2 Abstract and Applied Analysis
approaches [13–15]andtheactiveapproaches[16–18]. e
passive FTC approach designs a xed controller that is able
to tolerate only a limited range of predetermined faults, and,
once implemented, it compensates for the anticipated faults
without any online fault identication. However, the passive
FTChasaverylimitedfaulttolerancecapabilityandisoen
designed to be conservative. e active FTC compensates
for the eects of component faults by synthesizing a new
controller online or by selecting a predesigned controller
[19]. e active FTC requires a fault detection and diagnosis
(FDD) mechanism to detect and identify the faults in real
time, and then the controller is recongured based on the
identied faults. Errors in fault detection may cause the
controller to make wrong decisions. In this paper, we present
a fault-tolerant control strategy which does not use any fault
detection and isolation mechanism to detect, separate, and
identify the actuator faults online. Although the fault-tolerant
control methods have been used in a class of mechanical
systems such as spacecra [20] and near-space-vehicle [21],
few works have focused on the fault-tolerant control of
networkedEuler-Lagrangesystems.
Compared with the aforementioned works [2–12], this
paper addresses the distributed fault-tolerant tracking prob-
lem for networked unknown Lagrange systems on digraphs.
e distributed tracking control problem on digraph is more
challenging than that on undirected or balanced graphs.
We assume that the inertia matrix, the Coriolis/centrifugal
matrix, the friction term, and the gravity term are all
unknown for all Lagrange systems. A Lyapunov technique
is utilized to design the distributed tracking controllers.
By applying the universal approximation ability of fuzzy
logic systems [22–24], an adaptive coordinated fuzzy con-
troller is developed rstly when the actuators are fault-
free. en, an active fault-tolerant controller is developed,
which can compensate for both the actuator bias faults and
thelossofactuatoreectiveness.eproposedcontrollaw
does not require any FDD mechanism to detect the faults,
which reduces the computation burden and decreases the
response time of the controller. e analysis shows that
if the designed nominal controller can ensure the stability
of the fault-free distributed system, the proposed fault-
tolerant controller guarantees the stability of the distributed
system in the presence of faults. Numerical simulation
results are given to show the eectiveness of the proposed
method.
To our best knowledge, the fault-tolerant coopera-
tive control of networked uncertain Lagrange systems on
digraphs had not been fully investigated and it is still a
challenging task. We present a solution to this problem in this
paper. e main contributions of this research are described
as follows.
(1) A novel nite-time observer based cooperative con-
trol method is proposed for the networked nonlinear
Lagrange systems. e dynamic leader and all the fol-
lowers have unknown nonidentical dynamics. Com-
pared with the results in [2–12],theproposedmethod
is suitable for the general directed communication
topology.
(2) A robust fault-tolerant cooperative control scheme
is presented to achieve distributed tracking control
eveniftheactuatorbiasfaultandthelossofactuator
eectiveness coexist. In addition to the nominal
controller, an auxiliary control input is designed to
compensatefortheactuatorfaults.Itisprovedthatthe
proposed approach guarantees the convergence based
on Lyapunov stability theory.
e remainder of this paper is organized as follows.
Section 2 introduces the problem formulation and some
preliminary results. Section 3 provides the proposed nominal
controller and the robust fault-tolerant controller. e simu-
lations are shown in Section 4. Section 5 concludes this paper.
2. Problem Formulation and Preliminaries
2.1. Graph eory. Let G =(V,,A)be a directed graph of
order ,whereV =[V
1
,...,V
𝑛
]is the set of nodes, ⊆V ×V
is the set of edges, and A =[
𝑖𝑗
]is called the adjacency matrix
with weights
𝑖𝑗
>0if (V
𝑗
,V
𝑖
)∈and
𝑖𝑗
=0otherwise. We
assume that the graph is simple; that is, (V
𝑖
,V
𝑖
)∉∀,with
no self-loops. us,
𝑖𝑖
=0. Dene the in-degree of node V
𝑖
as the th row sum of ;thatis,
𝑖𝑛
(V
𝑖
)=
∑
𝑛
𝑗=1
𝑖𝑗
.Denethe
diagonal in-degree matrix =diag{
𝑖𝑛
(V
𝑖
)}and the graph
Laplacian matrix =−. e set of neighbors of a node
V
𝑖
is
𝑖
={V
𝑗
:(V
𝑗
,V
𝑖
)∈}, which is the set of nodes with
edges incoming to V
𝑖
. A directed path is a sequence of nodes
V
1
,V
2
,...,V
𝑟
such that (V
𝑖+1
,V
𝑖
)∈, ∈{1,2,...,−1}.A
directed tree is a directed graph, where every node, except
the root, has exactly one parent. A spanning tree of a digraph
is a directed tree that connects all the nodes of the graph.
2.2. Fuzzy Logic Systems. A fuzzy logic system (FLS) consists
of four parts: the knowledge base, the fuzzier, the fuzzy
inference engine, and the defuzzier. e knowledge base
for FLS comprises a collection of fuzzy if-then rules of the
following form:
𝑙
:If
1
is
𝑙
1
,
2
is
𝑙
2
...,
𝑛
is
𝑙
𝑛
,
en is
𝑙
, =1,2,...,,
(1)
where
=[
1
,
2
,...,
𝑛
]
𝑇
∈⊂
𝑛
and denote
the FLS input and output, respectively.
𝑙
𝑖
, =1,2,...,
and = 1,2,...,, are fuzzy sets and
𝑙
is the fuzzy
singletonfortheoutputintheth rule. Fuzzy sets
𝑙
𝑖
and
𝑙
are, respectively, associated with the membership functions
𝐴
𝑙
𝑖
(
𝑖
)=exp(−((
𝑖
−
𝑙
𝑖
)/
𝑙
𝑖
)
2
)and
𝐵
𝑙
(
𝑙
)=1,where
𝑙
𝑖
isthecenterofthereceptiveeldand
𝑙
𝑖
denotes the
width of the Gaussian function. is the rules number. By
applying singleton function, center average defuzzication,
and product inference, the FLS can be expressed as
(
)
=
∑
𝑀
𝑙=1
𝑙
∏
𝑛
𝑖=1
𝐴
𝑙
𝑖
𝑖
∑
𝑀
𝑙=1
∏
𝑛
𝑖=1
𝐴
𝑙
𝑖
𝑖
, (2)