High Technology Readiness Level Techniques for Brushless Direct Current Motors Failures Detection: A Systematic Review

ene gies Re iew High Technology Readiness Le el Techniques o B ushless Di ec Cu en Mo o s Failu es De ec ion: A Sys ema ic Re iew Vi o Ma io Fico 1,* , Ma ía Ángeles Ma ín P a s 2and Ca melina Ie a di 3 1Skyli e Enginee ing, 41092 Se ille, Spain 2Escuela Técnica Supe io de Ingenie ía, Elec onics Enginee ing Depa men , Uni e sidad de Se illa, 41092 Se ille, Spain; [email p o ec ed] 3Enginee ing Depa men , Uni e sidad Loyola Andalucía, 41704 Se ille, Spain; cie a [email protected] *Co espondence: [email p o ec ed] Recei ed: 3 Feb ua y 2020; Accep ed: 18 Ma ch 2020; Published: 1 Ap il 2020   Abs ac : Many pape s ela ed o his opic can be ound in he bibliog aphy; howe e , jus a modes pe cen age o he in oduced echniques a e de eloped o a Technology Readiness Le el (TRL) su icien ly high o be implemen able in indus ial applica ions. This pape is ocused p ecisely on he e iew o his speci ic opic. The in es iga ion on he s a e o he a has been ca ied ou as a sys ema ic e iew, a e y igo ous and eliable s anda dised scien i ic me hodology, and ies o collec he a icles which a e close o a possible implemen a ion. This selec ion has been ca e ully done wi h he de ini ion o a se ies o ules, d awn o ep esen he adequa e le el o eadiness o aul de ec ion echniques which he a ious a icles p opose. Keywo ds: ailu e; PMSM; de ec ion; diagnosis; BLDC; b ushless; sys ema ic e iew 1. In oduc ion The opic o he p esen e iew is subjec ed o a g owing in e es , bo h om he academic and he indus ial wo lds, due o he pa allel inc ease o he usage o elec ic machines o high eliabili y asks as mo o ing o elec ic ehicles and ac ua ion o ligh su ace o he u u e Mo e Elec ic Ai c a s. Abo e all, when ae ospace applica ions a e in ol ed, eliabili y becomes o i al impo ance; indeed, he pe o mance o ligh ac ua o s on a damaged ai c a is no as impo an as ensu ing ha he emaining ac ua o s con inue ope a ion un il he ai c a can land sa ely. In mos cases, an adequa e le el o eliabili y can be eached only by using diagnos ic ools [1]. The a ailabili y o an accu a e and e icien mean o condi ion moni o ing and machine aul diagnosis can be o pa amoun impo ance, as i imp o es he eliabili y and s abili y o he plan and a he same ime i educes cos s, ideally leading o a sys em wi hou p og ammed main enance. S a is ical s udies [ 2 ] show ha expec ed eliabili y can be imp o ed up o 5–6 pe cen age poin s wi h he use o moni o ing. In his con ex a ises he need o p ecisely know he e olu ion and he cu en s a e o he li e a u e abou he aul de ec ion and diagnosis echniques o B ushless DC (BLDC) and in pa icula which echniques a e close o a possible implemen a ion, i.e., which echniques ha e he highes Technology Readiness Le el (TRL). In he a ea o enginee ing, and especially o ae ospace enginee ing, he na a i e e ision is usually p e e ed. In his kind o e iew, he au ho s decide which pape s include in he su ey, based on hei wide knowledge and expe ience and o e ing a pe sonal poin o iew and in e p e a ion o he chosen heme. Ene gies 2020,13, 1573; doi:10.3390/en13071573 www.mdpi.com/jou nal/ene gies Ene gies 2020,13, 1573 2 o 24 The SR is a igo ous s anda dised scien i ic me hodology, used o p oduce eliable li e a u e e iews, mainly ecognised by i s objec i i y. I is employed wi h excep ional esul s in many a eas, including bio-science [ 3 , 4 ], compu e science [ 5 ] and in ecen yea s pa icula ly in so wa e enginee ing [ 6 , 7 ]. In pa icula , hese las pape s and he wo k p oposed by [ 8 ] ha e been used as guides o unde ake he sys ema ic e iew in his wo k. In ac , he i s objec i e o his wo k is p ecisely o adap he guidelines men ioned abo e o ou ield, while he second objec i e is o apply he sys ema ic e iew o a speci ic opic: high TRL echniques o BLDC mo o s ailu es de ec ion. To be mo e p ecise, he au ho s a e in e es ed in hose de ec ion echniques ha a e no es ic ed o a pa icula machine o wi h special se -ups, con igu a ions, loads o mo o manoeu es. Fu he mo e he echniques shall ha e been es ed a a ious ope a ion poin and he algo i hm shall ha e es ed wi h success o a leas one o some cases la e desc ibed in he inclusion c i e ia. Concluding, in o de o be accep ed in he sys ema ic e iew, he pape shall demons a e ha he p oposed algo i hm is capable o disce n be ween heal hy and aul y mo o . These cons i u e, mainly, he inclusion c i e ia o he s udies appea ing in he e iew. The undamen al scope o his e iew is indeed o de ec which echniques a e p esen ly being p o i ably used o mo o aul de ec ion and diagnosis and o p o ide he indus y wi h some high eadiness le el and es ed echniques. In his p ospec , mos o he inclusion and exclusion c i e ia ha e been de ined o ocus he in es iga ion on hose echniques wi h demons a ed aul de ec ion pe o mances a a ious ope a ion poin s and easily au oma able o al eady au oma ed. An addi ional ques ion conside ed in his SR has been he possibili y o embed in he mo o body, he ha dwa e needed o aul de ec ion. Al hough mos a icles a e ocused on he de ec ion by u ilising commonly measu ed a iables (mainly speed, cu en , ol age) some au ho s ha e elabo a ed aul de ec ion echniques based on he analysis o images om ex e nal came as o sensi i e accele ome e s. Those echniques a e app op ia e o be implemen ed only in pa icula applica ions [1] and ha e been disca ded om he scope o he p esen wo k. A e he sc eening o mo e han 3000 possible pape s, only 44 p ima y s udies ha e been ound o sa is y he a o emen ioned c i e ia. The au ho s ha e ca e ully e ised hose pape s and ha e collec ed he ollowing da a: he ype o aul de ec ed, he echnique ha was used o he de ec ion and he senso s used, he inclusion o expe imen s o simula ions, i he echnique has been es ed a di e en ope a ion poin s (di e se speed o loads o in he bes case a combina ion o bo h), he wo king condi ion (s a iona y o no ) and some o he limi a ions/ad an ages. This in o ma ion is hen ga he ed in a ea u e able, which is an use ul s a egy o ge a comple e, igo ous and objec i e iew o he chosen opic. A he end o he p ocess, he esea ch ques ions ini ially o mula ed a e answe ed, p o iding a ull pe spec i e o he opic [8]. The SR consis s o h ee sequen ial phases, each o which is subdi ided in u n in o sub-phases, as de ailed below [7]: 1. Planning he e iew •iden i ica ion o he need • esea ch ques ions • e iew p o ocol •e alua ing p o ocol 2. Conduc ing he e iew •selec ion o p ima y s udies •s udy quali y assessmen •ex ac ion and syn hesis o da a 3. Repo ing he e iew •speci ying dissemina ion mechanisms • o ma ing he main epo Ene gies 2020,13, 1573 3 o 24 •e alua ing he epo The es o he pape is o ganised as ollows.. Sec ion 2p esen s he desc ip ion and he adap a ion o ou case o planning phase o he SR. Conduc ing and epo ing o he sys ema ic e iew a e gi en in Sec ions 3and 4 espec i ely. Finally, he conclusions a e d awn in Sec ion 5. Addi ionally, in Appendix Ais gi en a de ini ion o he main e minology used along he documen . 2. Planning The i s s ep o he sys ema ic e iew consis s in planning, which is he ounda ion o he en i e e ision. I is a his s age ha he main ools a e de eloped, such as he Boolean unc ion, he inclusion and exclusion c i e ia, he choice o he di e en da abases in which o ca y ou he esea ch and abo e all he de elopmen and e alua ion o a p o ocol ha egula es all he phases. The need o unde ake a sys ema ic e iew, a ises i s o all because he esea ch opic is e y wide and a igo ous me hod was needed o co ec ly ex ac he needed in o ma ion. As said, in he enginee ing ield his ype o me hodology is no usual because, e en by being scien i ic and igo ous, i is di icul and complica ed o ca y ou . Cu en ly he e is no sys ema ic e iew on he aul de ec ion echniques o b ushless DC mo o s, and in eali y he e is no e en a adi ional e ision so de ailed on he chosen heme (Should be cla i ied ha his sen ence is e e ed o he e iews aking in o accoun he eadiness le el o he echnique.). 2.1. Resea ch Ques ions Once he conc e e opic has been iden i ied, he e a e some c i e ia ha help o clea ly o mula e esea ch ques ions. Among he mos used c i e ia in o he sec o s he e a e he c i e ia called PICOC (Popula ion, In e en ion, Compa ison, Ou come, Con ex ). In his wo k we ha e conside ed hose p esen ed in [8] and adap ed o ou case. In his case, jus some o hese c i e ia ha e been used o o mula e and p ocess he ques ions ha his SR is ying o answe . In he inal sec ion o he e iew, de ined as epo ing, he e is a sub-sec ion called Discussion (Sec ion 4.1) whe e he ela i e answe s a e discussed and analysed. The ques ions o mula ed o he p esen wo k a e lis ed below: RQ.1: Which a e he mos common aul s o BLDC mo o s? RQ.2: Which pa ame e s a e used o aul de ec ion in BLDC mo o s? RQ.3: Which ype o ailu e can be de ec ed by each echnique? RQ.4: Which echnique equi es less compu a ional powe ? RQ.5: Which echnique equi es less senso s? RQ.6: Which echnique gi es he bes esul s o each ype o ailu e? 2.2. Re iew P o ocol The e ision p o ocol is no hing mo e han a se o ules and c i e ia o be ollowed du ing all he s ages, in o de o educe he bias and make he SR as objec i e as possible. In he bioscience ield, he p o ocol is some imes eco ded in a p ospec i e egis e , such as PROSPERO (h ps://www.c d. yo k.ac.uk/p ospe o/). Un o una ely, hese ype o egis e s do no exis in he ae ospace ield. A e y impo an aspec o be conside ed o he SR is he cla i y whe ewi h he p o ocol is exposed and elabo a ed, as a leas wo pe sons a e in ol ed in he e iew d a ing. A common, bu e y ime consuming, app oach consis s in he implemen a ion o he SR by wo independen pe sons, who ca y ou he pa o he conduc ing and epo ing sepa a ely and hen compa e and discuss he ob ained esul s. Ano he me hod, ha is he one used in his wo k, is ha a pe son pe o ms all he phases indi idually and a second pe son andomly checks some da a, as o example, some o he ows o he ea u es ables (Tables 1–5). Ene gies 2020,13, 1573 4 o 24 Table 1. Fea u es able (a). Ci e Yea Faul Type Technique Used Senso s Used Expe imen s o Simula ions Va ious Speed/Loads Wo king Condi ion Limi a ions/ Ad an ages [9] 2019 A ma u e aul s Pa ame e s Es ima ion Vol age, Cu en and Posi ion senso s Bo h Bo h S a iona y condi ions I p opose indica o s deduced om symme ical componen o phase cu en s in he e e ence ame. The me hod has been alida e a a ious speed, loads and sho ci cui esis ance magni ude o ITSC and a cons an speed, load, esis ance magni ude o PPSC. The algo i hm compu a ional load is nos speci ied. [10] 2019 Pe manen Magne ic aul s Model, AI and neu al-ne wo k-based echniques Vol age and Cu en senso s Bo h Va ious Speed S a iona y condi ions Expe imen ally es ed wi h 5 mo o condi ions (1 heal hy, 4 aul y) wi h good de ec ion pe o mances. P oposes wo ailu e ex ac ion me hods and compa es hem. T aining ime and compu a ional load no speci ied. [11] 2019 Mechanical aul s MCSA Vol age and Cu en senso s Bo h Bo h S a iona y condi ions Uses wa ele decomposi ion o he cu en signal and an adap i e il e o es ima e and emo e he undamen al componen . Tes ed using wo case s udies, i.e., b oken magne and eccen ici y aul , au oma ic aul classi ica ion wi h SVM and a e age accu acy o 96%. T aining ime and compu a ional load no speci ied. [12] 2019 Pe manen Magne ic aul s O he (Hall E ec Senso s lux analysis) Hall E ec Senso s Expe imen s Va ious Speed Non-S a iona y condi ions Me hod capable o de ec ing bea ing and pe manen magne s aul s by analysing espec i ely he cascade DWT-CWT ans o m o he speed signal and he ku osis index o he du y cycle signal o he hall senso ou pu . Elec ically independen om he mo o . In luence o load no speci ied. Compu a ional load no speci ied. [1] 2019 Pe manen Magne ic aul s O he (Signals Simila i y Analysis) Vol age and Cu en senso s Bo h Va ious Speed S a iona y Condi ions The me hod has been es ed wi h FEM simula ions and expe imen ally wi h good esul s. The es e ec o o que on he me hod has no benn e alua ed. The me hod can be used only o mul ipole mo o s. [13] 2019 A ma u e aul s Elec omagne ic ield moni o ing Tunneling Magne o esis i e senso s Bo h Bo h S a iona y Condi ions The me hod is capable o de ec ing bo h loca ion and se e i y o in e - u n sho -ci cui by sensing he s ay magne ic ield ou side he s a o yoke. I needs he ins alla ion o TM senso s a ound he mo o body. Compu a ional load no speci ied. [14] 2018 Mechanical aul s O he (Angula Resample) Vol age and Cu en senso s Expe imen s Bo h Non-S a iona y condi ions Me hod based on he angula esample o speed ob ained wi h a senso less obse e . Compu a ional load no speci ied. To que a ia ion no speci ied. [15] 2018 Pe manen Magne ic aul s O he (Vol age Angle) Vol age and Cu en senso s Bo h Va ious Speed S a iona y Condi ions The me hod akes ad an age om he a ia ions o he ol age angle obse ed du ing demagne isa ion and in e - u ns sho aul s o iden i y hei p esence. The me hod is empe a u e-dependan . A clea de ec ion h eshold is no de ined. Compu a ional load no speci ied. Ene gies 2020,13, 1573 5 o 24 Table 2. Fea u es able (b). Ci e Yea Faul Type Technique Used Senso s used Expe imen s o Simula ions Va ious Speed/Loads Wo king Condi ion Limi a ions/ Ad an ages [16] 2018 A ma u e aul s Pa ame e s Es ima ion Cu en and Vol age senso s Bo h Bo h S a iona y condi ion The me hod is a ec ed by he magni ude o he s a o cu en , should be used in cons an o que condi ions. [17] 2017 Mechanical aul s Elec omagne ic ield moni o ing, sea ch coils, coils wound a ound mo o sha s Sea ch coil Simula ions Va ious Speeds S a iona y Condi ions The me hod is independen om mo o a iables, bu needs he sea ch coil o be ins alled on he s a o . [18] 2017 A ma u e aul s Model, AI, and neu al-ne wo k -based echniques, Pa ame e s Es ima ion Cu en , Vol age and Speed senso s Expe imen s Bo h S a iona y Condi ions This me hod has been es ed expe imen ally on an ae onau ical mo o , bu he expe imen se -up has no been p esen ed. The algo i hm is e y as ( ≈ 15 msec), bu he compu a ion ime o he ea u es is no aken in o accoun . Also i needs a la ge da abase o aining he algo i hm. [19] 2017 Pe manen Magne ic and Mechanical aul s Model, AI, and neu al -ne wo k-based echniques Cu en senso Expe imen s Va ious Loads S a iona y Condi ions Two ailu es in oduced on an expe imen al pla o m and a 10- old alida ion o he algo i hm is execu ed. The algo i hm is as (30 msec), bu he aining ime is no speci ied. [20] 2017 A ma u e aul s Pa ame e s Es ima ion Cu en and Vol age senso s Bo h Bo h Non-S a iona y Condi ions The p oposed me hod is jus sligh ly dependen om he load and speed. The au ho s also demons a ed obus ness agains pa ame e s a ia ion and inaccu acies by in oducing a h eshold, bu he allowed ole ance is no speci ied and his quan i y can also depend on he mo o . [21] 2017 A ma u e aul s MCSA Cu en and Vol age senso s Bo h Bo h S a iona y Condi ions The p oposed me hod is pa icula ised o in e mi en aul s. The es se -up is no p esen ed. Tes s a a ious loads and speeds ha e been execu ed, bu hei in luence on he me hod is no speci ied. [22] 2017 Mechanical aul s O he (Hall E ec Senso s lux analysis) Analogue o Digi al Hall senso s Bo h Va ious Loads S a iona y Condi ions The me hod is independen om speed and demons a es only a sligh dependence om loads. The bes accu acy is ob ained wi h analogue Hall e ec senso which a e no common, e en i he au ho s p o ide an al e na i e based on digi al Hall e ec senso s. This me hod can be used only i he Hall senso s a e placed in he adial di ec ion. [23] 2017 Pe manen Magne ic aul s O he (To que Ripple Analysis) To que senso Bo h Va ious Loads S a iona y Condi ions No in o ma ion is gi en abou how he mo o speed a ec s he p oposed me hod. A o que ansduce is needed o apply he algo i hm. Ene gies 2020,13, 1573 6 o 24 Table 3. Fea u es able (c). Ci e Yea Faul Type Technique Used Senso s Used Expe imen o Simula ions Va ious Speed/Loads Wo king Condi ion Limi a ions/ Ad an ages [24] 2016 A ma u e aul s Pa ame e s es ima ion Cu en , Vol age and Speed senso Bo h Va ious Speeds Non-S a iona y Condi ions The me hod is based on a modi ied mo o model. I is based on compu a ions in he o a ing ame (dq). No in o ma ion abou how he load a ec s he de ec ion me hod is gi en. [25] 2016 A ma u e aul s Pa ame e s Es ima ion Cu en and Speed senso s Expe imen s Va ious speeds S a iona y Condi ions The p oposed aul index has a e y educed dependence om mo o speed. [26] 2016 Mechanical aul s MCSA, Model, AI and NN-based echniques Cu en Senso Expe imen s Va ious Speeds S a iona y Condi ions The implemen a ion is e y close o a eal scena io, bu in some s udied condi ions he ail a e o he classi ie is ela i ely high. [27] 2016 Pe manen Magne ic Faul s Pa ame e s Es ima ion Cu en , Speed and Angle Senso s Bo h Bo h Non-S a iona y Condi ions The me hod needs he knowledge o a ious mo o pa ame e s and hei a ia ion (o inco ec ness) can esul in poo diagnosis pe o mances. The au ho s demons a ed good pe o mances wi h a ious demagne isa ion le els and wo king condi ions. [28] 2016 A ma u e aul s O he (PWM Ripple Cu en Measu emen s) Cu en and Vol age senso s Bo h Bo h S a iona y Condi ions The me hod needs an elec ic model o he mo o alid o high equencies. The au ho demons a ed good sensi i i y also a low speed. [29] 2015 Mechanical aul s Noise and Vib a ion Moni o ing, Model, AI, NN based echniques, MCSA Cu en senso and Accele ome e Bo h Bo h Non-S a iona y Condi ions The me hod is capable o de ec and dis inguish di e en bea ing ailu es. Need an accele ome e o be placed close o he bea ing. [30] 2015 Mechanical aul s Model, AI, NN-based echniques Cu en senso Expe imen s Bo h S a iona y Condi ions The au ho s pe o med an ex ensi e expe imen campaign wi h good esul s. F om he images in he a icle, he damages ep oduced on he bea ing appea o be conside able. [31] 2015 A ma u e aul s Model, AI and NN-based echniques, MCSA Cu en senso Expe imen s Bo h S a iona y Condi ions High de ec ion a io. The equency analysis is did wi h he FFT, his means ha du ing he 12 s o signal acquisi ion he mo o speed and load shall be cons an . [32] 2015 A ma u e aul s Elec omagne ic ield moni o ing, Sea ch Coils, Coils wound a ound mo o sha Sea ch coil Bo h Va ious Speeds S a iona y Condi ions The p esen ed me hod is in asi e o an al eady buil mo o . The de ec ion ime is e y sho (3–5) ms bu i seems be dependen on mo o speed; u he mo e he ha dwa e used o compu a ions is no p esen ed. [33] 2015 A ma u e aul s Model, AI and NN-based echniques, MCSA Cu en senso s Bo h Bo h S a iona y Condi ions The mo o used o expe imen s has an inhe en anomaly, bu he algo i hm is capable o disce n i om he sho -ci cui . The p esen ed me hod does no imply any p e ious knowledge on he mo o . They uses FFT, i implies ha du ing he signal acquisi ion he condi ions need o be s a iona y. Ene gies 2020,13, 1573 7 o 24 Table 4. Fea u es able (d). Ci e Yea Faul Type Technique Used Senso s Used Expe imen o Simula ions Va ious Speed/Loads Wo king Condi ion Limi a ions/ Ad an ages [34] 2013 Pe manen Magne s aul s Model, AI and NN-based echniques, Pa ame e s Es ima ion Cu en , Vol age and Speed senso s Expe imen s Va ious loads S a iona y Condi ions P oposes a me hod o demagne isa ion. Compa ison o he p oposed me hod wi h a ious o he es ablished me hods. [35] 2013 A ma u e, Pe manen Magne s and Mechanical Faul s Model, AI, NN-based echniques, MCSA Cu en senso s Simula ions Bo h S a iona y Condi ions The pape p esen s a good a ie y o mo o s, aul s and wo king condi ions. The de ec ion accu acy ob ained is e y high, bu i can be due o he use o clean signals om simula ions. [36] 2013 A ma u e aul s Model, AI and NN-based echniques Cu en and Vol age Senso s Bo h Bo h S a iona y Condi ions The aining o he AI has been execu ed wi h da a om bo h expe imen s and simula ions. The me hod uses cu en measu emen s in ime domain wi h no need o equency domain ans o ma ion. The me hod is capable o de ec ing aul se e i y and loca ion. [37] 2013 A ma u e aul s Pa ame e s es ima ion Cu en , Vol age and Speed senso s Bo h Va ious loads S a iona y Condi ions The me hod compa es an es ima ed back-EMF wi h a e e ence one o aul de ec ion. The e e ence is ob ained om a FEM model o om an heal hy machine. This me hod can be e y sensible o mo o pa ame e s change. The in luence o he load is no discussed. [38] 2013 A ma u e aul s MCSA Cu en , Vol age and Speed senso s Bo h Bo h Non-S a iona y Condi ions The p oposed me hod has a low compu a ional bu den, bu needs access o he mo o neu al poin o be applied. [39] 2011 Mechanical aul s Model, AI and NN-based echniques Cu en , Speed and To que senso s, Simula ions Bo h Non-S a iona y Condi ions The me hod ies o de ec mechanical aul s by es ima ing he bea ing heal h s a us. A o que senso is used, which is no usually moun ed in mo o s and he alida ion is ca ied ou by simula ion wi hou added noise. [40] 2011 A ma u e aul s Pa ame e s es ima ion Cu en , Vol age and Speed senso s Bo h Bo h Non-S a iona y Condi ions The back-EMF is es ima ed when he machine is heal hy and hen ozen, which causes dependence on mo o pa ame e s changes. The e is a model o compensa e he in e e losses compensa ion. [41] 2011 A ma u e aul s O he (High F equency Injec ion) Cu en and Angula Posi ion senso s Bo h Va ious loads S a iona y Condi ions The me hod has a e y good esolu ion, bu he de ec ion is based on a look-up able. This makes he algo i hm igno e all he ailu es (i any) p esen be o e he able c ea ion. [42] 2011 A ma u e aul s MCSA Cu en senso s Expe imen s Bo h S a iona y Condi ions Capable o de ec ing wo ailu es. Use he FFT o he equency analysis, bu de ec pe iods o s a iona i y o he mo o . Use linea in e pola ion o de ine he heal hy compa ison e m. Ene gies 2020,13, 1573 8 o 24 Table 5. Fea u es able (e). Ci e Yea Faul Type Technique Used Senso s Used Expe imen o Simula ions Va ious Speed/Loads Wo king Condi ion Limi a ions/ Ad an ages [43] 2011 A ma u e and Mechanical aul s Model, AI NN-based echniques Cu en and Vol age seno s Expe imen s Va ious Loads S a iona y Condi ions Pe o m an in e es ing mul i-class classi ica ion based on se en pa ame e s. The algo i hm does no seem capable o classi y ailu es no p esen he aining se . [44] 2010 Mechanical aul s Model, AI, NN-based echniques, MCSA Cu en senso s Bo h Va ious loads (Load independence demons a ed analy ically) S a iona y Condi ions S udy on he impac o SNR. [45] 2010 A ma u e aul s Model, AI, NN-based echniques, MCSA Cu en senso s Bo h Va ious loads (Load independence demons a ed analy ically) S a iona y Condi ions S udy on he impac o SNR. Two ailu es s udied, wi h aul se e i y es ima ion. [46] 2008 A ma u e aul s Model, AI and NN-based echniques Cu en , Vol age and Speed senso s Expe imen s Va ious Loads Non-S a iona y Condi ions T aining o a neu al-ne wo k o p edic cu en and include ini ial asymme ies. The p edic ed alue o he cu en is used as a e e ence o de ec ailu es unde load a ia ions. [47] 2007 Mechanical aul s MCSA Cu en and Vol age senso s Bo h Bo h Non-S a iona y Condi ions The expe imen al se -up is no desc ibed. The e is a compa ison be ween h ee echnique o ime- equency analysis and ela i e aul de ec ion. [48] 2007 Mechanical aul s MCSA Cu en and Vol age senso s Bo h Bo h Non-S a iona y Condi ions The expe imen al se -up is no desc ibed. [49] 2007 Mechanical aul s MCSA Cu en and Vol age senso s Bo h Va ious Speeds Non-S a iona y Condi ions The in luence o load is no aken in o accoun . The expe imen al se -up is no desc ibed. [50] 2006 A ma u e, Pe manen Magne s and Mechanical aul s MCSA Cu en and Speed senso s Expe imen s Bo h Non-S a iona y Condi ions Me hod o acking he aul equencies du ing a iable speed ope a ions. The es se -up is no desc ibed. Va ious aul s ha e been implemen ed. [51] 2006 Mechanical aul s MCSA Cu en and Speed senso s Bo h Bo h Non-S a iona y Condi ions Two me hods p esen ed based on di e en equency acking algo i hms. Real- ime implemen a ion wi h p ocesso execu ion ime is also included. Ene gies 2020,13, 1573 9 o 24 One o he basic s eps o he p o ocol is he c ea ion o a Boolean unc ion ha comp ehensi ely includes all he e ms ela ed o he chosen heme, including all he synonyms and e ms ha may be ela ed o he wo ds o in e es o he opic. To ca y ou his esea ch based on keywo ds, i is app op ia e o deeply ead abou he heme o de ec which wo ds a e mos equen ly used by he au ho s. The a icula e Boolean unc ion c ea ed o his wo k is as ollows: ((("b ushless DC" OR "pe manen magne elec ical") AND (mo o OR machine)) OR BLDC OR PMSM) AND (((condi ion OR heal h) AND moni o ing) OR ((diagnosis OR de ec ion) AND ( aul OR ailu e))) The i s pa o he Boolean unc ion de ines he ype o mo o , while he second one de ines he de ec ion o he de ec . A di icul y encoun e ed du ing he esea ch is ha he di e en bibliog aphic da abases a e no p epa ed o his kind o e ision, as hey do no allow ce ain esea ches o o sea ch in ce ain ields o he pape s. Indeed, he Boolean unc ion based esea ch was ca ied ou in he i le, abs ac and keywo ds o he pape s. Due o he esea ch es ic ions o he da abases, as speci ied in [ 52 ], and hanks o he good co e age o he edi o ials ob ained shown in Table 6, he ollowing da abases ha e been used: •IEEE Xplo e Digi al Lib a y •Scopus •ACM Digi al Lib a y •Science Di ec •Web o Science Table 6. Da abases co e age wi h espec o he con en o he publishe s: IE = IEEE, IT = IET, PE = Pegamon-Else ie , ES = Else ie Science, WB = Wiley Blackwell, TF = Taylo & F ancis, SP = Sp inge , SI = SIAM Publica ions, OX = Ox o d Uni e si y P ess, KO = Ko ean Ins . Elec ical Eng., SA = Sage Publica ions, AS = ASME, MP = Mic o ome Publica ions [52]. IE IT PE ES WB TF SP SI OX KO SA AS MP IEEEX ACM Scopus WoS SD Once he esea ch ques ions ha e been iden i ied and he ela i e Boolean unc ion c ea ed, i mus be in oduced in he di e en bibliog aphic da abases, adap ing i acco ding o he sea ch language o each da abase. In his wo k, he esea ch has been ca ied ou by sea ching only in he abs ac , i le and keywo ds o he pape s, ob aining a o al o 3167 i ems un il o No embe 2019, as de ailed in Table 7. Ene gies 2020,13, 1573 16 o 24 Figu e 4is no ep esen a i e o he whole li e a u e, bu i is possible o use i o in es iga e he p og esses on he opic. Du ing he las yea s, he echniques based on a i icial in elligence, pa ame e s es ima ion and models, a e being u ilised wi h inc easing equency, o en as classi ie s, in conjunc ion wi h es ablished me hods like he MCSA. On he o he hand, he numbe o a icles p esen ing de ec ion echniques based on MCSA has d as ically educed, p obably because hese echniques ha e been in ensi ely s udied in he pas yea s and he e is less space le o inno a ions. S a ing om 2016 echniques agged wi h O he , i.e., he echniques no classi iable in he p e iously de ined ca ego ies, ha e s eadily inc eased in numbe . This indica es ha p e iously unexplo ed phenomena a e being used o BLDC aul de ec ion and ou lines ha he esea ch on he chosen opic is in u moil. Figu e 5shows he o e all dis ibu ion o he pape s acco ding o he used echnique. The ollowing echniques ha e been omi ed om he g aph because hey ha e no been ound: • adio- equency emissions moni o ing, • empe a u e measu emen s, •in a ed ecogni ion, •chemical analysis. 6% 2% 2% 40% 32% 6% 12% Elec omagne ic ield moni o ing Radio- equency (RF) emissions moni o ing Noise and ib a ion moni o ing MCSA Model, AI, NN-based echniques Pa ame e s Es ima ion O he Figu e 5. Dis ibu ion a icles acco ding o he used echniques The MCSA is he mos used echnique, ollowed by he AI algo i hms. I is impo an o poin ou ha equen ly he echniques based on A i icial In elligence a e used as classi ie s o esul s ob ained wi h o he , al eady es ablished, me hods o ailu e de ec ion. This associa ion demons a ed o be ha e a g ea impac in imp o ing he de ec ion a e and in ex ending he use o he echnique o a wide ange o bo h speed and load. One o he key aspec in he g aph, is he p esence o a good amoun o pape s using echniques which we e no p e iously classi ied (g ouped unde he ag o he ). Be ween hem i is possible o ind inno a i e echniques based on High F equency Injec ion [ 41 ], hall e ec senso s measu emen s [ 22 ] o inno a i e mo o signals analysis [1,14,15]. An in e es ing al e na i e o he echniques based on Elec omagne ic ield moni o ing is ep esen ed by [ 13 ]. The au ho s use ex e nal senso o sense he s ay magne ic ields ou side he s a o o de ec a ma u e ailu es se e i y and loca ion, sol ing one o he bigges d awback o his powe ul echniques ca ego y, i.e., he in asi e p ocedu e o placing addi ional windings inside he s a o co e. The nex pa ag aphs a e dedica ed o answe o he p e iously o mula ed esea ch ques ions by using he selec ed pape s. Ene gies 2020,13, 1573 17 o 24 4.1.1. Rq.1: Mos Common Failu es o Bldc Mo o s Figu e 6shows he dis ibu ion o he pape s in ela ion o he ype o ailu e discussed. The esul s a e in acco dance wi h he ailu e dis ibu ion p esen ed in a ious pape s [ 29 , 31 , 36 , 54 ], and, in u n, his means ha he esea ch e o s a e consis en wi h he ailu es occu ence. A ma u e Faul 48% Pe manen Magne ic Faul s (pa ial o comple e) 21% Mechanical Faul s (bea ing ailu e and eccen ici y) 31% Figu e 6. Dis ibu ion o he pape s acco ding he ype o ailu e 4.1.2. Rq.2: Pa ame e s Used o Failu e De ec ion in Bldc Mo o s Due o he in ense esea ch in his ield, many o he mo o pa ame e s ha e been used o aul de ec ion pu poses. In he ollowing, he a iables used will be lis ed, di iding hem be ween hose di ec ly measu able and hose es ima ed. Di ec ly measu able quan i ies The quan i ies lis ed below a e di ec ly measu able by using speci ic senso s. Ou pu o que To que-me e s shall be used o measu e his a iable and i can p o ide e y use ul in o ma ion. The p oblem esides in he ac ha his ype o senso s a e o en big and expensi e. Cu en The cu en is always al eady measu ed by he mo o con olle and he e a e an immense quan i y o ailu e de ec ion algo i hms based on his a iable. Vol age The ol age is also commonly measu ed by he mo o con olle . Vib a ions By placing accele ome e s on he mo o , i is possible o measu e i s ib a ion le el. The algo i hms based on ib a ion analysis could p esen p oblems when used in mo ing sys ems, like ai c a , due o he coupling o ex e nal and unp edic able ib a ions. Magne ic lux The magne ic lux gi es a deep insigh on how he mo o is wo king. In o de o measu e i , i is usually necessa y o include in he mo o winding so called sea ch coils, i.e., some addi ional windings no connec ed o he phases. The inclusion o hese addi ional coils is no common and, al hough being a simple p ocedu e, i need o unmoun he mo o , ewound i and o ex ac om he in e io as many pai s o wi es as many sea ch coils as a e inse ed. An al e na i e o his p ocedu e is o place ex e nal magne ic senso s on he s a o o sense he s ay magne ic ields. Es ima ed quan i ies The p ocedu es based on pa ame e s es ima ion can iden i y ailu es by e alua ing he changes wi hin he measu ed mo o pa ame e s as well as e alua ing ac o s which a e no s aigh o wa dly quan i iable, such as: Ene gies 2020,13, 1573 18 o 24 •Back-EMF, •Magne ic lux, •Winding esis ance, •Winding induc ance. Es ima ion could be a e ec i e ins umen which allows he use o a iables s aigh o wa dly ela ed o he aul and some hing else no measu able. The d awback is ha i depends on models which can be cons ained o pa icula wo king poin s and a ec ed by he shi o some pa ame e . 4.1.3. Rq.3: Type o Failu es De ec able by Each Technique Figu e 7 ep esen s he dis ibu ion o he di e en pape s acco ding o he a ious echniques p oposed o de ec ing di e en ypes o aul s. This allows o e alua e which echniques a e mos sui able o de ec ing and dis inguishing be ween di e en ypes o ailu e o i some echniques a e mo e sui able o de ec ing speci ic aul s o can be used as a b oad spec um analysis ool. The echniques a e widely dis ibu ed among he ypes de ec ion me hods, wi h he excep ion o ib a ion moni o ing which appea s limi ed o he de ec ion o mechanical aul s; howe e , his speci ic i em can be biased due o he p esence o only one single pape in he e iew o his ca ego y. A ma u e Faul A ma u e Faul , Mechanical Faul s (bea ing ailu e and eccen ici y) A ma u e Faul , Pe manen Magne ic Faul s (pa ial o comple e), Mechanical Faul s (bea ing ailu e and eccen ici y) Mechanical Faul s (bea ing ailu e and eccen ici y) Pe manen Magne ic Faul s (pa ial o comple e) Pe manen Magne ic Faul s (pa ial o comple e), Mechanical Faul s (bea ing ailu e and eccen ici y) Pe manen Magne ic Faul s (pa ial o comple e), A ma u e Faul 0 5 10 15 20 21 2 1 3 3 1 2 1 1 1 5 1 3 1 2 1 2 2 7 1 2 1 Elec omagne ic ield moni o ing Model, AI, and NN-based echniques Model, AI, and NN-based echniques, MCSA Model, AI, and NN-based echniques, Pa ame e s Es ima ion MCSA O he Pa ame e s Es ima ion Noise and ib a ion moni o ing, Model, AI, and NN-based echniques, MCSA Figu e 7. Dis ibu ion o he pape s acco ding he ype o ailu e and he used echnique. 4.1.4. Rq.4: Compu a ional Powe Needed o Each Technique Only a ew pape s ([ 18 , 19 , 55 ]) o e ed a clea quan i ica ion o he compu a ional powe needed o implemen he p oposed echnique, and he e o e his ques ion can only be answe ed in a quali a i e manne . By wha has eme ged i can be seen ha by a heo e ical poin o iew he mos cos ly s a egies a e hose ocused on models. Tha is because o he necessi y o unning he compu e model pa allel o he machine i sel when compa ing he ou pu s. In addi ion, he complexi y inc eases wi h he le el o de ail o he model, he pa ame e s in ol ed, e c. The ollowing echniques in e ms o compu a ional cos a e hose based on he es ima ion o he pa ame e and hen hose ha use he NNs. In any case, i mos ly depends on he way he algo i hms a e implemen ed. The leas expensi e echniques a e he MCSA and o he echniques which di ec ly analyse senso da a. 4.1.5. Rq.5: Senso s Needed o Each Technique Gene ally he echniques ha equi e less senso s a e based on cu en o ol age analysis, such as MCSA. The nea - o ali y o he e iewed me hods mus a leas measu e cu en consump ion and mo o ol age, al hough his is no an issue as hese quan i ies a e al eady a ailable in mos d i e s. Ene gies 2020,13, 1573 19 o 24 Many me hods also employ he mo o speed o diagnose he aul . I can be speed can be es ima ed, om back-EMF measu emen s, o ob ained di ec ly om Hall e ec equency senso s o om a esol e [1]. 4.1.6. Rq.6: Bes De ec ion Resul s Only some selec ed pape s ([ 10 , 11 , 18 , 19 , 26 , 29 – 31 , 35 , 43 , 45 ]) p o ide s a is ics on he a e o e o de ec ion and a e mos ly based on he use o AI. I is e y complex o compa e all hese esul s, since he es condi ions a e no uni o m. This is conside ed a weak poin in his opic, which, al hough being e y ich in ideas and p oposed echniques, lacks alida ion and e i ica ion o he same. A possible solu ion o his p oblem, would be o p opose a minimum s anda d se o es s o be pe o med in o de o alida e a aul de ec ion algo i hm and gene a e a se o minimum compa able ou comes [1]. 4.2. Gene al Conside a ions Abou The Techniques In Table 12 he main cha ac e is ics o he de ec ion me hods ha e been g ouped. As p e iously men ioned, he echniques based on AI a e e y e ec i e. These can be used ei he as s and-alone aul de ec o s o in combina ion wi h o he echniques o signi ican ly imp o e hei pe o mance in de ec ion. I mus be no ed, in any case, ha hei success elies on an in ense p ocess o lea ning and hey ake conside able ime be o e wo king p ope ly [1]. Re e ing o he s a is ics lis ed abo e, in pa icula Figu e 4, i is possible o no ice how aul de ec ion echniques based on pa ame e es ima ion ha e also seen an inc ease in numbe . Such echniques can p o ide con inuous access o o he wise unobse able a iables such as back-EMF o magne ic lux, acili a e he ask o aul de ec ion o iden i y mo e explici ly obse able aul indica o s. No ewo hy is hei cha ac e is ic o being able o wo king while he mo o is unning in non-s eady-s a e condi ions o speed and/o o que. On he o he hand, he po en ial p oblem wi h hese echniques is ha hey a e based on assump ions, models and measu emen s o mo o a iables whose limi ed alidi y and inaccu acy could hinde aul de ec ion. Acco ding o he esul s o his esea ch, abou one hi d o he selec ed a icles p opose a echnique able o wo k in non-s a iona y condi ions and mos o hem a e based on AI, NN and pa ame e s es ima ion. The need o ope a e he mo o unde s eady-s a e condi ions can be a signi ican limi a ion, abo e all i his is necessa y o measu e signals o e a long pe iod o ime. Such condi ion may be achie ed wi h la ge indus ial machines wo king a cons an load bu a ely in ai c a ac ua o s [1]. Table 12. Techniques summa y [1]. Noise and Vib a ion Moni o ing Elec omagne ic Field Moni o ing Mo o Cu en Signa u e Analysis Model and AI based echniques Pa ame e s Es ima ion Ad an ages Mos sui able me hod o de ec ing mechanical aul s, as he accele ome e s can be placed close o he ib a ion sou ce Can di ec ly measu e he elec omagne ic ield inside he mo o , does no need complica ed algo i hm o de ec ailu es, can i ually de ec all he mo o ailu es Does no need addi ional senso s, can de ec a la ge a ie y o ailu es, is he mos used echnique Can be used du ing non-s a iona y mo o ope a ion, can be used in conjunc ion wi h o he echniques Can be used du ing non-s a iona y mo o ope a ion, can i ually moni o e e y mo o pa ame e Disad an ages Need o ins all accele ome e s on he mo o , measu emen s can be co up ed by en i onmen al ib a ions, di icul o use in non-s a iona y mo o ope a ion Need o ewind he s a o and o ex ac as many addi ional cables as many coils inse ed Need o ans o m he signal in he equency domain, he mo o cu en depend on he load, canno be used du ing non-s a iona y mo o ope a ion Need ex ensi e aining The me hod depends on he knowledge o a ious mo o pa ame e s and on he accu acy o he model, hei a ia ion (o inco ec ness) can esul in poo diagnosis pe o mance Ene gies 2020,13, 1573 20 o 24 5. Conclusions This pape p esen s a sys ema ic e iew o high TRL echniques o BLDC mo o s ailu es de ec ion, ha ha e been published in he pe iod o ime om he ea ly 1990s o No embe 2019. In addi ion, he a icle i sel can be conside ed as a p oo o concep o applying he SR o a pa icula s udy case in he ae ospace ield, in o de o demons a e i s easibili y. The s udies p esen ed in his wo k, ha e been analysed o espond o he esea ch ques ions posed, ha is, wha a e he echniques applied o aul de ec ion, he senso s used, he wo king condi ion, wha a e hei ad an ages and limi a ions. These esul s ha e been included in mul iple ables o illus a e he indings and ease he consul a ion. The g ea es di icul y encoun e ed du ing his s udy has been he impossibili y o compa ing he he di e en p oposed algo i hms in e ms o pe o mance, due o he lack o uni o mi y in es s, ea u es measu emen and es ima ion and p esen a ion o he esul s. The au ho s would sugges , as a possible solu ion o his issue, a s udy o in oduce a s anda dised benchma k and a se o pa ame e s o be p esen ed in o de o ha monise he e alua ion o he aul de ec ion algo i hms. Au ho Con ibu ions: Concep ualisa ion, V.M.F.; Funding acquisi ion, M.Á.M.P.; In es iga ion, V.M.F. and C.I.; Me hodology, V.M.F. and C.I.; Supe ision, M.Á.M.P.; Valida ion, C.I.; W i ing—o iginal d a , V.M.F.; W i ing— e iew & edi ing, V.M.F. and C.I. All au ho s ha e ead and ag eed o he published e sion o he manusc ip . Funding: The esea ch leading o hese esul s has been pa ly unded by he Eu opean Resea ch Council unde he Eu opean Union’s Se en h F amewo k P og amme (FP7/2007-2013)/ ERC g an ag eemen n. 785332 Con lic s o In e es : The au ho s decla e no con lic o in e es . Abb e ia ions The ollowing abb e ia ions a e used in his manusc ip : AI A i icial In elligence BLDC B ushless Di ec Cu en FEM Fini e Elemen Me hod HF High F equency MCSA Mo o Cu en Signa u e Analysis NN Neu al Ne wo k SR Sys ema ic Re iew SVM Suppo Vec o Machine TRL Technology Readiness Le el Appendix A Nomencla u e By going h ough he li e a u e, he e minology in his ield appea s non-uni o m. This is due o he ac ha aul de ec ion and diagnosis is usually dis ibu ed o e many di e en disciplines. The de ini ion o he ollowing e ms is speci ied in he glossa y sec ion and is based on [ 1 , 56 ]. This e minology will be used along he en i e documen . Faul : Unpe mi ed de ia ion o a leas one ea u e (cha ac e is ic p ope y) o he sys em ou o he accep able s anda d condi ion h eshold. The aul is a s a e o he sys em and can be o a ious ypes (manu ac u ing, assembly, main enance, so wa e, ope a o s, w ong ope a ion). I may no a ec he co ec unc ioning o he o e all sys em Failu e : Pe manen in e up ion o a sys em’s abili y o pe o m a equi ed unc ion unde de e mined ope a ing condi ions. Mal unc ion : In e mi en i egula i y in he ul ilmen o a sys em’s unc ion. I can a ise om one o mo e aul s. F om he desc ip ion i is possible o d aw he ela ionship be ween aul s, ailu es and mal unc ions (Figu e A1). Ene gies 2020,13, 1573 21 o 24 Fea u e Time e No mal S a e Faul Func ion Time 1 e 0 Failu e Func ion Time 1 e 0 Mal unc ion Figu e A1. Scheme o he ela ion be ween aul s, ailu es and mal unc ions [1]. Re e ences 1. Fico, V.M.; Vázquez, A.L.R.; P a s, M.Á.M.; Be nelli-Zazze a, F. Failu e de ec ion by signal simila i y measu emen o b ushless DC mo o s. Ene gies 2019,12. [C ossRe ] 2. Basak, D.; Tiwa i, A.; Das, S.P. Faul diagnosis and condi ion moni o ing o elec ical machines-A e iew. In P oceedings o he IEEE In e na ional Con e ence on Indus ial Technology, Mumbai, India, 15–17 Decembe 2006; pp. 3061–3066. [C ossRe ] 3. Buchwald, H.; A ido , Y.; B aunwald, E.; Jensen, M.D.; Po ies, W.; Fah bach, K.; Schoelles, K. Ba ia ic su ge y: A sys ema ic e iew and me a-analysis. JAMA 2004,292, 1724–1737. [C ossRe ] [PubMed] 4. Renehan, A.G.; Tyson, M.; Egge , M.; Helle , R.F.; Zwahlen, M. Body-mass index and incidence o cance : A sys ema ic e iew and me a-analysis o p ospec i e obse a ional s udies. Lance 2008 ,371, 569–578. [C ossRe ] 5. Ki chenham, B.; Cha e s, S. Guidelines o pe o ming sys ema ic li e a u e e iews in so wa e enginee ing. In EBSE Technical Repo ; Else ie : Ams e dam, The Ne he lands, 2007; pp. 1–65. 6. Hall, T.; Beecham, S.; Bowes, D.; G ay, D.; Counsell, S. A sys ema ic li e a u e e iew on aul p edic ion pe o mance in so wa e enginee ing. IEEE T ans. So w. Eng. 2011,38, 1276–1304. [C ossRe ] 7. Ki chenham, B.; B e e on, O.P.; Budgen, D.; Tu ne , M.; Bailey, J.; Linkman, S. Sys ema ic li e a u e e iews in so wa e enginee ing—A sys ema ic li e a u e e iew. In . So w. Technol. 2009,51, 7–15. [C ossRe ] 8. Ie a di, C.; O ihuela, L.; Ju ado, I. Dis ibu ed Es ima ion Techniques o Cybe -Physical Sys ems: A Sys ema ic Re iew. Senso s 2019,19, 4720. [C ossRe ] 9. Ge, Y.; Song, B.; Pei, Y.; Molle , Y.A.; Gyselinck, J.J. Analy ical Exp essions o Isola ion Indica o s o Pe manen -Magne Synch onous Machines Unde S a o Sho -Ci cui Faul s. IEEE T ans. Ene gy Con e s. 2019,34, 984–992. [C ossRe ] 10. Kao, I.H.; Wang, W.J.; Lai, Y.H.; Pe ng, J.W. Analysis o Pe manen Magne Synch onous Mo o Faul Diagnosis Based on Lea ning. IEEE T ans. Ins um. Meas. 2019,68, 310–324. [C ossRe ] 11. Heyda zadeh, M.; Za a ani, M.; Nou ani, M.; Akin, B. A Wa ele -Based Faul Diagnosis App oach o Pe manen Magne Synch onous Mo o s. IEEE T ans. Ene gy Con e s. 2019,34, 761–772. [C ossRe ] 12. Zandi, O.; Posh an, J. Faul Diagnosis o B ushless DC Mo o s Using Buil -In Hall Senso s. IEEE Senso s J. 2019,19, 8183–8190. [C ossRe ] 13. Liu, X.; Miao, W.; Xu, Q.; Cao, L.; Liu, C.; Pong, P.W.T. In e -Tu n Sho -Ci cui Faul De ec ion App oach o Pe manen Magne Synch onous Machines Th ough S ay Magne ic Field Sensing. IEEE Sens. J. 2019 , 1. [C ossRe ] 14. Ye, M.; Huang, J. Bea ing Faul Diagnosis unde Time-Va ying Speed and Load Condi ions ia Speed Senso less Algo i hm and Angula Resample. In P oceedings o he 2018 23 d In e na ional Con e ence on Elec ical Machines (ICEM 2018), Alexand oupoli, G eece, 3–6 Sep embe 2018; pp. 1775–1781. [C ossRe ] Ene gies 2020,13, 1573 22 o 24 15. Ullah, Z.; Lee, S.; Hu , J. A No el Faul Diagnosis Technique o IPMSM Using Vol age Angle. In P oceedings o he 2018 IEEE Ene gy Con e sion Cong ess and Exposi ion (ECCE), Po land, OR, USA, 23–27 Sep embe 2018; pp. 3236–3243. [C ossRe ] 16. Moon, S.; Jeong, H.; Lee, H.; Kim, S.W. In e u n Sho Faul Diagnosis in a PMSM by Vol age and Cu en Residual Analysis Wi h he Faul y Winding Model. IEEE T ans. Ene gy Con e s. 2018 ,33, 190–198. [C ossRe ] 17. Ahsanullah, K.; Jeyasanka , E.; Vignesh, A.; Panda, S.; Shanmukha, R.; Nada ajan, S. Eccen ici y aul analysis in PMSM based ma ine p opulsion mo o s. In P oceedings o he 20 h In e na ional Con e ence on Elec ical Machines and Sys ems, Sydney, NSW, Aus alia, 11–14 Augus 2017; pp. 1–6. 18. B eune al, R.; Cle c, G.; Nahid-Moba akeh, B.; Mansou i, B. Hyb id diagnosis o in e n- u n sho -ci cui o ai c a applica ions using SVM-MBF. In P oceedings o he 2017 IEEE In e na ional Con e ence on Fuzzy Sys ems (FUZZ-IEEE), Cincinna i, OH, USA, 1–5 Oc obe 2017; pp. 1–6. 19. Heyda zadeh, M.; Za a ani, M.; Ugu , E.; Akin, B.; Nou ani, M. A model-based signal p ocessing me hod o aul diagnosis in PMSM machine. In P oceedings o he Ene gy Con e sion Cong ess and Exposi ion (ECCE), Naples, I aly, 9–12 July 2017; pp. 3160–3164. 20. Mazzole i, M.A.; Bossio, G.R.; De Angelo, C.H.; Espinoza-T ejo, D.R. A model-based s a egy o in e u n sho -ci cui aul diagnosis in PMSM. IEEE T ans. Ind. Elec on. 2017,64, 7218–7228. [C ossRe ] 21. Obeid, N.H.; Ba is on, A.; Boileau, T.; Nahid-Moba akeh, B. Ea ly In e mi en In e u n Faul De ec ion and Localiza ion o a Pe manen Magne Synch onous Mo o o Elec ical Vehicles Using Wa ele T ans o m. IEEE T ans. T ansp. Elec i . 2017,3, 694–702. [C ossRe ] 22. Pa k, Y.; Fe nandez, D.; Lee, S.B.; Hyun, D.; Jeong, M.; Kommu i, S.K.; Cho, C.; Reigosa, D.; B iz, F. On-line de ec ion o o o eccen ici y o PMSMs based on hall-e ec ield senso measu emen s. In P oceedings o he Ene gy Con e sion Cong ess and Exposi ion (ECCE), Cincinna i, OH, USA, 1–5 Oc obe 2017; pp. 4678–4685. 23. Zhu, M.; Hu, W.; Ka , N.C. To que- ipple-based in e io pe manen -magne synch onous machine o o demagne iza ion aul de ec ion and cu en egula ion. IEEE T ans. Ind. Appl. 2017 ,53, 2795–2804. [C ossRe ] 24. Gu, B.G. S udy o IPMSM in e u n aul s pa II: Online aul pa ame e es ima ion. IEEE T ans. Powe Elec on. 2016,31, 7214–7223. [C ossRe ] 25. Jeong, H.; Moon, S.; Lee, J.; Kim, S.W. In e - u n sho aul diagnosis o pe manen magne synch onous machines using nega i e sequence componen s. In P oceedings o he 2016 IEEE In e na ional Con e ence on Indus ial Technology (ICIT), Taipei, Taiwan, 14–17 Ma ch 2016; pp. 170–174. 26. Mbo’o, C.P.; Hameye , K. Faul diagnosis o bea ing damage by means o he linea disc iminan analysis o s a o cu en ea u es om he equency selec ion. IEEE T ans. Ind. Appl. 2016,52, 3861–3868. [C ossRe ] 27. Moon, S.; Lee, J.; Jeong, H.; Kim, S.W. Demagne iza ion aul diagnosis o a PMSM based on s uc u e analysis o mo o induc ance. IEEE T ans. Ind. Elec on. 2016,63, 3795–3803. [C ossRe ] 28. Sen, B.; Wang, J. S a o in e u n aul de ec ion in pe manen -magne machines using PWM ipple cu en measu emen . IEEE T ans. Ind. Elec on. 2016,63, 3148–3157. [C ossRe ] 29. Abed, W.; Sha ma, S.; Su on, R.; Mo wani, A. A obus bea ing aul de ec ion and diagnosis echnique o b ushless DC mo o s unde non-s a iona y ope a ing condi ions. Con ol. Au om. Elec . Sys . 2015,26, 241–254. [C ossRe ] 30. Aka , M.; Hekim, M.; O han, U. Mechanical aul de ec ion in pe manen magne synch onous mo o s using equal wid h disc e iza ion-based p obabili y dis ibu ion and a neu al ne wo k model. Elec . Eng. Compu . Sci. 2015,23, 813–823. [C ossRe ] 31. Çi a, F.; A kan, M.; Gümü¸s, B. A new app oach o de ec s a o aul in pe manen magne synch onous 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. 316–321. 32. Lee, S.T.; Kim, K.T.; Hu , J. Diagnosis echnique o s a o winding in e - u n aul in BLDC mo o using de ec ion coil. In P oceedings o he 2015 9 h In e na ional Con e ence on Powe Elec onics and ECCE Asia (ICPE-ECCE Asia), Seoul, Ko ea, 1–5 June 2015; pp. 2925–2931. 33. Moosa i, S.; Dje di , A.; Ai -Ami a , Y.; Khabu i, D. ANN based aul diagnosis o pe manen magne synch onous mo o unde s a o winding sho ed u n. Elec . Powe Sys . Res. 2015 ,125, 67–82. [C ossRe ] 34. Chak abo y, S.; Kelle , E.; Ray, A.; Maye , J. De ec ion and es ima ion o demagne iza ion aul s in pe manen magne synch onous mo o s. Elec . Powe Sys . Res. 2013,96, 225–236. [C ossRe ] Ene gies 2020,13, 1573 23 o 24 35. Haddad, R.Z.; S angas, E.G. Faul de ec ion and classi ica ion in pe manen magne synch onous machines using Fas Fou ie T ans o m and Linea Disc iminan Analysis. In P oceedings o he 2013 9 h IEEE In e na ional Symposium on Diagnos ics o Elec ic Machines, Powe Elec onics and D i es (SDEMPED), Valencia, Spain, 27–30 Augus 2013; pp. 99–104. 36. Nyan eh, Y.D.; S i as a a, S.K.; Ed ing on, C.S.; Ca es, D.A. Applica ion o a i icial in elligence o s a o winding aul diagnosis in Pe manen Magne Synch onous Machines. Elec . Powe Sys . Res. 2013,103, 201–213. [C ossRe ] 37. Sa ikhani, A.; Mohammed, O.A. In e - u n aul de ec ion in PM synch onous machines by physics-based back elec omo i e o ce es ima ion. IEEE T ans. Ind. Elec on. 2013,60, 3472–3484. [C ossRe ] 38. U es y, J.C.; Riba, J.R.; Rome al, L. Diagnosis o in e u n aul s in PMSMs ope a ing unde nons a iona y condi ions by applying o de acking il e ing. IEEE T ans. Powe Elec on. 2013,28, 507–515. [C ossRe ] 39. Abu-Rub, H.; Ahmed, S.M.; Iqbal, A.; Toliya , H.A.; Rahimian, M.M. Incipien bea ing aul de ec ion o h ee-phase B ushless DC mo o d i e using ANFIS. In P oceedings o he IEEE In e na ional Symposium on Diagnos ics o Elec ic Machines, Powe Elec onics & D i es, Bologna, I aly, 5–8 Sep embe 2011; pp. 620–625. 40. Leboeu , N.; Boileau, T.; Nahid-Moba akeh, B.; Cle c, G.; Meibody-Taba , F. Real- ime de ec ion o in e u n aul s in PM d i es using back-EMF es ima ion and esidual analysis. IEEE T ans. Ind. Appl. 2011,47, 2402–2412. [C ossRe ] 41. A ellano-Padilla, J.; Sumne , M.; Ge ada, C. Winding condi ion moni o ing scheme o a pe manen magne machine using high- equency injec ion. IET Elec . Powe Appl. 2011,5, 89–99. [C ossRe ] 42. Kim, K.H.; Gu, B.G.; Jung, I.S. Online aul -de ec ing scheme o an in e e - ed pe manen magne synch onous mo o unde s a o winding sho ed u n and in e e swi ch open. IET Elec . Powe Appl. 2011,5, 529–539. [C ossRe ] 43. Ondel, O.; Bou leux, E.; Cle c, G. Diagnosis by pa e n ecogni ion o PMSM used in mo e elec ic ai c a . In P oceedings o he IECON 2011-37 h Annual Con e ence on IEEE Indus ial Elec onics Socie y, Melbou ne, VIC, Aus alia, 7–10 No embe 2011; pp. 3452–3458. 44. Eb ahimi, B.; Faiz, J.; A aabi, B. Pa e n iden i ica ion o eccen ici y aul diagnosis in pe manen magne synch onous mo o s using s a o cu en moni o ing. IET Elec . Powe Appl. 2010,4, 418–430. [C ossRe ] 45. Eb ahimi, B.M.; Faiz, J. Fea u e ex ac ion o sho -ci cui aul de ec ion in pe manen -magne synch onous mo o s using s a o -cu en moni o ing. IEEE T ans. Powe Elec on. 2010,25, 2673–2682. [C ossRe ] 46. Qui oga, J.; Ca es, D.; Ed ing on, C.; Liu, L. Neu al ne wo k based aul de ec ion o PMSM s a o winding sho unde load luc ua ion. In P oceedings o he 13 h Powe Elec onics and Mo ion Con ol Con e ence, Poznan, Poland, 1–3 Sep embe 2008; pp. 793–798. 47. Rose o, J.; Cusido, J.; Ga cia, A.; Rome al, L.; O ega, J. Faul de ec ion o eccen ici y by means o join ime- equency analysis in PMSM unde dynamic condi ions. In P oceedings o he In e na ional Symposium on In elligen Signal P ocessing, Alcala de Hena es, Spain, 3–5 Oc obe 2007; pp. 1–6. 48. Rose o, J.; Cusido, J.; O ega, J.; Rome al, L.; Ga cia, A. PMSM Bea ing Faul De ec ion by means o Fou ie and Wa ele ans o m. In P oceedings o he IECON 2007—33 d Annual Con e ence o he IEEE Indus ial Elec onics Socie y, Taipei, Taiwan, 5–8 No embe 2007; pp. 1163–1168. 49. Rose o, J.; Cusido, J.; Ga cia, A.; Rome al, L.; O ega, J. De ec ion o s a o sho ci cui s in PMSM by mean o join ime- equency analysis. In P oceedings o he IEEE In e na ional Symposium on Diagnos ics o Elec ic Machines, Powe Elec onics and D i es (SDEMPED 2007), C acow, Poland, 6–8 Sep embe 2007; pp. 420–425. 50. Ba endse, P.; Pillay, P. A new algo i hm o he de ec ion o aul s in pe manen magne machines. In P oceedings o he IECON 2006 - 32nd Annual Con e ence on IEEE Indus ial Elec onics, Pa is, F ance, 6–10 No embe 2006; pp. 823–828. 51. Rajagopalan, S.; Alle , J.M.; Res epo, J.A.; Habe le , T.G.; Ha ley, R.G. De ec ion o o o aul s in b ushless DC mo o s ope a ing unde nons a iona y condi ions. IEEE T ans. Ind. Appl. 2006 ,42, 1464–1477. [C ossRe ] 52. Ie a di, C.; O ihuela, L.; Ju ado, I. Guidelines o a sys ema ic e iew in sys ems and au oma ic enginee ing. Case s udy: Dis ibu ed es ima ion echniques o cybe -physical sys ems. In P oceedings o he Eu opean Con ol Con e ence, Limassol, Cyp us, 12–15 June 2018; pp. 2230–2235. 53. Mohe , D.; Libe a i, A.; Te zla , J.; Al man, D. Rep in -p e e ed epo ing i ems o sys ema ic e iews and me a-analyses: The PRISMA s a emen . Phys. The . 2009,89, 873–880. [C ossRe ] [PubMed] Ene gies 2020,13, 1573 24 o 24 54. Nandi, S.; Toliya , H.A.; Li, X. Condi ion moni o ing and aul diagnosis o elec ical mo o s-A e iew. IEEE T ans. Ene gy Con e s. 2005,20, 719–729. [C ossRe ] 55. Heyda zadeh, M.; Za a ani, M.; Akin, B.; Nou ani, M. Au oma ic aul diagnosis in PMSM using adap i e il e ing and wa ele ans o m. In P oceedings o he 2017 IEEE In e na ional Elec ic Machines and D i es Con e ence (IEMDC), Miami, FL, USA, 21–24 May 2017; pp. 1–7. 56. Ise mann, R. Faul -Diagnosis Sys ems: An In oduc ion om Faul De ec ion o Faul Tole ance; Sp inge : Be lin, Ge many, 2006; pp. 1–475, a Xi :1011.1669 3. [C ossRe ] c 2020 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 p://c ea i ecommons.o g/licenses/by/4.0/).