PID_slides.pdf

Embed Size (px)

Citation preview

  • 8/10/2019 PID_slides.pdf

    1/36

    PID Controls

    Issues on Control Design

  • 8/10/2019 PID_slides.pdf

    2/36

    Res onse S ecs Review

    MAAE 4500: Feedback Control 2

  • 8/10/2019 PID_slides.pdf

    3/36

    jtan

    tann n

    s

    x djn

    2

    r

    d

    np

    d

    n

    p 21e

    s cos cosnn

    (5%); (2%)

    MAAE 4500: Feedback Control 3

  • 8/10/2019 PID_slides.pdf

    4/36

    Movin Poles1,2 ds j

    ( ) cosd

    y t Ae t

    const.

    MAAE 4500: Feedback Control 4

  • 8/10/2019 PID_slides.pdf

    5/36

    1,2

    ( ) cos

    d

    t

    d

    s j

    y t Ae t

    const.

    MAAE 4500: Feedback Control 5

  • 8/10/2019 PID_slides.pdf

    6/36

    21

    pM e

    const.

    MAAE 4500: Feedback Control 6

  • 8/10/2019 PID_slides.pdf

    7/36

    Effect of

    Pole(s)

    Location(s)

    Time

    Response

    MAAE 4500: Feedback Control 7

  • 8/10/2019 PID_slides.pdf

    8/36

    Effect of

    Pole(s)

    Location(s)

    Time

    Response

    MAAE 4500: Feedback Control 8

  • 8/10/2019 PID_slides.pdf

    9/36

    Effect of

    Pole(s)

    Location(s)

    Time

    Response

    MAAE 4500: Feedback Control 9

  • 8/10/2019 PID_slides.pdf

    10/36

    Effect of

    Pole s

    Location(s)

    on System

    Response

    MAAE 4500: Feedback Control 10

  • 8/10/2019 PID_slides.pdf

    11/36

    11p d

    i

    K T sT s

    Plant

    1

    d

    Plant

    pK

    1

    i

    MAAE 4500: Feedback Control 11

  • 8/10/2019 PID_slides.pdf

    12/36

    Ziegler-Nichols Rules

    Industrial applications: engineers can tunethe response characteristics of a plant on site

    Based on experimental step responses

    Useful when a good mathematical modeldoes not exist (or it is too complicated)

    MAAE 4500: Feedback Control 12

  • 8/10/2019 PID_slides.pdf

    13/36

    r e e e e -

    If the plant does not involve integrator(s) nor dominant

    comp ex-con uga e po es, e un s ep response can e

    approximated by a S-shape curve as follows (first-orderres onse lus a time dela :

    K

    ( )c t ( )u t

    1

    ( ) LsC s Ke

    ttangent to inflection point

    t

    MAAE 4500: Feedback Control 13

    TL

  • 8/10/2019 PID_slides.pdf

    14/36

    First Method for Tuning the Controller

    controllerP

    pK

    iT

    dT

    T

    PI

    L 0

    T L

    PID

    0.9L 0.3

    1.2L 2L 2

    MAAE 4500: Feedback Control 14

  • 8/10/2019 PID_slides.pdf

    15/36

    o e a or e , e rs e o g ves or e

    compensator TF:

    1 1( ) 1 1.2 12 2

    c p d

    i

    T LG s K T s sT s L Ls

    21

    0.6

    s

    LT

    s

    one ole at the ori in and one double zero at s = 1L

    Q. Not good for type I systems. Why?

    MAAE 4500: Feedback Control 15

  • 8/10/2019 PID_slides.pdf

    16/36

    Ste 1: Desi n a ro ortional controller T = 0

    Ti=) and increase the gain until marginalabsolute stability is reached (pure imaginary. , p= cr

    The method does not apply if the condition does not

    exist for a finite gain

    Step 2: Determine the period of the correspondingundamped oscillations, Pcr (P=2/)

    Step 3: Use the next table to tune the controller Fine-tune analyzing the time response

    MAAE 4500: Feedback Control 16

  • 8/10/2019 PID_slides.pdf

    17/36

    Second Method for Tuning the Controller

    controllerP

    pK

    iT

    dT

    K

    PI

    2

    cr 0

    P

    PID

    .cr

    1.2

    cr

    P

    . cr 2

    cr

    . cr

    MAAE 4500: Feedback Control 17

  • 8/10/2019 PID_slides.pdf

    18/36

    o e a or e , e econ e o g ves or

    the compensator TF:

    1 2( ) 1 0.6 1 0.125c p d cr cr

    i cr

    G s K T s K P sT s P s

    2

    4

    cr

    s

    P

    . cr cr s

    one ole at the ori in and one double zero at s = 4Pcr

    MAAE 4500: Feedback Control 18

  • 8/10/2019 PID_slides.pdf

    19/36

    Example 1 (PID, Ziegler-Nichols):

    11

    p dK T s

    1

    1 5s s s

    ( )C s( )R s

    excluded. Use the second method and set Td= 0, Ti=:

    3

    ( ) 1 5p

    pR s s s s K

    2 6

    30

    ps K

    K

    3 26 5p

    ps s s K

    0

    6

    p

    s

    s K

    cr

    MAAE 4500: Feedback Control 19

  • 8/10/2019 PID_slides.pdf

    20/36

    using the critical value of Kinto the closed-loop

    characteristic equation:

    3 26 5 30 0s s s From the table:

    0.6 18K K

    3 2

    2

    6 5 30 0j j j

    0.5 1.405

    0.125 0.3512

    i crT P

    T P

    2 22.81

    crP

    24

    2.81s

    2

    . .

    6.322 1.423

    c s

    s

    MAAE 4500: Feedback Control 20

    s

  • 8/10/2019 PID_slides.pdf

    21/36

    Unit step response

    too large overshoot

    2

    4 3 2

    ( ) 6.322 18 12.811

    ( ) 1 6 11.322 18 12.811

    c

    c

    G GC s s s

    R s G G s s s s

    MAAE 4500: Feedback Control 21

  • 8/10/2019 PID_slides.pdf

    22/36

    26.322 1.423s

    cs

    Tr modif onl the controller ain. Draw root-locus lot:

    2 1 C s.K

    s 1 5s s s

    B s

    2

    1.423( ) K sB sKG s G s H s

    ( ) 1 5R s s s s

    MAAE 4500: Feedback Control 22

    .

  • 8/10/2019 PID_slides.pdf

    23/36

    Root-locus

    1 = 4.1328

    6.322 = 0.3737 1.6213

    s

    K s

    Overshoot is first

    insensitive to K,

    4 = 1.1197s

    increase!

    Lines of constant overshoot, for:= . =cons

    MAAE 4500: Feedback Control 23

  • 8/10/2019 PID_slides.pdf

    24/36

    For an overshoot of 20%:

    21

    20.2 ln 0.2 1.61 0.456

    1p

    M e

    ne w ave o c ange e ou e zeroes pos on: s

    found by try and error that by moving the zeroes to0.6:

    2

    0.6sK

    s

    1

    1 5s s s

    ( )C s( )R s

    ( )B s

    2

    0.6( ) K sB s

    2( ) 1 5c

    R s s s s

    the next root-locus lot is roduced:

    MAAE 4500: Feedback Control 24

  • 8/10/2019 PID_slides.pdf

    25/36

    Root-locus

    must be acceptable

    Mp < 20%Double pole

    at s=0

    Single pole at

    s=0.6

    2

    0.6s 1 ( )C s( )R s

    s=5

    s

    1 5s s s

    ( )B s

    MAAE 4500: Feedback Control 25

  • 8/10/2019 PID_slides.pdf

    26/36

    Unit step response

    , a ou

    2

    2

    .30

    1 5( )

    ( ) 1 0.6 1

    c

    c

    s

    s s s sG GC s

    R s G G s

    2

    1 5

    30 36 10.8

    s s s s

    s s

    .s s s s

    MAAE 4500: Feedback Control 26

  • 8/10/2019 PID_slides.pdf

    27/36

    Example 2 (PID, Nise): Given the system below,

    design a PID controller so that the system operates

    with 2/3 of the uncompensated peak time at 20%

    -

    21 3K K s

    s 3 6 10s s s

    Step 1: Evaluate performance of uncompensatedsystem operating at 20% overshoot (= 0.456).

    a n e roo - ocus p o an raw eovershoot line to determine the closed-loop dominantpoles:

    MAAE 4500: Feedback Control 27

  • 8/10/2019 PID_slides.pdf

    28/36

    Step 2: Design a PD

    compensa or o meethe transient response

    specs (determine the

    zero location and the

    PD controller gain).

    parts of the dominant

    pole must be:0.297

    10.569p

    d

    t

    .

    2 3 0.297

    8.13

    d

    d

    MAAE 4500: Feedback Control 28

    an .

  • 8/10/2019 PID_slides.pdf

    29/36

    ,compensator. The

    contribution from the

    8.13 15.867

    8

    3 6 10s j

    s

    s s s

    compensator zero must

    be:

    .

    . .

    From the Figure:

    15.867

    tan 18.3718.13

    55.909

    cz

    z

    1( ) 55.909G s K s

    MAAE 4500: Feedback Control 29

  • 8/10/2019 PID_slides.pdf

    30/36

  • 8/10/2019 PID_slides.pdf

    31/36

    Step 3: Design a PI

    2 1

    55.909 0.5 8( ) ( ) ( )

    3 6 10

    K s s sG s G s G s

    s s s s

    compensa or o meethe steady-state error

    (determine the zero

    location arbitrarily near

    the origin [e.g.0.5]

    0.5cs z s

    the origin):

    2s s

    Re-draw the root locus

    to determine the finalgain and the static

    dis lacement error and

    2 10

    lim (0) (0) (0)

    0

    ps

    ss

    K G G G

    e

    (type 1)

    MAAE 4500: Feedback Control 31

    the steady-state error)

  • 8/10/2019 PID_slides.pdf

    32/36

    Step 4 Check the specs:

    uncompensated PD PID

    Dominant poles 5.415 10.569j 8.13 15.867j 7.519 14.676j

    Gain, K121.510 5.339 4.604

    Damping ratio, 0.456 0.456 0.456

    Natural freq., n 11.875 17.829 16.49

    Max. overshoot,Mp0.2 0.2 0.2

    Peak time, tp 0.267 0.198

    (=2/3* 0.267)

    0.214

    (~4/5 * 0.267)

    Static dis l. error K

    Steady-state error, ess 0.156 0.070 0

    MAAE 4500: Feedback Control 32

    If not satisfactory redesign the PI controller

  • 8/10/2019 PID_slides.pdf

    33/36

    Unit step response

    PID compensated

    Compensation yields the original system

    faster and with no steady-state error; the

    peak time is still a little large for the specs

    .

    MAAE 4500: Feedback Control 33

  • 8/10/2019 PID_slides.pdf

    34/36

    2K 8K s ( )C s( )R s

    1 3s 3 6 10s s s

    2 1

    4.604 55.909 0.5( ) ( )

    s sG s G s

    s

    2 12

    259.74.604 56.409 27.954

    128.7

    4.604

    Ks s

    Ks

    K

    (One pole at the origin, one zero at 0.5 and one zero at

    MAAE 4500: Feedback Control 34

    .

  • 8/10/2019 PID_slides.pdf

    35/36

    Summary of Controllers and Their Usage

    MAAE 4500: Feedback Control 35

  • 8/10/2019 PID_slides.pdf

    36/36

    MAAE 4500: Feedback Control 36