Neural and neuro-fuzzy networks in fault diagnosis of dynamical systems

haremboingAI and Robotics

Oct 20, 2013 (3 years and 10 months ago)

95 views

N
eural

and neuro
-
fuzzy
networks in fault diagnosis

of dynamical systems





Józef Korbicz

University of Zielona Góra
, Poland

Institute of Control and Computation Engineering

e
-
mail:

j.korbicz@issi.uz.zgora
.pl




The lecture starts with the discussion of
the
methodology of Fault Detection and Isolation
(FDI) for dynamic systems. Then recent model
-
based approaches to FDI


analytical
ones

and
those based on
soft computing



are surveyed. Taking into account
many

limitations of
analytical methods
,

the main attention is
focused

on the use of neural
and neuro
-
fuzzy
networks in FDI for solving specific tasks such as fault isolation
,

but mainly fault detection.
Two kind
s

of dynamic neural networks


the
M
ulti
L
ayer

P
erceptron (MLP) and the Group
Method of Data Handling (GMDH)


are discussed for
the purpose of
modelling
the
diagnosed systems. Irrespective of the neural networks used, there is always the problem of
neural model uncertainty, i.e.
,

the model

r
eality mi
smatch. Therefore, the
neural network
-
based
fault

diagnosis scheme should provide robustness to model uncertainty. It will be
shown how to determine the structure and parameters of the GMDH network as well as how
to estimate modelling uncertainty of the re
sulting neural model using a Bounded
-
Error
Approach (BEA). Such

an

approach gives
the
possibility

of formulating
an algorithm that
allows obtaining a neural network with relatively small modelling uncertainty. The
presentation describes how to develop an a
daptive threshold with the GMDH model using
some knowledge regarding its uncertainty
,

and
how to
increas
e the robustness of
GMDH
-
based fault diagnosis.

To illustrate
the

effectiveness of the GMDH network
and the Takagi

Sugeno neuro
-
fuzzy network,
in

fault
diagnosis
,

several powerful examples


a
sugar factory
value actuator (DAMADICS benchmark problem) and a laboratory
DC motor
system


are
presented.


Keywords:

fault detection, soft computing, neural networks, neuro
-
fuzzy networks,
applications