Fault Detection for Markovian Jump Systems with Random Packet
Dropout and Delay using Delta Operator
Jianxun Zhou and Duanjin Zhang
Abstract—This paper addresses the problem of fault
detection for networked control systems with random packet
dropout and delay via delta operator. The packet dropout and
delay which in the networked control systems are both random,
and described by Markov chain process. So the system model is
converted to Markovian jump systems. A fault detection filter
based on the observer is proposed to generate the residual
signal, thus making the residual systems to be asymptotically
stable and satisfy the H-infinity performance in delta domain.
The parameters of the fault detection filter are obtained by
linear matrix inequality(LMI). The numerical example is
provided to show the effectiveness of the proposed method.
I. I
NTRODUCTION
In recent years, fault detection (FD) has been an active
field of research due to an increasing demand for higher
performance, higher safety, and reliability standards.
Networked control systems (NCSs) are feedback control
systems with closed-loops based on real-time shared network,
which have several advantages such as its low cost, easy
operation and simpler maintenance [1]-[4]. But, due to the
complexity of the network environment, there must exist time
delay [5]-[6] and packet dropout [7]-[8] in NCSs. The
network induced issues will inevitably degrade performances
of NCSs and even lead to system alarm or even collapse.
Therefore, the fault detection based on networked control
systems with packet dropout and delay has been received lots
of consideration.
During past years, the study on fault detection of the
networked control systems has been investigated in many
previous studies. The problem of adaptively adjusted event-
triggering mechanism-based fault detection for a class of
discrete-time NCSs with applications to aircraft dynamics is
studied in [9]. An input delay approach to fault detection of
networked control systems with network-induced delay and
data dropout is proposed in [10]. A fault detection filter
design problem for networked control systems is investigated
in [11], using the Markovian jump systems with model
uncertainty by traditional shift operator. The packet dropout
in NCSs is described by Bernoulli in [12].
Most of existing researches focused on the discrete-time
systems are formulated by traditional shift operator. However,
the method based on traditional shift operator may bring the
problems of crowding poles located in the boundary of
stability circle when sampling is fast, and even lead systems
to lose stability. The delta operator is a unified formulation of
continuous-time models and discrete-time models, which is a
new discretization method [13] proposed by Middleton and
Goodwin in 1986. The delta operator model can make
discrete-time models approach to the corresponding
continuous-time models and avoid the ill-conditioned
problem caused by the traditional shift operator when the
sampling rate is high.
In the past decades, there have been rich achievements
concerned with delta operator in NCSs [14]-[15]. For
example, the problem of
∞
fault detection for delta operator
formulated systems with time-delay, data dropouts and
communication constraints is investigated in [16]. Focusing
on the networked control systems with packet dropout, the
problems of observer-based fault detection are addressed in
[17]. The packet dropouts of the NCSs are modeled as a
stochastic variable satisfying Bernoulli distribution and the
fixed short-delays are considered. However, quite few
researches are involved in fault detection of networked
control systems with random packet dropout and delay based
on delta operator, which is the motivation of this paper.
This paper proposes the method for fault detection
problem of networked control systems with random packet
dropout and delay based on delta-operator. The random delay
and random packet dropout are described by Markov chain
process, so the system model is transformed into Markovian
jump systems. A fault detection filter based on the observer
is presented to generate residual signal for fault detection, and
the residual system is asymptotically stable and satisfies the
∞
performance in delta domain. The parameters of the filter
are derived by LMI via constructing Lyapunov-Krasovskii
functional. The numerical example is used to verify the filter
performance.
II. P
ROBLEM
F
ORMULATION
Consider the following linear time invariant system:
() () () () (),
() (),
xt AxtButDftEdt
yt Cxt
=+++
=
(1)
where,
()
n
xt R∈ , ()
u
ut R∈ , ()
r
yt R∈ denote the system
state vector, control vector and output vector, respectively.
The
()
m
dt R∈
is the external disturbance which belongs to
2
[0, )l +∞
,
()
tR∈
is the fault signal.
,,, ,
BCDE
are the
constant matrices with the appropriate dimension.
Then the definition of delta operator is given as follows:
* This work was supported by the National Natural Science Foundatio
of China under Grant 61471323.
J. Zhou and D. Zhang are with the School of Information Engineering,
Zhengzhou University, Zhengzhou, 450001, China (corresponding author
D. Zhang, phone: +86-371-67781545; fax: +86-371-67781270; e-mail:
djzhang@zzu.edu.cn).
2017 11th Asian Control Conference (ASCC)
Gold Coast Convention Centre, Australia
December 17-20, 2017
978-1-5090-1572-6/17/$31.00 ©2017 IEEE 1912