1
projsl Scilab Group Scilab Function projsl
3
projsl - linear system projection
10
Q,M : matrices (projection factorization)
13
slp= projected model of sl where Q*M is the full rank factorization of
16
If (A,B,C,D) is the representation of sl, the projected model is given by
19
Usually, the projection Q*M is obtained as the spectral projection of an
20
appropriate auxiliary matrix W e.g. W = product of (weighted) gramians or
21
product of Riccati equations.
24
rand('seed',0);sl=ssrand(2,2,5);[A,B,C,D]=abcd(sl);poles=spec(A)
25
[Q,M]=pbig(A,0,'c'); //keeping unstable poles
26
slred=projsl(sl,Q,M);spec(slred('A'))
27
sl('D')=rand(2,2); //making proper system
28
trzeros(sl) //zeros of sl
29
wi=inv(sl); //wi=inverse in state-space
30
[q,m]=psmall(wi('A'),2,'d'); //keeping small zeros (poles of wi) i.e. abs(z)<2
31
slred2=projsl(sl,q,m);
32
trzeros(slred2) //zeros of slred2 = small zeros of sl
33
// Example keeping second order modes
35
sl=syslin('c',A,rand(3,2),rand(2,3));[nk2,W]=hankelsv(sl)
36
[Q,M]=pbig(W,nk2(2)-%eps,'c'); //keeping 2 eigenvalues of W
37
slr=projsl(sl,Q,M); //reduced model