Académique Documents
Professionnel Documents
Culture Documents
Chapter 6
Chapter 6
Chapter 6
Introduction
& x = Ax + Bu y = Cx
G ( s ) = C ( sI A) 1 B
y1 ( s ) = g11 ( s )u1 ( s ) + g12 ( s )u 2 ( s ) + ..... + g1 p ( s )u p ( s ) y 2 ( s ) = g 21 ( s )u1 ( s ) + g 22 ( s )u 2 ( s ) + ..... + g 2 p ( s )u p ( s ) .................................................................................... .................................................................................... y p ( s ) = g p1 ( s )u1 ( s ) + g p 2 ( s )u 2 ( s ) + ..... + g pp ( s )u p ( s )
We see that every input controls more than one output and that every output is controlled by more than one input. Because of this phenomenon which is called interaction it is generally phenomenon, interaction, very difficult to control a multivariable system.
3
Ali Karimpour July 2012
Chapter 6
Topics to be covered
Decoupling Pre and post compensators and the SVD controller Decoupling by State Feedback Diagonal controller (decentralized control) Uncertainty in MIMO Systems
4
Ali Karimpour July 2012
Chapter 6
Decoupling
Definition 6-1 A multivariable system is said to be decoupled if its transfer-function matrix is diagonal and nonsingular.
A conceptually simple approach to multivariable control is given by a two-steps procedure in which 1. We first design a compensator to deal with the interactions in G(s) and
G s ( s ) = G ( s )Ws ( s )
Decoupling
2. 2 Then design a diagonal controller using methods similar to those for SISO systems. systems
K s (s )
K ( s ) = Ws ( s ) K s ( s )
5
Ali Karimpour July 2012
Chapter 6
Decoupling
1. We first design a compensator to deal with the interactions in G(s) and
G s ( s ) = G ( s )Ws ( s )
Decoupling p g
Dynamic decoupling
Steady-state decoupling
Gs (0) is diagonal.
This may be obtained by selecting a constant pre compensator Approximate decoupling at frequency 0
G0 is a real approximation of Gs ( j0 )
Ws = G 1 (0)
Gs ( j0 ) is as diagonal as possible.
W s = G 01
Ali Karimpour July 2012
Chapter 6
Decoupling
The idea of using a decoupling controller is appealing, but there are several difficulties. a. We cannot in general choose Gs freely. For example, Ws(s) must not cancel any
RHP-zeros and RHP poles in G(s) b. As we might expect, decoupling may be very sensitive to modeling errors and
uncertainties. c. The requirement of decoupling may not be desirable for disturbance rejection. One popular design method, which essentially yields a decoupling controller, is the internal model control (IMC) approach (Morari and Zafiriou). Another A h common strategy, which avoids most of the problems just mentioned, is to hi h id f h bl j i d i use partial (one-way) decoupling where Gs(s) is upper or lower triangular. 7
Ali Karimpour July 2012
Chapter 6
Topics to be covered
Decoupling Pre and post compensators and the SVD controller Decoupling by State Feedback Diagonal controller (decentralized control) Uncertainty in MIMO Systems
8
Ali Karimpour July 2012
Chapter 6
Gs ( s ) = Wsp ( s )G ( s )Ws ( s )
The overall controller is then
K ( s ) = Ws ( s ) K s ( s )Wsp ( s )
9
Ali Karimpour July 2012
Chapter 6
Topics to be covered
Decoupling Pre and post compensators and the SVD controller Decoupling by State Feedback Diagonal controller (decentralized control) Uncertainty in MIMO Systems
10
Ali Karimpour July 2012
Chapter 6
G ( s )G ( s ) 1 diagonal
G ( s ) 1 = C ( sI A) 1 B + D
= D 1C ( sI A + BD 1C ) 1 BD 1 + D 1
But i th B t in the case of |D|=0 f |D| 0 Static state feedback Static t t f db k St ti output feedback Dynamic output feedback
u (t ) = Kx (t ) + Tr (t ) u (t ) = K ( t ) + T (t ) Ky Tr
11
Ali Karimpour July 2012
Chapter 6
) G ( s ) = C ( sI A + BE 1 F ) 1 BE 1
We h ll d i i h f ll i W shall derive in the following the condition on G( ) under which the system can be h di i G(s) d hi h h b decoupled by state feedback.
12
Ali Karimpour July 2012
Chapter 6
with the transfer function matrix G(s) can be decoupled by state feedback of the form
u (t ) = E 1 (Fx (t ) r (t ) )
if and only if the constant matrix E is nonsingular. Furthermore the new system is in the form:
s d1 Gnew ( s ) = 0
E1 s d1 E2 0 E = . = lim O G ( s ) s d 0 . s p E p Proof: See Linear system theory and design Chi-Tsong Chen
0 O d m s C1 A d1 d2 C 2 A F= . . dp C p A
13
Chapter 6
s 2 + 6s + 11 3 s + 6s 2 + 11s + 6 1 G ( s) = C ( sI A) B = 6 s 2 + 5s + 6
s+6 s 3 + 6s 2 + 11s + 6 s6 s 2 + 5s + 6
The differences in degree of the first row of G(s) are 1 and 2, hence d1=1 and s 2 + 6s + 11 s+6 E1 = lim s 3 = [1 0] 2 3 2 s s + 6s + 11s + 6 s + 6s + 11s + 6 The differences in degree of the second row of G(s) are 2 and 1, hence d2=1 and g () , 6 s6 E 2 = lim s 2 = [0 1] 14 s s + 5s + 6 s 2 + 5s + 6
Chapter 6
0 1 r 0
15
Ali Karimpour July 2012
Chapter 6
34-
No transmission zero in decoupled system. i i i d l d Transmission zero of the system are deleted .
16
Ali Karimpour July 2012
Chapter 6
Exercise 2: Decouple following system and find the decoupled transfer function. p g y p
0 1 & x= 0 0 0 0 0 0 0 0 0 1 x + 0 0 0 1 0 1 0 0 1 0 u 1 0
0 1 0 0 y= x 0 0 0 1
Exercise 3: Use state feedback to decouple the following system and put the poles of new system on s=-3.
1 0 1 0 & x = 0 0 1 x + 0 6 11 6 0 0 1 u 0 1 y= 0 0 1 0 x 1
17
Ali Karimpour July 2012
Chapter 6
Topics to be covered
Decoupling Pre and post compensators and the SVD controller Decoupling by State Feedback Diagonal controller (decentralized control) Uncertainty in MIMO Systems
18
Ali Karimpour July 2012
Chapter 6
Clearly, this works well if G(s) is close to diagonal, because then the plant to be controlled is essentially a collection of independent sub plants and each element in plants, K(s) may be designed independently. , g () g , performance with However, if off diagonal elements in G(s) are large, then the p decentralized diagonal control may be poor because no attempt is made to counteract the interactions.
19
Ali Karimpour July 2012
Chapter 6
The d i Th design of decentralized control systems involves two steps: fd t li d t l t i l t t 1_ 1 2_ The choice of pairings (control configuration selection) The design (tuning) of each controller ki(s)
20
Ali Karimpour July 2012
Chapter 6
Input-Output Pairing
Definition of RGA (Relative Gain Array)
Physical Meaning of RGA: Let Relative gain?
ij = g ij / hij
uk =0, k j yk =0, k i
g ij hij
yi relation between yi and u j if other inputs = 0 or u j yi relation between yi and u j if other outputs = 0 or u j
u2 =
g 21 u1 g 22
g 21 ) u1 y1 = g11 + g12 ( g 22
21
Ali Karimpour July 2012
(G ) = RGA(G ) = G G T
Chapter 6
Input-Output Pairing
Example: Let
y1 = g11u1 + g12u2 y2 = g 21u1 + g 22u2
(G ) = G G
T
1 = 1
=1 Open loop and closed loop gains are the same, so interactions has no effect. =0 g11=0 so u1 has no effect on y1. 0< Closing second loop leads to change the gain between y1 and u1. <0 Closing second loop leads to changing the sign of the gain between y1 and u1.(Very Bad) In this section we provide two useful rules for pairing inputs and outputs. outputs
1_ To avoid instability caused by interactions in the crossover region one should prefer pairings for which the RGA matrix in this frequency range is close to identity. 2_ To avoid instability caused by interactions at low frequencies one should avoid 22 pairings with negative steady state RGA elements.
Ali Karimpour July 2012
Chapter 6
Input-Output Pairing
RGA property: 11 It is independent of input and output scaling. scaling 2- Its rows and columns sum to 1. 3- The RGA is identity matrix if G is upper or lower triangular. 4- Plant with large RGA elements are ill conditioned. 5- Suppose G(s) has no zeros or poles at s=0. If ij() (0) exist and have different signs then one of the following must be true. * G(s) has an RHP zeros. * Gij(s) has an RHP zeros. * gij(s) has an RHP zeros. 6- If gij gij(1-1/ij) then the perturbed system is singular. 7- Changing two columns/rows of G leads to same changes to Karimpour July 2012 Ali its RGA
23
Chapter 6
24
Ali Karimpour July 2012
Chapter 6
Absolute Relative Gain Array (ARGA), Relative Normalized Gain Array (RNGA), (RNGA)
[2] ARGA Loop Pairing Criteria for Multivariable Systems A. Balestrino, E. Crisostomi, A. Landi, and A. Menicagli ,2008 [3] RNGA based control system configuration for multivariable processes Mao-Jun He, Wen-Jian Cai *, Wei Ni, Li-Hua Xie Journal of Process Control 19 (2009) 10361042
25
Ali Karimpour July 2012
Chapter 6
Chapter 6
off-diagonal
diagonal
27
Ali Karimpour July 2012
Chapter 6
Topics to be covered
Decoupling Pre and post compensators and the SVD controller Decoupling by State Feedback Diagonal controller (decentralized control) Uncertainty in MIMO Systems
28
Ali Karimpour July 2012
Chapter 6
where
K r = R 1 B T X
Where X=XT 0 is the unique positive-semidefinite solution of the algebraic Riccati equation
AT X + XA XBR 1 B T X + M T QM = 0
29
Ali Karimpour July 2012
Chapter 6
S = I + K r (sI A) B
1
satisfies
(S ( j ) ) 1,
Nyquist plot in MIMO case
30
Ali Karimpour July 2012
Chapter 6
1 0 1 & x= x + 1u 0 2 y = [1 1]x
(y
+ Ru 2 dt
Let R = 0.0001
K r = R 1 B T X = [86.7008 - 75.3816]
( A bk ) = 7.1596 6.9828i
Chapter 6
The pre compensator approach may be extended by introducing a post compensator 1 0 7 8 7 8 s + 1 Gs ( s ) = Wsp ( s )G ( s )Ws ( s ) = Gd ( s ) G(s) = 6 7 2 6 7 0 s + 2 The overall controller is then h ll ll i h
K ( s ) = Ws ( s ) K s ( s )Wsp ( s )
7 8 k 0 7 8 k 0 K (s) = 0 k 6 7 = 0 k 6 7
Chapter 6
Chapter 6
34
Ali Karimpour July 2012
Chapter 6
Now let
1 G p ( s) = G0 ( s ) min p max ps +1
p = (1 r )
= ( min + max )
G0 ( s ) G0 ( s ) r s 1 G p ( s) = w2 ( s ) = = , w1 ( s ) = 1 1 + s r s 1 + s 1 w2 ( s )w1 ( s ) 1+ s
G p ( s ) = G ( s )(1 w2 ( s ) w1 ( s ) ) 1
Nonparametric uncertainty has more conservativeness.
35
Ali Karimpour July 2012
1 2
rt =
( min max ) / 2
, <1
Chapter 6
G p ( s) =
k e s 2 k , , 3 s +1
36
Ali Karimpour July 2012
Chapter 6
37
Ali Karimpour July 2012
Chapter 6
z P = 11 P21
P w 12 P22 u
u = K
z = P + P K ( I P22 K ) 1 P21 w = Nw 11 12
N = Fl ( P, K )
System with uncertainty y y
N structure
Pull t P ll out uncertainty Suitable for robust performance analysis
38
Chapter 6
z P = 11 P21
P w 12 P22 u
u = K
z = P + P K ( I P22 K ) 1 P21 w = Nw 11 12
N = Fl ( P, K )
System with uncertainty N structure y y
y N11 z = N 21
N12 u N 22 w
u = y
z = N 22 + N 21 ( I N11 ) 1 N12 w = Fw
F = Fu ( N , )
39
Ali Karimpour July 2012
Chapter 6
Robust stability in parametric uncertainty MIMO case Robust stability in parametric uncertainty
40
Ali Karimpour July 2012
Chapter 6
N structure
z = P + P K ( I P22 K ) 1 P21 w = Nw 11 12
z = N 22 + N 21 ( I N11 ) 1 N12 = Fw
Chapter 6
N structure
Suitable for nominal performance analysis Suitable for robust performance analysis
N11, N12, N21 and N22 are stable Suitable for controller design Suitable for robust stability analysis42
Ali Karimpour July 2012
M = N11
Chapter 6
Theorem: RS for unstructured(full) perturbation. Assume that the nominal system M(s) is stable (NS) and that the perturbations (s) are stable. Then The M-structure is stable M < 1/ (M ( j)) < 1/ for all satisfying |||| The M-structure is stable
M
<1
M structure
Chapter 6
Gp = G + w2w1
System with additive uncertainty
y = Mu
M = w1K (I + GK)1 w2
Robust stability condition: In obust stab ty co d t o : the case of |||| 1
M
(K (I + GK)1 )< 1
= w1K (I + GK)1 w2
<1
44
M structure
Chapter 6
Gp = G(I + w2w1 )
System with multiplicative input uncertainty
y = Mu
M = w1K (I + GK)1G 2 Gw
Robust stability condition: In the case of |||| 1
M = w1K (I + GK)1Gw2
45
(K (I + GK)1G)< 1
<1
M structure
Chapter 6
Gp = (I + w2w1 )G
System with multiplicative output uncertainty t t t i t
y = Mu
M = w1GK ( I + GK ) 1 w2
= w1GK(I + GK)1 w2
46
<1
M structure
Chapter 6
Gp = G(I w2w1G)1
System with inverse additive uncertainty
y = Mu
M = w1G(I + KG)1 w2
Robust stability condition: In the case of |||| 1
M = w1G(I + KG)1 w2
<1
47
M structure
Chapter 6
Gp = G(I w2w1 )1
System with inverse multiplicative input uncertainty
w2w1 < 1
y = Mu
M = w1 ( I + KG ) 1 w2
<1
48
M structure
Chapter 6
Gp = (I w2w1 )1G
System with inverse multiplicative output uncertainty
w2w1 < 1
M = w1 ( I + GK ) 1 w2
<1
49
M structure
Chapter 6
Uncertainty
Additive uncertainty Multiplicative input uncertainty Multiplicative output uncertainty Inverse additive uncertainty y
Perturbed Plant
M in M-structure
Gp = G + w2w1
M = w1 K ( I + GK ) 1 w2
M = w1K (I + GK)1Gw2 M = w1GK(I + GK)1 w2
1
M = w1G ( I + KG ) 1 w2
Inverse multiplicative input uncertainty Gp = G(I w2w1 )1 Inverse multiplicative output uncertainty Gp = (I w2w1 )1G
M = w1 ( I + KG ) 1 w2 M = w1 ( I + GK ) 1 w2
50
Ali Karimpour July 2012
M structure
Chapter 6
G = Ml Nl Gp = (Ml + M )1( Nl + N ) = [ N M ]
K M = ( I + GK ) 1 M l1 I
RS : N
< 1/
51
Ali Karimpour July 2012
Chapter 6
Gp = G(I + )
52
Ali Karimpour July 2012
M structure
Chapter 6
() < 1/ (K (I + GK)1G)
53
Ali Karimpour July 2012
Chapter 6
1, and i
1 )
Suppose we chose G(s)=3/(2s+1) with multiplicative uncertainty. Derive suitable scaling 54 Matrix.
Ali Karimpour July 2012