Mathematical Problems in Engineering 3
To facilitate the control design, the following lemmas are
applied.
Lemma 3 (Young’s inequality). For ∀(,) ∈ R
2
, ≤
(
𝑝
/)||
𝑝
+(1/
𝑞
)||
𝑞
holds, where >0, , > 1,and
1/+1/=1.
Lemma 4 (see [26]). For real variables ≥0and >0;then,
≤+(/)
𝑚
((−1)/)
𝑚−1
,where≥1is a real number.
Lemma 5 (see [31]). Let and be real variables; then,
for any real numbers ,, > 0 and continuous function
(⋅) ≥ 0,onehas(⋅)
𝑚
𝑛
≤ ||
𝑚+𝑛
+(/(+))((+
)/)
−𝑚/𝑛
(⋅)
(𝑚+𝑛)/𝑛
−𝑚/𝑛
||
𝑚+𝑛
.
Lemma 6 (see [31]). For ,∈R,where≥1is a constant,
the following inequalities hold: |+|
𝑝
≤2
𝑝−1
|
𝑝
+
𝑝
|and
(||+||)
1/𝑝
≤||
1/𝑝
+||
1/𝑝
.
In the sequel, radial basis function neural network (RBF
NN) is to be applied to estimate the unknown nonlinear
functions. By choosing suciently large node number, for any
continuous function ()over a compact set
𝑥
⊂R
𝑞
,there
is a RBF NN
∗
𝑇
()such that, for an ideal level of accuracy
(0<<1),
(
)
=
∗
𝑇
(
)
+
(
)
,
|
(
)
|
≤,
(5)
where ()is the approximation error and ()=[
1
(),...,
𝑁
()]
𝑇
is the known function vector with >1being the
RBF NN node number. e basis functions
𝑖
()(1≤≤)
are chosen as
𝑖
() = exp[−(−
𝑖
)
𝑇
(−
𝑖
)/
2
],whereis
the width of the function and
𝑖
=[
𝑖1
,...,
𝑖𝑛
]
𝑇
is the center
of the receptive eld.
∗
is the ideal constant weight vector
with the form
∗
=arg min
𝑊∈R
𝑁
{sup
𝑥∈𝑆
𝑥
|()−
𝑇
()|},
where arg min is the value of variable when the objective
function sup
𝑥∈𝑆
𝑥
|() −
𝑇
()| is minimum with =
[
1
,...,
𝑁
]
𝑇
being the weight vector.
3. Design of State-Feedback Controller
e objective of this paper is to design an adaptive NN state-
feedback controller for system (1) under weaker conditions
such that the closed-loop system is SGUUB. To achieve the
above objective, we need the following assumptions.
Assumption 7. e time-varying delays
𝑖
(), =1,...,,in
system (1) satisfy 0≤
𝑖
()≤
𝑖
and
𝑖
()≤
𝑖
<1for positive
constants
𝑖
and
𝑖
.
Assumption 8. () =0is unknown sign and takes value in
the unknown closed interval :=[
−
,
+
]with 0∉.
Assumption 9. Nonlinear functions
𝑖
and
𝑖
satisfy
𝑖
,
(
)
,−
𝑖
(
)
≤
𝑖1
+
𝑖2
−
𝑖
(
)
,
𝑖
,
(
)
,−
𝑖
(
)
≤
𝑖1
+
𝑖2
−
𝑖
(
)
,
(6)
for = 1,...,,where
𝑖1
,
𝑖2
,
𝑖1
,and
𝑖2
are positive
functions with
𝑖𝑗
()=
𝑖𝑗
()and
𝑖𝑗
()=
𝑖𝑗
()for =1,2.
Remark 10. We emphasize two points. (i) To the best of our
knowledge, only [33] consider the unknown control direc-
tions for stochastic high-order time-delay systems. However,
in [33], the unknown control directions are of known signs
and are bounded by positive constants. We allow the sign
of ()to be unknown and remove the bounds limitations
in Assumption 8. (ii) Motivated by [14, 15] for stochastic
time-delay systems, the restrictions on
𝑖
and
𝑖
are greatly
relaxed in Assumption 9 compared with the existing results
in [29, 30, 32–34].
To simplify the design process, dene
=max
𝑖
∗
𝑖
2
,=1,...,,
(7)
where
1
,...,
𝑛
are the number of RBF NN nodes and
∗
1
,...,
∗
𝑛
are the ideal constant weight vectors. Before
the design procedure, introduce the following coordinate
transformation:
1
=
1
,
𝑖
=
𝑖
−
𝑖𝑓
, =2,...,,
(8)
where
2𝑓
,...,
𝑛𝑓
are the outputs of the rst-order lter with
virtual control laws
2
,...,
𝑛
being inputs. Now, we give the
backstepping design procedure by utilizing the technique of
DSC and RBF NN approximation approach.
Step 1. Choosing the 1st Lyapunov function candidate as
1
(
1
,
)=(
1
/(−
1
+4))
𝑝−𝑝
1
+4
1
+(1/2Γ)
and using (1),
(3), and (8), one has
L
1
≤
1
𝑝−𝑝
1
+3
1
𝑝
1
2
+
1
(
⋅
)
+
−
1
+3
2
1
𝑝−𝑝
1
+2
1
1
(
⋅
)
𝑇
1
(
⋅
)
−
1
Γ
,
(9)
where =max{
𝑖
, = 1,...,},
1
, Γ>0are design
constants,
istheestimateof,and
=−
is the estimation
error of .
In the sequel, we estimate the terms of (9) by using
Lemmas 4-5 and Assumption 9 as
1
𝑝−𝑝
1
+3
1
1
(
⋅
)
≤
1
𝑝−𝑝
1
+3
1
11
+
12
−
1
(
)
≤
1
𝑝−𝑝
1
+4
1
11
+
11
𝑝−𝑝
1
+4
1
+
11
𝑝−𝑝
1
+4
12
−
1
(
)
≤
1
𝑝−𝑝
1
+4
1
11
+
11
𝑝−𝑝
1
+4
1
+
11
+
11
𝑝+3
12
−
1
(
)
,