518 IEEE TRANSACTIONS ON FUZZY SYSTEMS, VOL. 21, NO. 3, JUNE 2013
at time t, and d
k
denotes the uncertain impulse input delay at
impulse instant t
k
. We assume that there exists a scalar d ≥ 0
such that d
k
∈ [0,d] for all k ∈ N.
From (6), the overall output of this fuzzy impulsive controller
is given by
u(t)=
r
i=1
h
i
(z(t))
∞
k=1
δ(t − t
k
)K
i
x((t − d
k
)
−
) (7)
where h
i
(z(t)) is the same as that of the ith rule of the fuzzy
system (5). Then, the closed-loop system of (5) and (7) is shown
as follows:
˙x(t)=
r
i=1
h
i
(z(t))
A
i
x(t)+A
di
x(t − τ (t))
+ B
i
K
i
∞
k=1
δ(t − t
k
)x((t − d
k
)
−
)
. (8)
Integrating (8) on both sides from t
k
− h to t
k
+ h yields
x(t
k
+ h) − x(t
k
− h)
=
t
k
+h
t
k
−h
r
i=1
h
i
(z(t))
A
i
x(t)+A
di
x(t − τ (t))
+B
i
K
i
∞
k=1
δ(t − t
k
)x((t − d
k
)
−
)
dt
where h>0 is sufficiently small. Let h → 0
+
; then, we can
obtain
Δx(t
k
)=
r
i=1
h
i
(z(t
k
))B
i
K
i
x((t
k
− d
k
)
−
)
where Δx(t)=x(t
+
) − x(t
−
), and x(t
+
) denotes the limit
from the right at time t. Here, we assume that x(t
+
)=x(t).
Therefore, we can represent the fuzzy system (8) in the fol-
lowing equivalent form:
⎧
⎪
⎨
⎪
⎩
˙x(t)=
r
i=1
h
i
(z(t)) (A
i
x(t)+A
di
x(t − τ (t))) ,t= t
k
Δx(t)=
r
i=1
h
i
(z(t))B
i
K
i
x((t −d
k
)
−
),t= t
k
,k∈ N.
(9)
We introduce the definition of exponential stability for system
(9).
Definition 1: For a given impulsive time sequence {t
k
},the
zero solution of T–S fuzzy impulsive system (9) is said to be
exponentially stable (ES) if there exists a scalar
0
> 0 such that,
for every >0, there is a positive scalar
1
such that φ≤
1
implies
|x(t, t
0
,φ)| <e
−
0
(t−t
0
)
, ∀ t ≥ t
0
. (10)
Definition 2: Let S denote a class of admissible impulsive
time sequences. If for any {t
k
}∈S, condition (10) is satisfied
with the same
1
,, and
0
, then we say that the zero solution
of system (9) is uniformly exponentially stable (UES) over S.
For illustration simplicity, system (9) is said to be ES (or UES)
if its zero solution is ES (or UES). For given positive scalars
σ
1
and σ
2
satisfying σ
1
≤ σ
2
, we use the notation S(σ
1
,σ
2
) to
denote the class of impulsive time sequences that satisfy σ
1
≤
t
k
− t
k−1
≤ σ
2
,k ∈ N. The notation S(0,σ
2
) denotes the class
of impulsive time sequences that satisfy ≤ t
k
− t
k−1
≤ σ
2
for some scalar ∈ (0,σ
2
],k ∈ N. In addition, the following
notation will be used in the next section:
I
1
=[I
n
0
n
], I
2
=[0
n
I
n
], A
i
=[A
i
A
di
]
κ
1
= max
1≤i≤r
{|A
i
| + |A
di
|},κ
2
= max
1≤i≤r
{|B
i
K
i
|}.
III. M
AIN RESULTS
A. Exponential Stability of Takagi–Sugeno Fuzzy
Impulsive System
In this section, we first consider the exponential stability of
T–S fuzzy impulsive system (9). For this purpose, we need the
following lemma to give an estimate of the solutions of system
(9) on the interval [t
0
− ¯τ,t
0
+ d].
Lemma 1: Consider system (9). Assume that {t
k
}∈
S(σ
1
, ∞) and ( − 1)σ
1
<d≤ σ
1
for some positive integer
.Let
0
=(1+κ
2
)
e
κ
1
d
. Then, for any φ ∈ PC([−¯τ,0])
|x(t, t
0
,φ)|≤
0
φ,t∈ [t
0
− ¯τ,t
0
+ d].
Proof: Set x(t)=x(t, t
0
,φ). Since ( − 1)σ
1
<d≤ σ
1
,the
maximum number of impulse times on the interval (t
0
,t
0
+ d]
is . We assume that the impulsive instants on (t
0
,t
0
+ d] are
t
i
,i= 1, 2,...,
0
,
0
≤ .
For t ∈ [t
0
,t
1
) and θ ∈ [−¯τ,0], when t + θ ≤ t
0
,wehave
|x(t + θ)|≤φ. When t + θ>t
0
, we get
|x(t + θ)|
=
x(t
0
)+
t+ θ
t
0
˙x(s)ds
=
x(t
0
)+
t+ θ
t
0
r
i=1
h
i
(z(s))[A
i
x(s)+A
di
x(s − τ (s))]ds
≤φ +
t
t
0
κ
1
x
s
ds.
It follows that
x
t
≤φ +
t
t
0
κ
1
x
s
ds, t ∈ [t
0
,t
1
).
Applying the Gronwall inequality gives
x
t
≤φe
κ
1
(t−t
0
)
,t∈ [t
0
,t
1
).
Moreover
|x(t
1
)| = |x(t
−
1
)+Δx(t
1
)|
=
x(t
−
1
)+
r
i=1
h
i
(z(t
1
))B
i
K
i
x((t
1
− d
1
)
−
)
≤ (1 + κ
2
)φe
κ
1
(t
1
−t
0
)
.