NPßs-PeHoro.
-
Pe- Po
C(ost = l' 8)!loro.- Po NP
R3
Ø
Ø
Ø.1ØØ Ø.4ØØ Ø.5Øø
Turbine eorn
SI pressure
Ps Boiler
Rel¡eoter Control rod
Reoclor
ïhrotlle
volve
AFis.
Bs
Load
disturbance
E¿
Grid
FrequencyGenerotor Grld
NPump
llot gos feed
rote
aPump
|-
Ed
Pd d
i
u a
I
N
Ps
e
a As
i v
Linearized Plant
Controller
trig.
qWo StobLe-H V2
29- 7-1989
m ø
lc
Bø
4Ø
-4ø
-2
1 Ø2
Rod/Sec ( Powers of 1Ø) 3 4
1
WL
W tjtl¡
w
L\Àl2.
-8Ø
-4 -3
Fic.
t0N
co'r
?o P-,r'¡Ps (b"')
ù ( t?. ¡-u-)
As (1?" ¡'u.)
Resp. to +2% step LN Ed stobLe-H V2 29- 7-1989
4
2
Ø
-2
-4
-6
1ØA
Seco nd s -8
Ø
36 72 144 18,Ø
PTl
K2
oR2
K1 K3
PT2
(
NA s P
s