TimeVarying Systems and Computations
Lecture 3
Klaus Diepold
1. November 2012
Linear TimeVarying Systems
StateSpace System Model
We aim to derive the matrix containing the timevarying impulse responses by inspection of a generic
causal timevarying system. Recall the state equations
x
k
+1
=
A
k
·
x
k
+
B
k
·
u
k
y
k
=
C
k
·
x
k
+
D
k
·
u
k
,
(1)
or in matrix form,
x
k
+1
y
k
=
A
k
B
k
C
k
D
k
·
x
k
u
k
,
(2)
These statespace equations are similar to the equations for the timeinvariant case except that now the
entries
{
A
k
, B
k
, C
k
, D
k
}
of the realization matrix are depending on the time index
k
, which means that
these matrices can change from time step to time step, incorporating the timevariation that we are
looking for. A generic causal timevarying system is drawn as shown in Figure 1.
Z
Z
u
−
1
B
−
1
D
−
1
y
−
1
x
0
Z
u
0
A
0
C
0
D
0
B
0
y
0
x
1
A
1
C
1
D
1
B
1
u
1
y
1
x
2
Z
Z
Z
C
2
A
2
D
2
B
2
u
2
y
2
x
3
Z
Z
C
3
D
3
u
3
y
3
Abbildung 1: A timevarying system
Besides the timevarying system parameters
{
A
k
, B
k
, C
k
, D
k
}
we can see that the dimension of the
statespace, i.e. the number of delay elements may change from time step to time step. Similarly, the
dimension of the input signal vector
u
k
and the dimension of the output signal vector
y
k
may change
with time. However, we will not dwell in more detail on the the latter point.
1
This preview has intentionally blurred sections. Sign up to view the full version.
View Full Document
Lecture 3
2
TimeVarying ImpulseResponse
We seek to determine the I/O operator, i.e. the Toeplitz operator for this timevarying causal system.
We proceed in a similar fashion as with the timeinvariant case, by applying an impulse at time instant
k
and noting the output of the system (only in timevarying case, the response depends on
k
as well).
Taking the impulse responses as the columns of the matrix
T
we arrive at
T
=
k
= 0
k
= 1
k
= 2
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
0
.
.
.
.
.
.
.
.
.
D
0
0
.
.
.
.
.
.
C
1
B
0
D
1
0
.
.
.
C
2
A
1
B
0
C
2
B
1
D
2
.
.
.
.
.
.
C
3
A
2
A
1
B
0
C
3
A
2
B
1
C
3
B
2
.
.
.
C
4
A
3
A
2
A
1
B
0
C
4
A
3
A
2
B
1
C
4
A
3
B
2
.
.
.
.
.
.
C
5
A
4
A
3
A
2
B
1
C
5
A
4
A
3
B
2
.
.
.
.
.
.
.
.
.
C
6
A
5
A
4
A
3
B
2
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
(3)
We can note from the above that
•
i
th
column represents the impulse response of the system for an impulse input at
k
=
i
, i.e.
u
i
=
1
i
=
k,
0
else
•
T
has the same functionality as Toeplitz matrix of the timeinvariant case. In fact, if the system is
timeinvariant, all the matrices
A
k
, B
k
, C
k
, D
k
are identical, hence
T
will exhibit Toeplitz structure.
•
We can identify the system to be causal by lowertriangular structure of
T
.
•
The clear difference is that
T
has no Toeplitz structure.
•
Because of its timevarying nature, the dimension of the matrix
T
may be finite.
This is the end of the preview.
Sign up
to
access the rest of the document.
 Winter '12
 TUM
 Matrices, UK, realization, Transfer Operator

Click to edit the document details