Académique Documents
Professionnel Documents
Culture Documents
I.D. Landau
Emeritus Research Director at C.N.R.S
Laboratoire d’Automatique de Grenoble, (INPG/CNRS), France
April 2004,
Valencia
I.D. Landau A course on robust discrete time control, part1 1
Outline
Part I
Introduction, examples and basic concepts
Part II
Design of robust discrete time controllers
Part III
Special topics
IDENTIFICATION
Ref. + u y
CONTROLLER PLANT
+
CLOCK
r(t) u(t)
Computer D/A y(t)
+ PLANT A/D
(controller) ZOH
Discretized Plant
−d −1 − d −1 −1
q B( q ) q B * ( q )
Plant Model: G ( q −1
) = H ( q −1
) = =
A(q −1 ) A( q −1 )
A(q −1 ) = 1 + a1q −1 + ... + an q −n B(q −1 ) = b1q −1 + ... + bn q −n = q −1 B * ( q −1 )
A
A
B
B
Nominal stability : closed loop as. stable for the “nominal model”
Robust stability: closed loop as. Stable for all the plant models
belonging to the family (set) P
Robust stability:
-1 1 The closed loop poles remain inside the unit circle
(discrete time case) for all the models belonging to
the family P
-1 r 1
Robust performance:
c
The closed loop poles remain inside the circle (c,r)
for all the models belonging to the family P
Level to be controlled
Shaking
(to prevent adherence)
+
u(t) -d + y(t)
q B
A +
Specifications:
1. No attenuation of the sinusoidal disturbance (0.25 Hz)
2. Attenuation band in low frequencies: 0 à 0.03 Hz
3. Disturbance amplification at 0.07 Hz: < 3dB
4. Modulus margin > -6 dB and Delay margin > TS
5. No integrator in the controller
Finished
Steel product
strip
Measurement of
Air knife
Preheat oven deposited mass
Zinc
bath
air air
zinc
steel strip
Desired
deposited zinc mass per cm2 Measured
PRBS deposited zinc mass per cm2
(± ∆P)
m*
P0 +
Measu-
RST Pressure Pressurized Coating
rement
controller Gaz supply process
controller +
delay
m ± ∆m
q −7 (b q −1 )
Plant model for a type of product : 1
1 + a1 q −1
Model : TS = 12 sec; b1 = 0.3; a1 = −0.2(− 0.3) )
Specifications (performance) :
•Modulus margin: ∆M ≥ 0.5
•Delay margin: ∆τ ≥ 2TS
•Integrator
Φm
axis
Position
position
d.c. transducer
motor
y(t) A
D
D u(t) C
Controller R-S-T
A
C
controller Φref
RIGID FRAMES
MIRROR
ALUMINIUM
DETECTOR
COMPUTER
TACH. LOCAL
POSITION
POT. SERVO.
ENCODER
All four sensitivity functions should be stable ! (see book pg.102 - 103)
I.D. Landau A course on robust discrete time control, part1 27
All four sensitivity functions must be stable !
z − d B ( z −1 ) −1
Plant model: −1
A ( z ) is unstable
A( z )
Supose : R ( z −1 ) = A( z −1 ) (poles compensation by the controller zeros)
−1 A( z −1 ) S ( z −1 ) S ( z −1 )
S yp ( z ) = −1 −1 −1 −1
=
A( z ) S ( z ) + B( z ) A( z ) S ( z −1 ) + B ( z −1 )
−1 A( z −1 ) A( z −1 ) A( z −1 )
Sup ( z ) = − −1 −1 −1 −1
=−
A( z ) S ( z ) + B ( z ) A( z ) S ( z −1 ) + B( z −1 )
−1 B( z −1 ) A( z −1 ) B ( z −1 )
S yb ( z ) = − −1 −1 −1 −1
=−
A( z ) S ( z ) + B( z ) A( z ) S ( z −1 ) + B ( z −1 )
−1 B( z −1 ) S ( z −1 ) B( z −1 ) S ( z −1 )
S yv ( z ) = =
A( z ) S ( z ) + B( z ) A( z ) A( z −1 )[ S ( z −1 ) + B ( z −1 )]
−1 −1 −1 −1
S yp ( z −1 ) − S yb ( z −1 ) = S yp ( z −1 ) + S yr ( z −1 ) = 1
-1 H (e -j ω )
S yp = 1 + H (e -jω ) BO
BO
A( z −1 ) S ( z −1 ) + B( z −1 ) R( z −1 )
S yp ( z ) = 1 + H OL ( z ) =
−1 −1 −1
ω =0
A( z −1 ) S ( z −1 )
Remarks:
Im H -The controller poles may become
ω = 2π
1 unstable pole in Open Loop unstable if high performances are
Stable Closed Loop (a) required without using an appropriate
design method
ω=π Re H
-1
-The Nyquist plot from 0.5fS to fS is the
ω=π Stable Open Loop
symmetric with respect to the real axis
Unstable Closed Loop(b)
of the Nyquist plot from 0 to 0.5fS
ω =0 ω =0
1 -Gain margin ∆G
∆G -Phase margin ∆φ
-1 1 Re H
-Delay margin ∆τ
∆Μ
∆Φ -Modulus margin ∆M
Crossover |H OL |=1
frequency
ωCR
Gain margin
1
∆G = pour ∠φ (ω180 ) = −180 o
H BO ( jω 180 )
Phase margin
∆φ = 180 0 − ∠φ (ω cr ) pour H BO ( jω cr ) = 1
∆φ = min ∆ φ i If there several intersections with the unit circle
i
Delay margin
∆φ ∆φ i
∆τ = Several intersections points: ∆τ = min
ω cr i ω cr
i
Modulus margin
−1
∆M = 1 + H BO ( jω ) min = −1
S yp ( jω ) = S yp ( jω )
min max
I.D. Landau A course on robust discrete time control, part1 33
Robustness margins – typical values
Good gain and phase margin Good gain and phase margin
Bad modulus margin Bad delay margin
I.D. Landau A course on robust discrete time control, part1 35
Modulus margin and sensitivity function
−1
−1
A( z ) S ( z ) −1
pour z −1 = e − j 2πf
A( z ) S ( z ) + B( z ) R( z ) max
−1 −1 −1 −1
S yp (e − jω ) dB = ∆M −1dB = − ∆M dB
max
dB
-1
S yp
S yp = - ∆M
max
0
ω
-1
S yp = - S yp = ∆M
S yp min max
S yp = - ∆M
-1 Re H max
1 + H OL 0
HOL ω
'
H OL -1
S yp = - S yp = ∆M
S yp min max
H’OL HOL H’ H
B ′( z −1 ) B ( z −1 ) A( z −1 ) S ( z −1 ) + B ( z −1 ) R ( z −1 ) P ( z −1 ) (**)
− < = = S up−1 ( z −1 )
A′( z ) A( z )
−1 −1 −1
A( z ) R ( z ) −1 −1
A( z ) R ( z )−1
H’ H S
dB
up
Limitation of the actuator stress
0
0.5f e
S up -1
(Size of tolerated uncertainties)
I.D. Landau A course on robust discrete time control, part1 38
Tolerance to plant normalized uncertainty
(multiplicative uncertainty)
B′( z −1 ) B ( z −1 )
−
A′( z −1 ) A( z −1 ) A( z −1 ) S ( z −1 ) + B ( z −1 ) R ( z −1 ) P ( z −1 )
−1
< −1 −1
= −1 −1
= S yb−1 ( z −1 )
B( z ) B( z ) R( z ) B( z ) R( z )
A( z −1 )
H '−H
H ' = H + ( H '−H ) = H (1 + )
H
I.D. Landau A course on robust discrete time control, part1 39
Passivity (Hyperstability) Theorem
u1 H1 y1
strictly
- positive real
H2
y2 passive u2
Then :
lim
t →∞
x(t ) = 0 ; lim
t →∞
u1 (t ) = 0 ; lim
t →∞
y1 (t ) = 0
I.D. Landau A course on robust discrete time control, part1 40
Strictly positive real transfer function (SPR)
Frequency domain
continuous discrete
Im z
jω Im H Im H
+j jω
s s =j ω H(j ω)
z
jω H(e )
z =e
Re H Re H
σ -1 1
Re z
j jω
Re H < 0 Re H > 0 -
z =e Re H < 0 Re H > 0
- asimptotycally stable
- Re H (e ) > 0 for all e = 1, (0 < ω < π )
jω jω
(discrete time case)
ku2 - k1
-1 Re H (e -j ω )
u2
Re H ( z ) > −1 / k ; ∀ z = 1
H (e -j ω )
Im H (e -j ω )
critical circle
β u
u2 1 -1
Re H (e -j ω )
α u -α β
y2 -1
H (e -j ω )
Circle criterion
Im H ( e -j ω )
critical circle
u βu
αu -1
H (z ) - 1 -1
BO α -1
β Re H ( e -j ω )
-
∆M
H ( e -j ω )
KM=1
-1 1
Km=-1
H(z) should lie inside the unit circle
The modulus of the H(z) should be smaller than 1 at all frequencies i.e.:
H ( z) ∞ < 1
This is the “small gain theorem”
I.D. Landau A course on robust discrete time control, part1 44
Small gain theorem
u1 y1
S1
- S1 ∞ < 1
S2
y2 S2 ∞
≤1 u2
Then:
lim
t →∞
x(t ) = 0 ; lim
t →∞
u1 (t ) = 0 ; lim
t →∞
y1 (t ) = 0
H −1
If H is passive, S∞= ≤1
H +1 ∞
If S ∞ ≤ 1 ⇒ H = 1 + S is passive
1− S
H −1
If H is strictly passive , S ∞ = <1
H +1 ∞
1+ S
If S < 1 ⇒ H = is strictly passive
∞
1− S
Hint for a proof:
H (e − j ω ) − 1 H − 2 Re H + 1
2 2
4 Re H
S (e ) = = = −
− jω2
1
H (e − j ω ) + 1 H + 2 Re H + 1 H (e − jω ) + 1
2 2 2
Remark:
Correct writing in the general operator case S ∞ = ( H − 1)( H + 1) −1 ∞ ≤ 1
S1 S1 ∞ < 1
I
u1
H1 -
2I strictly
+ y1
- + - positive real
I
I
- +
0.5I H2
y2 + passive - u2
I
S2 S2 ∞
≤1
Re H
Uncertainty disk
(at a certain frequency)
1) It needs a description by a transfer function which may have any phase but a modulus < 1
2) The size of the radius will vary with the frequency and is characterized by a transfer function
δ Wa
+ − Sup Wa
K H -
+
-
δ
Apply small gain theorem
K = R / S; H = z B / A
−d
Wa ( z −1 ) = H ( z −1 )Wm ( z −1 )
δ Wm
+ − S yb Wm
K H -
- +
δ
δ Wm
- − S yp Wr
K H -
-
δ
+
• Multiplicative uncertainties
S yb ( z −1 )Wm ( z −1 ) ∞ < 1 S yb (e − jω ) < Wm (e − jω ) 0≤ω ≤π
−1
−1 −1
•The functions W ( z ) (the inverse of the size of the uncertainties)
define an “upper” template for the sensitivity functions
• Conversely the frequency profile of S xy (e − jω ) can be interpreted in
terms of tolerated uncertainties
dB
-1 S
S yp up dB
S yp = - ∆M
max
0 0
ω 0.5f e
-1
S = - S yp = ∆M S up -1
yp min max
S yp
f ( z −1 ) = 1, z −1 , z − 2 ,...,
2
Examples of families of plant models for which robust stability is guaranteed:
1
H ′( z −1 ) = H ( z −1 )
1 − λ∆ M
1
H ′( z −1 ) = H ( z −1 )
1 − λ∆ M z −1
I.D. Landau A course on robust discrete time control, part1 54
Delay margin and robust stability
z −d B ( z −1 ) z − d −1B ( z −1 )
∆τ = 1.Ts H (z ) = −1
−1
−1
H ′( z ) =
A( z ) A( z −1 )
H ′( z −1 ) − H ( z −1 )
−1
= z −1 − 1 Can be interpreted as a multiplicative uncertainty
H (z )
H ′( z −1 ) = H ( z −1 )[1 + δ ( z −1 )Wm ( z −1 ) ] = H ( z −1 )[1 + ( z −1 − 1)]
δ ( z −1 ) = 1 ; Wm ( z −1 ) = ( z −1 − 1)
Robust stability condition:
S yb (e − jω ) < Wm (e − jω ) 0 ≤ ω ≤ π
−1
1
S yb ( z ) < −1 ; z = e − jω , 0 ≤ ω ≤ π
−1
z −1
S yb−1 ( z −1 ) dB < −20 log 1 − z −1 ; z = e − jω
Define a template on Syb Syb
I.D. Landau A course on robust discrete time control, part1 55
Delay margin template on Syp
−1 A( z −1 )S ( z −1 ) −1 − B( z −1 ) R( z −1 )
S yp ( z ) = S yb ( z ) =
A( z −1 )S ( z −1 ) + B( z −1 )R( z −1 ) A( z −1 )S ( z −1 ) + B( z −1 ) R( z −1 )
S yp ( z −1 ) = 1 + S yb ( z −1 )
1 − S yb ( z −1 ) ≤ S yp ( z −1 ) ≤ 1 + S yb ( z −1 )
1
S yb ( z −1 ) < ; z = e − jω
,0≤ω ≤π
z −1
−1
−1 −1
1 − 1 − z −1 ≤ S yp ( z −1 ) ≤ 1 + 1 − z −1
Approximate
condition
Upper template:
Lower template:
nominal delay
perform. margin Output Sensitivity
0 Function
0.5f s
S up dB
actuator effort
Input Sensitivity
0 Function
0.5f s
S up -1
size of the tolerated additive uncertaintyaW
( G ’ = G + δW a )
I.D. Landau A course on robust discrete time control, part1 59
Templates for the Sensitivity Functions
S yp dB S yp max = - ∆ M
Output Sensitivity
0 Function
0,5f e
opening
the loop
Attenuation zone
S up dB
0 Input Sensitivity
Zone with uncertainty
Function
upon the modelquality