wort el bokkker
TRANSCRIPT
-
8/13/2019 Wort El Bokkker
1/20
Model Reductionwithin Control Systems
Design
Pepijn WortelboerPHILIPS OPTICAL STORAGE
January 2001
-
8/13/2019 Wort El Bokkker
2/20
Introduction
Control problem formulation Need for order reduction algorithms Introduction of CD-player mechanism Standard reduction techniques: modal and balanced Closed-loop reduction Iterative low-order control design for CD-player Concluding remarks
-
8/13/2019 Wort El Bokkker
3/20
Control problem formulation
Linear time-invariant model based standard plant formulation tuning by weighting function design
Goal:
simplest K achieving small z
G
K
w K M G z ),,(I=
z w
-
8/13/2019 Wort El Bokkker
4/20
Need for order reduction algorithms
High-speed high-accuracy servo-mechanisms implyhigh-order models
Highest frequency modes in model are not reliable
High-order model is not suited for optimal control Controller order exceeds model orderBut
order reduction not goal in itself
Part of design
-
8/13/2019 Wort El Bokkker
5/20
Rotating compact disc swing arm for radial tracking with optical pick-up unit
(vertically suspended lens)for focussing
mounting plate, disc, armnot infinitely stiff
2x2 servo controldesign problem
Introduction of CD-player mechanism
-
8/13/2019 Wort El Bokkker
6/20
Radial control loop
K G
distw
track w e i
No weights on input signals output weights
Ws to penalize low-frequenterror e
Wt to penalize high-frequentcurrent i
-
8/13/2019 Wort El Bokkker
7/20
Standard reduction techniques:
Modal A diagonal+ invariant for input & output+ direct+ no stability constraint
balanced Gramians P & Q diag+ input-output significant
Both suitable for truncation and singular perturbation Not optimal, how about closed-loop?
-
8/13/2019 Wort El Bokkker
8/20
Transform
model state controller state
to get closed-loop systemGramians with
closed-loop G-HSV closed-loop K-HSV
Closed-loop balancing
K
M
G
T
I
T
(
(
K
G
G
K
M
G
K
M
-
8/13/2019 Wort El Bokkker
9/20
Closed-loop balanced reduction
Solve P, Q from Lyapunov equations of I (G,M,K)
extract G -part: P G & Q G find balancing transformation:
T * Q G T = T-1 P G T
-* = diag( HSV G )
apply T to G : G bal =(T -1 AG T, T -1 B G , C G T ,D G ) truncate G
bal
-
8/13/2019 Wort El Bokkker
10/20
Reduction in twin feedback configuration
),,( bal K M GG hmm R I=(
G
K
M
Tuning freedom:
selection of order
controller or model
-
8/13/2019 Wort El Bokkker
11/20
Separate weighting in the form of power spectra
wh z mm K M GG R =
),,( bal I
(
( )
piecewise quadratic interpolationG
K
z w
-
8/13/2019 Wort El Bokkker
12/20
Iterative low-order control design
Available: High order model G Weights in M First guess controller K
Model reduction
Controller reduction
),,(
),,(),,(
K M G
K M G K M G
h
mh
I
II =
),,( r K M GI=
Optimal control
-
8/13/2019 Wort El Bokkker
13/20
CD-player example
Model G120 Ws order 1, Wt order 2 First guess
order-3 controller:
-
8/13/2019 Wort El Bokkker
14/20
-
8/13/2019 Wort El Bokkker
15/20
Closed-loop evaluation
10
2
10
3
10
4-80
-60
-40
-20
0
20
40
60
open-loop -G b
*K b
[ d B ]
10
2
10
3
10
4-1 5
-1 0
-5
0
5
10
15
20
controller K b
[ d B ]
102
103
104
-50
-40
-30
-20
-10
0
10
20sensitivity
Frequency [Hz]
[ d B ]
102
103
104
-8 0
-6 0
-4 0
-2 0
0
20complementary sensitivity
Frequency [Hz]
[ d B ]
ComplementarySensitivity
Sensitivity
bb K G b K
76.91323 = 77.91317 =
No visual difference
Model reduction OK
K23 K17
-
8/13/2019 Wort El Bokkker
16/20
HSV and performance measures
10 20 30 4010
-6
10-5
10-4
10-3 s tep aa
d e
l t a
& e
s t i m a
t e s
reduced order model4 6 8 10 1 2 14 1 6 1 8 20 2 2
0
500
1000
1500
2000
2500s t ep cc
g a m m a
& e
s t i m a
t e s
reduced order controller
5 10 15 2010
-6
10-4
10-2
100 step dd
d e
l t a
& e
s t i m a
t e s
reduced order model4 6 8 10 12 1 4 16
0
500
1000
1500
2000
2500step ff
g a m m a
& e
s t i m a
t e s
reduced order controller
HSV(G)
HSV(G)
HSV(K)
HSV(K)
-
8/13/2019 Wort El Bokkker
17/20
102
103
104
-120
-100
-8 0
-6 0
-4 0
-2 0
0
20
40
G120G120-G20G20-G14
G120
G20-G14
G120-G20
Models
-
8/13/2019 Wort El Bokkker
18/20
102
103
104
- 20
- 15
- 10
- 5
0
5
1 0
1 5
2 0K23K9K3
Controllers
K3
K23K9
-
8/13/2019 Wort El Bokkker
19/20
102
1 03
104-4 0
-3 0
-2 0
-1 0
0
10
20
30
40
max. sv of G23, G13, and difference
Direct controller reductionwith manually shaped
frequency weights
102
103
104-15
-10
-5
0
5
10
15
20
102
103
104-4 0
-3 0
-2 0
-1 0
0
10
20
30
40
max. sv of G23, G11, and difference
Plain balanced reduction
23
12
Interval weightedbalanced reduction23 to 11
Interval weightedbalanced reduction23 to 13
K23Kreduced
error
weighting
-
8/13/2019 Wort El Bokkker
20/20
Recommendations from Experience: See order reduction as a tool for control design Use robust and fast algorithms Exploit specific structure of model Apply iterative scheme (with frequency tuning) Evaluate results in more than one measure (graphics) Choose odd or even order
Do not rely on a priori HSV-based error bounds Customize and automate in later stage
Further reading:
P.M.R. Wortelboer, M. Steinbuch and O.H. BosgraIterative Model and Controller Reduction using Closed-loop Balancing,
with application to a Compact Disc Mechanism Int. J. Robust Nonlinear Control 9, 123-142 (1999)