ene gies
A icle
U iliza ion o Two Senso s in O line Diagnosis o
Squi el-Cage Ro o s o Asynch onous Mo o s
Pe Kaco * , Pe Be na and Pe Mold ik


Ci a ion: Kaco , P.; Be na , P.;
Mold ik, P. U iliza ion o Two Senso s
in O line Diagnosis o Squi el-Cage
Ro o s o Asynch onous Mo o s.
Ene gies 2021,14, 6573. h ps://
doi.o g/10.3390/en14206573
Academic Edi o : Maciej Sułowicz
Recei ed: 4 Sep embe 2021
Accep ed: 10 Oc obe 2021
Published: 13 Oc obe 2021
Publishe ’s No e: MDPI s ays neu al
wi h ega d o ju isdic ional claims in
published maps and ins i u ional a il-
ia ions.
Copy igh : © 2021 by he au ho s.
Licensee MDPI, Basel, Swi ze land.
This a icle is an open access a icle
dis ibu ed unde he e ms and
condi ions o he C ea i e Commons
A ibu ion (CC BY) license (h ps://
c ea i ecommons.o g/licenses/by/
4.0/).
Depa men o Elec ical Powe Enginee ing a VSB, Technical Uni e si y o Os a a, 17. Lis opadu 2172/15,
708 00 Os a a-Po uba, Czech Republic; pe [email p o ec ed] (P.B.); pe [email p o ec ed] (P.M.)
*Co espondence: pe [email p o ec ed]
Abs ac :
In he manu ac u e squi el-cage o o s o asynch onous mo o s, a high s anda d o quali y
is equi ed in e e y pa o he p oduc ion cycle. The die cas ing p ocess usually c ea es po osi y in
he o o ba s. This mos common de ec in he o o o en emains hidden du ing he en i e assembly
o he machine and is usually only de ec ed du ing inal es ing o he mo o , i.e., a he end o he
p oduc ion p ocess. This leads o unnecessa y p oduc ion cos s. The e o e, he aim is o conduc
a con inuous con ol immedia ely a e he o o has been cas be o e u he p ocessing. In ou
pape , we a e in e es ed in selec ing a sui able me hod o o line o o diagnos ics o an asynch onous
mo o ha would be e ec i e o hese needs. In he i s s ep, he selec ion o he me hod and i s
in eg a ion in o he o e all manu ac u ing p ocess is ca ied ou . The a angemen o he senso s and
hei calib a ion is hen simula ed on a 2D Fini e Elemen Model o he o o . The p oposed o line
measu emen p ocedu es and echnologies a e inally alida ed by es ing measu emen s on a o o
ha simula es he mos equen ly occu ing aul s. A es sys em is also de eloped ha p o ides
he ope a o con inuous in o ma ion abou he unning o o measu emen s and makes i easie o
e alua e he quali y o he cas o o by means o g aphical isualiza ion o he aul s.
Keywo ds:
squi el-cage; aul de ec ion; b oken ba ; o line diagnosis; FEM simula ion; signal
p ocessing; o al pa e ns
1. In oduc ion
In he p oduc ion o asynch onous machine o o s, we o en encoun e he need o es
hei manu ac u ing quali y du ing p oduc ion. This ype o es ing can de ec a p oduc ion
de ec in he o o ea ly on, immedia ely a e he o o has been cas and emo ed om he
die cas ing mold. O he wise, his de ec would only appea a he end o he manu ac u ing
p ocess o he elec ic mo o du ing i s ou pu es ing [
1
,
2
]. In his case, he mo o is s ill
epai able by o al o o eplacemen . This me hod adds eno mous ex a manu ac u ing
cos s. Mo eo e , a comple e o o eplacemen is e y di icul o pe o m in he mass
p oduc ion o elec ic machines. Addi ional cos s a e incu ed in all he echnological
ope a ions ha he o o goes h ough a e cas ing. These include he p essing o he sha
in o he s a o pack, he adjus men o he o o dimensions and i s su ace, he o e all
balancing o he o a ing masses, he p essing o he bea ings on o he sha , and i s inal
ixing o he ame wi h he s a o [2].
Ro o manu ac u ing ailu es can be di ided in o wo ca ego ies acco ding o he
signi icance o he ailu e: majo and mino [
3
,
4
]. Majo de ec s can be de ec able by he
naked eye o by using image p ocessing [
5
–
7
]. These include b oken ba s, ba s sepa a ed
om he ings, bulging ma e ial on he o o su ace, o comple ely missing ma e ial in
he ing o ba . Mino de ec s in cas o o s a e mainly caused by ai bubbles in he ba s
and end- ings o in e nal ma e ial de o ma ion caused by in ensi e cooling o he mol en
me al [
8
]. A o o wi h such de ec s may seem isual o be pe ec ly ine, bu om an
elec ical iew, i has educed esis ance, which may be une enly dis ibu ed be ween he
Ene gies 2021,14, 6573. h ps://doi.o g/10.3390/en14206573 h ps://www.mdpi.com/jou nal/ene gies
Ene gies 2021,14, 6573 2 o 23
ba s and end- ings [
8
,
9
]. In he case o die-cas o o cages, whe e he ings a e in di ec
con ac wi h he o o packing, isual inspec ion e en by an expe ienced echnician may be
unsuccess ul and mino de ec s p ac ically unde ec able [2,4].
The in e nal shape o he o o ba s and hei o e all a angemen can also be e y
a ied, e.g., single s aigh ba s, skewed ba s, and single o double cage o o s. The shape
o a owed o o ba s wi h a middle ing is also o en ound [
10
,
11
]. The o o slo s can be
open, semi-closed, o ully closed [
12
–
14
]. The a iabili y o he o o s, in gene al, makes
he iden i ica ion o ailu es di icul and is qui e insu icien in mul i-se ies p oduc ion o
cas o o s.
Du ing se e al decades, e y high-quali y me hods ha e been de eloped and elab-
o a ed, which a e oday commonly used o iden i y machine ailu es based on online
diagnos ics [
15
–
20
]. These me hods a e use ul du ing ope a ion when disassembly o he
elec ic mo o is no equi ed. Faul measu emen s can be pe o med a a sa e dis ance
om he machine [
21
]. Widely used echniques a e, e.g., single-phase es wi h educed
machine supply, MSCA (Mo o Signa u e Cu en Analysis), o diagnos ics om he s ay
ield in adial o axial di ec ion [
21
,
22
]. Al hough hese me hods can de ec mino o o
aul s, such a sys em is no sui able o diagnosing a single o o a e cas ing because i
equi es a machine s a o . In an assembled machine, he es esul s can also be a ec ed
by he condi ion o o he mo o componen s such as he s a o , i s magne ic ci cui wi h
windings, bea ings, e c. [23,24].
To de ec ailu es in he squi el-cage o o immedia ely a e die cas ing, i is necessa y
o sea ch in an o -line es g oup ha does no include any mo o componen s o he han he
o o i sel . In his a ea, we can ind, o example, magne ic oil diagnos ics, dye pene an ,
DLRO (Digi al Low Resis ance Ohmme e ) measu emen s, o ul asonic measu emen s o
de e mine he quali y o he end- ings [
25
]. The e also exis s a me hod ha equi es high
cu en exci a ion (HCE) in which local ho spo s on he o o su ace a e measu ed using
an IR came a. The empe a u e can each high alues and he lowing cu en s can dis o
he measu ing equipmen [3,4].
The G owle es is his o ically he oldes o -line me hod o measu ing o o quali y.
I was o iginally used o winding o o s o commu a o machines o loca e in e - u n
connec ions o in e up ed connec ions o he winding ou pu s o he commu a o [
26
].
The G owle yoke exci es an AC magne ic ield in he o o , which induces a ol age in
he ba s (windings). When a pa o he winding is in e up ed o sho -ci cui ed, a local
change in he magne ic ield is p oduced on he o o su ace. The change in he ield can be
de ec ed by e omagne ic iden i ica ion ape, magne ic oil, Hall senso , o by compa ing
he change in supply cu en as o he o o slowly o a es. Fo bes sensi i i y, he o o is
usually in di ec con ac wi h he poles o he G owle yoke [27].
O he me hods use di ec analysis o he magne ic ield esponse o measu emen
o he magne ic induc ion alue nea he o o su ace o es he o o . The o o usually
o a es a a slow speed du ing he measu emen and he exci a ion magne ic ield is
gene a ed by an ex e nal sou ce using a co e coil o pe manen magne s [
28
,
29
]. He e,
he o o de ec is ep esen ed by a educed o missing magne ic lux ampli ude. Di ec
magne ic measu emen s ha e hei indispu able ad an ages, bu he echnology used is
qui e expensi e and equi es a good quali y measu emen sys em wi h as con e e s.
Addi ionally, Hall p obes a e e y agile o handling and he e is a isk o mechanical
damage in no mal echnological ope a ion.
The main mo i a ion is o ind and choose a sui able o -line me hod o o o diagnos-
ics. O e all, he diagnos ic sys em should be simple and loca ed in he o o manu ac u ing
p ocess be ween he die cas ing echnology and he ollowing ope a ions be o e he assem-
bly wi h he s a o . F om he e e ence li e a u e used [
28
,
30
], i is mo e o less e iden
ha he main e alua ion me hod o he iden i ica ion o o o de ec s is he cu en ly used
FFT analysis. Especially in he case o he use o FFT analysis, he diagnos ic pos in he
manu ac u ing company needs o be co e ed by a esea che o a pe son wi h a highe
quali ica ion. Howe e , such a solu ion is no possible in mass p oduc ion o o o s. This
Ene gies 2021,14, 6573 3 o 23
led us o he idea o ying o isualize he de ec o he es ed o o in a di e en way han
he abo e men ioned p ocedu es. The main aim o ou wo k is o p o ide he in o ma ion
abou he de ec o he manu ac u ed o o o he ope a o so ha he iden i ica ion o he
de ec is based on a g aphical pa e n. In addi ion, he li e a u e [
28
,
30
] also show he
ou pu s o p ac ical implemen a ion o measu emen sys ems ha can be used in p ac ice in
mass p oduc ion o o o s. Howe e , he e is no de ailed discussion o he use o 2-senso
measu emen s, which we belie e can p o ide new in o ma ion in o o diagnos ics.
The pape is o ganized as ollows. In Sec ion 2, he inding o a sui able echnique
o o -line o o diagnos ics is desc ibed and he a angemen o senso s o measu ing
o o de ec s is p oposed. The design o he senso s is also desc ibed. In Sec ion 3, a
nume ical simula ion o a 2D elec omagne ic model o he o o is pe o med and he
o e all unc ionali y o he p oposed measu emen sys em is alida ed. In Sec ion 4, he
cons uc ion o a es o o designed o simula e o o aul s is desc ibed. In Sec ion 5, he
p ocessing o he senso signals, hei handling, and analysis is desc ibed. In
Sec ion 6
,
measu emen s on a eal o o o an asynch onous mo o a e made and alida ion o
he measu emen esul s is pe o med. The discussion and conclusion is p esen ed in
Sec ions 7and 8.
2. Senso Con igu a ion and Measu emen Me hod
F om he poin o iew o simplici y, eliabili y, and sensi i i y, a senso whose basic
p inciple is based on Fa aday’s law o elec omagne ic induc ion was conside ed o o o
diagnos ics [
31
]. The magne izing yoke ha gene a es he s a ic magne ic ield is loca ed
nea he o o [
28
,
29
]. The magne ic ield lines om he exci a ion yoke a e closed by an
ai gap ac oss he o o ee h a ound he o o ba slo s. When he o o o a es, a ol age
is induced in he o o ba s. All he ba s a e sho -ci cui ed a bo h end- ings and hen
cu en lows h ough he o o ba s. This cu en c ea es i s magne ic ield and a ec s he
o iginal exci a ion s a ic ield.
The gene a ed esponse in he magne ic ield will also p oduce a ime change in he
magne ic lux. This change is de ec ed and a ol age is induced in he sensing coil. A
heal hy o o gene a es a egula pe iodic ol age wa e o m, while a de ec o o shows a
de o ma ion in he ol age wa e o m depending on he ype o aul . The magni ude and
shape o his ol age a e he e o e in di ec p opo ion o he physical s a e o he o o [
30
].
In he sea ch o a sui able a angemen , se e al possible designs and senso a angemen s
we e selec ed, see Figu e 1.
Ene gies 2021, 14, x FOR PEER REVIEW 3 o 24
pos in he manu ac u ing company needs o be co e ed by a esea che o a pe son wi h
a highe quali ica ion. Howe e , such a solu ion is no possible in mass p oduc ion o
o o s. This led us o he idea o ying o isualize he de ec o he es ed o o in a di -
e en way han he abo e men ioned p ocedu es. The main aim o ou wo k is o p o ide
he in o ma ion abou he de ec o he manu ac u ed o o o he ope a o so ha he
iden i ica ion o he de ec is based on a g aphical pa e n. In addi ion, he li e a u e
[28,30] also show he ou pu s o p ac ical implemen a ion o measu emen sys ems ha
can be used in p ac ice in mass p oduc ion o o o s. Howe e , he e is no de ailed dis-
cussion o he use o 2-senso measu emen s, which we belie e can p o ide new in o -
ma ion in o o diagnos ics.
The pape is o ganized as ollows. In Sec ion 2, he inding o a sui able echnique o
o -line o o diagnos ics is desc ibed and he a angemen o senso s o measu ing o o
de ec s is p oposed. The design o he senso s is also desc ibed. In Sec ion 3, a nume ical
simula ion o a 2D elec omagne ic model o he o o is pe o med and he o e all unc-
ionali y o he p oposed measu emen sys em is alida ed. In Sec ion 4, he cons uc ion
o a es o o designed o simula e o o aul s is desc ibed. In Sec ion 5, he p ocessing o
he senso signals, hei handling, and analysis is desc ibed. In Sec ion 6, measu emen s
on a eal o o o an asynch onous mo o a e made and alida ion o he measu emen
esul s is pe o med. The discussion and conclusion is p esen ed in Sec ions 7 and 8.
2. Senso Con igu a ion and Measu emen Me hod
F om he poin o iew o simplici y, eliabili y, and sensi i i y, a senso whose basic
p inciple is based on Fa aday’s law o elec omagne ic induc ion was conside ed o o o
diagnos ics [31]. The magne izing yoke ha gene a es he s a ic magne ic ield is loca ed
nea he o o [28,29]. The magne ic ield lines om he exci a ion yoke a e closed by an
ai gap ac oss he o o ee h a ound he o o ba slo s. When he o o o a es, a ol age
is induced in he o o ba s. All he ba s a e sho -ci cui ed a bo h end- ings and hen
cu en lows h ough he o o ba s. This cu en c ea es i s magne ic ield and a ec s he
o iginal exci a ion s a ic ield.
The gene a ed esponse in he magne ic ield will also p oduce a ime change in he
magne ic lux. This change is de ec ed and a ol age is induced in he sensing coil. A
heal hy o o gene a es a egula pe iodic ol age wa e o m, while a de ec o o shows a
de o ma ion in he ol age wa e o m depending on he ype o aul . The magni ude and
shape o his ol age a e he e o e in di ec p opo ion o he physical s a e o he o o
[30]. In he sea ch o a sui able a angemen , se e al possible designs and senso a -
angemen s we e selec ed, see Figu e 1.
(a)
(b)
(c)
Figu e 1. The selec ed design o ac i e senso s: (a) DC supply exci a ion; (b) wi h pe manen magne s posi ioned on he
yoke poles; (c) wi h a pe manen magne posi ioned in one pa o he yoke only.
Figu e 1a shows a senso exci ed by a DC cu en sou ce, Figu e 1b shows a senso
whe e exci a ion is p o ided by pe manen magne s glued o he poles o he senso ’s
magne ic ci cui , and inally Figu e 1c shows a a ian wi h a pe manen magne on one
pole o he senso . In all cases, he magne ic senso ci cui was made in a U-shape o
Figu e 1.
The selec ed design o ac i e senso s: (
a
) DC supply exci a ion; (
b
) wi h pe manen magne s posi ioned on he
yoke poles; (c) wi h a pe manen magne posi ioned in one pa o he yoke only.
Figu e 1a shows a senso exci ed by a DC cu en sou ce, Figu e 1b shows a senso
whe e exci a ion is p o ided by pe manen magne s glued o he poles o he senso ’s
magne ic ci cui , and inally Figu e 1c shows a a ian wi h a pe manen magne on one
pole o he senso . In all cases, he magne ic senso ci cui was made in a U-shape o simpli y
he handling o he sensing coil. Al hough such senso s a e qui e simple o manu ac u e,
Ene gies 2021,14, 6573 4 o 23
we ound a numbe o p oblems when aligning hem o he o o . These ac i e senso s
showed a high endency o ib a e due o he pulsa ing magne ic pull as he o o o a es.
To elimina e he ib a ions ans e ed o he sensing pa o he senso , we pa ially
sepa a e he exci a ion and sensing unc ions by di iding he senso in o 2 pieces, ac i e
and passi e. The ac i e senso con ains exci a ion pe manen magne s NdFeB ype N42. I s
sensing coil is placed longi udinal be ween he pole ex ensions and wound on an insula ing
co e ame. The coil con ains app oxima ely N = 150 u ns o insula ed wi e and is o ien ed
in he axis o o a ion. The pole ex ensions and he co e o he sensing coil a e assembled
om high g ade o ien ed elec o echnical shee M165-35. The ac i e senso assembly is
clamped wi h b ass sc ews be ween aluminum holde s, see Figu e 2.
Ene gies 2021, 14, x FOR PEER REVIEW 4 o 24
simpli y he handling o he sensing coil. Al hough such senso s a e qui e simple o
manu ac u e, we ound a numbe o p oblems when aligning hem o he o o . These
ac i e senso s showed a high endency o ib a e due o he pulsa ing magne ic pull as
he o o o a es.
To elimina e he ib a ions ans e ed o he sensing pa o he senso , we pa ially
sepa a e he exci a ion and sensing unc ions by di iding he senso in o 2 pieces, ac i e
and passi e. The ac i e senso con ains exci a ion pe manen magne s NdFeB ype N42.
I s sensing coil is placed longi udinal be ween he pole ex ensions and wound on an in-
sula ing co e ame. The coil con ains app oxima ely N = 150 u ns o insula ed wi e and
is o ien ed in he axis o o a ion. The pole ex ensions and he co e o he sensing coil a e
assembled om high g ade o ien ed elec o echnical shee M165-35. The ac i e senso
assembly is clamped wi h b ass sc ews be ween aluminum holde s, see Figu e 2.
Figu e 2. Design con igu a ion o he senso s—Ac i e senso wi h he sensing coil be ween pole
ex ende s and Passi e senso wi h wo coils con igu a ion.
The passi e senso in Figu e 2 consis s o a magne ic ci cui also made o o ien ed
shee s and ben in o a U-shape. Two sensing coils a e a ached o i s poles. Each o he
coils con ain app oxima ely N = 500 u ns o insula ed wi e and he wo coils a e elec i-
cally connec ed in se ies o inc ease he alue o he induced ol age. The passi e senso
assembly is ixed in a non-magne ic insula ing enclosu e.
The con igu a ion o he measu emen sys em shown in Figu e 3 was ound o be he
mos con enien om a p ac ical poin o iew. Iden i ica ion o o o aul s uses a com-
bina ion o ac i e and passi e senso s o a ed 90 deg ela i e o each o he . The ac i e
senso gene a es he necessa y s a ic magne izing ield and senses i s change as he o o
o a es. The passi e senso cap u es he s ay ield nea he o o su ace.
Figu e 2.
Design con igu a ion o he senso s—Ac i e senso wi h he sensing coil be ween pole
ex ende s and Passi e senso wi h wo coils con igu a ion.
The passi e senso in Figu e 2consis s o a magne ic ci cui also made o o ien ed
shee s and ben in o a U-shape. Two sensing coils a e a ached o i s poles. Each o
he coils con ain app oxima ely N = 500 u ns o insula ed wi e and he wo coils a e
elec ically connec ed in se ies o inc ease he alue o he induced ol age. The passi e
senso assembly is ixed in a non-magne ic insula ing enclosu e.
The con igu a ion o he measu emen sys em shown in Figu e 3was ound o be
he mos con enien om a p ac ical poin o iew. Iden i ica ion o o o aul s uses a
combina ion o ac i e and passi e senso s o a ed 90 deg ela i e o each o he . The ac i e
senso gene a es he necessa y s a ic magne izing ield and senses i s change as he o o
o a es. The passi e senso cap u es he s ay ield nea he o o su ace.
Ene gies 2021, 14, x FOR PEER REVIEW 5 o 24
Figu e 3. Basic con igu a ion o he measu ing sys em wi h wo independen senso s.
The ol age ou pu s o bo h senso s a e ed o a DAQ ca d connec ed o a compu e .
The measu ed ol ages a e hen adjus ed, ampli ied, il e ed, and p ocessed by so wa e.
The modi ied signals a e hen analyzed and displayed g aphically o he ope a o o
e alua e.
3. 2D FEM Elec omagne ic Simula ion
A 2D FEM elec omagne ic model was used as a basic ool o e i y he unc ionali y
o he designed o o measu emen sys em. In he ini ial phase, he FEM model was used
o ind he mos sui able a angemen o senso s and o de e mine hei op imal posi ions
and dimensions. The FEM model con ains a o o shee wi h slo s in which he o o ba s
a e placed in isola ion, see Figu e 4.
Figu e 4. 2D FEM model o he es o o assembly wi h mesh in he senso egions.
The ac i e senso is loca ed a he bo om, while he passi e senso is loca ed on he
igh along he di ec ion o o o o a ion. In he ai gap and o o ba egions, he model
is buil wi h a p ecision mesh o elimina e alse induced ol age wa e o ms due o poo
Figu e 3. Basic con igu a ion o he measu ing sys em wi h wo independen senso s.
Ene gies 2021,14, 6573 5 o 23
The ol age ou pu s o bo h senso s a e ed o a DAQ ca d connec ed o a compu e . The
measu ed ol ages a e hen adjus ed, ampli ied, il e ed, and p ocessed by so wa e. The
modi ied signals a e hen analyzed and displayed g aphically o he ope a o o e alua e.
3. 2D FEM Elec omagne ic Simula ion
A 2D FEM elec omagne ic model was used as a basic ool o e i y he unc ionali y
o he designed o o measu emen sys em. In he ini ial phase, he FEM model was used
o ind he mos sui able a angemen o senso s and o de e mine hei op imal posi ions
and dimensions. The FEM model con ains a o o shee wi h slo s in which he o o ba s
a e placed in isola ion, see Figu e 4.
Ene gies 2021, 14, x FOR PEER REVIEW 5 o 24
Figu e 3. Basic con igu a ion o he measu ing sys em wi h wo independen senso s.
The ol age ou pu s o bo h senso s a e ed o a DAQ ca d connec ed o a compu e .
The measu ed ol ages a e hen adjus ed, ampli ied, il e ed, and p ocessed by so wa e.
The modi ied signals a e hen analyzed and displayed g aphically o he ope a o o
e alua e.
3. 2D FEM Elec omagne ic Simula ion
A 2D FEM elec omagne ic model was used as a basic ool o e i y he unc ionali y
o he designed o o measu emen sys em. In he ini ial phase, he FEM model was used
o ind he mos sui able a angemen o senso s and o de e mine hei op imal posi ions
and dimensions. The FEM model con ains a o o shee wi h slo s in which he o o ba s
a e placed in isola ion, see Figu e 4.
Figu e 4. 2D FEM model o he es o o assembly wi h mesh in he senso egions.
The ac i e senso is loca ed a he bo om, while he passi e senso is loca ed on he
igh along he di ec ion o o o o a ion. In he ai gap and o o ba egions, he model
is buil wi h a p ecision mesh o elimina e alse induced ol age wa e o ms due o poo
Figu e 4. 2D FEM model o he es o o assembly wi h mesh in he senso egions.
The ac i e senso is loca ed a he bo om, while he passi e senso is loca ed on he
igh along he di ec ion o o o o a ion. In he ai gap and o o ba egions, he model
is buil wi h a p ecision mesh o elimina e alse induced ol age wa e o ms due o poo
disc e iza ion. The FEM model con ains app oxima ely 60,000 elemen s and ansien ime
analysis was used o sol e he model.
The elec omagne ic model was sol ed in Ansys Maxwell, which also allows o de ine
an ex e nal elec ical ci cui wi h a di ec link o he elec omagne ic FEM model. The
ex e nal ci cui was used o desc ibe he squi el cage o he o o , see Figu e 5.
The ci cui diag am is ully enclosed and consis s o induc ances and esis ances. The
ci cui elemen s a e a anged o ep esen he elec ical coupling o he o o ba s (Rbx,
Lbx) wi h he end- ing segmen s (Re _xy, Le _xy) [
32
,
33
]. By changing he alues o he
ci cui elemen s, he cha ac e is ics o he ba s and end- ings can be desc ibed in he case o
a simula ed aul , e.g., a b oken ba , a ing segmen wi h educed conduc i i y, o o e all a
combina ion o aul s.
As he o o o a es, he cu en s in he ba s c ea e a eac ion ield ha dis o s he
o iginally symme ical ield o he magne iza ion yoke. Figu e 6shows he magne ic ield
dis ibu ion in he o o as he aul y ba passes by he wo senso s.
The aul y ba is shown he e in black colo . No ol age is induced in his b oken ba
and he cu en alue Iba = 0 A. No eac ion lux is hen gene a ed in he su oundings o
he ba and a pa o he magne ic lux passes h ough he adjacen oo h. A his poin , a
de o ma ion occu s in he o al magne ic ield, which is cap u ed by he sensing coil o he
ac i e senso and also by he coil o he passi e senso .

Ene gies 2021,14, 6573 6 o 23
Ene gies 2021, 14, x FOR PEER REVIEW 6 o 24
disc e iza ion. The FEM model con ains app oxima ely 60,000 elemen s and ansien
ime analysis was used o sol e he model.
The elec omagne ic model was sol ed in Ansys Maxwell, which also allows o de-
ine an ex e nal elec ical ci cui wi h a di ec link o he elec omagne ic FEM model. The
ex e nal ci cui was used o desc ibe he squi el cage o he o o , see Figu e 5.
Figu e 5. Equi alen elec ical ci cui o he o o cage.
The ci cui diag am is ully enclosed and consis s o induc ances and esis ances.
The ci cui elemen s a e a anged o ep esen he elec ical coupling o he o o ba s
(Rbx, Lbx) wi h he end- ing segmen s (Re _xy, Le _xy) [32,33]. By changing he alues o
he ci cui elemen s, he cha ac e is ics o he ba s and end- ings can be desc ibed in he
case o a simula ed aul , e.g., a b oken ba , a ing segmen wi h educed conduc i i y, o
o e all a combina ion o aul s.
As he o o o a es, he cu en s in he ba s c ea e a eac ion ield ha dis o s he
o iginally symme ical ield o he magne iza ion yoke. Figu e 6 shows he magne ic ield
dis ibu ion in he o o as he aul y ba passes by he wo senso s.
Figu e 6. Magne ic lux densi y dis ibu ion on he 2D FEM model o he es o o assembly.
The aul y ba is shown he e in black colo . No ol age is induced in his b oken ba
and he cu en alue Iba = 0 A. No eac ion lux is hen gene a ed in he su oundings o
he ba and a pa o he magne ic lux passes h ough he adjacen oo h. A his poin , a
Figu e 5. Equi alen elec ical ci cui o he o o cage.
Ene gies 2021, 14, x FOR PEER REVIEW 6 o 24
disc e iza ion. The FEM model con ains app oxima ely 60,000 elemen s and ansien
ime analysis was used o sol e he model.
The elec omagne ic model was sol ed in Ansys Maxwell, which also allows o de-
ine an ex e nal elec ical ci cui wi h a di ec link o he elec omagne ic FEM model. The
ex e nal ci cui was used o desc ibe he squi el cage o he o o , see Figu e 5.
Figu e 5. Equi alen elec ical ci cui o he o o cage.
The ci cui diag am is ully enclosed and consis s o induc ances and esis ances.
The ci cui elemen s a e a anged o ep esen he elec ical coupling o he o o ba s
(Rbx, Lbx) wi h he end- ing segmen s (Re _xy, Le _xy) [32,33]. By changing he alues o
he ci cui elemen s, he cha ac e is ics o he ba s and end- ings can be desc ibed in he
case o a simula ed aul , e.g., a b oken ba , a ing segmen wi h educed conduc i i y, o
o e all a combina ion o aul s.
As he o o o a es, he cu en s in he ba s c ea e a eac ion ield ha dis o s he
o iginally symme ical ield o he magne iza ion yoke. Figu e 6 shows he magne ic ield
dis ibu ion in he o o as he aul y ba passes by he wo senso s.
Figu e 6. Magne ic lux densi y dis ibu ion on he 2D FEM model o he es o o assembly.
The aul y ba is shown he e in black colo . No ol age is induced in his b oken ba
and he cu en alue Iba = 0 A. No eac ion lux is hen gene a ed in he su oundings o
he ba and a pa o he magne ic lux passes h ough he adjacen oo h. A his poin , a
Figu e 6. Magne ic lux densi y dis ibu ion on he 2D FEM model o he es o o assembly.
When he o o is o a ed by 90 deg, he b oken ba eaches he passi e senso . He e,
he magne ic lux lines a e locally closed h ough a egion ac oss he o o ee h ha
su ound he aul y ba and he con ac b idge o he closed slo . This local dis o ion o
he magne ic lux eappea s in he s ay ield and is again de ec ed by he passi e senso .
The induced ol age wa e o ms on he sensing coils o he ac i e and passi e senso s
o he FEM model a e shown in Figu e 7a. A ime window = (0.4–1.0) s is selec ed o
plo ing, whe e he ol age wa e o ms a e in s eady s a e al eady. The wa e o ms a e o se
in he g aph o a oid o e lapping and o make hem clea e . The ed colo co esponds o
he induced ol age o he ac i e senso , he blue colo o he induced ol age on he passi e
senso . The ol age ange o bo h signals is also no malized in he in e al (
−
1,1). The
ime wa e o m o bo h ol ages is no pu e sinusoidal bu shows a shape de o ma ion. The
de o ma ion is mo e isible in he ol age wa e o m o he passi e senso . A heal hy o o
wi hou de ec s shows a egula wa e o m o bo h induced ol ages wi hou de ia ions.
Figu e 7b shows he FFT analysis o bo h senso signals. The no malized ampli ude
spec um is shown, whe e he equency x = 84 Hz isibly domina es. The alue o his
equency co esponds o he se o o o a ion speed n = 180 pm and he numbe o i s
slo s (ba s) Nd = 28. The sidebands also appea and do no exceed he alue o
−
40 dB in
ampli ude. The sidebands a e much mo e signi ican in a passi e senso .
Ene gies 2021,14, 6573 7 o 23
Ene gies 2021, 14, x FOR PEER REVIEW 7 o 24
de o ma ion occu s in he o al magne ic ield, which is cap u ed by he sensing coil o
he ac i e senso and also by he coil o he passi e senso .
When he o o is o a ed by 90 deg, he b oken ba eaches he passi e senso . He e,
he magne ic lux lines a e locally closed h ough a egion ac oss he o o ee h ha
su ound he aul y ba and he con ac b idge o he closed slo . This local dis o ion o
he magne ic lux eappea s in he s ay ield and is again de ec ed by he passi e senso .
The induced ol age wa e o ms on he sensing coils o he ac i e and passi e sen-
so s o he FEM model a e shown in Figu e 7a. A ime window = (0.4–1.0) s is selec ed
o plo ing, whe e he ol age wa e o ms a e in s eady s a e al eady. The wa e o ms a e
o se in he g aph o a oid o e lapping and o make hem clea e . The ed colo co e-
sponds o he induced ol age o he ac i e senso , he blue colo o he induced ol age
on he passi e senso . The ol age ange o bo h signals is also no malized in he in e al
(−1,1). The ime wa e o m o bo h ol ages is no pu e sinusoidal bu shows a shape de-
o ma ion. The de o ma ion is mo e isible in he ol age wa e o m o he passi e senso .
A heal hy o o wi hou de ec s shows a egula wa e o m o bo h induced ol ages
wi hou de ia ions. Figu e 7b shows he FFT analysis o bo h senso signals. The no -
malized ampli ude spec um is shown, whe e he equency x = 84 Hz isibly domina es.
The alue o his equency co esponds o he se o o o a ion speed n = 180 pm and
he numbe o i s slo s (ba s) Nd = 28. The sidebands also appea and do no exceed he
alue o −40 dB in ampli ude. The sidebands a e much mo e signi ican in a passi e
senso .
(a)
(b)
Figu e 7. FEM model o a heal hy o o wi hou any de ec s: (a) no malized wa e o ms o he induced ol ages in he
senso s; (b) FFT analysis o induced ol age wa e o ms.
Figu e 8a shows he induced ol age wa e o ms o he FEM simula ion o 1 b oken
ba . The e is a egion on he induced ol age wa e o m o he ac i e senso whe e pa o
i is comple ely missing (g een a ow). This emp y egion is pe iodically epea ed a he
same ime in e als as he o o o a es. This indica es a aul in he ba , whe e he ol age
is induced, bu no elec ic cu en lows h ough he ba due o he missing elec ical
connec ion o he end- ing. The lack o a eac i e magne ic lux will c ea e a signi ican
dis o ion in he o he wise egula ol age wa e o m on he ac i e senso . On he ol age
wa e o m o he passi e senso , he in o ma ion abou he b oken o o ba appea s a
he same ime poin as o he ac i e senso ( 1 = 480 ms). Howe e , he ol age wa e o m
is di e en on he passi e senso due o i s di e en design na u e. Wi h a ime delay o
abou Δ = 85 ms (in ime 2 = 565 ms), he in o ma ion abou he b oken ba eappea s by
he oscilla ion o he induced ol age on he passi e senso . This is he poin a which he
b oken ba is close o he passi e senso and he o o has o a ed 90 deg (+7 ba dis-
placemen ).
Figu e 7.
FEM model o a heal hy o o wi hou any de ec s: (
a
) no malized wa e o ms o he induced ol ages in he
senso s; (b) FFT analysis o induced ol age wa e o ms.
Figu e 8a shows he induced ol age wa e o ms o he FEM simula ion o 1 b oken
ba . The e is a egion on he induced ol age wa e o m o he ac i e senso whe e pa
o i is comple ely missing (g een a ow). This emp y egion is pe iodically epea ed
a he same ime in e als as he o o o a es. This indica es a aul in he ba , whe e
he ol age is induced, bu no elec ic cu en lows h ough he ba due o he missing
elec ical connec ion o he end- ing. The lack o a eac i e magne ic lux will c ea e a
signi ican dis o ion in he o he wise egula ol age wa e o m on he ac i e senso . On
he ol age wa e o m o he passi e senso , he in o ma ion abou he b oken o o ba
appea s a he same ime poin as o he ac i e senso ( 1 = 480 ms). Howe e , he ol age
wa e o m is di e en on he passi e senso due o i s di e en design na u e. Wi h a ime
delay o abou
∆
= 85 ms (in ime 2 = 565 ms), he in o ma ion abou he b oken ba
eappea s by he oscilla ion o he induced ol age on he passi e senso . This is he poin
a which he b oken ba is close o he passi e senso and he o o has o a ed 90 deg
(+7 ba displacemen ).
Ene gies 2021, 14, x FOR PEER REVIEW 8 o 24
(a)
(b)
Figu e 8. FEM model o a o o wi h a b oken ba : (a) no malized wa e o ms o he induced ol ages in he senso s; (b)
FFT analysis o induced ol age wa e o ms.
Figu e 8b shows he FFT analysis o he ol ages o bo h senso s. In addi ion o he
main ha monic x = 84 Hz, he e a e signi ican sidebands wi h a equency sepa a ion o
Δ = 3 Hz indica ing a ba aul . Fo he ac i e senso , he signal o se is app oxima ely
−24 dB. The passi e senso is mo e sensi i e due o he wo de ec ed dis o ions and he
o se om he main ha monic is less han −18 dB.
In he FEM simula ion o a b oken end- ing segmen be ween he ba s, his de ec
again p esen s simila ly on bo h senso induced ol age wa e o ms. I can be seen in
Figu e 9a ha he b oken segmen be ween he ba s causes mul iple oscilla ions jus as i
passes o e he exci a ion poles o he ac i e senso . The wa e o m is hen u he de-
o med in ime by he DC componen .
(a)
(b)
Figu e 9. FEM model o a o o wi h a b oken end- ing segmen : (a) no malized wa e o ms o he induced ol ages in he
senso s; (b) FFT analysis o induced ol age wa e o ms.
A segmen ailu e is epea edly isible on he induced ol age wa e o m o he
passi e senso , simila o he case o a b oken ba . The in o ma ion abou he aul y
segmen is a he same ime ins an as o he ac i e senso ( 1 = 480 ms). I appea s as an
oscilla ion and a signi ican wa e o m dis o ion. Then, he ampli ude o he oscilla ion
dec eases and hen slowly inc eases un il he aul y segmen (2 adjacen ba s wi h a
aul y segmen ) is close o he passi e senso . He e again, he ampli ude o he induced
Figu e 8.
FEM model o a o o wi h a b oken ba : (
a
) no malized wa e o ms o he induced ol ages in he senso s; (
b
) FFT
analysis o induced ol age wa e o ms.
Figu e 8b shows he FFT analysis o he ol ages o bo h senso s. In addi ion o he
main ha monic x = 84 Hz, he e a e signi ican sidebands wi h a equency sepa a ion o
∆
= 3 Hz indica ing a ba aul . Fo he ac i e senso , he signal o se is app oxima ely
Ene gies 2021,14, 6573 8 o 23
−
24 dB. The passi e senso is mo e sensi i e due o he wo de ec ed dis o ions and he
o se om he main ha monic is less han −18 dB.
In he FEM simula ion o a b oken end- ing segmen be ween he ba s, his de ec
again p esen s simila ly on bo h senso induced ol age wa e o ms. I can be seen in
Figu e 9a ha he b oken segmen be ween he ba s causes mul iple oscilla ions jus as i
passes o e he exci a ion poles o he ac i e senso . The wa e o m is hen u he de o med
in ime by he DC componen .
Ene gies 2021, 14, x FOR PEER REVIEW 8 o 24
(a)
(b)
Figu e 8. FEM model o a o o wi h a b oken ba : (a) no malized wa e o ms o he induced ol ages in he senso s; (b)
FFT analysis o induced ol age wa e o ms.
Figu e 8b shows he FFT analysis o he ol ages o bo h senso s. In addi ion o he
main ha monic x = 84 Hz, he e a e signi ican sidebands wi h a equency sepa a ion o
Δ = 3 Hz indica ing a ba aul . Fo he ac i e senso , he signal o se is app oxima ely
−24 dB. The passi e senso is mo e sensi i e due o he wo de ec ed dis o ions and he
o se om he main ha monic is less han −18 dB.
In he FEM simula ion o a b oken end- ing segmen be ween he ba s, his de ec
again p esen s simila ly on bo h senso induced ol age wa e o ms. I can be seen in
Figu e 9a ha he b oken segmen be ween he ba s causes mul iple oscilla ions jus as i
passes o e he exci a ion poles o he ac i e senso . The wa e o m is hen u he de-
o med in ime by he DC componen .
(a)
(b)
Figu e 9. FEM model o a o o wi h a b oken end- ing segmen : (a) no malized wa e o ms o he induced ol ages in he
senso s; (b) FFT analysis o induced ol age wa e o ms.
A segmen ailu e is epea edly isible on he induced ol age wa e o m o he
passi e senso , simila o he case o a b oken ba . The in o ma ion abou he aul y
segmen is a he same ime ins an as o he ac i e senso ( 1 = 480 ms). I appea s as an
oscilla ion and a signi ican wa e o m dis o ion. Then, he ampli ude o he oscilla ion
dec eases and hen slowly inc eases un il he aul y segmen (2 adjacen ba s wi h a
aul y segmen ) is close o he passi e senso . He e again, he ampli ude o he induced
Figu e 9.
FEM model o a o o wi h a b oken end- ing segmen : (
a
) no malized wa e o ms o he induced ol ages in he
senso s; (b) FFT analysis o induced ol age wa e o ms.
A segmen ailu e is epea edly isible on he induced ol age wa e o m o he passi e
senso , simila o he case o a b oken ba . The in o ma ion abou he aul y segmen is a
he same ime ins an as o he ac i e senso ( 1 = 480 ms). I appea s as an oscilla ion and a
signi ican wa e o m dis o ion. Then, he ampli ude o he oscilla ion dec eases and hen
slowly inc eases un il he aul y segmen (2 adjacen ba s wi h a aul y segmen ) is close o
he passi e senso . He e again, he ampli ude o he induced ol age inc eases suddenly
(
2 = 565 ms
). Figu e 9b shows he FFT analysis o he ol ages o bo h senso s. Wi h he
main ha monic x = 84 Hz, signi ican sidebands wi h equency spacing
∆ =3Hz
appea .
Fo he ac i e senso , he sideband is concen a ed a ound he main ha monic and also
a ound he equency 2 = 28 Hz, which co esponds o he numbe o o o slo s. In a
passi e senso , he sidebands a e dis ibu ed o e he en i e displayed equency ange
(0–125 Hz). The lowes alue o he o se signal is in he o de o −21 dB.
The 2D FEM nume ical model desc ibes he p oblem only in he plane o he o o
sec ion and hus is no able o p o ide all he ele an in o ma ion, such as he o o slo
skew, he dis ibu ion o de ec s in he ba along he leng h, he axial o a ion o he senso s,
e c. The FEM simula ion o he induced ol age wa e o ms on he senso s showed ha he
basic p inciple o he designed induc i e senso s wo ks eliably. The measu emen sys em
is able o de ec a wide ange o o o manu ac u ing de ec s wi h su icien esolu ion.
Wi h he help o he pe o med elec omagne ic ield simula ions, he op imum loca ions
o senso placemen we e ound and also he op imiza ion o he ac i e and passi e senso
design was ealized.
4. Tes Ro o Assembly and Adjus men o Ro o Faul Ve i ica ion
To e i y he designed measu emen echnique and FEM simula ions, a es o o was
manu ac u ed. The es o o consis s o high g ade non-o ien ed elec o echnical shee s
ype M530-65A wi h hickness h = 0.65 mm and o al leng h o packe L = 150 mm. The
shee con ains Nd = 28 closed slo s in o which he b ass o o ba s a e inse ed. The o o
Ene gies 2021,14, 6573 9 o 23
shee s a e no p ocessed in any echnological way be o e assembling in o a packe and
we e andomly selec ed om he con aine o he cu ing machine.
The o o ba s a e made o b ass wi h a diame e o d = 4 mm and a e h eaded a
bo h ends. The end- ings a e milled om coppe shee wi h a hickness o 2 mm. All o o
ba s a e insula ed in he o o slo s by using insula ing ubes. Simila ly, he end- ings a e
sepa a ed om he o o by an insula ing pla e and a e no elec ically connec ed o he
o o shee s. Figu e 10 shows he indi idual pa s o he o o be o e i is assembled and
pa ial iew o he manu ac u ed es o o .
Ene gies 2021, 14, x FOR PEER REVIEW 9 o 24
ol age inc eases suddenly ( 2 = 565 ms). Figu e 9b shows he FFT analysis o he ol ages
o bo h senso s. Wi h he main ha monic x = 84 Hz, signi ican sidebands wi h equency
spacing Δ = 3 Hz appea . Fo he ac i e senso , he sideband is concen a ed a ound he
main ha monic and also a ound he equency 2 = 28 Hz, which co esponds o he
numbe o o o slo s. In a passi e senso , he sidebands a e dis ibu ed o e he en i e
displayed equency ange (0–125 Hz). The lowes alue o he o se signal is in he o de
o −21 dB.
The 2D FEM nume ical model desc ibes he p oblem only in he plane o he o o
sec ion and hus is no able o p o ide all he ele an in o ma ion, such as he o o slo
skew, he dis ibu ion o de ec s in he ba along he leng h, he axial o a ion o he sen-
so s, e c. The FEM simula ion o he induced ol age wa e o ms on he senso s showed
ha he basic p inciple o he designed induc i e senso s wo ks eliably. The measu e-
men sys em is able o de ec a wide ange o o o manu ac u ing de ec s wi h su icien
esolu ion. Wi h he help o he pe o med elec omagne ic ield simula ions, he op i-
mum loca ions o senso placemen we e ound and also he op imiza ion o he ac i e
and passi e senso design was ealized.
4. Tes Ro o Assembly and Adjus men o Ro o Faul Ve i ica ion
To e i y he designed measu emen echnique and FEM simula ions, a es o o
was manu ac u ed. The es o o consis s o high g ade non-o ien ed elec o echnical
shee s ype M530-65A wi h hickness h = 0.65 mm and o al leng h o packe L = 150 mm.
The shee con ains Nd = 28 closed slo s in o which he b ass o o ba s a e inse ed. The
o o shee s a e no p ocessed in any echnological way be o e assembling in o a packe
and we e andomly selec ed om he con aine o he cu ing machine.
The o o ba s a e made o b ass wi h a diame e o d = 4 mm and a e h eaded a
bo h ends. The end- ings a e milled om coppe shee wi h a hickness o 2 mm. All o o
ba s a e insula ed in he o o slo s by using insula ing ubes. Simila ly, he end- ings a e
sepa a ed om he o o by an insula ing pla e and a e no elec ically connec ed o he
o o shee s. Figu e 10 shows he indi idual pa s o he o o be o e i is assembled and
pa ial iew o he manu ac u ed es o o .
Figu e 10. The es o o se wi h indi idual componen s.
The o o ba s a e ixed o he end- ings wi h nu s. The o e all o o a angemen
was chosen o make aul simula ion easie . The o o allows es ing, o example, a
change in he conduc i i y o he end- ing, a b eak in he end- ing o pa o he end- ing,
a educed conduc i i y o he ba , a b oken o o ba o a combina ion o hese aul s. A
basic o o measu ing se up wi h senso a angemen is shown in Figu e 11.
Figu e 10. The es o o se wi h indi idual componen s.
The o o ba s a e ixed o he end- ings wi h nu s. The o e all o o a angemen was
chosen o make aul simula ion easie . The o o allows es ing, o example, a change in
he conduc i i y o he end- ing, a b eak in he end- ing o pa o he end- ing, a educed
conduc i i y o he ba , a b oken o o ba o a combina ion o hese aul s. A basic o o
measu ing se up wi h senso a angemen is shown in Figu e 11.
Ene gies 2021, 14, x FOR PEER REVIEW 10 o 24
Figu e 11. O e all iew o he measu ing sys em wi h es o o .
The o o o a ion is ensu ed by he NEMA17 s epped mo o ope a ed by a d i e
ia he A duino module. The mo o o a ion speed is ixed o n = 180 pm (3 ps), and i
needs abou 10 s o each he se speed in linea accele a ion mode. The s epped mo o is
a ached o he o o by iming pulleys and a ubbe iming bel o elimina e he mo o
ib a ions as well as o ensu e exac e olu ion wi hou slip. The sha wi h he es o o
is placed in bea ing housing.
Figu e 12 shows selec ed modi ica ions o he es o o end- ing o simula e i s
ailu e. Figu e 12 shows he se -up o simula ing he conduc i i y o he ing. He e, he
o iginal solid c oss-sec ion o he end- ing is eplaced by a se o coppe pla es. The
numbe o shee s de e mines he inal alue o he conduc i i y. Figu e 12 also shows a
simula ion o a b oken segmen o he end- ing be ween he ba s. The ai gap causes a
disconnec ion o he ini ially solid ing. The o o ba s emain conduc i ely connec ed o
he es o he ing.
Figu e 12. Cons uc ion adjus men s o he es o o wi h o e all ing conduc i i y dec ease and
disconnec ing in he end- ing segmen be ween he ba s.
To simula e he de ec s o he o o ba s, geome ic modi ica ions we e made o hei
c oss-sec ion as shown in Figu e 13. Many o he au ho s [18,21,27] use di ec d illing o
o o ba s o c ea e a de ec . In ou case, he educ ion o he ac i e c oss-sec ion o he ba
is achie ed by u ning he c oss-sec ion a abou 1/5 o i s o al leng h. The ba s so p e-
pa ed a e sequen ially inse ed in o he slo and he e ec o he conduc i i y d op can be
measu ed.
Figu e 11. O e all iew o he measu ing sys em wi h es o o .
The o o o a ion is ensu ed by he NEMA17 s epped mo o ope a ed by a d i e ia
he A duino module. The mo o o a ion speed is ixed o n = 180 pm (3 ps), and i needs
abou 10 s o each he se speed in linea accele a ion mode. The s epped mo o is a ached
o he o o by iming pulleys and a ubbe iming bel o elimina e he mo o ib a ions as
well as o ensu e exac e olu ion wi hou slip. The sha wi h he es o o is placed in
bea ing housing.
Figu e 12 shows selec ed modi ica ions o he es o o end- ing o simula e i s ailu e.
Figu e 12 shows he se -up o simula ing he conduc i i y o he ing. He e, he o iginal
solid c oss-sec ion o he end- ing is eplaced by a se o coppe pla es. The numbe o
shee s de e mines he inal alue o he conduc i i y. Figu e 12 also shows a simula ion o
a b oken segmen o he end- ing be ween he ba s. The ai gap causes a disconnec ion o
he ini ially solid ing. The o o ba s emain conduc i ely connec ed o he es o he ing.
Ene gies 2021,14, 6573 16 o 23
Figu e 22 shows he change in he o e all de o ma ion o he o al pa e n when a
ailu e occu s in a segmen o he end- ing. The change in he conduc i i y o he connec ion
be ween he wo ba s o he ini ially sepa a ed segmen was simula ed. As he conduc i i y
o he segmen dec eases, he p opo ions o he o al emain almos iden ical, bu i s
inclina ion in he symme y longi udinal axis changes signi ican ly. Wi h a ully b oken
segmen be ween he ba s, see Figu e 22d, he plo ing poin mo es along a se o ellip ic
cu es ha a e o a ed ela i e o each o he . The measu emen sensi i i y is lowe in his
case, which is also e iden om he FFT analysis o he spec a. This aul is be e isible
in he passi e senso , which shows a as e inc ease in sideband ampli udes.
Ene gies 2021, 14, x FOR PEER REVIEW 17 o 24
Figu e 21. Compa ison o FFT and o al pa e ns o he selec ed numbe o b oken ba s: (a)
Re e ence o al pa e n o o o wi hou de ec ; (b) 1 B oken ba ; (c) 2 B oken ba s; (d) 3 B oken
ba s.
Figu e 22 shows he change in he o e all de o ma ion o he o al pa e n when a
ailu e occu s in a segmen o he end- ing. The change in he conduc i i y o he con-
nec ion be ween he wo ba s o he ini ially sepa a ed segmen was simula ed. As he
conduc i i y o he segmen dec eases, he p opo ions o he o al emain almos iden i-
cal, bu i s inclina ion in he symme y longi udinal axis changes signi ican ly. Wi h a
ully b oken segmen be ween he ba s, see Figu e 22d, he plo ing poin mo es along a
se o ellip ic cu es ha a e o a ed ela i e o each o he . The measu emen sensi i i y is
lowe in his case, which is also e iden om he FFT analysis o he spec a. This aul is
be e isible in he passi e senso , which shows a as e inc ease in sideband ampli udes.
Figu e 22.
Compa ison o FFT and o al pa e ns o ailu es in segmen o end- ing: (
a
) Re e -
ence o al pa e n o o o wi hou de ec ; (
b
) 33% C oss-Sec ion Loss; (
c
) 67% C oss-Sec ion Loss;
(d) 100% C oss-Sec ion Loss.

Ene gies 2021,14, 6573 17 o 23
Combined o o ailu es a e shown in Figu e 23. A b oken ba in combina ion wi h a
b oken segmen o he end- ing o mul iple ailu es such as 2 b oken ba s wi h a b oken
segmen , e c., cause signi ican de o ma ion o he o iginally smoo h o al pa e n. The
plo ing poin g adually sepa a es om he o iginal o al pa e n in mul iple sequen ial
s eps. The pa e ns also il and o m in e nal loops. Thei inal o m is in ac a composi e
o he pa e ns ypical o he elemen a y aul s in Figu es 21 and 22. The sidebands in
he FFT analysis comple ely ill he space o he le and igh o he dominan ha monic
x = 84 Hz and illus a e he signi icance o he aul simula ed on he es o o .
Ene gies 2021, 14, x FOR PEER REVIEW 18 o 24
Figu e 22. Compa ison o FFT and o al pa e ns o ailu es in segmen o end- ing: (a) Re e ence
o al pa e n o o o wi hou de ec ; (b) 33% C oss-Sec ion Loss; (c) 67% C oss-Sec ion Loss; (d)
100% C oss-Sec ion Loss.
Combined o o ailu es a e shown in Figu e 23. A b oken ba in combina ion wi h a
b oken segmen o he end- ing o mul iple ailu es such as 2 b oken ba s wi h a b oken
segmen , e c., cause signi ican de o ma ion o he o iginally smoo h o al pa e n. The
plo ing poin g adually sepa a es om he o iginal o al pa e n in mul iple sequen ial
s eps. The pa e ns also il and o m in e nal loops. Thei inal o m is in ac a composi e
o he pa e ns ypical o he elemen a y aul s in Figu es 21 and 22. The sidebands in he
FFT analysis comple ely ill he space o he le and igh o he dominan ha monic x =
84 Hz and illus a e he signi icance o he aul simula ed on he es o o .
Figu e 23. Compa ison o FFT and o al pa e ns o combined o o ailu es: (a) 1 b oken ba + 1
b oken end- ing segmen ; (b) 1 b oken ba + 2 b oken end- ing segmen s; (c) 2 b oken ba s + 1
b oken end- ing segmen .
6. O -Line Measu emen o Real Squi el Cage Ro o s a e Cas ing P ocess
The designed measu emen p ocedu e and senso con igu a ion was e i ied on
se e al eal o o p o o ypes. Figu e 24 shows he basic a angemen o he measu emen
sys em wi h he o o ixed be ween he aluminum conical cen e ing ips. Figu e 24 also
Figu e 23.
Compa ison o FFT and o al pa e ns o combined o o ailu es: (
a
)
1 b oken ba + 1 b oken
end- ing segmen ; (
b
) 1 b oken ba + 2 b oken end- ing segmen s; (
c
) 2 b oken ba s + 1 b oken end- ing
segmen .
6. O -Line Measu emen o Real Squi el Cage Ro o s a e Cas ing P ocess
The designed measu emen p ocedu e and senso con igu a ion was e i ied on
se e al eal o o p o o ypes. Figu e 24 shows he basic a angemen o he measu emen
sys em wi h he o o ixed be ween he aluminum conical cen e ing ips. Figu e 24 also
shows ep esen a i e o o samples p epa ed o de ec de ec ion. The o o samples a e
unmachined, wi hou sha s, and we e collec ed a e cas ing and cooling.
Ene gies 2021,14, 6573 18 o 23
Ene gies 2021, 14, x FOR PEER REVIEW 19 o 24
shows ep esen a i e o o samples p epa ed o de ec de ec ion. The o o samples a e
unmachined, wi hou sha s, and we e collec ed a e cas ing and cooling.
Figu e 24. Measu emen sys em o eal o o es ing.
Measu emen s we e in gene al he same as in he p e ious es o o . Howe e , he
main ask was di e en and was o use he isualized o al pa e n o de e mine he aul
in he end- ing o o o ba s. Mos o he es ed o o s ep esen ed by sample R1 dis-
played a pa e n simila o ha shown in Figu e 25. The o al shape is ela i ely smoo h,
i s hickness being a ec ed by eccen ici y. FFT analysis indica es only a hin sideband
concen a ed a ound he dominan ha monic x = 84 Hz. The measu ed p o o ype o o s
con ain Nd = 20 ully closed slo s, and he pm had o be se highe han he es o o .
The equency band o se hus eaches Δ = 4.2 Hz.
Figu e 25. Ou pu s o a o o R1 (no de ec ).
In he g oup o o o s, a sample labeled R2 was ound ha showed a di e en o al
shape, see Figu e 26. FFT analysis indica ed he appea ance o a sideband in he lowe
equency ange. The o al pa e n is sepa a ed and he plo ing poin ci cula es along
mu ually il ed ellip ical shapes. F om his ou pu , we assumed ha his is a aul in he
end- ing.
Figu e 24. Measu emen sys em o eal o o es ing.
Measu emen s we e in gene al he same as in he p e ious es o o . Howe e ,
he main ask was di e en and was o use he isualized o al pa e n o de e mine he
aul in he end- ing o o o ba s. Mos o he es ed o o s ep esen ed by sample R1
displayed a pa e n simila o ha shown in Figu e 25. The o al shape is ela i ely smoo h,
i s hickness being a ec ed by eccen ici y. FFT analysis indica es only a hin sideband
concen a ed a ound he dominan ha monic x = 84 Hz. The measu ed p o o ype o o s
con ain
Nd = 20
ully closed slo s, and he pm had o be se highe han he es o o .
The equency band o se hus eaches ∆ = 4.2 Hz.
Ene gies 2021, 14, x FOR PEER REVIEW 19 o 24
shows ep esen a i e o o samples p epa ed o de ec de ec ion. The o o samples a e
unmachined, wi hou sha s, and we e collec ed a e cas ing and cooling.
Figu e 24. Measu emen sys em o eal o o es ing.
Measu emen s we e in gene al he same as in he p e ious es o o . Howe e , he
main ask was di e en and was o use he isualized o al pa e n o de e mine he aul
in he end- ing o o o ba s. Mos o he es ed o o s ep esen ed by sample R1 dis-
played a pa e n simila o ha shown in Figu e 25. The o al shape is ela i ely smoo h,
i s hickness being a ec ed by eccen ici y. FFT analysis indica es only a hin sideband
concen a ed a ound he dominan ha monic x = 84 Hz. The measu ed p o o ype o o s
con ain Nd = 20 ully closed slo s, and he pm had o be se highe han he es o o .
The equency band o se hus eaches Δ = 4.2 Hz.
Figu e 25. Ou pu s o a o o R1 (no de ec ).
In he g oup o o o s, a sample labeled R2 was ound ha showed a di e en o al
shape, see Figu e 26. FFT analysis indica ed he appea ance o a sideband in he lowe
equency ange. The o al pa e n is sepa a ed and he plo ing poin ci cula es along
mu ually il ed ellip ical shapes. F om his ou pu , we assumed ha his is a aul in he
end- ing.
Figu e 25. Ou pu s o a o o R1 (no de ec ).
In he g oup o o o s, a sample labeled R2 was ound ha showed a di e en o al
shape, see Figu e 26. FFT analysis indica ed he appea ance o a sideband in he lowe
equency ange. The o al pa e n is sepa a ed and he plo ing poin ci cula es along
mu ually il ed ellip ical shapes. F om his ou pu , we assumed ha his is a aul in
he end- ing.
Ene gies 2021, 14, x FOR PEER REVIEW 19 o 24
shows ep esen a i e o o samples p epa ed o de ec de ec ion. The o o samples a e
unmachined, wi hou sha s, and we e collec ed a e cas ing and cooling.
Figu e 24. Measu emen sys em o eal o o es ing.
Measu emen s we e in gene al he same as in he p e ious es o o . Howe e , he
main ask was di e en and was o use he isualized o al pa e n o de e mine he aul
in he end- ing o o o ba s. Mos o he es ed o o s ep esen ed by sample R1 dis-
played a pa e n simila o ha shown in Figu e 25. The o al shape is ela i ely smoo h,
i s hickness being a ec ed by eccen ici y. FFT analysis indica es only a hin sideband
concen a ed a ound he dominan ha monic x = 84 Hz. The measu ed p o o ype o o s
con ain Nd = 20 ully closed slo s, and he pm had o be se highe han he es o o .
The equency band o se hus eaches Δ = 4.2 Hz.
Figu e 25. Ou pu s o a o o R1 (no de ec ).
In he g oup o o o s, a sample labeled R2 was ound ha showed a di e en o al
shape, see Figu e 26. FFT analysis indica ed he appea ance o a sideband in he lowe
equency ange. The o al pa e n is sepa a ed and he plo ing poin ci cula es along
mu ually il ed ellip ical shapes. F om his ou pu , we assumed ha his is a aul in he
end- ing.
Figu e 26. Ou pu s o a o o R2 (de ec in end- ing).
The assump ion was inally con i med by wo me hods, see Figu e 27. X- ay pho-
og aphy shows he p esence o ai bubbles ha almos comple ely ill he olume o he
Ene gies 2021,14, 6573 19 o 23
end- ing. The des uc i e me hod o cu ing h ough he end- ing illus a ed he scale o
his de ec , which signi ican ly educes he conduc i i y o he end- ing segmen s.
Ene gies 2021, 14, x FOR PEER REVIEW 20 o 24
Figu e 26. Ou pu s o a o o R2 (de ec in end- ing).
The assump ion was inally con i med by wo me hods, see Figu e 27. X- ay pho-
og aphy shows he p esence o ai bubbles ha almos comple ely ill he olume o he
end- ing. The des uc i e me hod o cu ing h ough he end- ing illus a ed he scale o
his de ec , which signi ican ly educes he conduc i i y o he end- ing segmen s.
Figu e 27. X- ay pho og aphy o es ed o o R2 and mechanical cu o end- ing.
The o o labeled R3 p esen s a sample wi h a combined aul in he o o ba s, see
Figu e 28. The in e nal c ossing o he ajec o y o he plo ing poin is cha ac e is ic o a
aul in he o o ba s.
Figu e 28. Ou pu s o a o o R3 (mul iple de ec ed o o ba s).
The FFT analysis indica es a signi ican p esence o sidebands on bo h sides o he
dominan ha monic and also a be e sensi i i y o he passi e senso o his ype o aul .
A cu made a 1/3 o he o o leng h exposed ai bubbles and deepe holes in se e al
o o ba s, see Figu e 29.
Figu e 29. Cu in 1/3 o he leng h o he sample R3.
Figu e 27. X- ay pho og aphy o es ed o o R2 and mechanical cu o end- ing.
The o o labeled R3 p esen s a sample wi h a combined aul in he o o ba s, see
Figu e 28. The in e nal c ossing o he ajec o y o he plo ing poin is cha ac e is ic o a
aul in he o o ba s.
Ene gies 2021, 14, x FOR PEER REVIEW 20 o 24
Figu e 26. Ou pu s o a o o R2 (de ec in end- ing).
The assump ion was inally con i med by wo me hods, see Figu e 27. X- ay pho-
og aphy shows he p esence o ai bubbles ha almos comple ely ill he olume o he
end- ing. The des uc i e me hod o cu ing h ough he end- ing illus a ed he scale o
his de ec , which signi ican ly educes he conduc i i y o he end- ing segmen s.
Figu e 27. X- ay pho og aphy o es ed o o R2 and mechanical cu o end- ing.
The o o labeled R3 p esen s a sample wi h a combined aul in he o o ba s, see
Figu e 28. The in e nal c ossing o he ajec o y o he plo ing poin is cha ac e is ic o a
aul in he o o ba s.
Figu e 28. Ou pu s o a o o R3 (mul iple de ec ed o o ba s).
The FFT analysis indica es a signi ican p esence o sidebands on bo h sides o he
dominan ha monic and also a be e sensi i i y o he passi e senso o his ype o aul .
A cu made a 1/3 o he o o leng h exposed ai bubbles and deepe holes in se e al
o o ba s, see Figu e 29.
Figu e 29. Cu in 1/3 o he leng h o he sample R3.
Figu e 28. Ou pu s o a o o R3 (mul iple de ec ed o o ba s).
The FFT analysis indica es a signi ican p esence o sidebands on bo h sides o he
dominan ha monic and also a be e sensi i i y o he passi e senso o his ype o aul .
A cu made a 1/3 o he o o leng h exposed ai bubbles and deepe holes in se e al o o
ba s, see Figu e 29.
Ene gies 2021, 14, x FOR PEER REVIEW 20 o 24
Figu e 26. Ou pu s o a o o R2 (de ec in end- ing).
The assump ion was inally con i med by wo me hods, see Figu e 27. X- ay pho-
og aphy shows he p esence o ai bubbles ha almos comple ely ill he olume o he
end- ing. The des uc i e me hod o cu ing h ough he end- ing illus a ed he scale o
his de ec , which signi ican ly educes he conduc i i y o he end- ing segmen s.
Figu e 27. X- ay pho og aphy o es ed o o R2 and mechanical cu o end- ing.
The o o labeled R3 p esen s a sample wi h a combined aul in he o o ba s, see
Figu e 28. The in e nal c ossing o he ajec o y o he plo ing poin is cha ac e is ic o a
aul in he o o ba s.
Figu e 28. Ou pu s o a o o R3 (mul iple de ec ed o o ba s).
The FFT analysis indica es a signi ican p esence o sidebands on bo h sides o he
dominan ha monic and also a be e sensi i i y o he passi e senso o his ype o aul .
A cu made a 1/3 o he o o leng h exposed ai bubbles and deepe holes in se e al
o o ba s, see Figu e 29.
Figu e 29. Cu in 1/3 o he leng h o he sample R3.
Figu e 29. Cu in 1/3 o he leng h o he sample R3.
Howe e , his does no au oma ically imply ha he p esence o ailu es in his c oss-
sec ion is comple e. Acco ding o he o al pa e n, he e a e p obably mo e damaged ba s,
o ba s wi h a high con en o ai bubbles o holes, in he es ed o o R3.
Ene gies 2021,14, 6573 20 o 23
7. Discussion
The p oposed sys em o aul de ec ion by using a g aphical pa e n is based on he
p inciple o ol age induc ion in a conduc o (ba ) mo ing in a s a ic magne ic ield. The
2-senso s measu emen s we e used o de e mine he quali y o he manu ac u ed o o .
The o e all summa y o he ad an ages and disad an ages o he me hodology used is
o ganized in he ollowing sec ions:
(A)
Senso design
The magne ic ield is gene a ed by pe manen magne s in he ac i e senso enclosu e.
The ad an age o his solu ion is he absence o exci a ion cu en , which is o he wise
necessa y when using a classical g owle yoke. To su icien ly magne ize he o o ee h,
a he s ong NdFeB magne s a e needed. The exci a ion canno be adjus ed, he e o e
he ai gap be ween he exci a ion yoke and he o o su ace mus be ela i ely small.
This makes he exci a ion sensi i e o s icking o e omagne ic pa icles and sawdus . In
labo a o y condi ions his is no a p oblem in p inciple, bu i he measu ing sys em is
implemen ed in a p oduc ion cycle, i mus be emembe ed ha sawdus can e en ually
c ea e a magne ic sho ci cui be ween he poles o he ac i e senso .
(B)
Tes sys em assembly
I is impo an o keep in mind he in luence o he su ounding en i onmen . The
e omagne ic mass enclosing he o o o be measu ed can cause high dis o ion in he
sensed signals. In ou case, he massi e clamping head showed o be he bigges p oblem.
The clamping sys em has o be made wi h non-magne ic ods ha a e a enough away
om he ac ual o o body.
(C)
Calib a ion o senso s using a es o o
The adjus men and alida ion o he me hodology used was e i ied on a es o o .
The design o he es o o should be as close as possible o he eal o o . This was no
ully me in ou case. The eal squi el cage o o does no con ain isola ed o o ba s
as in he case o he es o o . Du ing die cas ing, he o o slo s a e illed wi h mol en
aluminum and he o o ba comes in o di ec con ac wi h he o o pla es. The e is a
high p obabili y o in e ba cu en s, whe e he elec ic cu en lows pa ially h ough he
o o pla es and can bypass a mino de ec inside he ba . The c oss-sec ion o he ba in a
eal o o is no ci cula . The eal o o con ains a double cage, see Figu e 29. In addi ion,
he hickness o he end- ing o he eal o o is la ge, see Figu e 24. I he ai bubbles
occupy a signi ican pa o he end- ing c oss-sec ion, hen his de ec is qui e de ec able.
Howe e , i he po osi y o he ai bubbles is smalle , he p oposed measu emen sys em
shows lowe sensi i i y.
(D)
C i e ia o e alua ing pa e ns, sensi i i y
Fo a eal o o wi hou de ec s, he ajec o y o he plo ing poin in X-Y mode o ms a
egula and smoo h o al. The shape and dimensional p opo ions o his o al a e iden ical
o iden ical o o s wi hou de ec s. Such a o o hen plo s a ajec o y in X-Y mode which
is he e e ence o o he measu ed o o s o iden ical design. The ou e and inne bo de
o he o al is de ined by he ajec o y o he plo ing poin , which uns in a speci ic ange
a ound i s pe ime e . This p oduces he ypical wid h o he o al, he magni ude o which is
de e mined by he eccen ici y o a o o wi hou de ec s. The eccen ici y can be adjus ed
by using clamping conical ips inse ed in o he sha bo e and e i ied on he induced
ol age ime his o ies included in he es window.
Faul iden i ica ion is based on g aphical and isual compa ison o he plo ed pa e n
wi h a e e ence o al shape. The isual assessmen is (i) he inc ease in he wid h o he
o al om he e e ence pa e n, (ii) he sepa a ion o he ajec o y o he plo ing poin
om he e e ence wid h o he o al, (iii) he endency o he plo ing poin o o m in e nal
loops inside he o al, (i ) he numbe o in e nal loops o med.
Ene gies 2021,14, 6573 21 o 23
In he case o a mino de ec in he o o ba , he de o ma ion o he plo ing poin
ajec o y may no be comple ely e iden o an accu a e speci ica ion o he ype o ailu e.
By looking a he FFT analysis window, which is also pa o he es panel, he o o aul
can be speci ied mo e p ecisely. A double-sided sideband indica es a aul in he ba , a
le -sided sideband ep esen s a aul in he ing. Un o una ely, he sensi i i y o de ec s in
he end- ing is less o he ab ica ed es sys em han o de ec s in he ba s.
(E)
Measu emen ou pu s om es and eal o o s
As can be seen om Figu e 20b, a de ec in he od (ai bubble, educed conduc i i y)
is only de ec able a app oxima ely 25% loss in od c oss-sec ion. I was obse ed ha a
lowe aul anges <25% o od c oss-sec ion loss, he pa e n only sligh ly expands and
he aul esolu ion is wo se. The aul is much mo e iden i iable on FFT analysis o bo h
signals. In he case o a comple ely c acked ba o se o ba s, he pa e ns on Figu e 21b–d
show a signi ican de ia ion o he plo ing poin om he o iginal o al shape and hese
ypes o aul s a e e y well iden i iable. In he second phase o e i ica ion, he de ec s
in he end- ing we e measu ed. Again, i is e iden ha he ange o damage (po osi y)
had o achie e a leas 30% c oss sec ion loss o his de ec o be easonably isible, see
Figu e 22b.
Compa ing he ba and end- ing de ec , he di e ence in he ajec o y o he plo ing
poin is e iden . Wi h he ba aul , he hickness o he o al only sligh ly expands, and he
ajec o y o he plo ing poin o ms an in e nal c ossing. The o al g adually sepa a es in
he uppe le pa . In an end- ing aul , he indi idual o als a e sepa a ed and sligh ly
o a ed. The e ec o he segmen po osi y is e lec ed by he ajec o y de ia ion in he
lowe pa o he o al. The passi e senso is loca ed app oxima ely in he middle o
he o o leng h du ing he measu emen , and hus ac ually cap u es an image o he
ins an aneous alue o he cu en in he ba de e mined by he aul condi ion in he
end- ing segmen . The de ec abili y o a aul in he end- ing can be imp o ed i he passi e
senso is loca ed close o he end- ing and o ien ed pa allel o he axis o o a ion o he
o o being measu ed.
Fo a eal o o wi hou de ec s, he ajec o y o he ende ing poin in X-Y mode
o ms a egula and smoo h o al, see Figu e 25. The de ec in he eal o o can no longe be
explici ly se , as in he case o he es o o , bu he aul iden i ica ion is pe o med di ec ly.
As can be seen om Figu es 26 and 28, he de ec s in he eal o o a e mo e likely o occu
in a combined o m. The o al in Figu e 26 ends o inc ease in hickness and he e is only a
sligh c ossing o he plo ing poin ajec o y. The equency band is cen e ed o he le o
he dominan ha monic x = 84 Hz. The shape o he o al shows a cha ac e consis en wi h
an end- ing de ec . A se ies o c ossings and inne loops appea in Figu e 28. The loops in
he cen e o he o al a e smoo h and do no con ain sha p shapes. The ba s in he o o
will no be comple ely in e up ed, bu conside able po osi y can be expec ed. One o he
ods has a e y signi ican ly educed conduc i i y. The equency band appea s on ei he
side o he dominan ha monic x = 84 Hz.
8. Conclusions
The pape deal wi h o -line diagnos ics o he o o o an asynch onous machine
using wo senso s. The unc ionali y o he whole sys em o measu ing and iden i ying
aul s in he o o was i s ly e i ied using a 2D FEM elec omagne ic model. The FEM
analysis was also used o ind he app op ia e a angemen o he senso s and hei op imal
posi ion nea he o o . To alida e he measu emen sys em, a es o o was buil o
simula e he speci ic ypes o aul s and hei possible combina ions ha can occu in
squi el cage o o s. The e alua ion sys em includes a g aphical aul isualiza ion in e ace
ha displays ol age wa e o ms, FFT signal analysis, and o al pa e ns and is used o
iden i y speci ic o o aul s. I was impo an o ind cha ac e is ic pa e n shapes whose
shape co esponds o he simula ed aul . The g aphical ou pu s o he pa e ns se e as
addi ional in o ma ion o he ope a o o e alua e he quali y o he manu ac u ed o o . I
he plo poin de ia es signi ican ly om he e e ence shape o he egula o al, he o o

Ene gies 2021,14, 6573 22 o 23
can be conside ed as de ec i e and can be emo ed om he nex p oduc ion p ocess. The
inal decision can be made by he ope a o o by he au oma ed sys em.
Au ho Con ibu ions:
Concep ualiza ion, P.K. and P.B.; o mal analysis, P.B. and P.M.; in es iga ion,
P.B. and P.M.; me hodology, P.K. and P.B.; so wa e, P.K.; supe ision, P.K. and P.B.; alida ion, P.K.;
isualiza ion, P.K., P.B. and P.M.; w i ing—o iginal d a , P.K.; w i ing— e iew and edi ing, P.K. and
P.M. All au ho s ha e ead and ag eed o he published e sion o he manusc ip .
Funding:
This esea ch was unded by in e nal unding p og am SGS o he VSB—Technical Uni e -
si y o Os a a, iden i ica ion no: SP2021/20.
Ins i u ional Re iew Boa d S a emen : No applicable.
In o med Consen S a emen : No applicable.
Da a A ailabili y S a emen : No applicable.
Con lic s o In e es : The au ho s decla e no con lic o in e es .
Re e ences
1.
Yamamo o, S. O e iew o he la es esea ch and de elopmen o coppe die-cas squi el-cage o o s. In P oceedings o he
2018 In e na ional Powe Elec onics Con e ence (IPEC-Niiga a 2018 -ECCE Asia), Niiga a, Japan, 20–24 May 2018; pp. 1949–1954.
[C ossRe ]
2.
Bonne , A.H.; Soukup, G.C. Cause and analysis o s a o and o o ailu es in h ee-phase squi el-cage induc ion mo o s. IEEE
T ans. Ind. Appl. 1992,28, 921–937. [C ossRe ]
3. Bishop, T. Squi el cage o o es ing. In P oceedings o he EASA Con en ion, San F ancisco, CA, USA, 30 June 2003.
4.
Cla k, S.W.; S e ens, D. Induc ion mo o o o ba damage e alua ion wi h magne ic ield analysis. IEEE T ans. Ind. Appl.
2016
,
52, 1469–1476. [C ossRe ]
5.
Lopez-Pe ez, D.; An onino-Da iu, J. Applica ion o in a ed he mog aphy o ailu e de ec ion in indus ial induc ion mo o s:
Case s o ies. IEEE T ans. Ind. Appl. 2017,53, 1901–1908. [C ossRe ]
6.
Bai, T.; Zhang, L.; Duan, L.; Wang, J. NSCT-based in a ed image enhancemen me hod o o a ing machine y aul diagnosis.
IEEE T ans. Ins um. Meas. 2016,65, 2293–2301. [C ossRe ]
7.
Fe ucho-Al a ez, E.R.; Cabal-Yepez, E.; Ledesma-Ca illo, L.M.; Ma inez-He e a, A.L. B oken o o ba de ec ion by image
ex u e ea u es and uzzy logic. In P oceedings o he IECON 2019–45 h Annual Con e ence o he IEEE Indus ial Elec onics
Socie y, Lisbon, Po ugal, 14–17 Oc obe 2019; pp. 934–938. [C ossRe ]
8.
Yun, J.; Lee, S.; Jeong, M.; Lee, S.B. In luence o die-cas o o ill ac o on he s a ing pe o mance o induc ion machines. IEEE
T ans. Magn. 2018,54, 1–4. [C ossRe ]
9.
Va ghese, S.T.; Singh, B.; Rajagopal, K.R. Faul in es iga ions on die-cas coppe o o s. IEEE T ans. Ind. Appl.
2016
,54, 184–194.
[C ossRe ]
10.
Fe ko a, Z.; Kindl, V. In luence o skewed squi el cage o o wi h in e media e ing on magne ic ield o ai gap in induc ion
machine. Elek on. Elek o echnika 2017,23, 26–30. [C ossRe ]
11.
Xu, W.; Bao, X.; Cheng, S.; Xu, Y.; Lu, Q. Ro o eddy cu en analysis and op imisa ion design o in e media e ing in no el double
squi el-cage induc ion mo o . IET Elec . Powe Appl. 2020,14, 375–382. [C ossRe ]
12.
Swann, S.A.; Salmon, J.W. E ec i e esis ance and eac ance o a ec angula conduc o placed in a semi-closed slo . P oc. Ins .
Elec . Eng. 1963,110, 1656. [C ossRe ]
13.
Ra ajdus, P.; H abo co a, V.; Lehocky, P.; Makys, P.; Holub, F. E ec o sa u a ion on ield o ien ed con ol o he new designed
eluc ance synch onous mo o . Ene gies 2018,11, 3223. [C ossRe ]
14.
Zhao, H.; Liu, C.; Zhan, Y.; Xu, G. Loss and S a ing Pe o mance o In e e -Fed Induc ion Mo o s Conside ing Semi-Closed
E ec o Closed Slo . In P oceedings o he 2019 In e na ional Applied Compu a ional Elec omagne ics Socie y Symposium in
Miami (2019 ACES-Miami), Miami, FL, USA, 14–18 Ap il 2019; pp. 1–2.
15.
Thomson, W.T.; Fenge , M. Cu en signa u e analysis o de ec induc ion mo o aul s. IEEE Ind. Appl. Mag.
2001
,7, 26–34.
[C ossRe ]
16.
Yang, C.; Pa k, S.; Lee, S.B.; Jang, G.; Kim, S.; Jung, G.; Lee, J.; Shim, S.; Lim, Y.K.; Kim, J. S a ing cu en analysis in medium
ol age induc ion mo o s: De ec ing o o aul s and eac o s a ing de ec s. IEEE Ind. Appl. Mag. 2019,25, 69–79. [C ossRe ]
17.
A igao, E.; Koukou a, S.; Hon ubia-Esc ibano, A.; Ca oll, J.; McDonald, A.; Gómez-Láza o, E. Cu en signa u e and ib a ion
analyses o diagnose an in-se ice wind u bine d i e ain. Ene gies 2018,11, 960. [C ossRe ]
18.
Tang, J.; Yang, Y.; Chen, J.; Qiu, R.; Liu, Z. Cha ac e is ics analysis and measu emen o in e e - ed induc ion mo o s o S a o
and o o aul de ec ion. Ene gies 2019,13, 101. [C ossRe ]
19.
Puche-Panade o, R.; Pineda-Sanchez, M.; Rie a-Guasp, M.; Roge -Folch, J.; Hu ado-Pe ez, E.; Pe ez-C uz, J. Imp o ed esolu ion
o he MCSA me hod ia Hilbe ans o m, enabling he diagnosis o o o asymme ies a e y low slip. IEEE T ans. Ene gy
Con e s. 2009,24, 52–59. [C ossRe ]
Ene gies 2021,14, 6573 23 o 23
20.
Sun, L.; Xu, B. An imp o ed me hod o disce ning b oken o o ba aul and load oscilla ion in induc ion mo o s. Ene gies
2018
,
11, 3130. [C ossRe ]
21.
Zamudio-Ramí ez, I.; Oso nio-Ríos, R.A.; An onino-Da iu, J.A.; Quijano-Lopez, A. Sma -senso o he au oma ic de ec ion
o elec omechanical aul s in induc ion mo o s based on he ansien s ay lux analysis. Senso s
2020
,20, 1477. [C ossRe ]
[PubMed]
22.
Vi ek, O.; Janda, M.; Hajek, V.; Baue , P. De ec ion o eccen ici y and bea ings aul using s ay lux moni o ing. In P oceedings o
he 8 h IEEE Symposium on Diagnos ics o Elec ical Machines, Powe Elec onics & D i es, Bologna, I aly, 5–8 Sep embe 2011;
pp. 456–461. [C ossRe ]
23.
Kindl, V.; H uska, K.; Pechanek, R.; Sob a, J.; Skala, B. The e ec o space ha monic componen s in he ai gap magne ic lux
densi y on o que cha ac e is ic o a squi el-cage induc ion machine. In P oceedings o he 2015 17 h Eu opean Con e ence on
Powe Elec onics and Applica ions (EPE’15 ECCE-Eu ope), Gene a, Swi ze land, 8–10 Sep embe 2015; pp. 1–5. [C ossRe ]
24.
Shin, S.; Kim, J.; Lee, S.B.; Lim, C.; Wiedenb ug, E.J. E alua ion o he in luence o o o magne ic aniso opy on condi ion
moni o ing o wo-pole induc ion mo o s. IEEE T ans. Ind. Appl. 2015,51, 2896–2904. [C ossRe ]
25.
S one, G.; Boul e , E.A.; Culbe , I.; Dhi ani, H. O Line Ro o and S a o Winding Tes s. In Elec ical Insula ion o Ro a ing
Machines: Design, E alua ion, Aging, Tes ing, and Repai ; John Wiley & Sons, Inc.: Hoboken, NJ, USA, 2004; pp. 235–283. [C ossRe ]
26.
Kim, B.; Lee, K.; Yang, J.; Lee, S.B.; Wiedenb ug, E.J.; Shah, M.R. Au oma ed de ec ion o o o aul s o in e e - ed induc ion
machines unde s ands ill condi ions. IEEE T ans. Ind. Appl. 2011,47, 55–64. [C ossRe ]
27.
Jeong, M.; Yun, J.; Pa k, Y.; Lee, S.B.; Gy akis, K. O -line lux injec ion es p obe o sc eening de ec i e o o s in squi el cage
induc ion machines. In P oceedings o he 2017 IEEE 11 h In e na ional Symposium on Diagnos ics o Elec ical Machines,
Powe Elec onics and D i es (SDEMPED), Tinos, G eece, 29 Augus –1 Sep embe 2017; pp. 233–239. [C ossRe ]
28.
Nau, S.L.; Schmi z, D.; de Lima Pi es, W. Me hods o e alua e he quali y o s a o and o o o elec ic mo o s. In P oceedings o
he 2015 IEEE 10 h In e na ional Symposium on Diagnos ics o Elec ical Machines, Powe Elec onics and D i es (SDEMPED),
Gua da, Po ugal, 1–4 Sep embe 2015; pp. 64–70. [C ossRe ]
29.
Kang, T.-J.; Kim, J.; Lee, S.B.; Yung, C. Expe imen al e alua ion o low- ol age o line es ing o induc ion mo o o o aul
diagnos ics. IEEE T ans. Ind. Appl. 2015,51, 1375–1384. [C ossRe ]
30.
Va ghese, S.T.; Rajagopal, K.R.; Singh, B. Design and de elopmen o o o quali y es sys em o die-cas coppe o o s. IEEE
T ans. Ind. Appl. 2018,54, 2105–2114. [C ossRe ]
31. Tumanski, S. Induc ion coil senso s—A e iew. Meas. Sci. Technol. 2007,18, R31–R46. [C ossRe ]
32.
Innes, A.G. Condi ion Moni o ing o Induc ion Mo o s: The De ec ion o B oken Ro o Ba s in Va iable Speed Induc ion Mo o
De ices. Ph.D Thesis, Uni e si y o Tasmania, Hoba , Aus alia, 1999.
33.
Salah, L.; Adel, G.; Khaled, K.; Ahmed, B.; Issam, A. B oken o o ba aul diagnos ic o DTC Fed induc ion mo o using
s a o ins an aneous complex appa en powe en elope signa u e analysis. In . J. Powe Elec on. D i e Sys .
2019
,10, 1187–1196.
[C ossRe ]
34.
Gy akis, K.N.; Pla e o, C.A.; Zhang, Y.; Be nal, S. Diagnosis o s a ic eccen ici y in 3-phase synch onous machines using a
pseudo ze o-sequence cu en . Ene gies 2019,12, 2476. [C ossRe ]
35.
Li, S.; Li, Y.; Sa lioglu, B. Ro o unbalanced magne ic o ce in lux-swi ching pe manen magne machines due o s a ic and
dynamic eccen ici y. Elec . Powe Compon. Sys . 2016,44, 336–342. [C ossRe ]
36.
Liao, Y.X.; Zhu, T.Y.; Sun, Y.Q.; Zhang, J.; Cheng, T.; Wang, Y. Load in luence on lissajous igu e o online ans o me winding
diagnosis. In P oceedings o he 2016 IEEE In e na ional Con e ence on High Vol age Enginee ing and Applica ion (ICHVE),
Chengdu, China, 19–22 Sep embe 2016; pp. 1–4. [C ossRe ]
37.
Vilheka , T.G.; Ragha end a, R.; Ballal, M.S. De ec ion o winding aul s in auxilia y and main windings o single phase squi el
cage induc ion mo o by moni o ing Lissajous pa e ns. In P oceedings o he 2016 IEEE In e na ional Con e ence on Powe
Elec onics, D i es and Ene gy Sys ems (PEDES), T i and um, India, 14–17 Decembe 2016; pp. 1–6. [C ossRe ]
38.
Kugels ad , T. Ac i e Fil e Design Techniques. In Op Amps o E e yone; Else ie : Ams e dam, The Ne he lands, 2003; pp. 261–323,
ISBN 9780750677011. [C ossRe ]
39.
Webs e , J.G. Elec ical Measu emen , Signal P ocessing, and Displays; CRC P ess: Boca Ra on, FL, USA, 2003; ISBN 9780849317330.