Robust Control Control Spring, 2018 Instructor: Prof. Masayuki Fujita (S5-303B) 2nd class Tue., 17th...
Transcript of Robust Control Control Spring, 2018 Instructor: Prof. Masayuki Fujita (S5-303B) 2nd class Tue., 17th...
![Page 1: Robust Control Control Spring, 2018 Instructor: Prof. Masayuki Fujita (S5-303B) 2nd class Tue., 17th April, 2018, 10:45 ~12:15, S423 Lecture Room 2. Nominal Performance Reference:](https://reader031.fdocuments.in/reader031/viewer/2022021513/5b04633e7f8b9a41528c64a1/html5/thumbnails/1.jpg)
Robust Control
Spring, 2018Instructor: Prof. Masayuki Fujita (S5-303B)
2nd classTue., 17th April, 2018, 10:45~12:15,
S423 Lecture Room
![Page 2: Robust Control Control Spring, 2018 Instructor: Prof. Masayuki Fujita (S5-303B) 2nd class Tue., 17th April, 2018, 10:45 ~12:15, S423 Lecture Room 2. Nominal Performance Reference:](https://reader031.fdocuments.in/reader031/viewer/2022021513/5b04633e7f8b9a41528c64a1/html5/thumbnails/2.jpg)
2. Nominal Performance
Reference:[SP05] S. Skogestad and I. Postlethwaite,
Multivariable Feedback Control; Analysis and Design,Second Edition, Wiley, 2005.
2.1 Weighted Sensitivity [SP05, Sec. 2.8, 3.3, 4.10, 6.2, 6.3]
2.2 Nominal Performance [SP05, Sec. 2.8, 3.2, 3.3]
2.3 Sensitivity Minimization
2.4 Remarks on Fundamental Limitations
[SP05, Sec. 3.2, 3.3, 9.3]
[SP05, Sec. 6.2]2.5 1st Report
![Page 3: Robust Control Control Spring, 2018 Instructor: Prof. Masayuki Fujita (S5-303B) 2nd class Tue., 17th April, 2018, 10:45 ~12:15, S423 Lecture Room 2. Nominal Performance Reference:](https://reader031.fdocuments.in/reader031/viewer/2022021513/5b04633e7f8b9a41528c64a1/html5/thumbnails/3.jpg)
3
Disturbance AttenuationOpen-loop Closed-loop
small: Good Feedback Performance
: Sensitivity
Sensitivity as Feedback Performance in SISO Systems
![Page 4: Robust Control Control Spring, 2018 Instructor: Prof. Masayuki Fujita (S5-303B) 2nd class Tue., 17th April, 2018, 10:45 ~12:15, S423 Lecture Room 2. Nominal Performance Reference:](https://reader031.fdocuments.in/reader031/viewer/2022021513/5b04633e7f8b9a41528c64a1/html5/thumbnails/4.jpg)
4
Insensitivity to Plant Variations
small : Good Feedback Performance (Absolute Value)
[SP05, p. 23]Sensitivity as Feedback Performance in SISO Systems
MIMO ?
![Page 5: Robust Control Control Spring, 2018 Instructor: Prof. Masayuki Fujita (S5-303B) 2nd class Tue., 17th April, 2018, 10:45 ~12:15, S423 Lecture Room 2. Nominal Performance Reference:](https://reader031.fdocuments.in/reader031/viewer/2022021513/5b04633e7f8b9a41528c64a1/html5/thumbnails/5.jpg)
5
Norm as System Gain
[Ex.]
hinfG = normhinf(G)MATLAB Command
[SP05]
System Gain(p. 158)
: Proper stable system
- plot
〜1 2
![Page 6: Robust Control Control Spring, 2018 Instructor: Prof. Masayuki Fujita (S5-303B) 2nd class Tue., 17th April, 2018, 10:45 ~12:15, S423 Lecture Room 2. Nominal Performance Reference:](https://reader031.fdocuments.in/reader031/viewer/2022021513/5b04633e7f8b9a41528c64a1/html5/thumbnails/6.jpg)
6
Difference between the and norms [SP05, pp. 75, 159]
normPush down“peak of maximum singular value”
Minimizing
Worst direction, worst frequency
Multiplicative property
norm (LQG)Push down “whole thing”
(all singular values over all frequencies)
Minimizing
Average direction, average frequency
Multiplicative property
?
![Page 7: Robust Control Control Spring, 2018 Instructor: Prof. Masayuki Fujita (S5-303B) 2nd class Tue., 17th April, 2018, 10:45 ~12:15, S423 Lecture Room 2. Nominal Performance Reference:](https://reader031.fdocuments.in/reader031/viewer/2022021513/5b04633e7f8b9a41528c64a1/html5/thumbnails/7.jpg)
7
Optimization in Feedback Control
−
+
Sensitivity
“Feedback Performance = Sensitivity”Sensitivity optimization
from Reference to Error
(System Gain)
with norm
( -parameterization)
![Page 8: Robust Control Control Spring, 2018 Instructor: Prof. Masayuki Fujita (S5-303B) 2nd class Tue., 17th April, 2018, 10:45 ~12:15, S423 Lecture Room 2. Nominal Performance Reference:](https://reader031.fdocuments.in/reader031/viewer/2022021513/5b04633e7f8b9a41528c64a1/html5/thumbnails/8.jpg)
Sugihara “Sempo” Chiune, consular officer of Japan, helped him a lot. Leaving for Canada
The World War II occurred. 1939Escaping to Europe through Lithuania
1941 Arrival in Kobe Through Russia, Siberia and Japan sea,Witnessed by Soviet’s tank
George Zames (1934-1997)G.Zames, IEEE TAC, 26, 1981
Frustration with LQG control
• Formulation of the optimization problem not on time domain but on frequency domain
control∞H
8
( control)2H
![Page 9: Robust Control Control Spring, 2018 Instructor: Prof. Masayuki Fujita (S5-303B) 2nd class Tue., 17th April, 2018, 10:45 ~12:15, S423 Lecture Room 2. Nominal Performance Reference:](https://reader031.fdocuments.in/reader031/viewer/2022021513/5b04633e7f8b9a41528c64a1/html5/thumbnails/9.jpg)
Bode Sensitivity Integrals (Waterbed Effects) for Stable Plant
9
There exists a frequency range over which the magnitude of the sensitivity function exceeds 1 if it is to be kept below 1 at the other frequency range.
[SP05, p. 167, p. 223]
100 101
10
10−
0
]dB[
Frequency [rad/s][SP05, Ex., p. 170]
(unstable)
Dirt
RHP(Right-Half Plane) Zero
![Page 10: Robust Control Control Spring, 2018 Instructor: Prof. Masayuki Fujita (S5-303B) 2nd class Tue., 17th April, 2018, 10:45 ~12:15, S423 Lecture Room 2. Nominal Performance Reference:](https://reader031.fdocuments.in/reader031/viewer/2022021513/5b04633e7f8b9a41528c64a1/html5/thumbnails/10.jpg)
10
Sensitivity [SP05, p. 60]
10
10−
0
]dB[
Frequency [rad/s]
Waterbed Effects
SISO Case
[SP05, pp. 62, 80]: Performance weight transfer function matrix
Small?
+
−
Small! Intractable Tractable!
Weighted
How to specify ?
![Page 11: Robust Control Control Spring, 2018 Instructor: Prof. Masayuki Fujita (S5-303B) 2nd class Tue., 17th April, 2018, 10:45 ~12:15, S423 Lecture Room 2. Nominal Performance Reference:](https://reader031.fdocuments.in/reader031/viewer/2022021513/5b04633e7f8b9a41528c64a1/html5/thumbnails/11.jpg)
11
[SP05, pp. 62, 80]Performance Weight
: the frequency at which the asymptote of crosses 1, and the bandwidth requirement approximately
First-order Performance Weight
: at low frequencies
: at high frequencies ( : Rule of thumb)
Feedback Effect
![Page 12: Robust Control Control Spring, 2018 Instructor: Prof. Masayuki Fujita (S5-303B) 2nd class Tue., 17th April, 2018, 10:45 ~12:15, S423 Lecture Room 2. Nominal Performance Reference:](https://reader031.fdocuments.in/reader031/viewer/2022021513/5b04633e7f8b9a41528c64a1/html5/thumbnails/12.jpg)
12
Stabilization and Performance
Rise time
00.1
0.91
Unstable PlantReal RHP Poles:
Imaginary Poles:
Complex RHP Poles:
[SP05 Sec 5.9]
[QZ07] L. Qiu and K. Zhou (2007) Introduction to Feedback Control, Prentice Hall.
First-order System
Rise time
Second-order System
Rise time
Stable Plant
×
×
×
×
×
![Page 13: Robust Control Control Spring, 2018 Instructor: Prof. Masayuki Fujita (S5-303B) 2nd class Tue., 17th April, 2018, 10:45 ~12:15, S423 Lecture Room 2. Nominal Performance Reference:](https://reader031.fdocuments.in/reader031/viewer/2022021513/5b04633e7f8b9a41528c64a1/html5/thumbnails/13.jpg)
13
Nominal Performance (NP) [SP05, p. 81]
Nominal Performance (NP) TestGiven a controller ,
+
−
3
![Page 14: Robust Control Control Spring, 2018 Instructor: Prof. Masayuki Fujita (S5-303B) 2nd class Tue., 17th April, 2018, 10:45 ~12:15, S423 Lecture Room 2. Nominal Performance Reference:](https://reader031.fdocuments.in/reader031/viewer/2022021513/5b04633e7f8b9a41528c64a1/html5/thumbnails/14.jpg)
14
[SP05, p. 60]
[Ex.]
(NP)
Nominal Performance Test in SISO Systems
: small: fast
1)■
2)■
3)■
(NP)
(NP): large
(NP)
![Page 15: Robust Control Control Spring, 2018 Instructor: Prof. Masayuki Fujita (S5-303B) 2nd class Tue., 17th April, 2018, 10:45 ~12:15, S423 Lecture Room 2. Nominal Performance Reference:](https://reader031.fdocuments.in/reader031/viewer/2022021513/5b04633e7f8b9a41528c64a1/html5/thumbnails/15.jpg)
Nominal Performance Test in SISO Systems
15
[SP05, p. 60]
[Ex.]
(NP)
: small: fast
Norm Condition
1)■
2)■
3)■
(NP)
(NP): large
(NP)
![Page 16: Robust Control Control Spring, 2018 Instructor: Prof. Masayuki Fujita (S5-303B) 2nd class Tue., 17th April, 2018, 10:45 ~12:15, S423 Lecture Room 2. Nominal Performance Reference:](https://reader031.fdocuments.in/reader031/viewer/2022021513/5b04633e7f8b9a41528c64a1/html5/thumbnails/16.jpg)
16
Spinning Satellite: Performance Weight[Ex.]Performance Weight
Specifications
Ms = 2; A = 1e-2; wb = 11.5;wP = tf([1/Ms wb], [1 wb*A]);WP = eye(2)*wP;figuresigma(WP)hold on; grid on;
MATLAB Command
11.5rad/s10 rad/s
Poles on the imaginary axis10
the steady state error
Gain crossover frequency
Phase stabilization [SP05, p. 194]System bandwidth of Actuator/Sensor/Controller
11.5 rad/s
![Page 17: Robust Control Control Spring, 2018 Instructor: Prof. Masayuki Fujita (S5-303B) 2nd class Tue., 17th April, 2018, 10:45 ~12:15, S423 Lecture Room 2. Nominal Performance Reference:](https://reader031.fdocuments.in/reader031/viewer/2022021513/5b04633e7f8b9a41528c64a1/html5/thumbnails/17.jpg)
(Output) Sensitivity Function
0.8935
KI = inv(Pnom)*tf([1],[1 30 0])*diag([900 900]);FI = loopsens(Pnom,KI);sigma(FI.So) ;hinfSo = normhinf(WP*FI.So)
MATLAB Command
17
Spinning Satellite: Nominal Performance
Plant
[Ex.]
11.5rad/s
Target Loop Transfer Function
[rad/s]4811.5
30
-20dB/dec
-40dB/decController: Inverse-based Controller
21rad/s
NP
![Page 18: Robust Control Control Spring, 2018 Instructor: Prof. Masayuki Fujita (S5-303B) 2nd class Tue., 17th April, 2018, 10:45 ~12:15, S423 Lecture Room 2. Nominal Performance Reference:](https://reader031.fdocuments.in/reader031/viewer/2022021513/5b04633e7f8b9a41528c64a1/html5/thumbnails/18.jpg)
18
Sensitivity MinimizationOptimal Sensitivity Problem
Find a stabilizing controller which
tractableInminimizemake smaller
Given , find all stabilizing controllers such thatSensitivity Minimization Problem
-iteration
Control
dBno
Parameterization
+
− −
Linear Fractional Transformation (LFT)
![Page 19: Robust Control Control Spring, 2018 Instructor: Prof. Masayuki Fujita (S5-303B) 2nd class Tue., 17th April, 2018, 10:45 ~12:15, S423 Lecture Room 2. Nominal Performance Reference:](https://reader031.fdocuments.in/reader031/viewer/2022021513/5b04633e7f8b9a41528c64a1/html5/thumbnails/19.jpg)
19
Sensitivity MinimizationOptimal Sensitivity Problem
Find a stabilizing controller which
tractablemake smaller
Given , find all stabilizing controllers such thatSensitivity Minimization Problem
-iteration
Control
1no
Parameterization
−
Linear Fractional Transformation (LFT)
![Page 20: Robust Control Control Spring, 2018 Instructor: Prof. Masayuki Fujita (S5-303B) 2nd class Tue., 17th April, 2018, 10:45 ~12:15, S423 Lecture Room 2. Nominal Performance Reference:](https://reader031.fdocuments.in/reader031/viewer/2022021513/5b04633e7f8b9a41528c64a1/html5/thumbnails/20.jpg)
Sensitivity for MIMO Systems
Output Sensitivity Function:
Sensitivity to Output Disturbance
20
For SISO Systems but for MIMO Systems
Good disturbance rejection at output does not always mean good rejection at input
Input Sensitivity Function:
Sensitivity to Input Disturbance
[SP05, p. 70]
![Page 21: Robust Control Control Spring, 2018 Instructor: Prof. Masayuki Fujita (S5-303B) 2nd class Tue., 17th April, 2018, 10:45 ~12:15, S423 Lecture Room 2. Nominal Performance Reference:](https://reader031.fdocuments.in/reader031/viewer/2022021513/5b04633e7f8b9a41528c64a1/html5/thumbnails/21.jpg)
21
Standard Feedback Configuration with Weights
Sensitivity Minimization Problem
[SP05, p. 363]
![Page 22: Robust Control Control Spring, 2018 Instructor: Prof. Masayuki Fujita (S5-303B) 2nd class Tue., 17th April, 2018, 10:45 ~12:15, S423 Lecture Room 2. Nominal Performance Reference:](https://reader031.fdocuments.in/reader031/viewer/2022021513/5b04633e7f8b9a41528c64a1/html5/thumbnails/22.jpg)
22
“Respect the unstable”
Control Systems Magazine, 23(4):12-25, 2003.
Bode lecture, CDC, 1989
Gunter Stein
Re
Im
unstablepole
unstablezero
Remarks on Fundamental Limitations
• Time Delay• Wrong Sensor Placement
Unstable Zero
![Page 23: Robust Control Control Spring, 2018 Instructor: Prof. Masayuki Fujita (S5-303B) 2nd class Tue., 17th April, 2018, 10:45 ~12:15, S423 Lecture Room 2. Nominal Performance Reference:](https://reader031.fdocuments.in/reader031/viewer/2022021513/5b04633e7f8b9a41528c64a1/html5/thumbnails/23.jpg)
23
SISO Loop Shaping
Loop Shapinggives us graphical interpretation
[SP05, pp. 41, 42, 343]
Unstable Zero:Time Delay:
Unstable Pole:
• Bode Plot• System Gain
Performance
Robust Stability(+ Roll-off)
〜4 5
![Page 24: Robust Control Control Spring, 2018 Instructor: Prof. Masayuki Fujita (S5-303B) 2nd class Tue., 17th April, 2018, 10:45 ~12:15, S423 Lecture Room 2. Nominal Performance Reference:](https://reader031.fdocuments.in/reader031/viewer/2022021513/5b04633e7f8b9a41528c64a1/html5/thumbnails/24.jpg)
RHP Poles/Zeros, Time Delays and Sensitivity in SISO Systems
25
For systems with a RHP pole and RHP zero (or a time delay ), any stabilizing controller gives sensitivity functions with the property
)( ωjS )( ωjT
〜6 7
The zero and the pole must be sufficiently far apart
The product of RHP pole and time delay must be
sufficiently small
![Page 25: Robust Control Control Spring, 2018 Instructor: Prof. Masayuki Fujita (S5-303B) 2nd class Tue., 17th April, 2018, 10:45 ~12:15, S423 Lecture Room 2. Nominal Performance Reference:](https://reader031.fdocuments.in/reader031/viewer/2022021513/5b04633e7f8b9a41528c64a1/html5/thumbnails/25.jpg)
2. Nominal Performance
Reference:[SP05] S. Skogestad and I. Postlethwaite,
Multivariable Feedback Control; Analysis and Design,Second Edition, Wiley, 2005.
2.1 Weighted Sensitivity[SP05, Sec. 2.8, 3.3, 4.10, 6.2, 6.3]
2.2 Nominal Performance [SP05, Sec. 2.8, 3.2, 3.3]
2.3 Sensitivity Minimization
2.4 Remarks on Fundamental Limitations
[SP05, Sec. 3.2, 3.3, 9.3]
[SP05, Sec. 6.2]2.5 1st Report
![Page 26: Robust Control Control Spring, 2018 Instructor: Prof. Masayuki Fujita (S5-303B) 2nd class Tue., 17th April, 2018, 10:45 ~12:15, S423 Lecture Room 2. Nominal Performance Reference:](https://reader031.fdocuments.in/reader031/viewer/2022021513/5b04633e7f8b9a41528c64a1/html5/thumbnails/26.jpg)
3. Robustness and Uncertainty
3.2 Representing Uncertainty3.1 Why Robustness? [SP05, Sec. 4.1.1, 7.1, 9.2]
[SP05, Sec. 7.2, 7.3, 7.4]
Reference:[SP05] S. Skogestad and I. Postlethwaite,
Multivariable Feedback Control; Analysis and Design,Second Edition, Wiley, 2005.
[SP05, Sec. 8.1, 8.2, 8.3]
3.4 Systems with Structured Uncertainty[SP05, Sec. 8.2]
3.3 Uncertain Systems
![Page 27: Robust Control Control Spring, 2018 Instructor: Prof. Masayuki Fujita (S5-303B) 2nd class Tue., 17th April, 2018, 10:45 ~12:15, S423 Lecture Room 2. Nominal Performance Reference:](https://reader031.fdocuments.in/reader031/viewer/2022021513/5b04633e7f8b9a41528c64a1/html5/thumbnails/27.jpg)
27
Norm
Vector Norm [Ex.]
Key properties2. Positive iff3. , :scalar4.
1. Non-negative
HomogeneousTriangle inequality
[SP05, A.5]
(Euclidean Vector Norm)
(Induced) Matrix Norm
[Ex.]
1
![Page 28: Robust Control Control Spring, 2018 Instructor: Prof. Masayuki Fujita (S5-303B) 2nd class Tue., 17th April, 2018, 10:45 ~12:15, S423 Lecture Room 2. Nominal Performance Reference:](https://reader031.fdocuments.in/reader031/viewer/2022021513/5b04633e7f8b9a41528c64a1/html5/thumbnails/28.jpg)
AreaSignal Norm
Norm [SP05, A.5]
“Energy of signal”( -norm, : Lebesgue space)
Integral absolute error
“maximum value over time”
[Ex.]
Energy
33
2
System Norm MIMO
(System Gain)
![Page 29: Robust Control Control Spring, 2018 Instructor: Prof. Masayuki Fujita (S5-303B) 2nd class Tue., 17th April, 2018, 10:45 ~12:15, S423 Lecture Room 2. Nominal Performance Reference:](https://reader031.fdocuments.in/reader031/viewer/2022021513/5b04633e7f8b9a41528c64a1/html5/thumbnails/29.jpg)
should be away from by
Nyquist Plot
1−
29
1−
[SP05, p. 281]
Nominal Performance in SISO Systems3
![Page 30: Robust Control Control Spring, 2018 Instructor: Prof. Masayuki Fujita (S5-303B) 2nd class Tue., 17th April, 2018, 10:45 ~12:15, S423 Lecture Room 2. Nominal Performance Reference:](https://reader031.fdocuments.in/reader031/viewer/2022021513/5b04633e7f8b9a41528c64a1/html5/thumbnails/30.jpg)
Fundamental LimitationsBound on the Crossover Frequency RHP (Right half-plane) Zero
Time Delay
Im
Re0 z
betterworse
Step Response
Time [s]
Tight RestrictionsSlow RHP Zeros ( small):Fast RHP Zeros ( large): Loose Restrictions
Frequency [rad/s]
[SP05, pp. 183]
Unstable zero
4
30
![Page 31: Robust Control Control Spring, 2018 Instructor: Prof. Masayuki Fujita (S5-303B) 2nd class Tue., 17th April, 2018, 10:45 ~12:15, S423 Lecture Room 2. Nominal Performance Reference:](https://reader031.fdocuments.in/reader031/viewer/2022021513/5b04633e7f8b9a41528c64a1/html5/thumbnails/31.jpg)
Fundamental LimitationsBound on the Crossover Frequency
Im
Re0 p×××××××
worsebetter
RHP (Right half-plane) Pole
Fast RHP Poles ( large): Tight RestrictionsLoose RestrictionsSlow RHP Poles ( small):
Frequency [rad/s]
[SP05, pp. 192, 194]
Unstable pole
Poles on imaginary axis
5
31
![Page 32: Robust Control Control Spring, 2018 Instructor: Prof. Masayuki Fujita (S5-303B) 2nd class Tue., 17th April, 2018, 10:45 ~12:15, S423 Lecture Room 2. Nominal Performance Reference:](https://reader031.fdocuments.in/reader031/viewer/2022021513/5b04633e7f8b9a41528c64a1/html5/thumbnails/32.jpg)
RHP Poles/Zeros, Time Delays and Sensitivity in SISO Systems
34
All-pass system ( )
The zero and the pole must be sufficiently far apart
Frequency [rad/s]
Phas
e [d
eg]
or
The product of RHP pole and time delay must be
sufficiently small
6
Frequency [rad/s]Ph
ase
[deg
]
![Page 33: Robust Control Control Spring, 2018 Instructor: Prof. Masayuki Fujita (S5-303B) 2nd class Tue., 17th April, 2018, 10:45 ~12:15, S423 Lecture Room 2. Nominal Performance Reference:](https://reader031.fdocuments.in/reader031/viewer/2022021513/5b04633e7f8b9a41528c64a1/html5/thumbnails/33.jpg)
33
Fundamental Limitations: Sensitivity in MIMO Systems[SP05, Sec. 6.2]Algebraic Constraint
is large if and only if is large
,
One RHP Pole and One RHP Zero
Fundamental Limitations: Bounds on Peaks in MIMO Systems[SP05, Sec. 6.3]
:Pole and Zero Direction [SP05, 4.4, 4.5 ]
7