Information Technology Reference
In-Depth Information
means that the
i
j
th subsystem is active when
t
[
t
j
,t
j
+1
).
A
σ
,B
σ
are con-
stant matrices of appropriate dimensions. The uncertainties
ΔA
σ
and
ΔB
σ
are
assumed to satisfy the following assumption.
∈
Assumption 1.
The uncertainties are in the form:
ΔA
σ
=
D
σ
Δ
A
σ
(
t
)
L
σ
,ΔB
σ
=
E
σ
Δ
B
σ
(
t
)
M
σ
(2)
where
D
σ
,L
σ
,E
σ
and
M
σ
are the constant matrices with appropriate dimen-
sions, and
ΔA
σ
and
ΔB
σ
are the unknown, real and possible time-varying
matrices satisfying
Δ
A
σ
(
t
)
Δ
A
σ
(
t
)
0and
Δ
B
σ
(
t
)
Δ
B
σ
(
t
)
≤
I,t
≥
≤
I,t
≥
0,
respectively.
For the system (1), we design the state feedback controller as follows
u
(
t
)=(
K
σ
+
ΔK
σ
)
x
(
t
)
(3)
R
m×n
denotes the state feedback gain matrix and
ΔK
σ
denotes
the control gain perturbation matrix. In this paper, we consider the following
two forms of the control gain perturbations.
(i) the additive form:
where
K
σ
∈
ΔK
σ
=
F
σ
Δ
K
σ
(
t
)
N
σ
(4)
(ii) the multiplicative form:
ΔK
σ
=
F
σ
Δ
K
σ
(
t
)
N
σ
K
σ
(5)
where
F
σ
,N
σ
,F
σ
and
N
σ
are the known constant matrices,
Δ
K
σ
(
t
)and
Δ
K
σ
(
t
)
denote time-varying uncertainties satisfying
I, Δ
K
σ
(
t
)
Δ
K
σ
(
t
)
Δ
K
σ
(
t
)
Δ
K
σ
(
t
)
≤
≤
I.
To conclude this section, we recall the following lemmas which will be used in
the proof of our main results.
Lemma 1 (Schur Complement).
For any given constant real symmetric ma-
trix
P
=
P
11
P
12
P
12
P
22
, the following three arguments are equivalent
(
i
)
P<
0;
(
ii
)
P
22
<
0
,P
11
−
P
12
P
−
1
22
P
12
<
0;
P
12
P
−
1
(
iii
)
P
11
<
0
,P
22
−
11
P
12
<
0
.
(6)
Lemma 2.
Let
U
,
V
be real matrices of appropriate dimensions. Then, for any
matrix
Q>
0
of appropriate dimension and any scalar
ε>
0
, such that
UV
+
V
T
U
T
ε
−
1
UQ
−
1
U
T
+
εV
T
QV .
≤
(7)
Search WWH ::
Custom Search