Kontrol Teorisi

download Kontrol Teorisi

of 42

Transcript of Kontrol Teorisi

  • 7/26/2019 Kontrol Teorisi

    1/42

    Otomatik Kontrol I

    P(oransal)I(integral)D(trevsel) kontrol

    Dr. Vasfi Emre mrl

  • 7/26/2019 Kontrol Teorisi

    2/42

    By Vasfi Emre mrl, Ph.D., 2005 2

    PID kontrol matematii

    Doru akm motoru zerinde uygulama

  • 7/26/2019 Kontrol Teorisi

    3/42

    By Vasfi Emre mrl, Ph.D., 2005 3

    PID kontrol

    Kullanm kolayl dolaysyla endstride ounluklakullanlmaktadr.

    Oransal

    ntegral Trevsel PIDPID

    Oransal ntegral TrevselOransal ntegral Trevsel

  • 7/26/2019 Kontrol Teorisi

    4/42

    By Vasfi Emre mrl, Ph.D., 2005 4

    PID blok diagramPIDdenetleyicie(t) u(t)

    1

    TdsKp

    1/(Tis)

    e(t) u(t)

  • 7/26/2019 Kontrol Teorisi

    5/42

    By Vasfi Emre mrl, Ph.D., 2005 5

    Oransal kontrol

    Hatann sabit deeri iin sabit denetleyici k retilir. Hatanndevam etme durumunda kontrol k deimez.

    Hata miktarna ve Kp katsaysna bal olarak oransalkontrol, denetleyici kn retir

    Sistemin statik doruluunu ve dinamik cevabn artrr

    Hatann ve oransal kontrol katsaysnn dorudanfonksiyonudur

    up = Kp . (hata miktar)

  • 7/26/2019 Kontrol Teorisi

    6/42

    By Vasfi Emre mrl, Ph.D., 2005 6

    ntegral kontrol birikimli/kmlatif

    Hatann sfrdan farkl olma durumunda kontrol miktarn artrr

    Ki kontrol katsaysna ve hata miktarna gre denetleyicik ayarlanr.

    Dinamik cevapdan feragat ederek statik doruluk

    miktarn artrr Hata birikiminin ve integral kontrolc katsaysnnfonksiyonudur.

  • 7/26/2019 Kontrol Teorisi

    7/42

    By Vasfi Emre mrl, Ph.D., 2005 7

    Trevsel kontrol

    Hatann sabit olma durumunda, trevsel kontrol bir kretmez. Fakat kontrol hatann deime hzna bal olarakretilir.

    Hatann deime hzna ve Kd trevsel kontrolkatsaysna bal olarak, denetleyici k ayarlanr

    Dinamik cevab artrr veya gelitirir Hatann deime hznn ve trevsel kontrol katsaysnndorudan fonksiyonudur

  • 7/26/2019 Kontrol Teorisi

    8/42

    Oransal ntegral Trevsel

    up = Kp . (hata) uI =Ki (hata).dt uD =Kd .(de/dt)

    PID kontrol

  • 7/26/2019 Kontrol Teorisi

    9/42

    PID denetleyicinin uyarlamas daha mevcuttur

  • 7/26/2019 Kontrol Teorisi

    10/42

    k

    b

    x

    M

    F

    rnek

    Bu sistemin dinamik denklemi

  • 7/26/2019 Kontrol Teorisi

    11/42

    Bu sistemin dinamik denklemi

  • 7/26/2019 Kontrol Teorisi

    12/42

    Let,

    M=1kg

    B= 10 N.s/m

    K=20 N/m

    F(s)=1

  • 7/26/2019 Kontrol Teorisi

    13/42

    Ak evrim sistem davran yle olacaktr,

  • 7/26/2019 Kontrol Teorisi

    14/42

    Sadece oransal kontrol uygulayacak olursak

    X(s)F(s) KP

    s2 + 10s + (20+KP)

    Oransal kontrol katsays KP=300

  • 7/26/2019 Kontrol Teorisi

    15/42

    Oransal-Trevsel kontrol uygulayacak olursak

    PD kontrol ile beraber kapal evrim sistemin transfer fonksiyonu

    KP=300, KD =10

  • 7/26/2019 Kontrol Teorisi

    16/42

    PI kontrol uygulayacak olursak

    PI kontrol ile beraber sistemin transfer fonksiyonu

    KP=30, KI =70

    PID k t l

  • 7/26/2019 Kontrol Teorisi

    17/42

    PID kontrol

    PID kontrol ile beraber kapal evrim sistemin transfer fonksiyonu

    Kp=350, Ki=300, Kd=50

  • 7/26/2019 Kontrol Teorisi

    18/42

    Grafiklerin yorumu Oransal kontrol ykselme

    zamann ve kararl halhatasn azaltm, stam

    artrm ve yerlemezamann azaltmtr

    Trevsel kontrol stam veyerleme zamann drr.

    Fakat ykselme zamanna vekararl hal hatasna az etkisi

    vardr.

    ntegral kontrol kararl halhatasn azaltacaktr. Yalnz

    oransal kontrol katsaysdrlr ki integral kontroldeoransal kontrol gibi stam

    artrc etkiye sahiptir.

    PID kontrol uygulanmasylastamsz, hzl ykselme

    zamanl ve kararl hal hatasolmayan bir cevap eldeedilmitir.

  • 7/26/2019 Kontrol Teorisi

    19/42

    Gerekli olmad takdirde oransal, integral ve trevselkontroln nn ayn anda uygulanmasna gerek yoktur.

    Kontrolcnn mmkn olduunca basit tutulmasnda

    da fayda vardr.

    Sonu

  • 7/26/2019 Kontrol Teorisi

    20/42

    By Vasfi Emre mrl, Ph.D., 2005 20

    )(

    )()()(

    )()(

    )()(

    )()(

    )()(

    )()(

    )()(1

    )()(

    )()(

    )(1

    )()(

    )(

    )()()()(

    )()()()(

    )()()()(

    )(

    )(

    2

    )(

    )(

    2

    2

    )(

    sTKKBRsJRBLsJL

    sLRsV

    KKBRsJRBLsJL

    Ks

    sTK

    sLRs

    K

    KKBRsJRBLsJLsV

    sTK

    sLRsK

    K

    BsJsLRsV

    sKsTK

    sK

    BsJsLRsV

    sTK

    sK

    BsJsI

    sKsIsLRsV

    sTsBsIKssJ

    sKsIsLsIRsV

    TBiKJ

    vdt

    di

    LiRv

    L

    fon ksiyonutransferarasrasntorkuykvehzmotor

    sT

    s

    tvaaaa

    aaex

    fon ksiyonutransferarasrasnvoltajtahrikvehzmotor

    sV

    s

    tvaaaa

    t

    Lt

    aa

    t

    tvaaaa

    ex

    L

    t

    aav

    t

    aaex

    vL

    tt

    aaex

    Ltt

    a

    vaaaex

    Lat

    vaaaaex

    Lat

    tK

    bemf

    a

    aaaex

    Lex

    v

    PM DC Motor Modelleme

    Elektrikselksmndiferansiyel denklemi

    Mekanikksmndiferansiyel denklemi

    Yukardaki denklemlerin sfr balang artyla

    Laplacednmlerinin alnmas

    C k i bl k

  • 7/26/2019 Kontrol Teorisi

    21/42

    By Vasfi Emre mrl, Ph.D., 2005 21

    PM DC motorun ak evrim blokdiagram

    Yukardaki denklemler bu blok diagram ileifade edilebilir. Bu blok diagram zersenizayn diferansiyel denklemlerle karlarsnz

  • 7/26/2019 Kontrol Teorisi

    22/42

    By Vasfi Emre mrl, Ph.D., 2005 22

    PM DC Motor Kontrol Problem

    DC motorda neyi kontrol ederiz?1. Hz kontrol2. Sistem veriminin artrlmas3. Bozucu etkilerin etkisinin azaltlmas

  • 7/26/2019 Kontrol Teorisi

    23/42

    By Vasfi Emre mrl, Ph.D., 2005 23

    PM DC Motor Kontrol Problem

    Kapal evrim transfer fonksiyonu

  • 7/26/2019 Kontrol Teorisi

    24/42

    By Vasfi Emre mrl, Ph.D., 2005 24

    DC motor kontrol

  • 7/26/2019 Kontrol Teorisi

    25/42

    By Vasfi Emre mrl, Ph.D., 2005 25

    DC motorun kapal evrim kontrol

    Ak evrim ve kapal evrim sistem bozucu etkiye nasl cevap verir?

    LT = 0 , OL

    LT 0 , OL

    LT = 0 , CL

    L

    T 0 , CL V

  • 7/26/2019 Kontrol Teorisi

    26/42

    By Vasfi Emre mrl, Ph.D., 2005 26

    Oransal kontroln etkisi _Think of proportional controller whith proportionality constant of K

    Vex = K.r in steady_state, rAKyss .. if ryA

    K ss 1

    If 0LT

    _ Lss TBArKy ... ifA

    K1

    Lss TBry .

    IfL

    T = 0 , CL

    _ Vex = K.(r-y)

    11.

    1

    11

    .

    21

    21

    ss

    AK

    ss

    AK

    sR

    sY

    sTBsRKAsssY L...11 21

    ryAKrKA

    KAy ssss

    1

    .1

    .

    V_

    )(11

    )(11

    .

    2121

    sTss

    BsRss

    AKsY L

    Lss TAK

    Br

    AK

    AKy

    .1.1

    .

  • 7/26/2019 Kontrol Teorisi

    27/42

    By Vasfi Emre mrl, Ph.D., 2005 27

    PM DC Motor Control - Result

    OL CL

    TL= 0 0LT TL= 0 0LT

    ryss

    if disturbance is

    zero andmodel of

    the systemcorrect,

    selecting control

    constant

    Lss TBry .

    if disturbance is not

    zero, we will

    observe the

    amplified effect of

    disturbance of the

    output

    rKA

    KAyss

    .1

    .

    if no disturbance

    increasing gain K

    willresult inreducing

    steady_state error,

    ryss

    Lss TAK

    Br

    AK

    AKy

    .1.1

    .

    in case of disturbance,

    increasing K will decrease

    the effect of disturbance

  • 7/26/2019 Kontrol Teorisi

    28/42

    By Vasfi Emre mrl, Ph.D., 2005 28

    PM DC Motor Control - ResultFor the system, increasing gain K decreases steady_state error. However, is there a limit

    increasing gain K ?

    Physical system limitations, stability

    Check stability by locking at the pole locations of the system for increasing gain K

  • 7/26/2019 Kontrol Teorisi

    29/42

    By Vasfi Emre mrl, Ph.D., 2005 29

    PM DC Motor Control Problem

    For given example (DC motor speed control), proportional control does not result in 0

    sse Apply integral control

    t

    I

    dteT

    Kptu

    0

    .)( sT

    Kp

    sE

    sU

    I

    )(

    )(

    IT = integtal time (reset time)

    t

    I

    p

    dtyrT

    K

    Vex0

    feedback control

    Lt

    I

    pTBdtyr

    T

    KAyyy .)(

    0

    2121

    L

    Iy

    TBrT

    KpAy

    T

    KpAyyy .

    .)( 2121

    in steady_state ifL

    T =0 ryss

    PM DC Motor Control Problem

  • 7/26/2019 Kontrol Teorisi

    30/42

    By Vasfi Emre mrl, Ph.D., 2005 30

    PM DC Motor Control Problem -

    Stability

  • 7/26/2019 Kontrol Teorisi

    31/42

    By Vasfi Emre mrl, Ph.D., 2005 31

    Example

    If we have mechanic system with transfer function 1

    1)(

    ssT

    Kt G(s)+-

    (s)T

    PIDr(s)

    Kt+-

    (s)T

    r(s)

    sT

    sTsTTK

    i

    idip

    12

    1

    1

    s

  • 7/26/2019 Kontrol Teorisi

    32/42

    By Vasfi Emre mrl, Ph.D., 2005 32

    Example cont.

    For a DC motor we studied transfer function (relation) betweenexcitation voltage and output angular velocity (s)/Vex(s).However we know that armature current is directly (almost)related to generated torque.

    Kt: torque constant (Nm/A).

    If we think of only bearing damping and carried load, a conveyortransfer function is simply J: motor inertia

    B:bearing damping

    T:excitation torque

    This is a linear system, yet conveyor load is subject to change, sodisturbance changes.

    BJss

    T

    1)(

  • 7/26/2019 Kontrol Teorisi

    33/42

    By Vasfi Emre mrl, Ph.D., 2005 33

    ExampleOpen Loop Response

    0 1 2 3 4 5 6 7 8 9 100

    0.005

    0.01

    0.015

    0.02

    0.025

    time (sec)

    Angularvelocity(r

    ad/sec)

    Unit Step Response of the Open Loop DC motor System (Electrical Dynamicsare neglected)

  • 7/26/2019 Kontrol Teorisi

    34/42

    By Vasfi Emre mrl, Ph.D., 2005 34

    ExamplePerformance?

    ess = ? =%Mp

    tr = 1,8/nts = 4,6/ n.tp = / dIn standard 2nd order form

    If TD is selected to be 1;

    Char. eq. of the system

    sTss rs

    1...lim0

    itpiD

    tp

    itpiD

    iitp

    iiD

    tpiD

    tp

    TKKTT

    KKs

    TKKTT

    TTKKs

    sTsTTKKTT

    KK

    sT

    ...

    .

    ...

    ..

    1.....

    .

    )(2

    2

    )1..(

    .

    /1.1.

    .

    )(2

    2

    tpi

    tp

    i

    tp

    tp

    KKT

    KKss

    TssKK

    KK

    sT

    )1..(

    .,12,0

    )1..(

    . 22

    tpi

    tp

    nn

    tpi

    tp

    KKT

    KK

    KKT

    KKss

  • 7/26/2019 Kontrol Teorisi

    35/42

    By Vasfi Emre mrl, Ph.D., 2005 35

    ExampleLittle Talk

    If KpKt is very large compared to 1, then, numerator gain will be almost 1.Also, if Ti is close to 1 then, numerator has one zero at s=0.

    Q: If ts is able to be set?

    A: No, since ts is already set to 9,2 sec. Because of TD =1 and so 2n=1

    )1.(

    .2

    1

    tpi

    tp

    KKT

    KK

  • 7/26/2019 Kontrol Teorisi

    36/42

    By Vasfi Emre mrl, Ph.D., 2005 36

    Examplecont.

    In what condition ess will be minimum?

    In this case

    For r(s) = 1/s unit stepess = 0

    For r(s) = 1/s2 unit ramp

    ess = Type I

    If Ti is very small Kp is large enough, ess 0

    sTsse rsss 1...lim0

    )1.(

    .

    1.

    1

    1.

    1

    )](1[2

    2

    KtKpTi

    KtKpss

    sKtKp

    sKtKp

    sT

    KtKp

    Ti

    .

  • 7/26/2019 Kontrol Teorisi

    37/42

    By Vasfi Emre mrl, Ph.D., 2005 37

    Examplecont.

    Performance specs as, %Mp = %10, tr 1 sectr= 1,8/n 1 sec, n 1,8

    .

    )1.(718.0

    .

    )1.(24.3

    .

    )1.(718.0

    .

    )1.(718.0.

    59.0

    )1.(

    .2

    159.0

    3.5

    3.2

    )1,0(ln

    1,0ln

    )1.0(ln)1.0(ln,1/1.0ln,1.0

    )1.(24.3

    .),1.(24.3.,8.1

    )1.(

    .

    222

    2222221/2

    tp

    tp

    i

    tp

    tp

    i

    tp

    tp

    itpitp

    tpi

    tp

    tp

    tp

    itpitp

    tpi

    tp

    n

    KK

    KKT

    KK

    KKT

    KK

    KK

    TKKTKK

    KKT

    KK

    e

    KK

    KKTKKTKK

    KKT

    KK

    or

  • 7/26/2019 Kontrol Teorisi

    38/42

    By Vasfi Emre mrl, Ph.D., 2005 38

    ExampleClosed Loop Results

    0 5 10 150

    0.2

    0.4

    0.6

    0.8

    1

    1.2

    1.4

    time(sec)

    angularvelocity(rad/sec)

    Closed Loop PID control of a DC motor

    (Electrical Dynamics are Neglected)

    Kp=10, Td=1

    Ti=0.0293 black

    Ti=0.0351 brown

    Ti=0.0410 light blue

    Ti=0.0468 red

    Ti=0.0527 green

    Ti=0.0585 blue

  • 7/26/2019 Kontrol Teorisi

    39/42

    By Vasfi Emre mrl, Ph.D., 2005 39

    Example%Mp = %1, ts 6 secis required. Find

    constants K and b. a=1

  • 7/26/2019 Kontrol Teorisi

    40/42

    By Vasfi Emre mrl, Ph.D., 2005 40

    Example

    What is the system steady state error for unit step input and system type

    882.068.1

    )67.1)(1(67.1)(

    )(

    )(2

    ss

    sssT

    sR

    sY

    882.068.1

    )67.167.2(67.11

    1.lim)(1)(.lim

    2

    2

    00ss

    ss

    sssTsRse ssss

    16.2882.068.1

    9.178.267.0lim

    2

    2

    0

    ss

    sss type 0

  • 7/26/2019 Kontrol Teorisi

    41/42

    By Vasfi Emre mrl, Ph.D., 2005 41

    Example

    if = 0,1 is archievable for various K in the system, find it b = 1.67 a = 1

    various ways to solve1_ from roots of char. eq.

    2_ from standart nd2 order dif. eq.

    2_ 02 22 nnss assuming K>0

    01

    67.1

    1

    67.22

    K

    Ks

    K

    Ks 067.167.2

    1

    2

    mmssK

    Km

    mn 67.12 mn 67.22

    03.167.12

    67.2 m

    m

    m

    K

    K

    196.01.0

    K

    K

    1096.0 2

    K = 0.01

    Dependence on settling time on PID parameters

    K and bDependencies of Percent overshoot on PID

    parameters K and b

  • 7/26/2019 Kontrol Teorisi

    42/42

    By Vasfi Emre mrl, Ph.D., 2005 42

    0.1

    1.1

    1.7 0.10

    0.90

    1.700

    20

    40

    60

    80

    100

    settlin

    gtime(sec)

    b

    K

    80.00-100.00

    60.00-80.00

    40.00-60.00

    20.00-40.00

    0.00-20.00

    0.

    10

    0.

    50

    0.

    90

    1.

    30

    1.

    61

    1.

    65

    1.

    69

    2.

    00

    0.10

    1.300

    5

    10

    15

    20

    25

    30

    35

    40

    PercentOvershoot

    b

    K

    35.00-40.00

    30.00-35.00

    25.00-30.00

    20.00-25.00

    15.00-20.00

    10.00-15.00

    5.00-10.00

    0.00-5.00

    0.

    10

    0.

    50

    0.

    90

    1.

    30

    1.

    61

    1.6

    5

    1.

    69

    2.

    00

    0.10

    1.200.00

    10.00

    20.00

    30.00

    40.00

    50.00

    60.00

    70.00

    peaktime

    K

    b

    Dependencies of peak time on PID parameters K

    and b

    60.00-70.00

    50.00-60.00

    40.00-50.00

    30.00-40.00

    20.00-30.00

    10.00-20.00

    0.00-10.00