Systems & Control Letters 96 (2016) 132–140
Contents lists available at ScienceDirect
Systems & Control Letters
journal homepage: www.elsevier.com/locate/sysconle
Fault detection of a sandwich system with dead-zone based on robust
observer
Zupeng Zhou
a,∗
, Yonghong Tan
b
, Peng Shi
c
a
School of Mechanical and Electrical Engineering, Guilin University of Electronic Technology, Guilin, China
b
College of Information, Mechanical and Electrical Engineering, Shanghai Normal University, Shanghai 200234, China
c
School of Electrical and Electronic Engineering, University of Adelaide, Adelaide, SA 5005, Australia
a r t i c l e i n f o
Article history:
Received 18 December 2014
Received in revised form
3 August 2016
Accepted 16 August 2016
Keywords:
Robust observer
Dead-zone
Sandwich system
General disturbance
Fault detection
a b s t r a c t
Non-smooth sandwich systems with dead-zone widely exist in the real engineering applications. For
accurately detecting its faults, a novel robust observer has been proposed in this paper. With the
consideration of the model uncertainties, disturbances, and switching error which specially belongs to the
system, the so-called general disturbance is defined. After that, the conventional dynamic robust observer
design method can be applied to the system. Then, for saving the computing time and effectively reducing
the effect of the disturbances to the residual, the main frequencies of the disturbances have been identified
by the spectrum analysis of the residual created by the conventional observer and the zeros assignment
methodology has been applied to get one feedback matrix of the robust observer. Finally, the rest of the
feedback matrices of the robust observer can be obtained by solving an optimal problem with H
∞,F
/H
−,F
as the minimizing object. For verifying the effectiveness of this novel robust observer, the comparison
between the proposed robust non-smooth scheme and the conventional method has been made. The
final results show that the proposed robust fault detection approach can detect the actuator and sensor
faults of the system more accurately and timely with a lower missing and false alarm rates comparing
with the conventional one.
© 2016 Elsevier B.V. All rights reserved.
1. Introduction
A dead zone is a non-smooth and nonlinear character which
widely exists in all kinds of motors, mechanical transition systems,
hydraulic systems, and mechatronic systems [1]. Dead zone usually
does not exist independently. On the contrary, it usually connects
with other parts. For instance, in a DC motor servo system, the
linear parts of the power amplifier and the DC motor can be
regarded as the front linear subsystem while the load of the motor
can be regarded as the back linear subsystem. The dead zone of
the DC motor embedded between the two dynamic linear parts
and this structure can be regarded as sandwich systems with dead
zone. In the industry field, many systems can be described as
sandwich systems with dead zone such as a hydraulic actuator of
aircraft lift [2], position stage driven by a DC motor, and a hydraulic
actuator controlled by a pilot valve etc. [3].
In engineering practice, the dead zone does not exist indepen-
dently and it usually connects with other conventional parts. If the
∗
Corresponding author.
E-mail address: zhouzupeng@guet.edu.cn (Z. Zhou).
dead zone nonlinearity is sandwiched into two linear dynamic sub-
systems, this system can be defined as the sandwich system with
dead zone. In real application, the model uncertainties and distur-
bances always exist and how to design a robust fault prediction
observer to restrain the effect of the model uncertainties and
disturbances are of crucial importance [4,5]. A Luenberger-type
switching observer is designed for a class of hybrid linear sys-
tems [6,7], while an observer is proposed for a class of piecewise
affine systems, respectively in [8,9].
In [10], a novel fault detection and identification (FDI) scheme is
presented for a class of nonlinear systems with model uncertainty.
The heart of this approach is an on-line approximator, referred to
as fault tracking approximator (FTA). In [11], a new sensor fault
detection, isolation, and identification (FDII) strategy is proposed
using the multiple-model (MM) approach. The scheme is based
on multiple hybrid Kalman filters (MHKFs), which represents an
integration of a nonlinear mathematical model of the system with
a number of piecewise linear (PWL) models. The work in [12]
investigates a fault detection and accommodation (FDA) problem
of saturated actuators for trajectory tracking of underactuated
surface vessels (USVs) in the presence of nonlinear uncertainties.
http://dx.doi.org/10.1016/j.sysconle.2016.08.004
0167-6911/© 2016 Elsevier B.V. All rights reserved.