controller gain design

Upload: xzax-tornadox

Post on 04-Feb-2018

217 views

Category:

Documents


0 download

TRANSCRIPT

  • 7/21/2019 Controller Gain Design

    1/24

    221

    KD 4.5.5.14

    4.6 (Selecting controller gains)

    (PID-control) , (dominant time constant) (steadystateerror) (rise time),(settling time),(maximum overshoot) (bandwidth), (resonantfrequency) (peak amplitude)

    1) (Equilibrium specifications)- (stability)- (steady state error)

    2) (transient specifications)- (speed of response)- (degree

    of damping)

    3) (Sensitivity specification)

  • 7/21/2019 Controller Gain Design

    2/24

    Principles of Feedback Control

    222

    - (sensitivity to parameter variations)

    - (sensitvity to model inaccuracies)

    - (Noise rejection)

    (bandwidth)

    4) (nonlinear effects)- (stability)- (final control element capabilities)

    (standard industrial controller) (tunning)

    (Performance Index criteria)

  • 7/21/2019 Controller Gain Design

    3/24

    223

    (Optimal Control) (performance index) (performance index)

    (costing function)

    (optimal control system)

    0

    J e(t)dt

    = 4.6.1(Integral Absolute Error), IAE

    0

    J t e(t)dt

    = 4.6.2(Integral Time multiplied Absolute Error), ITAE

    2

    0

    J e (t)dt

    = 4.6.3(Integral Square Error), ISE2

    0

    J te (t)dt

    = 4.6.4(Integral Time multiplied Square Error), ITSE

    e(t) (e(t)) t

  • 7/21/2019 Controller Gain Design

    4/24

    Principles of Feedback Control

    224

    e(t) ft

    (IAE)

    (ITAE) t t (highlyunderdamped) (highly overdamped)

    ITAE (overshoot)(osillating) (well-damped oscillations)

    (ISE)( ITSE) (IAE) ( ITAE) 1)

    2) 3)

    4.6.1

  • 7/21/2019 Controller Gain Design

    5/24

    225

    (Advanced Control System)

    K

    p

    x

    x

    x( )0

    x x u+ = 4.6.5 u Kx= () 4.6.6

    (performance index)

    2 21 2

    0

    J w x (t) w x (t) dt

    = + 4.6.7

    w1w2(weighting factors).

    u Kx=

    x x Kx+ = 4.6.8

    (1 K)tx(t) x(0)e += 4.6.9

    x(t) (performance index)

  • 7/21/2019 Controller Gain Design

    6/24

    Principles of Feedback Control

    226

    2 2

    1 2

    1J x (0) w w (1 K)

    2(1 K) = + + +

    4.6.10

    K J

    J

    0K

    =

    2

    2

    J

    0x

    4.6.11

    1+K > 0

    1

    2

    wK 1

    w= + 4.6.12

    K w1w2

    w10 x(t)

    K 1 x(t)=x(0) ( )x t w2= 0 K

    x(t)0 K 0 < K < x x w1w2

    K

    K

    K w1 w2(performance

    index) (simple system)

    4.6.1

  • 7/21/2019 Controller Gain Design

    7/24

    227

    (optimization) (performance index) ITAE

    4.6.1 ITAE

    C s

    R s

    a

    s a s a s a s an

    n n n

    n n

    n

    ( )

    ( ) = + + + + + = 11

    2

    2

    1

    2

    , an

    s n+

    s sn n2 21 4+ +.

    s s sn n n

    3 2 2 31 75 215+ + +. .

    s s s sn n n n

    4 3 2 2 3 421 3 4 2 7+ + + +. . .

    s s s s sn n n n n5 4 2 3 3 2 4 52 8 5 0 5 5 3 4+ + + + +. . . .

    s s s s s sn n n n n n6 5 2 4 3 3 4 2 5 63 25 6 60 8 60 7 45 3 95+ + + + + +. . . . .

    4.6.2 (normalized form)

    1G(s)

    s(s 1)=

    +

    (PD-control) 4.6.1 KpKDITAE

  • 7/21/2019 Controller Gain Design

    8/24

    Principles of Feedback Control

    228

    1G(s)

    s(s 1)=

    +

    4.6.1

    2pDs (1 K )s K 0+ + + =

    4.6.1 ITAE

    2 2

    n ns 1.4 s 0 + + =

    2rad/sec n= 2rad/sec

    D D

    2P

    1 K 1.4 2 2.8 K 1.8

    K 2 4

    + = = =

    = =

    4.6.1 n (Closed-loopnatural frequency) P-control 4.6.1

  • 7/21/2019 Controller Gain Design

    9/24

    229

    4.7 -(Ziegler-Nichols) PID (Ziegler-Nichols Tuning of PID regulators)

    -(Ziegler-Nichols) PID(transient response) (steady state) - (Model-based control) --

    (dynamic equation) (transfer function)

    (empirically based rules) -(Ziegler-Nichols) 40(Classical Control)

    -

  • 7/21/2019 Controller Gain Design

    10/24

    Principles of Feedback Control

    230

    -

    ( )P PG s K=

    ( )PI p r1G s K 1 Ts

    = +

    ( ) ( )pPD dG s K 1 T s= +

    ( ) dpPDfilterD

    T sG s K 1

    s 1

    = + +

    ( ) pPID dr

    1G s K 1 T s

    T s

    = + +

    ( ) dpPIDfilterr D

    T s1G s K 1

    Ts s 1

    = + + +

    ( ) s ssPIDseriess s

    I D sG s K 1 1

    s D s 1

    = + + +

    ( ) p ppPIDparallelp p

    I D sG s K

    s D s 1= + +

    +

    (low pass filter) fast mode

    4.5.4.1 (setpoint change) (Saturated Control) (intergral wind-up) 7

  • 7/21/2019 Controller Gain Design

    11/24

    231

    - IAE (systemresponse) (unit-step input) -(Ziegler-Nichols)

    (quarter-decay)

    (overshoot)

    25%4.7.2 - (Processreaction method) (Ultimate cycle method)

    (Process reaction method)(Process reaction curve)

    (Signal Recorder)Storage Oscilloscope, Analog recoder, Digital recorder

    ( )sGOutputUnit Step

    - (input signal)

    s (Process reaction

  • 7/21/2019 Controller Gain Design

    12/24

    Principles of Feedback Control

    232

    curve)(FirstorderSystem)

    dt sY(s) Ke

    U(s) s 1

    =

    +

    4.7.1

    ( first ordersystem) (transportation lag, td)

    P,PI PID() RL P, PIPID4.7.1

  • 7/21/2019 Controller Gain Design

    13/24

    233

    (Untimate cycle method) 25%

    (quarter-decay) 4.7.2 =0.21 (Proportionalcontrol) 4.7.1 K(oscillate) (unstable) K (Closed-loop Characteristic equation) (Imaginary Axis) K Ku

    (period of oscillation)Pu

    (ultimateperiod) (amplitude) Ku Pu

    P, PI PID 4.7.1

    4.7.1

  • 7/21/2019 Controller Gain Design

    14/24

    Principles of Feedback Control

    234

    100%

    25%

    Period

    4.7.2

    4.7.1 -(The Ziegler-Nichols Rules)

    Ip pD DI

    K1G(s) K 1 T s K K s

    T s s

    = + + = + +

    (P-Control)

    p

    1K

    RL= p pu

    K 0.5K=

    (PI-Control)

    p 0.9K RL= p puK 0.45K=

    IT 3.3L= I uT 0.83P=

    (PID-Control)

    p

    1.2K

    RL= p pu

    K 0.6K=

    IT 2L= I uT 0.5P=

    DT 0.5L= uDT 0.125P=

  • 7/21/2019 Controller Gain Design

    15/24

    235

    4.7.1 (Third order plant) 0 1 2 3a , a ,a , a

    + + +3 2

    3 2 1 0

    1

    a s a s a s a

    3 23 2 1 0 pa s a s a s a K 0+ + + + = 4.7.2

    KpPu(ultimate period)

    Kpu

    Kp(sustainedoscillation) s (ultimate period) us j= u 2 2us = 3 3us j= s

    ( ) ( )2 30 pu 2 u 1 u 3 ua K a j a a 0 + + = 4.7.3

    4.7.3 (real part)(imaginary part)

  • 7/21/2019 Controller Gain Design

    16/24

    Principles of Feedback Control

    236

    2

    0 pu 2 ua K a 0+ = 4.7.4

    ( )2u 1 3 ua a 0 = 4.7.5

    4.7.5

    1u

    3

    aa

    = 4.7.6

    4.7.64.7.4 Kpu

    (ultimate gain)

    2 2 1pu 2 u 0 0

    3

    a aK a a a

    a= = 4.7.7

    Pu(ultimate period)

    uu

    2P

    = 4.7.8

    (sustained oscillation) (Real Axis) s(negativerealpart) 1 us j= 2 us j= s

    ( )( )( )u u 3s j s j s s 0 + =

    3 2 2 23 u 3 us s s s s 0 + = 4.7.9

  • 7/21/2019 Controller Gain Design

    17/24

    237

    233

    as

    a=

    3 2

    1G(s)

    s 2s s 1=

    + + +

    3 2 1a 1, a 2, a 1= = = 0a 1=

    -(The Ziegler-Nichols Rules)(PID-control)

    p

    I

    D

    K 0.6 1 0.6

    T 0.5 2 3.142

    T 0.125 2 0.785

    = =

    = =

    = =

    I

    p D

    K1G(s) 0.6 1 0.785s K K s

    3.142s s

    = + + = + +

    4.7.3 4.7.4 (PID-

    control) -(Ziegler-Nichols Rules)

    3 2

    1

    s 2s s 1+ + +

    10.6 1 0.785s

    3.142s

    + +

  • 7/21/2019 Controller Gain Design

    18/24

    Principles of Feedback Control

    238

    4.7.3 4.7.1 3 2

    1G(s)

    s 2s s 1=

    + + +

    4.7.4 4.7.1 3 2

    1G(s)

    s 2s s 1=

    + + +

    (PID-Control) = 10.6 1 0.785s3.142s

    + +

  • 7/21/2019 Controller Gain Design

    19/24

    239

    (Untimate cycle method) Matlab/Simulink Slider Gain Scope 4.7.5 Simulink Slider Gain Scope Scope

    (Autoscale)

    4.7.6 1 1.16 Scope Ku 1 4.7.1-(The Ziegler-Nichols Rules)-

    5 6 7 frequency response

    4.7.5 Simulink 4.7.1

  • 7/21/2019 Controller Gain Design

    20/24

    Principles of Feedback Control

    240

    Slider Gain= 1

    Scope

    Slider Gain= 1.16

    Scope4.7.6 Slider Gain Scope

    4.7.2-

    ( )( )

    3

    1G s

    s 1=

    +

    (Untimate cycle method) -(Imaginary axis)

  • 7/21/2019 Controller Gain Design

    21/24

    241

    ( ) ( )

    3

    Pu Pu u3

    11 K 0 K j 1

    s 1+ = = +

    +

    ( ) ( )3 3 2

    Pu u u u uK j 1 j 3 3 1 = + = + +

    puK 8= u 3 = rad/sec Pu(Ultimate period) 3.63 4.7.1

    p

    I

    D

    K 0.6 8 4.8

    T 0.5 3.63 1.81

    T 0.125 3.63 0.45

    =

    =

    =

    I

    p DK1G(s) 4.8 1 0.45s K K s

    1.81s s = + + = + +

    ( )3

    1

    s 1+

    14.8 1 0.45s

    1.81s

    + +

  • 7/21/2019 Controller Gain Design

    22/24

    Principles of Feedback Control

    242

    Matlab/Simulink

    ( ) dpPIDfilterI D

    T s1G s K 1

    T s s 1

    = + + +

    D d0.1 *T 0.045 = =

    ( ) ( )( )

    ( )( ) ( )( )

    2 DDd 2I I

    pPIDfilter 3 3

    D

    1T s 1 sT T 52.8s 109.23s 58.93G s G s K

    s s 1 s 1 s s 22.2 s 1

    + + + + + +

    = =+ + + +

    Matlab >> sys1=tf([52.8 109.23 58.93],poly([0 -22.2 -1 -1 -1]));

    >> syscl1=feedback(sys1,1);

    >> step(syscl1) ()

  • 7/21/2019 Controller Gain Design

    23/24

    243

    dT 1=

    ( ) ( )

    ( )

    ( )( ) ( )( )

    2 DDd 2

    I IpPIDfilter 3 3

    D

    1T s 1 s

    T T 52.8s 50.65s 26.52G s G s K

    s s 1 s 1 s s 10 s 1

    + + + + + + = =

    + + + +

  • 7/21/2019 Controller Gain Design

    24/24

    244

    -

    (fine tune)