Sie sind auf Seite 1von 6

Proceedings, 2nd IFAC Conference on

Proceedings, 2nd IFAC Conference


Modelling, Identification and Controlon
of Nonlinear Systems
Proceedings,
Modelling, 2nd IFAC Conference
Identification and Controlon
Available online
of Nonlinear at www.sciencedirect.com
Systems
Proceedings,Mexico,
Guadalajara,
Proceedings, 2nd IFAC
2nd IFAC Conference
June 20-22,
Conference on
2018
on
Modelling, Identification
Guadalajara, Mexico, June and Control
20-22, of Nonlinear Systems
2018
Modelling, Identification
Modelling, Identification and
and Control
Control of Nonlinear Systems
of Nonlinear Systems
Guadalajara, Mexico, June 20-22, 2018
Guadalajara,
Guadalajara, Mexico,
Mexico, June
June 20-22,
20-22, 2018
2018
ScienceDirect
IFAC PapersOnLine 51-13 (2018) 662–667
Simplified
Simplified Model
Model of
of a
a Three-Phase
Three-Phase Induction
Induction MotorMotor for
for Fault
Fault Diagnostic
Diagnostic Using
Using the
the
Simplified
Simplified Model
Model of
of a
a Three-Phase
Synchronous Reference
Three-Phase Induction
Frame
Induction DQ Motor
and
Motor for
Parity
for Fault Diagnostic
Equations
Fault Diagnostic Using the
Simplified Model of a Three-Phase
Synchronous ReferenceInduction
Frame DQ Motor for Fault
and Parity Diagnostic Using the
Equations Using the
Synchronous
Synchronous Reference
Reference Frame
Frame DQ and Parity Equations
Synchronous Reference
Edgar Chulines, Frame DQ DQ
Marco A. Rodríguez, Iván and
and Parity
Parity
Duran, Equations
Equations
Rafael Sánchez
Edgar Chulines, Marco A. Rodríguez, Iván Duran, Rafael Sánchez
Edgar
Edgar Chulines,Chulines, Marco A.
Marco A. Rodríguez,
Rodríguez, Iván Iván Duran,
Duran, Rafael
Rafael Sánchez
Sánchez
Universidad Autónoma Edgar Chulines,
del Carmen Marco A. Rodríguez,
UNACAR, Iván
Departamento Duran, Rafael
de Ingeniería Sánchez
Electrónica, Ciudad
Universidad Autónoma del Carmen UNACAR, Departamento de Ingeniería Electrónica, Ciudad
Universidad Autónoma del del
CarmenCarmen Campeche
UNACAR, México.
Departamento C.P 24155
Universidad
Universidad Autónoma Autónoma del deldel
Carmen
CarmenCarmen Campeche
UNACAR,
UNACAR, México. C.Pde
Departamento
Departamento de Ingeniería
de24155
Ingeniería Electrónica, Ciudad
Ingeniería Electrónica,
Electrónica, Ciudad Ciudad
del
E-mail: mrodriguez@pampano.unacar.mx,
del Carmen
Carmen Campeche
Campeche México. C.P 24155
chulines_mecatronica@hotmail.com
México. C.P 24155
del Carmen Campeche México.
E-mail: mrodriguez@pampano.unacar.mx, C.P 24155
chulines_mecatronica@hotmail.com
E-mail:
E-mail: mrodriguez@pampano.unacar.mx, chulines_mecatronica@hotmail.com
E-mail: mrodriguez@pampano.unacar.mx,
mrodriguez@pampano.unacar.mx, chulines_mecatronica@hotmail.com
chulines_mecatronica@hotmail.com
Abstract: This work is a simplified analysis of a nonlinear model for a three-phase induction motor in
Abstract: This work is a simplified analysis of a nonlinear model for a three-phase induction motor in
steady
Abstract: stateThisbehaviorwork toisdiagnose
aa simplifiedof additive
analysis andofparametric
aa nonlinear faults.
model Multiple
for faults are also
aa three-phase considered
induction motor using
in
steady
Abstract:
Abstract: stateThisbehavior
This work
work tois
is diagnose
a of additive
simplified
simplified analysis
analysis andof
ofparametric
a nonlinear
nonlinear faults.
model
model Multiple
for
for a faults are also
three-phase
three-phase considered
induction
induction motor
motor using
in
in
the
steady paritystate equation
behavior approach
to diagnose and some
of logical
additive assumptions.
and parametric The
faults. simplified
Multiple analysis
faults are of the
also nonlinear
considered model
using
the
steady
steady paritystate
state equation
behavior
behavior approach
to diagnose
to diagnose and some
of
of logical
additive assumptions.
and parametric The
faults. simplified
Multiple analysis
faults are of the
also nonlinear
considered model
using
in aparity
the synchronousequation reference
approach frame
and isadditive
some used andand
logical parametric
matched
assumptions. withfaults.
the linear
The Multiple
simplifiedmodel faults areofmotor,
of d.c.
analysis also
the considered
then the model
nonlinear using
parity
in
theaparity
the synchronous
parity equationreference
equation approach
approach frame
and the
and is used
some
some and matched
logical
logical assumptions.
assumptions. with Thethe linear
The simplified
simplifiedmodel of d.c.of
analysis
analysis ofmotor,
thethe
the then the model
nonlinear
nonlinear parity
model
space
in a is determined,
synchronous and finally,
reference frame is residual
used and equations
matched are
with obtained.
the linearThe performance
model of d.c. of
motor, fault the
then detection
space
in
system
is determined,
in aa synchronous
synchronous
validatedreference
isdetermined,
and
reference
byand using
finally,
frame
frame theis
the Matlab
residual
is used
used equations
and matched
and
software. matched with are obtained.
with thethe linear The
linear model performance
model of of d.c. of
d.c. motor, the
motor, then fault
then the the parity
detection
parity
parity
space
system
space is
is isdetermined,
validated byand usingfinally, the
the Matlab
finally, the residual
residual equations
software.
equations are
are obtained.
obtained. The
The performance
performance of
of the
the fault
fault detection
detection
space
system is determined, and finally, the residual equations are obtained. The performance of the fault detection
© 2018,is
Keywords:
system
Keywords:
system is validated
isIFAC
Fault
validated
Fault
validated
by
(International
diagnosis,
by usinginduction
by using
diagnosis,
the
the Matlab
Federation
the Matlab
usinginduction Matlab ofsoftware.
motor,Automatic
software.
motor,
software. Control) Hosting
parity equation,
parity equation,
modelling,
modelling, by Elsevier
residues,
residues, Ltd. All rights.. reserved.
insulation
insulation
Keywords:
Keywords: Fault diagnosis,
diagnosis, induction motor, motor, parity equation,equation, modelling, residues, residues, insulation..
Keywords: Fault Fault diagnosis, induction induction motor, parity parity equation, modelling,
modelling, residues, insulation insulation.
1. INTRODUCTION  the Human-Machine Interface (HMI) is commonly expensive.
1. INTRODUCTION  the Human-Machine Interface (HMI) is commonly expensive.
1. INTRODUCTION Although
the only the early
Human-Machine detection
Interface time is not very necessary
The Induction motor (IM)
The Induction motor (IM)
has been
1. INTRODUCTION
1. INTRODUCTION widely used in
has been widely used in the industry the
the industry Although
the only
Human-Machine
Human-Machine
for incipient
the
failures. Interface (HMI)
early detection
Interface (HMI)
(HMI) time is
is commonly
is is not veryexpensive.
commonly
commonly necessary
expensive.
expensive.
due to its simple construction, reliability, ruggedness, low cost, Although
for incipient
Although only only the
failures.
only the early
the early detection
early detection
detection time time
time is is not
is not very
very necessary
not very necessary
The
due
The Induction
to its
Inductionsimple motor (IM)
construction,
motor (IM) has
has been widely
reliability,
been widely used
ruggedness,
used in
in the
the industry
low cost,
industry Although necessary
The
and Induction
many motor
applications. (IM)Additionally,
has reliability,
been widely used in the
inruggedness,
contrast industry
tolowthecost,
d.c. Also, for
for incipient
an Adaptive
incipient failures.
failures.Neuronal Network (ANN) online technique
due
and
due to its
many
to its simple construction,
applications. Additionally, inruggedness,
contrast tolow thecost, for incipient failures.
d.c. Also, an Adaptive Neuronal Network (ANN) online technique
due
and to
motor,many it simple
its simple
can beconstruction,
construction,
used in volatile
applications.
reliability,
reliability, ruggedness,
or aggressive low
environment cost, to only detect bearings fault has been used, Nandi et al. (2005,
motor,
and
and many
many it can be used inAdditionally,
applications.
applications. volatile or aggressive
Additionally,
Additionally,
in
in
in
contrast
contrast
contrast
to
to
to
the
environment
the
the
d.c.
d.c. to
d.c.
Also,
onlyan
Also,
Also, an
Adaptive
andetect
Adaptive
Adaptive
Neuronal
bearings
Neuronal
Neuronalfault has Network
Network
Network
(ANN)
been used,
(ANN)
(ANN)
online
Nandi
online
onlineet technique
al. (2005,
technique
technique
since there
motor, it are be
can no used
problems in with corrosion
volatile or and spark,
aggressive although
environment with
to this
only technique
detect bearings seven faults
fault has has
been beenused,detected
Nandi but
et in offline
since
motor,
motor, there
it can
it
the reliability
are
can be no
be problems
used tends
used
degree in with
in volatile
volatile corrosion
to decreaseor and
or aggressive
aggressive
when
spark, although
environment
environment
the operating
with
to
to
way,
this
onlyKolla
only technique
detectetbearings
detect bearings seven
al. (2000).
faults
faultHowever,
fault has
has been
has been
been the used,
used,detected
two Nandi
Nandi et al.
but
et
previous al.
al.
(2005,
in offline
(2005,
(2005,
works
since
the there
sincereliability
there are are
are no
no problems
degree
problems with
tendswith
with corrosion
to decrease
corrosion and
when spark,
and spark, although
the operating
spark, although with
way, this
with thisKolla
this technique seven
et al. (2000).
technique seven faults has
However,
faults been
has been
been detected
thedetected
two previous
detected but
but in offline
works
in offline
offline
since
the there
conditions
reliability are no problems
extreme,
degree tends whichto corrosion
is
decreasemagnifiedand
when when
the although
the
operating IM with
result
way, in
Kollatechnique
a et high
al. seven
cost
(2000). faults
in has
storage,
However, the recursive
two but in
operations,
previous works
conditions
the reliability are degree
extreme, which is magnified when the IM result in aet et high cost in storage,the therecursive operations,
the reliability
operates
conditions a critical
are degree load,tends
extreme, tends
where
which
to
toa decrease
decrease
failure
is
when
when
could
magnified
the
the
bewhena riskoperating
operating
to staff
the IM
way,
way, Kolla
Kolla
computational
result in a
al.
al. (2000).
(2000).
complexity
high cost
However,
However,
andstorage,
in response two
time
recursive
previous
two delay.
previous works
works
Although
operations,
operates
conditions
conditions a critical
are
are load,
extreme,
extreme, where
which
which a failure
is
is could
magnified
magnified be a
when
when risk to
the
the staff
IM
IM computational
result
result in
in a
a complexity
high
high cost
cost and
in
in response
storage,
storage, time
recursive
recursive delay. Although
operations,
operations,
safety,
operates the environment and economy. for critical applications the first and second disadvantages are
safety,
operates
operates theaaa critical
environment
critical
critical
load,
load,
load,
where
where
aa failure
and economy.
where a failure
failure
could
could
could
be
be
be
aa risk
a risk
risk
to
to
to
staff
staff
staff
computational
for critical
computational
computational
complexity
applications
complexity
complexity theand
and
and
response
firstfault
and tolerant
response
response second time
time
time
delay.
delay.
delay.
Although
disadvantages
Although
Although are
safety, the environment and economy. not important.
for critical In addition,
applications the for
first and second systems
disadvantages the third
are
A program
safety, the of maintenance
the environment and and specific attention must be not
and economy. important. In addition, the for
firstfault
and tolerant systems the third
safety,
A program environment
of maintenance and specific attention must be for
economy. for critical
critical
drawback
not important.
applications
applications
cannot In be the
despised
addition,
first
for
and
fault
second
second disadvantages
tolerant disadvantages
systems the
are
are
third
given
A to
program be able
of to have
maintenance a correctand operation
specific of the
attention motors.must This
be drawback
not
not important.
important.cannot In
In be despised
addition,
addition, for
for fault
fault tolerant
tolerant systems
systems the
the third
third
given
A to
program be able
of to have
maintenance a correctand operation
specific of the
attention motors.must This
be
be drawback
A
canprogram
givenbe done of maintenance
through permanent andmonitoring
specific of attention
to detect mustThis
incipient The stator cannot
drawback winding
cannot be despised
fault
befault is one of the electrical faults that has
despised
can
given
givenbeto
to
to
be
done
be
be
able
through
able
able
to
to
to
have
have
have
aa correct
permanent
a correct
correct
operation
monitoring
operation
operation tothe
of
of detect
the
the
motors.
incipient
motors.
motors. This
This
drawback
The stator cannot
winding be despised
is one of the electrical faults that has
faults
can be and
done prevent
through them. Generally,
permanent the
monitoring most frequent faults in reported short detection delay time using model-based
faults
can
can be
be and
done
done prevent
through
through them. Generally,
permanent
permanent the mostto
monitoring
monitoring to
to
detect
frequent
detect
detect
incipient
faults in reported
incipient
incipient
The
The
The
stator
stator
stator
winding
short
winding
winding
fault
detection
fault
fault
is
is
is
one
delay
one
one
of
of
of
the
time
the
the
electrical
using faults
electrical
electrical faults
faults
that
model-based
that
that
has
has
has
IM are
faults and mechanical
prevent and
them. these
Generally, are related
the most to the
frequent electrical
faults in methods,
reported like
short the Winding
detection delay Function
time Approach
using (WFA)
model-based
IM are
faults
faults and
and mechanical
prevent
prevent and
them.
them. these
Generally,
Generally, are related
the
the most
most to the
frequent
frequent electrical
faults
faults in
in methods,
reported
reported like
short
short the Winding
detection
detection delay
delay Function
time
time Approach
using
using (WFA)
model-based
model-based
operation
IM are of the
mechanical motor andsuch as local overheating and inter-turn Bianchini et al. (2011). Another method based on the model is
operation
IM
IM are
are
short-circuit of the
mechanical
mechanical motor
stator and
andsuchthese
winding these
these
are
as Pineda-Sanchez
local
are
are
related
overheating
related
related
to
to
to et
the
and
the
the al.
electrical
inter-turn
electrical
electrical
(2011).
methods,
Bianchini
methods,
methods,
the use of
like
et
like
like
parity
the
al. (2011).
the
the
Winding
Another
Winding
Winding
equations, which
Function
method
Function
Function
may be
Approach
based on the to
Approach
Approach
adequate
(WFA)
model
(WFA)
(WFA)
detectis
operation
short-circuit
operation of
of the
the motor
stator
motor such
such as
winding asin local
local overheating
Pineda-Sanchez
overheating etand
andal.inter-turn
(2011). the
inter-turn Bianchini of et
use faults,
Bianchini al.
parity
et (2011).
al.but equations,
(2011). Another
Another which method based
maygeneral
method be
based on
on the
adequate the tomodel
detect
model is
is
operation
These faults
short-circuit of thecausemotor
stator such
changes
winding as localtheoverheating
basic parameters
Pineda-Sanchez etand al.inter-turn
of the Bianchini
(2011). several
the use of et al.
parity (2011). Another
the residuals
equations, which method
of the
may based
be on the
parity
adequate model
equation
to detectis
These faults
short-circuit
short-circuit cause
stator
stator changes
winding
winding in the basic
Pineda-Sanchez
Pineda-Sanchez parameters
et
et al.
al. of the
(2011).
(2011). several
the
the use
use faults,
of
of parity
paritybut the residuals
equations,
equations, which
which of the
may
may general
be
be parity
adequate
adequate equation
to
to detect
detect
motor faults
These such as cause resistances
changes and inductances;
in and can be severalmust befaults,
calculated and an accurate mathematic model of the
motor
These such as
These faults
faults cause
cause resistances
changes
changes andthe
in
in the
basic
basic parameters
theinductances;
basic parameters
parameters and can of the
be must
of and
of the
the several
several bemust
calculated
faults,
faults,
but
but the
andresiduals
the anChan
residualsaccurateof
of the
the general
mathematic
general For parity
model
parity equation
of the
equation
diagnosed
motor such in aas timely manner by
resistances and determining
inductances; the type,
and size
can be system
must be bebut
calculated the
obtained
and residuals
an accurateof the
et al. general
(2006).
mathematic parity
thisequation
model reason,
of the
diagnosed
motor
motor such
such in aasastimely manner by
resistances
resistances and
and determining
inductances;
inductances; the type,
and size
and can and
can be system
be must
must bemust
be be obtained
calculated
calculated and
and anChan
an accurate
accurateet al.mathematic
(2006). Formodel
mathematic this reason,
model of
of the
the
location
diagnosed ofinthe a fault
timely asmanner
well asby itsdetermining
time of detection.
the type, size and this
systemhas mainly
must be been
obtainedapplied Chan to linear
et al. systems,
(2006). where
For this accurate
reason,
location
diagnosed
diagnosed ofin the
in fault asmanner
aa timely
timely well asby
manner byitsdetermining
time of detection.
determining the
the type,
type, size
size and
and this
system
system
modelshas must
mainly
must
are be
be
more been
obtained
obtainedapplied
readily Chan
Chan to linear
available. et
et al.
al. systems,
(2006).
(2006).
However, where
For
For this
this
since accurate
reason,
reason,
accurate
location
The
location of
mechanical
of the fault as well as
vibration its
analysistime of is detection.
the most popular models this has mainly
are more been applied
readily to linear
available. systems,
However, where
since accurate
location
The of the
mechanicalthe fault as
as well
faultvibrationwell as its
its time
asanalysis time of is detection.
of the most popular this
detection. this
models
models
has mainly
has for
mainly
are more
been
been
nonlinear applied
applied
readilysystems to linear
to are
linear
available. more systems,
systems,
difficult
However,
where
where accurate
accurate
to obtain
since accuratein
technique
The to detectvibration
mechanical faults in the analysisIM, Bianchini
is the et al. popular
most (2011), models models for
are
are nonlinear
more
more readily
readilysystems are
available.
available. more difficult
However,
However, to
since
since obtain
accurate
accuratein
technique
The ismechanical to
mechanical detect faults in the
vibration magnitudes IM,
analysis is Bianchini
is and
thethemost et
most al. (2011),
popular practice,
models forin addition to the fact that the parity equation is
The
this
techniquedue to to detect vibration
the significant
faults in analysis
the IM, the
Bianchini etimmunity
al. popularto practice,
(2011), models
models
sensitive for
forin nonlinear
to addition
nonlinear
nonlinear
noise and
systems
to
systems
model
are
are
more
the uncertainties,
systems fact
are that
more
more thedifficult
parity
difficult
difficult
the
to
to
to
parity
obtain
equation
obtain
obtain
equation
in
is
in
in
this is
techniquedue to
to the
detect significant
faults in magnitudes
the IM, and
Bianchini the etimmunity
al. to
(2011), practice, in addition
technique
external
this is due to
phenomena
to detect
the faults
like the
significant inelectromagnetic
the IM, Bianchini
magnitudes and the etimmunity
interferenceal. (2011),
over
to sensitive
practice,
practice, to noise
ineasily
in addition
additionand to model
to
to
the
the
the
fact
fact
fact
that
that the
uncertainties,
that the
the
parity
the
parity
parity
equation
parity equation
equation
equation
is
is
is
external
this
this is
is due
duephenomena
to
to the
the like the electromagnetic
significant
significant magnitudes
magnitudes and
and interference
the
the immunity
immunity over
to
to cannot beto
sensitive noise applied
and model to uncertainties,
nonlinear systems,
the Chan
parity et al.
equation
sensors
external that
phenomena commonlylike the are accelerometers;
electromagnetic interference however,
over cannot
sensitive
sensitive be to
to easily
noise
noise applied
and
and model
model to nonlinear
uncertainties,
uncertainties, systems,
the
the Chan
parity
parity et
equation
equation al.
sensors
external that commonly
phenomena like are accelerometers; however, (2006). be easily applied to nonlinear systems, Chan et al.
external
sensors phenomena
accelerometers that have
commonly verythe
like the electromagnetic
electromagnetic
limited
are operation
accelerometers;
interference
interference
ranges. over
over (2006).
however,
cannot
cannot be
cannot be easily
easily applied
applied to to nonlinear
nonlinear systems,systems, Chan Chan et et al.
al.
accelerometers
sensors
sensors that
that have have
commonly
commonly very limited
are operation ranges.
accelerometers;
are operation
accelerometers; however,
however, (2006). Generally, the diagnostic procedure is based on the heuristic
accelerometers
Another technique very
widely limited
used for the ranges.
detection of mechanical (2006).
(2006).
Generally, the
accelerometers
accelerometers
Another technique have
have very
very limited
widely used foroperation
limited operation
the detection ranges.
ranges. of mechanical knowledge of diagnostic
the processprocedure and observed is based on the heuristic
analytical heuristic
and electrical
Another technique faults in theused
widely IMfor is the
thedetection
phase current of spectral Generally,
mechanical knowledge
Generally,
Generally,
the
of diagnostic
the
the the processprocedure
diagnostic
diagnostic and observed
procedure
procedure
is
is
is
based
based
based
on
on the
analytical
on the
the
heuristic
heuristic
heuristic
and electrical
Another
Another technique
technique faults in
widely
widely theused
usedIM is
for
for the
the
the phase
detection
detection current
of
of spectral
mechanical
mechanical symptom,
knowledge commonly
of the represented
process and in a
observed signal called
analytical residual.
heuristicIn
analysis,
and Loparo
electrical et
faults al.
in (2000)
the IMand is Strangas
the phaseet al. (2008).
current Other
spectral symptom,
knowledge
knowledge commonly
of the
of the represented
process
process in
and knowledge
and a signal
observed analytical
observed called
analytical residual.
heuristic
heuristicIn
analysis,
and
and Loparo
electrical
electrical et
faults
faults al.
in (2000)
the
in(2000)
theonIMIMand is Strangas
the
isStrangas phase
themotor
phaseet al. (2008).
current
current Other
spectral
spectral these
symptom, last two
commonly cases, a priori
represented in a signal of fault
called symptom
residual. In
techniques
analysis, Loparo are et based
al. and the et phase
al. (2008). current
Other these
symptom,
symptom, last commonly
two
commonly cases, a large
prioridata
represented
represented knowledge
in
in aa signal
signal of faultresidual.
called
called symptom
residual. In
In
techniques
analysis,
analysis, Loparo
Loparo are et
et based
al. (2000)
al. (2000) on and the motor
Strangas et phase
al. (2008). current
Other causalities,
these last as
two well
cases,as, aa priori bank
knowledge is necessary,
of fault Isermann
symptom
transformations,
techniques are Mendes
based etonal.and the Strangas
(2006). motor et al.phase
Others (2008).
works Other causalities,
evaluate
current these
these last
last as well
two
two cases,
cases,as, aaalarge
priori
priori data bank is necessary,
knowledge
knowledge of
of fault
fault Isermann
symptom
symptom
transformations,
techniques
techniques are Mendes
are based
based etonal. (2006).
on the
the Cash Others phase
motor
motor works evaluate
phase current
current et al. (1995).
causalities, as However,
well as, in
a large the
data diagnosis
bank is by using
necessary, analytical
Isermann
the neutral terminal
transformations, Mendes voltage et of(2006).
al. IM, et al.
Others works(1998) or by et
evaluate al. (1995).
causalities,
causalities, as
as However,
well
well as, in the
as, aa values
large
large data
data diagnosis
bank
bank is by using
is necessary,
necessary, analytical
Isermann
Isermann
the neutral
transformations, terminal Mendes voltage et of
al. IM,
(2006). Cash et
Others al.
works(1998) or
evaluateby symptoms
et al. (1995). with limit
However, in the of measurement
diagnosis by using signals and
analytical
transformations,
calculating
the neutral system
terminal Mendes
impedances,
voltage et al.
of (2006).
Klima
IM, CashOthers works
et alet(2003).
al. (1998) evaluate
or by symptoms
et al. (1995). with limit
However, values
in the of measurement
diagnosis by using signals and
analytical
calculating
the neutral system
neutral terminal impedances,
terminal voltage voltage of of IM,Klima
IM, Cash et
Cash etal (2003).
et al.
al. (1998)
(1998) or or by
by change et al.
change
symptoms (1995).
detectionHowever,
with have
limit ain
values the
minimal diagnosis
of measurement by
mathematical using analytical
effort
signals and
and
the
calculating system impedances, Klima et al (2003). symptoms detection
with have
limit a allows
valuesminimalof mathematical
measurement effort
signals and
The general
calculating problem
system of all
impedances, these techniques
Klima et al is
al (2003).
(2003).that the delay symptoms
without
change data data with
bank,
detection limit
which values ofthemeasurement
simplicity andsignals and
reliability
calculating
The generalsystem problem impedances,
of all theseKlima et
techniques is that the delay without bank, have whichaaa allows
minimal the mathematical
simplicity andeffort and
reliability
time
The of detection
general problem and localization
of all these of the fault is
techniques is that
verythe and change
longdelay change
without
detection
detection
data bank,
have
have
which
minimal
minimal
allows the
mathematical
mathematical
simplicity and
effort
effort and
and
reliability
time
The
The of detection
general
general problem
problem and localization
of
of all
all these
these of the
techniques
techniques fault is
is very
that
that long
the
the and
delay
delay without
without data
data bank,
bank, which
which allows
allows the
the simplicity
simplicity and
and reliability
reliability
time
time ofof detection
detection and
and localization
localization of
of the
the fault
faultofis very
very long
isAutomatic and
longControl)
and Hosting by Elsevier Ltd. All rights reserved.
time of
2405-8963 detection
Proceedings, © 2018,
2nd IFAC and
IFAC localization
(International
Conference onof the
Federationfault is very long and 662
Proceedings,
Modelling, 2nd responsibility
under IFAC Conference
Peer reviewIdentification on
of International
and Control 662Control.
Federation of Automatic
of Nonlinear
Proceedings,
Modelling, 2nd IFAC Conference
Identification and Controlon
10.1016/j.ifacol.2018.07.356 of Nonlinear 662
Proceedings,
Systems
Proceedings, 2nd IFAC Conference
2nd IFAC Conference on 662
Modelling,
Systems Identification and Controlon of Nonlinear 662
Modelling, Identification
Guadalajara, Mexico, June and Control
20-22, 2018of Nonlinear
2018 IFAC MICNON Edgar Chulines et al. / IFAC PapersOnLine 51-13 (2018) 662–667 663
Guadalajara, Mexico, June 20-22, 2018

when a relative large change of their feature is obtained, From these equations relating the rotor in the axes dq of the
Isermann et al (1997). magnetic flow links are obtained using the following equations
in the stator voltages.

𝑖𝑖𝑞𝑞𝑞𝑞 = − 𝑖𝑖
𝐿𝐿𝑚𝑚
In this paper, a new approach is presented not only to detect
𝐿𝐿𝑟𝑟 𝑞𝑞𝑞𝑞
faults, but to isolate faults for the nonlinear model of IM, (3)

𝑖𝑖𝑑𝑑𝑑𝑑 = − 𝑖𝑖
𝛹𝛹𝑟𝑟 𝐿𝐿𝑚𝑚
matching it with the linear model of a d.c. motor considering

𝐿𝐿𝑟𝑟 𝐿𝐿𝑟𝑟 𝑑𝑑𝑑𝑑


the performance during the steady state behaviour. A set of (4)
simplified equations is obtained, that allow an easy
implementation using modest digital processors or HMIs Substituting the currents of the rotor (3) and (4) in (1) and (2)
device, such as Arduino or Labview, respectively. it obtained the following expressions

𝑉𝑉𝑞𝑞𝑞𝑞 = (𝑅𝑅𝑠𝑠 + 𝜎𝜎𝐿𝐿𝑠𝑠 𝜌𝜌)𝑖𝑖𝑞𝑞𝑞𝑞 + 𝜎𝜎𝜎𝜎𝑠𝑠 𝐿𝐿𝑠𝑠 𝑖𝑖𝑑𝑑𝑑𝑑 + 𝜔𝜔𝑠𝑠 𝛹𝛹𝑟𝑟


𝐿𝐿𝑚𝑚
𝐿𝐿𝑟𝑟
2. NONLINEAR MODELLING OF INDUCTION (5)

𝑉𝑉𝑑𝑑𝑑𝑑 = (𝑅𝑅𝑠𝑠 + 𝜎𝜎𝐿𝐿𝑠𝑠 𝜌𝜌)𝑖𝑖𝑑𝑑𝑑𝑑 − 𝜎𝜎𝜎𝜎𝑠𝑠 𝐿𝐿𝑠𝑠 𝑖𝑖𝑑𝑑𝑑𝑑 + 𝜌𝜌 𝛹𝛹𝑟𝑟


𝐿𝐿𝑚𝑚
MOTOR IN DQ REFERENCE FRAME
𝐿𝐿𝑟𝑟
The IM is usually connected to an inverter for speed control (6)

Where 𝜎𝜎 is the leakage coefficient and is obtained when the


applications. However, in the industry there are more non-
critical applications where only constant operation in steady
state is necessary. In this way, a steady state analysis for fault flow component produced by the stator current is constant in
diagnosis in IM synchronous is important. the stable state. So that the derivatives of the stator current in
the d axis in the synchronous frame are.
𝑖𝑖𝑓𝑓 = 𝑖𝑖𝑑𝑑𝑑𝑑 𝑝𝑝 · 𝑖𝑖𝑑𝑑𝑑𝑑 = 0
The starting point to perform the analysis of the model of the
IM in the synchronous reference frame is to initially deduce
the transfer functions in the subsystems such as the mechanical The torque component produced by the stator current is the
and electrical part. For the mechanical and electrical parts to current in the q axis in the synchronous frame.
𝑖𝑖 𝑇𝑇 = 𝑖𝑖𝑞𝑞𝑞𝑞
be coupled, there is a link between the current produced by the
torque and the induced magnetic force.
It is also known that the rotor flow is given by.
𝛹𝛹𝑟𝑟 = 𝐿𝐿𝑚𝑚 𝑖𝑖𝑓𝑓
This link is implicit within the block diagram where the total
current loop in the interior is independent of the mechanical
part in the transfer function, Krishnan et al. (2001).
Replacing these values in the voltage equation.
𝐿𝐿𝑚𝑚 2
𝑉𝑉𝑞𝑞𝑞𝑞 = (𝑅𝑅𝑠𝑠 + 𝐿𝐿𝑎𝑎 𝜌𝜌)𝑖𝑖 𝑇𝑇 + 𝜔𝜔𝑠𝑠 𝐿𝐿𝑎𝑎 𝑖𝑖𝑓𝑓 + 𝜔𝜔𝑠𝑠 𝑖𝑖
NOMENCLATURE

𝐿𝐿𝑟𝑟 𝑓𝑓
B Friction coefficient

= (𝑅𝑅𝑠𝑠 + 𝐿𝐿𝑎𝑎 𝜌𝜌)𝑖𝑖 𝑇𝑇 + 𝜔𝜔𝑠𝑠 𝐿𝐿𝑠𝑠 𝑖𝑖𝑓𝑓


dq Park transformation dq (7)

Where 𝐿𝐿𝑎𝑎 is.


idr d axis rotor current in the synchronous frame
ids d axis stator current in the synchronous frames
if Flux-producing component of the stator current

𝐿𝐿𝑎𝑎 = 𝜎𝜎𝐿𝐿𝑠𝑠 = (𝐿𝐿𝑠𝑠 − )


𝐿𝐿𝑚𝑚 2
iqr q axis rotor current in the synchronous frame
𝐿𝐿𝑟𝑟
iqs q axis stator current in the synchronous frame (8)
iT Torque producing component of the stator current
Is Stator current

𝑖𝑖 𝑇𝑇 𝑅𝑅𝑟𝑟
J Moment of inertia
Now the stator frequency is represented by.

𝜔𝜔𝑠𝑠1 = ( )
𝑖𝑖𝑓𝑓 𝐿𝐿𝑟𝑟
Lm Magnetizing inductance
Lr Rotor inductance
Ls Stator inductance

𝜔𝜔𝑠𝑠 = 𝜔𝜔𝑟𝑟 + 𝜔𝜔𝑠𝑠1 = 𝜔𝜔𝑟𝑟 + ( )


𝑖𝑖𝑇𝑇 𝑅𝑅𝑟𝑟
Rs Stator resistance

𝑖𝑖𝑓𝑓 𝐿𝐿𝑟𝑟
Vqs q axis stator voltage in the synchronous frame (9)
Vds d axis stator voltage in the synchronous frame

substituting 𝜔𝜔𝑠𝑠 in equation (2).


λr Rotor flux linkage phasor
σ Leakage coefficient The equation of the electric part of the motor is obtained by
 Derivate

𝑉𝑉𝑞𝑞𝑞𝑞 = (𝑅𝑅𝑠𝑠 + + 𝐿𝐿𝑎𝑎 𝜌𝜌) 𝑖𝑖 𝑇𝑇 + 𝜔𝜔𝑟𝑟 𝐿𝐿𝑠𝑠 𝑖𝑖𝑓𝑓


𝑅𝑅𝑟𝑟 𝐿𝐿𝑠𝑠
ωr Rotor electrical speed

𝐿𝐿𝑟𝑟
ωs Slip speed (10)

The starting point for the deduction of the transfer functions Where the torque produced by the stator current can be derived
applied to the IM, is to take as constant the flow link in the from the following equation.

𝐼𝐼𝑇𝑇 = = (𝑉𝑉𝑞𝑞𝑞𝑞 − 𝜔𝜔𝑟𝑟 𝐿𝐿𝑠𝑠 𝑖𝑖𝑓𝑓 )


rotor. 𝑉𝑉𝑞𝑞𝑞𝑞 −𝜔𝜔𝑟𝑟 𝐿𝐿𝑠𝑠 𝑖𝑖𝑓𝑓 𝐾𝐾𝑎𝑎

𝛹𝛹𝑟𝑟 = 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶 𝑝𝑝𝛹𝛹𝑟𝑟 = 0


𝑅𝑅 𝐿𝐿
𝑅𝑅𝑠𝑠 + 𝑟𝑟 𝑠𝑠 +𝐿𝐿𝑎𝑎 𝜌𝜌 1+𝑠𝑠𝑇𝑇𝑎𝑎
(11)
𝐿𝐿𝑟𝑟

With 𝑝𝑝 is a derivate
Where

𝑅𝑅𝑎𝑎 = 𝑅𝑅𝑠𝑠 + 𝑅𝑅𝑟𝑟 𝐾𝐾𝑎𝑎 = 𝑇𝑇𝑎𝑎 =


𝐿𝐿𝑠𝑠 1 𝐿𝐿𝑎𝑎
Then, the stator equations are 𝐿𝐿𝑟𝑟 𝑅𝑅𝑎𝑎 𝑅𝑅𝑎𝑎

𝑉𝑉𝑞𝑞𝑞𝑞 = (𝑅𝑅𝑠𝑠 + 𝐿𝐿𝑠𝑠 𝜌𝜌)𝑖𝑖𝑞𝑞𝑞𝑞 + 𝜔𝜔𝑠𝑠 𝐿𝐿𝑠𝑠 𝑖𝑖𝑑𝑑𝑑𝑑 + 𝐿𝐿𝑚𝑚 𝜌𝜌𝑖𝑖𝑞𝑞𝑞𝑞 + 𝜔𝜔𝑠𝑠 𝐿𝐿𝑚𝑚 𝑖𝑖𝑑𝑑𝑑𝑑 (1)
𝑉𝑉𝑑𝑑𝑑𝑑 = (𝑅𝑅𝑠𝑠 + 𝐿𝐿𝑠𝑠 𝜌𝜌)𝑖𝑖𝑑𝑑𝑑𝑑 − 𝜔𝜔𝑠𝑠 𝐿𝐿𝑠𝑠 𝑖𝑖𝑞𝑞𝑞𝑞 + 𝐿𝐿𝑚𝑚 𝜌𝜌𝑖𝑖𝑑𝑑𝑑𝑑 − 𝜔𝜔𝑠𝑠 𝐿𝐿𝑚𝑚 𝑖𝑖𝑞𝑞𝑞𝑞
From this block, which converts the feedback voltage to torque
(2) current, then electromagnetic torque is written.

663
2018 IFAC MICNON
664 Edgar Chulines et al. / IFAC PapersOnLine 51-13 (2018) 662–667
Guadalajara, Mexico, June 20-22, 2018

𝜏𝜏𝑒𝑒 = 𝐾𝐾𝑓𝑓 𝑖𝑖 𝑇𝑇
−𝑅𝑅𝑎𝑎 −𝜓𝜓
(12) Where

1
𝐿𝐿𝑎𝑎 𝐿𝐿𝑎𝑎 0 1 0
Where the constant of torque is definite.

𝐾𝐾𝑓𝑓 =
3 𝑃𝑃 𝐿𝐿𝑚𝑚 2
𝑖𝑖𝑓𝑓 𝐴𝐴 = , 𝐵𝐵 = [𝐿𝐿𝑎𝑎 ] , 𝐶𝐶 = [ ]
𝐾𝐾𝑡𝑡 𝑁𝑁𝑝𝑝 −(𝐵𝐵 + 𝐵𝐵1 ) 0 1
0 0
2 2 𝐿𝐿𝑟𝑟
[ 𝐽𝐽 𝐽𝐽 ]
(13)

The dynamic loads can be represented taking the


electromagnetic torque and the torque of the load which is
−𝑅𝑅𝑎𝑎 −𝜓𝜓
Then

1
considered as friction in this particular case.

𝑗𝑗
𝑑𝑑𝜔𝜔𝑚𝑚
+ 𝐵𝐵𝜔𝜔𝑚𝑚 = 𝜏𝜏𝑒𝑒 − 𝜏𝜏𝐿𝐿 = 𝐾𝐾𝑓𝑓 𝑖𝑖 𝑇𝑇 − 𝐵𝐵𝑙𝑙 𝜔𝜔𝑚𝑚 𝐼𝐼 ̇ 𝐿𝐿𝑎𝑎 𝐿𝐿𝑎𝑎 𝐼𝐼 0 𝑉𝑉𝑞𝑞𝑞𝑞
[ 𝑞𝑞𝑞𝑞 ] = [ 𝑞𝑞𝑞𝑞 ] + [𝐿𝐿𝑎𝑎 ][ ]
−(𝐵𝐵 + 𝐵𝐵1 ) 𝜔𝜔𝑟𝑟
𝑑𝑑𝑑𝑑
𝜔𝜔𝑟𝑟̇ 𝐾𝐾𝑡𝑡 𝑁𝑁𝑝𝑝 0
(14)

0 0
(19)

[ 𝐽𝐽 𝐽𝐽 ]
1 0 𝐼𝐼𝑞𝑞𝑞𝑞
Where in terms of the electric speed of the rotor, it is derived

𝑌𝑌(𝑡𝑡) = [ ][ ]
0 1 𝜔𝜔𝑟𝑟
from the multiplication of both sides by the pair of poles.

𝑗𝑗 + 𝐵𝐵𝜔𝜔𝑟𝑟 = 𝐾𝐾𝑓𝑓 𝑖𝑖 𝑇𝑇 − 𝐵𝐵𝑙𝑙 𝜔𝜔𝑟𝑟


𝑑𝑑𝜔𝜔𝑟𝑟 𝑃𝑃
(20)

𝜓𝜓 = 𝐿𝐿𝑠𝑠 𝐼𝐼𝑑𝑑𝑑𝑑 , 𝑁𝑁𝑝𝑝 =


𝑑𝑑𝑑𝑑 2 𝑃𝑃
(15)
2
Where (21)
Then the transfer function between the velocity and the torque
produced current is. Note that, the structure obtained in (19) is similar to DC motor

=
𝜔𝜔𝑟𝑟 (𝑠𝑠) 𝐾𝐾𝑎𝑎
model shown in Chan et al. (2006) and Höfling et al (1996) but

𝐼𝐼𝑞𝑞𝑞𝑞̇ in (19) the magnetic flow 𝜓𝜓 is defined as the relation


𝐼𝐼𝑇𝑇 (𝑠𝑠) 1+𝑠𝑠𝑇𝑇𝑎𝑎
(16) no equal. An important difference is that the second term of

between the stator inductance 𝐿𝐿𝑠𝑠 and the d axis stator current
𝑃𝑃 𝐾𝐾𝑓𝑓 𝐽𝐽
Where

𝐾𝐾𝑎𝑎 = , 𝐵𝐵𝑡𝑡 = 𝐵𝐵 + 𝐵𝐵𝑙𝑙 , 𝑇𝑇𝑎𝑎 = that the first term of 𝜔𝜔𝑟𝑟̇ ,the magnetic flow 𝜓𝜓 of d.c. motor
2 𝐵𝐵𝑡𝑡 𝐵𝐵𝑡𝑡
in the synchronous frames Ids. Another important difference is

model is defined as the relationship among poles number P,


Once having the equations of the IM of the electrical and magnetic inductance Lm, rotor inductance Lr and flux-
mechanical part, from the equations (7) and (8) the block producing component of the stator current If. A way to add
diagram of the IM is made as shown in the Fig. 1. redundancy in the equations at the same instant tis by
introducing (17) in (18) with its respective derivatives like.
𝒀𝒀(𝑡𝑡) = 𝑻𝑻𝑥𝑥(𝑡𝑡) + 𝑸𝑸𝑸𝑸(𝑡𝑡)
Ls.if
(22)
B1 Where
𝑦𝑦(𝑡𝑡) 𝑪𝑪
-

𝑦𝑦̇ (𝑡𝑡)
Te

𝑪𝑪𝑪𝑪
Ka iT P - 1 ωr
+ Kf +

𝑦𝑦̈ (𝑡𝑡) = 𝑪𝑪𝑪𝑪𝟐𝟐 𝑥𝑥(𝑡𝑡)


Vqs 1+sTa Iqs 2 B+sJ

⋮ ⋮
[𝑦𝑦 𝑞𝑞 (𝑡𝑡)] [𝑪𝑪𝑪𝑪𝑞𝑞 ]
Electrical Mechanical
(23)

0 0 0 … 0 𝑢𝑢(𝑡𝑡)
Fig. 1. Block diagram of IM model with constant rotor flux

𝑪𝑪𝑪𝑪 0 0 … 0 𝑢𝑢̇ (𝑡𝑡)


linkages

+ 𝑪𝑪𝑪𝑪𝑪𝑪 𝑪𝑪𝑪𝑪 𝑪𝑪𝑪𝑪 … 0 𝑢𝑢̈ (𝑡𝑡)


⋮ ⋮ ⋮ ⋱ ⋮ ⋮
This model is so similar to CD motor model obtained in Chan

difference is that the input parameter is 𝑉𝑉𝑞𝑞𝑞𝑞 instead of the [𝑪𝑪𝑨𝑨𝒒𝒒−𝟏𝟏 𝑩𝑩 𝑪𝑪𝑨𝑨𝒒𝒒−𝟐𝟐 𝑩𝑩 … 𝑪𝑪𝑪𝑪 𝟎𝟎] [𝑢𝑢𝑞𝑞 (𝑡𝑡)]
et al. (2006) and Höfling et al (1996). Although the main

armature current.

3. RESIDUAL GENERATION Now, the residual vector based on state-space model for
continuous time is given in (24) which is deduced in from the
A simple model of IM in steady state behaviour which is little residual generation with parity equation for multiple-input
similar to d.c motor model mentioned in Krishnan et al. (2001) multiple-output (MIMO) process with transfer functions and
is used in this work, because the fault detection based on parity polynomial errors.

𝒓𝒓(𝑡𝑡) = 𝑾𝑾𝑾𝑾(𝑡𝑡) − 𝑾𝑾𝑾𝑾𝑾𝑾(𝑡𝑡)


equations for this model types is availability to detect several
parameters, Chan et al. (2006). (24)
Based on known ways of theoretically modelling the structure An important condition to satisfy that both of the first and
of a linear mathematical model in the continuous time without second term of (24) are zeros is that WT=0, Höfling et al
considering disturbances (17) and (18), the state space (1996) where W is called the null space of T and can be
representation obtained for the IM is displayed in (19) and obtained by proposing the greater number of zeros possible at
(20). the rows, taking into account that the lines are linearly

𝑋𝑋̇(𝑡𝑡) = 𝑨𝑨𝑥𝑥(𝑡𝑡) + 𝑩𝑩𝑢𝑢(𝑡𝑡)


independent. In our case of study, the W matrix obtained by
(17)
𝑌𝑌(𝑡𝑡) = 𝑪𝑪𝑥𝑥(𝑡𝑡)
the IM is (25).
(18)

664
2018 IFAC MICNON Edgar Chulines et al. / IFAC PapersOnLine 51-13 (2018) 662–667 665
Guadalajara, Mexico, June 20-22, 2018

𝑅𝑅𝑎𝑎 𝜓𝜓 𝐿𝐿𝑎𝑎 0 0 0
−𝛼𝛼 𝛽𝛽 0 𝐽𝐽 0 0
It is important to remember that the detection of these faults
𝑾𝑾 = [ ]
𝛾𝛾 0 𝛿𝛿 0 𝐽𝐽𝐿𝐿𝑎𝑎 0
are only in steady state behavior which allows to propose
(25)
0 𝛾𝛾 0 𝛿𝛿 0 𝐽𝐽𝐿𝐿𝑎𝑎
heuristically a pair of fixed thresholds close to the nominal
current of the IM, this condition is shown in Fig. 2 and is
Where represented in the expression (29) in ideal condition.

𝛼𝛼 = 𝐾𝐾𝑡𝑡 𝑁𝑁𝑝𝑝 , 𝛽𝛽 = 𝐵𝐵 + 𝐵𝐵1 ,


Table 2. Insolation of fault

𝛾𝛾 = 𝜓𝜓𝜓𝜓 + 𝑅𝑅𝑎𝑎 𝛽𝛽, 𝛿𝛿 = 𝐿𝐿𝑎𝑎 𝛽𝛽 + 𝐽𝐽𝑅𝑅𝑎𝑎


Table 1. Fault detection matrix in Rs and Ls
(26)
faults r1 r2 r3 r4 Failure in Is
𝐹𝐹𝑎𝑎 𝐹𝐹𝑏𝑏 𝐹𝐹𝑐𝑐
do not change, 𝑟𝑟(𝑡𝑡) = 0, then, a fault is detected when 𝒓𝒓(𝑡𝑡) ≠
Failure
By the assumption that in the healthy operation the parameter Rs I 0 0 I

0. The residuals obtained by the IM are.


Rr I 0 0 I 0 0 0 Healthy

parametric
operation

𝑟𝑟1 (𝑡𝑡) = 𝑅𝑅𝑎𝑎 𝐼𝐼𝑞𝑞𝑞𝑞 (𝑡𝑡) + 𝜓𝜓𝜔𝜔𝑟𝑟 (𝑡𝑡) + 𝐿𝐿𝑎𝑎 𝐼𝐼𝑞𝑞𝑞𝑞̇ (𝑡𝑡) − 𝑉𝑉𝑞𝑞𝑞𝑞 𝑡𝑡
Ls I 0 I I 0 0 I Failure in
phase c
Lr I I I I
̇
𝑟𝑟2 (𝑡𝑡) = −𝛼𝛼𝐼𝐼𝑞𝑞𝑞𝑞 (𝑡𝑡) + 𝛽𝛽𝜔𝜔𝑟𝑟 (𝑡𝑡) + 𝐽𝐽𝜔𝜔𝑟𝑟 (𝑡𝑡)
0 I 0 Failure in
B 0 I I I phase b

𝑟𝑟3 (𝑡𝑡) = 𝛾𝛾𝐼𝐼𝑞𝑞𝑞𝑞 (𝑡𝑡) + (𝐿𝐿𝑎𝑎 𝛽𝛽 + 𝐽𝐽𝑅𝑅𝑎𝑎 )𝐼𝐼𝑞𝑞𝑞𝑞̇ (𝑡𝑡) + 𝐽𝐽𝐿𝐿𝑎𝑎 𝐼𝐼𝑞𝑞𝑞𝑞̈ (𝑡𝑡)
0 I I Failure in
Bl 0 I I I phase b, c

− 𝛽𝛽𝑉𝑉𝑞𝑞𝑞𝑞 𝑡𝑡 − 𝐽𝐽𝑉𝑉𝑞𝑞𝑞𝑞̇ (𝑡𝑡)


I 0 0 Failure in
(27) Iqse I I I 0 phase a

additive
̇ + 𝐽𝐽𝐿𝐿𝑎𝑎 𝜔𝜔𝑟𝑟 (𝑡𝑡)
̈
Failure in

𝑟𝑟4 (𝑡𝑡) = 𝛾𝛾𝜔𝜔𝑟𝑟 (𝑡𝑡) + [𝐿𝐿𝑎𝑎 𝛽𝛽 + 𝐽𝐽𝑅𝑅𝑎𝑎 ]𝜔𝜔𝑟𝑟 (𝑡𝑡)


ωr I I 0 I I 0 I
phase a, c

− 𝛼𝛼𝑉𝑉𝑞𝑞𝑞𝑞
I I 0 Failure in
Vqse I 0 I I phase a, b

Note that during the steady state, the derived of 𝑥𝑥(𝑡𝑡) is zero,
Where “I” represents a significant I I I Failure in
change which can be positive or phase a,b,c

and 𝑉𝑉𝑞𝑞𝑞𝑞 𝑡𝑡 = 𝑉𝑉𝑞𝑞𝑞𝑞 , thus, the residual can be simplified, this is so


negative

suitable when the fault type is incipient; considering that is the


most common in the electrical machines, then, the residuals
can be reduced as.
𝑟𝑟1 (𝑡𝑡) = 𝑅𝑅𝑎𝑎 𝐼𝐼𝑞𝑞𝑞𝑞 (𝑡𝑡) + 𝜓𝜓𝜔𝜔𝑟𝑟 (𝑡𝑡) − 𝑉𝑉𝑞𝑞𝑞𝑞
𝑟𝑟2 (𝑡𝑡) = −𝛼𝛼𝐼𝐼𝑞𝑞𝑞𝑞 (𝑡𝑡) + 𝛽𝛽𝜔𝜔𝑟𝑟 (𝑡𝑡)
𝑟𝑟3 (𝑡𝑡) = 𝛾𝛾𝐼𝐼𝑞𝑞𝑞𝑞 (𝑡𝑡) − 𝛽𝛽𝑉𝑉𝑞𝑞𝑞𝑞
(28)

𝑟𝑟4 (𝑡𝑡) = 𝛾𝛾𝜔𝜔𝑟𝑟 (𝑡𝑡) − 𝛼𝛼𝑉𝑉𝑞𝑞𝑞𝑞

Likewise, as in the d.c. motor model, if an additive fault


occurs, all residuals except the decouple one are deflected as Fig. 2. Ideal current for fixes threshold close to nominal
current of IM

1→ 𝐼𝐼𝑛𝑛 > 𝐼𝐼ℎ𝑛𝑛+


shown in Table 1. This supports strongly to locate the sensor
faults, and thus this fault types are easy detectable. When a
𝑓𝑓𝑛𝑛 = { 1→ 𝐼𝐼𝑛𝑛 < 𝐼𝐼ℎ𝑛𝑛−
parametric fault occurs on Rs or Rr there is no a considerable
0→ 𝐼𝐼ℎ𝑛𝑛− > 𝐼𝐼𝑛𝑛 > 𝐼𝐼ℎ𝑛𝑛+
increase in r3, thus, a null value can be considered to simplify (29)
the fault detection matrix. Finally, a simple way to distinguish
the fault is by using classical limit-values detectors with a Where
suitable tuning, taking into account the behaviour of residual
obtained in Fig. 2. n represents phase a, b or c

Table 1 shows the parameters associated with the diagnosis of Additionally, multiple failures for only these two parameters
electrical faults in the IM model using parity equations based are involved. The above is valid for a three-phase IM
on the reference frame dq for its linearization. However, this connected in star and neutral terminal connected to ground,
information does not identify the damaged phase associated this is because the affected phase current will unbalance the
with the electrical parameter. Therefore, it is proposed to sum of the zero current in the neutral point n (Fig. 4). Then,
analyze in particular the stator currents to accurately identify the excess current will flow to earth without affecting the
the damaged element. On the other hand, it is important to phases in healthy condition. In Fig. 3 shown the simulation in
mention that, for practicality during corrective maintenance, it Matlab software of three-phase IM connected in star with and
is not necessary to know the damaged phase come from the neutral terminal connected to ground with parametric
rotor. The Table 2 shows the Insolation of fault in Rs and Ls to variation.
identify the affected phase in the fault diagnosis in Rs or Ls.

665
2018 IFAC MICNON
666 Edgar Chulines et al. / IFAC PapersOnLine 51-13 (2018) 662–667
Guadalajara, Mexico, June 20-22, 2018

Fig. 3. Three-phase IM connected in star with and neutral terminal connected to ground: (a) Rs nominal value -50%, (b) Rs
nominal value +50%, (c) Rr nominal value -50%, (d) Rr nominal value +50%, (e) Lr nominal value -50%, (f) Lr nominal value
+50%, (g) Ls nominal value -50%, (h) Ls nominal value +50%, (i) B nominal value -50%, (j) B nominal value +50%

and (29) obtained in this work, which allow to obtain tables 1


and 2, are extremely simple to implement in any digital
processor since the operators are addition, subtraction and
multiplication without recursion. The main problem is that the
parameters of the IM must be known a priori.

REFERENCES
Bianchini, C., Immovilli, F., Cocconcelli, M., Rubini, R., &
Bellini, A. (2011). Fault detection of linear bearings in
brushless AC linear motors by vibration analysis. IEEE
Transactions on Industrial Electronics, 58(5), 1684-
1694.
Cash, M. A., Habetler, T. G., & Kliman, G. B. (1998).
Insulation failure prediction in AC machines using line-
neutral voltages. IEEE Transactions on Industry
Applications, 34(6), 1234-1239.
Chan, C. W., Hua, S., & Hong-Yue, Z. (2006). Application
Fig. 4. Three-phase IM connected in star with and neutral of fully decoupled parity equation in fault detection and
terminal connected to ground identification of DC motors. IEEE Transactions on
Industrial Electronics, 53(4), 1277-1284.
Then with the above matrix of table 1 can be seen that the fault
Höfling, T., & Isermann, R. (1996). Fault detection based
detection probability for Rs, Rr, Ls, B, Bl and Vqse are 50%
on adaptive parity equations and single-parameter
and only for Lr, Iqse and ωr are 100%, but according to table
tracking. Control Engineering Practice, 4(10), 1361-
2, Rs and Ls the fault detection probability is 100%.
1369.
Isermann, R. (1995, June). Model base fault detection and
4. CONCLUSIONS
diagnosis methods. In American Control Conference,
The linearization of IM nonlinear model using the Proceedings of the 1995 (Vol. 3, pp. 1605-1609). IEEE.
synchronous reference frame dq is valid only when the IM is Isermann, R. (1997). Supervision, fault-detection and fault-
working in stable state operation. In this way, it is convenient diagnosis methods—an introduction. Control
to consider heuristically a diagnostic dead time at start-up and engineering practice, 5(5), 639-652.
fixed current thresholds, according to the stabilization time and Klima, J. (2003). Analytical investigation of an induction
input nominal current respectively. Since the IM model is motor drive under inverter fault mode operations. IEE
matched to the d.c.. motor model the analysis allows to assure Proceedings-Electric Power Applications, 150(3), 255-
the existence of parity space and therefore to obtain the 262.
advantages of the fault detection for this system type. Add a Kolla, S., & Varatharasa, L. (2000). Identifying three-phase
pair of fixed thresholds to the stator current and use Table 2, it induction motor faults using artificial neural
allows to insolation the affected phase for Lsn and Rsn. In networks. ISA transactions, 39(4), 433-439.
addition, the detection of multiple faults to these parameters is Krishnan, R. (2001). Electric motor drives: modeling,
obtained naturally, as long as the IM is connected in star and analysis, and control. Prentice Hall.
the neutral terminal connected to ground. The expression (28)

666
2018 IFAC MICNON
Guadalajara, Mexico, June 20-22, 2018 Edgar Chulines et al. / IFAC PapersOnLine 51-13 (2018) 662–667 667

Loparo, K. A., Adams, M. L., Lin, W., Abdel-Magied, M.


F., & Afshari, N. (2000). Fault detection and diagnosis
of rotating machinery. IEEE Transactions on Industrial
Electronics, 47(5), 1005-1014.
Mendes, A. M., & Cardoso, A. M. (2006). Fault-tolerant
operating strategies applied to three-phase induction-
motor drives. IEEE Transactions on Industrial
Electronics, 53(6), 1807-1817.
Nandi, S., Toliyat, H. A., & Li, X. (2005). Condition
monitoring and fault diagnosis of electrical motors—A
review. IEEE transactions on energy conversion, 20(4),
719-729.
Pineda-Sanchez, M., Riera-Guasp, M., Roger-Folch, J.,
Antonino-Daviu, J. A., Perez-Cruz, J., & Puche-
Panadero, R. (2011). Diagnosis of induction motor
faults in time-varying conditions using the polynomial-
phase transform of the current. IEEE Transactions on
Industrial Electronics, 58(4), 1428-1439.
Strangas, E. G., Aviyente, S., & Zaidi, S. S. H. (2008).
Time–frequency analysis for efficient fault diagnosis
and failure prognosis for interior permanent-magnet
AC motors. IEEE Transactions on Industrial
Electronics, 55(12), 4191-4199.

667

Das könnte Ihnen auch gefallen