1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
|
lqg2stan Scilab Group Scilab Function lqg2stan
NAME
lqg2stan - LQG to standard problem
CALLING SEQUENCE
[P,r]=lqg2stan(P22,bigQ,bigR)
PARAMETERS
P22 : syslin list (nominal plant) in state-space form
bigQ : [Q,S;S',N] (symmetric) weighting matrix
bigR : [R,T;T',V] (symmetric) covariance matrix
r : 1x2 row vector = (number of measurements, number of inputs)
(dimension of the 2,2 part of P)
P : syslin list (augmented plant)
DESCRIPTION
lqg2stan returns the augmented plant for linear LQG (H2) controller
design.
P22=syslin(dom,A,B2,C2) is the nominal plant; it can be in continuous
time (dom='c') or discrete time (dom='d').
.
x = Ax + w1 + B2u
y = C2x + w2
for continuous time plant.
x[n+1]= Ax[n] + w1 + B2u
y = C2x + w2
for discrete time plant.
The (instantaneous) cost function is [x' u'] bigQ [x;u].
The covariance of [w1;w2] is E[w1;w2] [w1',w2'] = bigR
If [B1;D21] is a factor of bigQ, [C1,D12] is a factor of bigR and
[A,B2,C2,D22] is a realization of P22, then P is a realization of
[A,[B1,B2],[C1,-C2],[0,D12;D21,D22]. The (negative) feedback computed by
lqg stabilizes P22, i.e. the poles of cl=P22/.K are stable.
EXAMPLE
ny=2;nu=3;nx=4;
P22=ssrand(ny,nu,nx);
bigQ=rand(nx+nu,nx+nu);bigQ=bigQ*bigQ';
bigR=rand(nx+ny,nx+ny);bigR=bigR*bigR';
[P,r]=lqg2stan(P22,bigQ,bigR);K=lqg(P,r); //K=LQG-controller
spec(h_cl(P,r,K)) //Closed loop should be stable
//Same as Cl=P22/.K; spec(Cl('A'))
s=poly(0,'s')
lqg2stan(1/(s+2),eye(2,2),eye(2,2))
SEE ALSO
lqg, lqr, lqe, obscont, h_inf, augment, fstabst, feedback
AUTHOR
F.D.
|
