You are on page 1of 14

OUTLINE

SLIDINGMODECONTROL Variablestructuresystems
Slidingmodecontrol
Motivatingexample(Khalil)
M.SAMIFADALI EquivalentControl
PROFESSOREBME
UNIVERSITYOFNEVADA,RENO

2
VARIABLESTRUCTURESYSTEMS
VARIABLESTRUCTURECONTROL
Dynamicsystemsoftheform
where has discontinuitieswithrespect Statedependentswitchingfeedbackcontrol
tosomearguments thatintentionallychangesthestructureofthe
Occurinproblemsinphysics,control system.
engineeringandmathematics. Origins:relaycontrol,bangbangcontrol.
Occurnaturallyinsomephysicalsystems,e.g. Variablestructurecontrolsystem: composedof
forsomeelectricmotorsandpower independentstructuresandaswitchinglogic.
converters.
Overallsystembehaviorisunlikeanyofits
Forsuchsystems,thecontrollawisnaturally structures.
discontinuous
3

4
EXAMPLE:LINEARPLANT Assume a=1and|k|=12
PLANT: LINEARSTATEFEEDBACK
Unstableequilibrium
CASE1:k 12 a 4
2
x1 (t ) x2 (t ) CONTROLLAW:
pointattheorigin.
x2 (t )ax2 (t )u (t ) u ( t ) kx 1 ( t ) 1, 2 0.5 j 3.4278
x2(t)
Discontinuous argument
3

2
Closedloopeigenvalues
1
x1(t)
a a 4k2
a a 4k 2
-1 -0.8 -0.6 -0.4 -0.2 0.2 0.4 0.6 0.8 1
1 1
2 2 -1
-2
For theeigenvalueshavedifferent -3
propertiesfordifferent and parameter values. -4

6
CASE2: k 12 a 2 4 The green dashed line equation candefined as
x 2 ( t) Saddlepointattheorigin.
1 3, 2 4
x 2 ( t)
4
4

s (x) x 2 (t ) 1 x1 (t )
3

3
2

2
x 2 (t ) 3 x1 (t ) 1

x 1 (t )
-1 - 0 .8 -0.6 - 0 .4 -0 . 2 0 .2 0 .4 0 .6 0 .8 1

1
-1

x 1 (t ) -2

-1 - 0 .8 -0. 6 - 0 .4 -0 .2 0 .2 0 .4 0 .6 0 .8 1
-3

-1
-4
s(x)=0
-2

-3 Choosethe discontinuouscontrollaw
-4

Greendashedline(eigenvectorofstableeigenvalue): 12 x1 (t ), s(x) x1 0
systemtrajectoriesconverge totheorigin. u (t )
VSSTHEORY: usethisstructure withadiscontinuous to 12 x1 (t ), s(x) x1 0
makethesystemstable.
7

8
OVERALLSYSTEMBEHAVIOR
SLIDINGMODES
Unlikeanyofitsstructures. s(x) x2 3 x1 0 switching function ischosen from system
trajectories.
VariableStructureSystem(VSS) canpossess Ingeneral,theswitchingfunctionischosenusingthe system
trajectories.These are known assliding modes.
newpropertiesnotpresentinanyofthe x2(t)

structuresused. 3

Newswitching 2

Intheexample,wehave function
1

Case1:UnstableEquilibrium x1(t)
-1 -0.5 0.5 1

Case 2:Saddle Point -1 x2+3x1=0


x2+c1x1=0 (0<c1<3)
VSSSystem:Asymptotically Stable
-2

10
-3

9
SLIDINGMODECONTROL (SMC) SMCTRAJECTORIES
SMCdesign involves twosteps: Atrajectorystartingfromanonzeroinitial
(i)Selectionofstablehyperplane(s) inthe condition,evolvesintwophases:
state/errorspaceonwhichmotion should a)Reachingmode,inwhichitreachesthe
berestricted,calledtheswitching slidingsurface,and
function,and
b)Slidingmode,inwhichthetrajectoryon
(ii)Synthesisofacontrollaw whichmakes reachingtheslidingsurface,remainsthere
theselectedslidingsurfaceattractive . foralltimesandthusevolvesaccordingto
thedynamicsspecifiedbythesliding
surface.
11

12
PLANTANDSWITCHEDCONTROL

I:DESIGNOF SWITCHINGFUNCTION Eachmatrixentry:continuouslydifferentiablew.r.t.


Switched(Corrective)Control:

SwitchingSurface: dimensionalmanifoldin
determinedby constraints.

13

14
PropertiesofSwitchingFunctions Importance ofSwitching Functions
a) Orderofswitchingfunctionislessthanorderofplant
Example:2ndordersystem 1storderswitchingfunction e2
III. IV.
c1<c2<c3 e1 e2
b)Slidingmodedoesnotdependonplantdynamicsandis
e2 ce2 ksign( s )
determinedbyparametersoftheswitchingfunctiononly e2e
e1(0),(0)
(intheexampleon only) e1
c)Switchingfunctiondoesnotdependonthecontrollaw.

II. c1 I.
0 2 c3 c2
UpperLimitc3 : dependsonthephysicalpropertiesofthe
1
systemandthetechnologyused.
LowerLimitc1 : dependsontheallowabletrackingtime
TradeoffbetweenPERFORMANCEandROBUSTNESS
15

16
Switching Function Design Switching Function Design
Constant TimeVarying
Nonlinear Linear e e
e e

e(0),(0)
e e(0),(0)
e
e(0),(0)
e e(0),(0)
e e e
e e

e e2
ADVANTAGES ADVANTAGES 0.5

0.4

+Appropriateforglobaldynamic +Easytoobtainthesurfaceparameters 0.3

propertiesofnonlinearsystems DISADVANTAGES 0.2

+Numerousdesignoptions Maynotbeappopriateforsystem e(0),(0)


e
0.1

e1
0

DISADVANTAGES dynamics, ingeneral. e -1 -0.8 -0.6 -0.4 -0.2

-0.1
0 0.2 0.4 0.6 0.8 1

Difficulttofindnonlinear functions Magnitudeofthecontrolsignalincreases -0.2

Difficulttoobtainthesurface directlyproportionaltothetrackingerror. -0.3

-0.4

parameters Fewer designoptions -0.5

17

18
Switching Function Design II:FINDACONTROLLAWTOREACH
ANDSTAYTHEREAFTER
MostcommonchoiceisLTI SwitchingSurfaces
AttractiveSurface(SlidingSurface):Trajectories
outsidethesurfaceconvergetoit.Onceonthe
surface,trajectoriesremainonit
u(t)=uc(t)+ueq(t)
e x0 CORRECTIVECONTROL EQUIVALENTCONTROL
2, 1 s2=0 3, 2 (compensatethedeviations (makesthederivative
fromthe sliding surface oftheslidingsurface
x0
e s1=0 to reach the sliding surface) equalzero tostayon
theslidingsurface)
19

20
s=0
EXISTENCEOF SLIDING MODE LYAPUNOVFUNCTIONAPPROACH
Asystemwith inputscanhave PositivedefiniteLyapunovfunction
switchingfunctionsand sliding
surfaces.
Thecontrollawdesignandexistenceof
Derivative
slidingmodearesurveyedin:
Hunget.al.,VariableStructureControl.A
Survey,IEEETran.IndustrialElectronics,40(1),
1993.
DirectSwitchingApproach
ReachingLawApproach Choose foranegativedefinite :
LyapunovFunctionApproach trajectoriesconvergetothesurface.

21

22
SWITCHINGSURFACE CORRECTIVECONTROL
Correctivecontrol :Usehighspeed
switchingtodrivethestatetrajectorytoa
specifiedswitchingsurface.
Choosetheinputamplitudesufficientlylarge
Attractivesurface:trajectoriesoutsidethe
tomake negativedefiniteforany :
surfaceconvergetoitandonesstartingonthe
Robustw.r.t.modelingerrors.
surfacestayonit.
Typicalchoice
Localcontrol :Designsurfacesothatthe
systemhasgooddynamicbehavioron
(e.g.bypoleplacementforalinear
surface).
23

24
CORRECTIVECONTROL : EQUIVALENTCONTROL
SISYSTEM/SCALARLINEARSURFACE
Motiontangenttotheswitchingsurface

CorrectiveControl:

Equivalentdynamics(onthesurface)

25

26
EQUIVALENTCONTROL :
LINEARSYSTEM/LINEARSURFACE POLEPLACEMENTDESIGN
LinearDynamics
Motiontangenttotheswitchingsurface
Similaritytransformation

EquivalentControl:
SlidingModeDynamics(equivalentsystem):order
zeroeigenvalues
nonsingular
Constraint:
27

28
LEMMA:CONTROLLABILITY LEMMA:CONTROLLABILITY
If iscontrollablethen is If iscontrollablethen is
controllable. controllable.
Proof:Controllabilityisinvariantunder Proof:Controllabilityisinvariantunder
similaritytransformation, similaritytransformation,

29

30
EQUIVALENTCONTROL ONSWITCHINGSURFACE

Usethetransformedsystem

Assign eigenvaluesusingpoleplacementtoselect
andselectanynonsingular
Eigenvaluesinvariantundersimilaritytransformation
31

32
EXAMPLE MATLAB
>>k=place(A11,A12,5)
k=
1.6000
3.2000
>>Ku=[k*A11+A21,k*A12+A22]/T
Ku=
1.20002.60001.4000
9.60003.20006.2000
>>eig(AB*Ku)
ans =
0.0000
5.0000
0

33

34
CHATTERING
EXAMPLE(SLOTINE)
Intheory,thetrajectoriesslidealongthe
switchingfunction.
Inpractice,thereishigh frequencyswitching.
Occursinthevicinityoftheswitchingsurface ModelUncertainty
duetononidealswitchinge.g.delays,
hysteresis,etc.
Calledchatteringbecauseofthesoundmade
byoldmechanicalswitches.
35

36
SLIDINGCONDITION
EXAMPLE(KHALIL)

unknown

Designastatefeedbacklawthat
stabilizestheorigin.
Choose

37

38
SLIDINGMANIFOLD
LYAPUNOVFUNCTION
Constrainthesystemtothesurface
(manifold)

Motiononmanifold

Stablefor forany .

Drivethetrajectoriestothesurfaceand
maintainthemonit.
39

40
SLIDINGMODECONTROL
COMPARISONPRINCIPLE
Assume

satisfies

Decreasestillitreachesitsminimum( )
>0 Themanifold isreachedinfinitetime.
Staystherebecause

41

42
REGIONOFATTRACTION
CONSTANTBOUND (invariantset)
Assume ,i.e.

(relay)
Relayleadstoafiniteregionofattraction
(invariantset)
Invariantset
43

44
TRAJECTORIESIN REGIONOFATTRACTION
Trajectoriesapproachtheinvariantset 4

Insidetheinvariantset,trajectories 1

approach if 0
-5 -3 -1 1 3 5

-1

-2

-3

-4
5 0 5

45

46
SPECIFICSYSTEM:PENDULUM
SIMULATIONDIAGRAM

Let

Simulatethesystemassumingnegligibleswitching
delays.
47

48
NOSWITCHINGDELAYS
SWITCHINGDELAYS
0

Chattering:highfrequencyswitching.
-0.5
Occursinthevicinityoftheswitchingsurface
-1
duetoswitchingdelays.

-1.5
Iftherelayswitchestopositivefromnegativeat
andfromnegativetopositiveat ,
-2 thesystemexhibitschatteringbehavior.

-2.5

0 0.5 1 1.5 2

49

50
CHATTERING(SWITCH
REDUCINGCHATTERING
0.2
Addacontinuouscontrolcomponent
0

-0.2 Changethesgn(.)functiontosat(.):linear
-0.4 controlinsideaboundarylayerofwidth
-0.6

-0.8

-1

-1.2

-1.4

-1.6

0 0.2 0.4 0.6 0.8 1


51

52
With signum function
0.2

EXAMPLE SWITCHINGDELAY0.1 0

-0.2

-0.4

-0.6

-0.8

Addacontinuouscontrolcomponent 0.2
-1

-1.2

0 Ultimatelybounded -1.4

-1.6

Changethesgn(.)functiontosat(.):linear butdoesnotconverge
0 0.2 0.4 0.6 0.8 1

-0.2
controlinsideaboundarylayerofwidth totheorigin.
-0.4

-0.6

-0.8

-1

-1.2

-1.4

-1.6
-0.2 0 0.2 0.4 0.6 0.8 1

53

54
CONCLUSION
SlidingModeControlis
REFERENCES
DETERMINISTIC(only bounds ofvariations are 1. R.A.DeCarlo,S.H.Zak,andG.P.Mathews,Variable
StructureControlofNonlinearMultivariableSystems:A
considered)
Tutorial,ProceedingsIEEE,Vol.76,No.3,March1988.
NONLINEAR(the corrective term isnonlinear)
ROBUST (onceontheslidingsurface,thesystemis 2. H.Khalil,NonlinearSystems,PrenticeHall,UpperSaddle
River,NJ,2003.
robusttoBOUNDEDPARAMETERSVARIATIONSand
BOUNDEDDISTURBANCES) 3. J.J.Slotine andW.Li,AppliedNonlinearControl,Prentice
Hall,EnglewoodCliffs,NJ,1991.
Switchingfunctiondesignandcontrollawdesign
determinetheperformance. 4. KarKeungYoung,P.Kokotovic,andV.Utkin,Asingular
Althoughchatteringisaproblem,variousmethods perturbationanalysisofhighgainfeedbacksystems,IEEE
Trans.Automat.Contr.,Vol.22,No.6,pp931938,1977.
areavailableformitigatingoreliminatingit.
5. S.H.Zak,SystemsandControl,OxfordUniv.Press,NY,2003.
55

56

You might also like