1 Linear Time-Varying Systems

LTV system in state space

x(t) = A(t)x(t) + B(t)u(t),
y(t) = C(t)x(t) + D(t)u(t).
1.1 Existence and uniqueness of solution
Dierential equation:
x(t) = f(x(t), t), a t b.
Sucient condition for existence and uniqueness of solution: f(x, t) is
Lipschitz, i.e.,
f(y(t), t) f(x(t), t) k(t)y(t) x(t), a t b,
where k() is (piecewise) continuous.
For LTV f(x, t) = A(t)x(t) + B(t)u(t) and
f(y(t), t) f(x(t), t) = A(t)y(t) + B(t)u(t) A(t)x(t) B(t)u(t),
= A(t)y(t) A(t)x(t),
A(t)y(t) x(t) Lipschitz!
Conclusion: LTV system has a solution and it is unique!
MAE 280A 1 Maurcio de Oliveira
1.2 Solution to LTV
Scalar homogeneous equation
x(t) = a(t)x(t), t 0, x(0) = x
0
.
Separation of variables
1
x
dx = a(t)dt.
Integrate on both sides
_
x(t)
x(0)
1
x
dx =
_
t
0
a()d, x(t) = e

t
0
a()d
x(0).
In the matrix case, assume
x(t) = e

t
0
A()d
x(0),
and compute
d
dt
x(t) =
d
dt
e
F(t)
x(0), F(t) =
_
t
0
A()d,
=

i=0
1
i!
d
dt
F
i
(t)x(0).
But
d
dt
F
2
(t) = F(t)
d
dt
F(t) + [
d
dt
F(t)]F(t),
=
_
t
0
A()dA(t) + A(t)
_
t
0
A()d = 2A(t)
_
t
0
A()d,
since A(t) and
_
t
0
A()d do not necessarily commute!
Therefore
d
dt
x(t) =
d
dt
e

t
0
A()d
x(0) = A(t)e

t
0
A()d
x(0) = A(t)x(t)
Remember that for LTI A(t) = A and
_
t
0
A()d = At commutes with A!
MAE 280A 2 Maurcio de Oliveira
1.2.1 Equivalent transformations for LTV systems
LTV in state space
x(t) = A(t)x(t) + B(t)u(t),
y(t) = C(t)x(t) + D(t)u(t).
Let P(t) be nonsingular for all t and dene
x(t) = P(t)z(t)
such that
x(t) =

P(t)z(t) + P(t) z(t) = A(t)P(t)z(t) + B(t)u(t),
P(t) z(t) =
_
A(t)P(t)

P(t)
_
z(t) + B(t)u(t).
Equivalent LTV system
z(t) = P(t)
1
_
A(t)P(t)

P(t)
_
x(t) + P(t)
1
B(t)u(t),
y(t) = C(t)P(t)z(t) + D(t)u(t).
MAE 280A 3 Maurcio de Oliveira
1.2.2 Fundamental Matrix
P(t) is called a fundamental matrix when

P(t) = A(t)P(t), |P(t

0
)| = 0.
If P(t) is a fundamental matrix then
z(t) = P(t)
1
B(t)u(t) z(t) = z(t
0
) +
_
t
t
0
P()
1
B()u()d.
and
x(t) = P(t)z(t
0
) +
_
t
t
0
P(t)P()
1
B()u()d,
= P(t)P(t
0
)
1
x(t
0
) +
_
t
t
0
P(t)P()
1
B()u()d,
= (t, t
0
)x(t
0
) +
_
t
t
0
(t, )B()u()d, (t, ) = P(t)P()
1
.
1.2.3 State Transition Matrix
(t, ) is called the state transition matrix
Properties
1) (t, t) = I,
2)
1
(t, ) = (, t),
3) (t
1
, t
2
) = (t
1
, t
0
)(t
0
, t
2
).
4)
d
dt
(t, ) = A(t, ), (, ) = I.
Proof:
1) (t, t) = P(t)P
1
(t) = I,
2)
1
(t, ) =
_
P(t)P
1
()

1
= P()P
1
(t) = (, t),
3) (t
1
, t
2
) = P(t
1
)P
1
(t
2
)P(t
2
)P
1
(t
3
) = P(t
1
)P
1
(t
3
) = (t
1
, t
3
).
4)
d
dt
(t, ) =
d
dt
P(t)P()
1
=

P(t)P() = A(t)P(t)P() = A(t)(t, ).
MAE 280A 4 Maurcio de Oliveira
1.2.4 Complete solution
y(t) = C(t)x(t) + D(t)u(t),
= C(t)(t, t
0
)x(t
0
) +
_
t
t
0
C(t)(t, )B()u()d + D(t)u(t).
For SIMO we have
y(t) = C(t)(t, t
0
)x(t
0
) +
_
t
t
0
[C(t)(t, )B() + D()(t )] u()d,
= C(t)(t, t
0
)x(t
0
) +
_
t
t
0
h(t, )u()d,
where
h(t, ) = C(t)(t, )B() + D()(t )
is the impulse response.
MAE 280A 5 Maurcio de Oliveira
1.2.5 Floquet theory
There exists

P(t) that transforms the LTV homogeneous system
x(t) = A(t)x(t)
into the equivalent LTI homogeneous system
z(t) =

Az(t)
We have already seen one case:

A = 0!
In general, for any

P(t) nonsingular we have
z(t) =

P(t)
1
_
A(t)

P(t)

P(t)
_
z(t)
Dening

P(t) = P(t)e

At
,
where P(t) is any fundamental matrix then

P(t)
1
_
A(t)

P(t)

P(t)
_
= e

At
P
1
(t)
_
A(t)P(t)e

At
+ P(t)e

At

A

P(t)e

At
_
,
= e

At
P
1
(t)P(t)e

At

A
+ e

At
P
1
(t)
_
A(t)P(t)

P(t)
_
e

At
,
=

A.
When

A = 0 then

P(t) = P(t)!
MAE 280A 6 Maurcio de Oliveira
1.3 Example
LTV system
x =
_
0 0
t 0
_
x
is equivalent to equations
_
x
1
= 0,
x
2
= tx
1
.

_
_
_
x
1
(t) = x
1
(t
0
)
x
2
(t) = x
2
(t
0
) +
1
2
(t
2
t
2
0
)x
1
(t
0
)
Fundamental matrix for t
0
= 0
_
x
1
(0)
x
2
(0)
_
=
_
1
0
_

_
x
1
(t)
x
2
(t)
_
=
_
1
1
2
t
2
_
,
_
x
1
(0)
x
2
(0)
_
=
_
0
1
_

_
x
1
(t)
x
2
(t)
_
=
_
0
1
_
,
so that
P(t) =
_
1 0
1
2
t
2
1
_
.
State transition matrix
(t, t
0
) = P(t)P
1
(t
0
),
=
_
1 0
1
2
t
2
1
_ _
1 0
1
2
t
2
0
1
_
1
,
=
_
1 0
1
2
t
2
1
_ _
1 0

1
2
t
2
0
1
_
,
=
_
1 0

1
2
(t
2
t
2
0
) 1
_
.
MAE 280A 7 Maurcio de Oliveira