You are on page 1of 16
A REVIEW OF ON - LINE CONDITION MONITORING TECHNIQUES FOR THREE - PHASE SQUIRREL - CAGE INDUCTION MOTORS -PAST PRESENT AND FUTURE ‘W. T. Thomson, Senior Member, IEEE ‘The Robert Gordon University, Schoolhill, Aberdeen, Scotland ABLO 1FR rhe main objective of this review is to provi industry with state of the art knowledge on condition ‘monitoring strategies for assessing the operational health of induction motor drive systems. An overview of typical failure ‘mechanisms and their causes is presented. ‘The essential caiteria for 2 reliable condition monitoring system are Citical speeds/shaft resonances. ‘+ Stator core vibration problems/teeth resonances ‘This list is not complete since there is the mechanical oad to consider. Pumps, fans, compressors (ete) can have their own unique mechanical problems and it is not ‘possible to cover all of these in this paper. Il. ON-LINE MONITORING TECHNIQUES FOR DETECTING SPECIFIC PROBLEMS AND FAULTS, A, Online Diagnosis of Broken Rotor Bars Inaroduction In general, on-line condition monitoring and diagnostics requires the sensing and analysis of signals that contain specific information, which is characteristic of the degradation process, problem, or fault to be detected. During the past twenty years various monitoring and diagnostic. strategies have been proposed for the diagnosis of problems in induction motors. With respect to detecting broken rotor bars in cage induction motors signals such as airgap fiux [18], axial flux [19], speed (20), torque [21,22], vibration [12] and current [13-15] have been sensed and analysed. It is accepted that all these signals will change due to broken rotor bars since there is @ change in the frequency content of the airgap flux signal. Figure (5) is presented as a diagrammatic aid for further discussion on the applicability of sensing certain signals. The field plot in figure (5) is only an aid and is not what would actually exist in the machine but itis useful to the reader with respect to understanding the different flux regions. There are various factors that nneed to be considered when selecting the most appropriate monitoring technique for application in the industrial environment, these are: Sensor should be non-invasive. 2. Sensor and instrumentation system must be highly reliable. 3. The diagnosis must be reliable, 4, ‘The severity ofthe problem should be quantified. $. Ideally, an estimation of remaining run-tife should be given. 6. Ideally, © prediction of the fundamental cause/s of the problem/fault should be provided via on-line information from sensors ete. Due to the complexity of degradation mechanisms, problems and faults it is extremely difficult and in most ‘cases impossible to achieve all the above criteria. Due to the worldwide interest in condition monitoring and diagnostics of drive systems substantial advances have been made during the past ten years. In many cases itis possible to achieve eriteria (1) to (4). However, riteia (S) and (6) are extremely difficult to achieve due to the numerous variables that effect the determination of the causes’s of a problem and prognosis for Future run-life, ‘These are challenges that have still to be overcome. Basic Theory For a balanced stator winding the fundamental m.m.t. can be expressed as: my = My (@~ PB} roe Lt) ‘Due to broken rotor bars and consequential variation of the rotor impedance the rotor mam.f. will be subject to modulation. For simplicity, consi only the fundamentals with reference to the stationary stator winding a rotor winding asymmetry will generate m.m.f. components given by 120] any Min (9-15) =390- 4 sn a=29)- p02]. ‘The second of these terms alone will induce a stator winding e.m.f, and hence a current which is @ measure of rotor asymmetry. The cyclic variation in current causes a torque pulsation at twice slip frequency (2sf,H2) and a speed variation which is a function of the system inertia, This speed oscillation produces a reduction in the current component at f,{l-2s)Hz and a component at f,(1+2s)Hz appears. This latter ‘component is also enhanced by modulation of the third time harmonic flux. It should also be mentioned that any cyclic speed oscillation by the mechanical load will also cause modulation of the current signal, This can be ‘confused with the twice slip frequency modulation of current caused by broken rotor bars. It is extremely ‘important that any diagnostic technique can differentiate between mechanical load oscillations and actual rotor ‘winding asymmetry [23,24]. Diagnostic techniques ‘A real measure of the success of any diagnostic technique is its reliability and usage in the industrial environment. ts ability to fulfil as many of the criteria (1-6) specified above has also to be assessed, The author accepts that signals such as speed, airgap flux, axial flux, vibration or current can be analysed to dotermine the presence of broken rotor bars [12,13,15,18-20,25-29]. Researchers have proved this ‘under Isboratory based test conditions but it now has to bbe accepted that current monitoring and analysis is the preferred method used by industry. Current monitoring hhas been commercialised and at least four to six companies are now offering eurrent monitoring systems that include the diagnosis of broken rotor bars, {tis also the method that is most applicable in industry and is hhighly likely to meet criteria one to four above. Nomerous on-site case histories have verified the reliability of current analysis to diagnose broken cotor bars. However, itis not simply a case of detecting the twice slip frequency sidebands in the current spectrum ‘and to then state that Figure (6) Current spectrum, twice components evident, broken rotor bars present. Motor pump drive, 3.5MW, 6.6KV, 3580r.p.m., 60H2, 46 rotor bars broken rotor bars are present. ‘The diagnosis from sensing only the current and its subsequent analysis is ‘much more complex than it at first appears [30]. This is particularly nue when a commercial diagnostic system has to cater for motors between a few kilowatts up to 20MW, of different designs, driving a diverse range of ‘mechanical loads [31]. The accurate on-tine prediction of slip, from onty @ current analysis, to cope with such range of induction motor drives operating at different load conditions is not @ simple task. The accurate prediction of slip is required for the diagnosis of the {wice slip frequency sidebands. This has been achieved ‘but any user should clearly establish with the suppliers of current monitoring systems that it can cope with all induction motor drives. The diagnosis of purely mechanical load oscilitions/disturbences. needs to be included in the diagnostic system since sidebands in the current spectrum can be produced by purely mechanical phenomena and are therefore independent of broken rotor bars (23, 24], This will avoid an incorrest diagnosis of broken rotor bars. For example, coal crusher induction motor drives in coal fired power stations, conveyor belt drives, and gearboxes in the drive train ete can ali produce sideband eurrent components around the supply frequency that can, under certain circumstances, be identified as coming from broken rotor bars (23). The intelligent diagnosis of the twice slip frequency sidebands due to broken rotor bars is therefore required. Certain magnetic circui'rotor shaft designs in large indvtion motors can also cause twice slip frequency modulation of the current, For example, {4 very poor rotor design that has the same number of rotor spiders Figure (7) Actual rotor after removal, 20 broken rotor bars between the shaft and rotor core as there are poles will result in twice slip frequency modulation of the magnetising current. This has caused confusion in a small number of eases and led to an incorrect diagnosis of broken rotor bars [23]. Of course this can now be avoided via an intelligent diagnostic system. The actual load operating condition of the motor also needs to be taken into account for the prediction of the severity of broken rotor bars [13]. For a given number of broken rotor bars tie magnitude of the sidebands can change the load is varied. ‘This is not true forall motors since it is a function of the motor rating, design and load characteristics. However, any diognostic system must include intelligent diagnosis to cater for this ‘phenomenon occurring in practice, ‘The reader is now referred to the references for actual on-site histories. A sample of a typical case history is given in Figures (6) and (7) and show the current spectra ‘and faulty rotors. Typical current monitoring systems ‘are given in the references and Figure (8) gives an example. Note that the CT is remote from the motor ‘environment. B, OnLine Diagnosis Of Stator Winding Insulation Degradation And Faults Introduction In sections A and B of II in this paper a clear distinction was given between stator winding problems/faults in stator windings operating at below(LV and MV). and abovo(HV) 4.1kV respectively. The on-line diagnosis of insulation degradation in HV motors that leads to stator ‘winding failure is quite different from the diagnosis of sctual faults (shorted tums within a coil) in LY stator phase ‘Spectrum Ansiyser ‘Dats Collectar Signal Condtioner Bel orercre —--___ cable I Leeman Figure (8) Typical current monitoring system — applicable inthe industrial environment ‘windings. [ts important to distinguish between the on- line monitoring techniques applicable to LV and HV stator windings. Online diagnosis of shorted turns within coils in LY Stator Windings With reference to Figure (5) it has been shown that the sensing and analysis of, for example, airgap flux or axia/siay leakage flux can identify components indicative of shorted tums in stator windings [18,19, 33,34]. Figure (9) shows the typical positions of flux sensors. The use of airgap search coils is highly invasive and is not a practieal option for machines already installed and it is not popular in new machines. Operators do not like invasive transducers. Axial flux ‘monitoring via a search coll around the shaft can be used to identify shorted tums, although it is less invasive than airgap search coils, it is still invasive to the motor’s environment, Typical frequency components in the axial flux signal that are indicative of stator winding faults are given in references [33, 34]: Studies carried out in laboratories have shown that axial ‘lux monitoring can detect shorted tums, however, there is distinct lack of industrial case histories to fully assess the reliability of the technique when applied to a ‘wide range of LV induction motors of different designs ‘and ratings. It should also be noted that to ensure a high ‘quality signal the axial flux coil should be inside the on of Stray Flux sensors — note they are invasive to the motor's environment rmotor's outer frame when the fame is steel. It then of ‘course becomes highly invasive, Any outer stec! frame ‘acts as a magnetic shunt and with thick stel frames the flux penetration depth is also another relevant factor. Results have also shown that components in the axial flux that are indicative of shorted tums are also very load dependent {33}. In fact the change in magnitude of ‘these axial fux components can be much higher due to load changes wien no shorted turns exist in comparison, to the changes in magnitude that occur due to a fault ata given load condition — see reference (33) for quantifiable data. This is clearly a problem for reliable cliagnosis. It has also been proposed (a commercial system is available) that stray flux sensing (Smart sensor) around the outer frame can be used to diagnose stator and rotor winding faults. Users should be aware of the application of this method is resily only suitable to barrel type motors and not large motors with complex steel structures as shown in (Figure (10)). The relatively thick stel frames will AB I tg Spe 3s Se Cae eta Figure (10) Photo : Courtesy of Rolls Royce, Peebles Electric Machines, Edinburgh, Scotland severely attenvate leakage flux . The position of the transducer is another variable that will effect the magnitude of the signal. 1 tis accepted that it is feasible to diagnose shorted tums from leakage fix monitoring bout there isa lack of industrial case histories to validate the technique and the signal will also be influenced by load variations. Results have not been published that confirm that une shorted turn (minimum fault) within a coil can be diagnosed via stray flux monitoring. in essence the early inception of a stator winding feilure has not been proved for a wide range of LV induction motors operating at different loads. The prediction of the severity of shorted tums in LV stator windings via an analysis of flux or any other signal has not yet been achieved. Since current monitoring has been proved to be highly successful at diagnosing rotor winding faults and the fact that a CT is non-invasive the operators are keen that the same signal can be analysed to diagnose stator winding fauks in LV induction motors. Results have been published which show certain current components can bo identified that are functions of shorted turns in stator coils [35]. With eight tums (10%) shorted in an eighty turn coil ( for a smal, phase, 418¥. 3BW, Apo 1A, [3Scpm, 50H SCIM. rotor and stor slots equal 27 and 36 respectively), one of the rotor slotting components (702Hz) in the current signal increased by 2648 (20 times increase) at full-lond operating condition (36). However, it was also shown that these components are & function of load on the motor when no faults exist, as was the case with axial flux monitoring. More recent tests ( in 1999 ) on 2 larger motor ((1kW, 415V, dela ~ sonnecied, S0H2, 20A, W4dOrpm, SCIM, ReSlsiats, S=36sot: commerillyprodeced motor) have confirmed that with only fone turn shorted in a 15 tum coil that one of the rotor slot passing frequency components (1274HHz at 1440r.pm. ~ the fall load speed) in the current signal increased by 204B (factor of 10). This is withthe actual current in the shorted tum limited to 20A (fall-loed) via an external short circuit current of 30A (limited by an external short circuit resistance to avoid burn ou). Figures (11) and (12) show the increase of the 1274Hz component and these current spectra are produced after high pass filtering to remove the dominance of the supply frequency component. Figure (11) Current spectrum — note the level of the component at 1274Hz.: no stator fault Figure (12) Current spectrum ~ note the increase in level ‘of the component at [274Hz : one shorted tum in a 15 umn coil ‘One shorted tum can therefore be diagnosed and it should be recognised that the actual short circuit eurrent ‘would be much greater with an actual short circuit in the ‘end winding. Full short circuit tests of turns have been completed but the results are of a confidential nature, The components ate still a function of load with no shorted tums but with full shorted tums the change is greatly in excess compared to load changes, In any ‘monitoring system it is still wise to include the load {influence in the diagnosis of shorted tums. These results demonstrate the real potential of current monitoring to diagnose shorted tums in LV, 3-phase, induction motors prior to a phase to phase or phase to ground fault developing. as Sos wfileene 3 2h na 6) Equation [3] is used to calculate the rotor slot passing frequency components. It should be stated that these components are also a function of nirgap eceentrety but 4 unique signature patter can be identified that is able to discriminate between the diagnosis of the two ifforent types of problems. However, this is commercially confidential and cannot be published. On-Line Diagnosis of Insulation degradation/Fallure in HY Stator Windlings There is a fundamental difference between on-line monitoring of the health of HV (4.1kV and above) stator windings compared to the diagnosis of actual shorted tumis in LV windings. in the former, itis the diagnosis of insulation degradstion prior to the development of actual faults such as, shorted tums/phase to phase/phase to ground developing that has to be achieved. In an HV ‘machine itis pointless monitoring for the diagnosis of a shorted tum/s it is widely accepted that the subsequent time taken to failure is very small (secs or ‘mins?). Iti for that reason that on-line partial discharge {PD) monitoring is used as a means of assessing the insulation health of an HY stator winding. Research bas proved that PD monitoring can identify insulation Gegradation prior to complete breakdown and a subsequent fault that activates the normal protection equipment for an HV machine [4,5]. The technique is now widely applied in industry and successful case histories have verified its validity (36,37), Several ‘commercial PD monitoring systems are now available. ‘There are different PD sensing and analysis systems, for ‘example, using capacitive couplers or Rowgoski coils ‘and the reader is referred to the published literature for further details. Both these techniques have been successfully applied in industry. The main point the author of this paper is making is that there is a clear distinction between the monitoring techniques, which are applicable to HY and LV stator windings, ©. On line Diagnosis of Airgap Eccentricity in 3- Phase Induction Motors ‘A discussion on airgap eccentricity and fts causes has already been given in this paper. Various methods have teen proposed for the on-line diagnosis of sirgap eccentricity in 3-phase induction motors. These include the monitoring and spectrum analysis of airgap flux {18}, axial flux [19}, stator core vibration of current [16,17]. Ail these signals will show changes in their respective spectra due to an increase in airgap eccentricity. Itis suffice to state that current monitoring is now the preferred method by industry. Industrial case histories have confirmed that abnormal levels of aitgap eccentricity can be diagnosed via an analysis of the ‘current signal [16,17]. Of particular importance in the analysis i to take account of the combined influence of static and dynamic eccentricity since both will always ‘occur simultaneously in an cleetrical mechine [38]. The ‘changes in the signal (cutrent, flux or vibration ) are @ function of the interaction between static and dynamic eccentricity flux waves [39]. For reliable diagnosis, several sections of the cutrent spectrum should be interpreted based on equation (3) above and the following equations: massing tng 22nap2ngp)- Sa =f the Equation (4) gives the pole-paits of the respective flux ‘waves predicted from equation (3) and a compatibility analysis between these pole-pairs and the harmonic ppole-pairs for the stator winding of the particular motor being monitored must be carried out (39). This is required to establish whether the particular flux components predicted by equation (3) will actually induce a voltage and current in the stator winding. Equation (5) also predicts frequency components that are a function of airgap eccentricity [38]. It was originally thought that these components were only due to dynamic eccentricity and had pl pole-pairs and could not therefore induce current components at these frequencies with » p pole-pair winding [18,40]. This ‘was theoretically correct if only dynamic eccentricity is being considered in isolation from static eccentricity ‘The author of this paper ( who considers thit precise and conirled experimental tens o> commercial motors to be of egtl inportnce ss thonctical models ) Was the first to show that with inherent dynamic eccentricity and controlled changes to static eccentricity that the components given by equation (4) were present and increased in magnitude as static eccentricity increased [41 ]. A subsequent theoretical analysis was catied out to explain the experimental phenomenon [39]. This proved that due to modulation effects between static and dynamic eccentricity waves that flux components at frequencies Figures (13) and (14) — Current spectra [17] for a L45MW, SCI IM with normel airgap eccentricity. Note there are no ‘components at fj = f, and the dynamic eccentricity component at 1013Hz is zero. Figures (15) and (16)~ Current speetra [17] for a 1.45MW, SCIM with abnormally hi (35%) airgap eccentricity. Note there are obviously components at f; + f, and both dynamic eccentricity components are evident and of equal magnitude given by equation (3) occurred with p21 pole-pairs hence these flux waves can induce curents in the stator ‘winding. The results shown in Figures (13) to (16) confirm the case where abnormal levels of airgap eccentricity were diagnosed in a large (14SMW) induction motor by examining the spectra for the ‘components predicted from equations (3) to (5) 173]. Recent rescarch has shown that the Finite Element method can be used to predict the magnitude of current components as a function of the level of airgap eccentricity [42]. This means the severity of airgap ceecentricity can be predicted and this has now been applied to a large (J.45MW) induction motor operating. in industry. D. Online Diagnosis of Mechanical Problems in Induction Motor Drives It is universally accepted that vibration monitoring and analysis can be highly successful at detecting mechanical problems in electrical drives. The author does not intend presenting details on this since itis well covered in the literature. Case histories on the diagnosis of bearing weariprobiems, rotor imbalance, and shafvcoupling misalignment etc are well documented. ‘Numerous commercial vibration monitoring systems are available and these include advanced intelligent techniques. With respect to new advancements in detecting mechanical problems in Induction motors there has been considerable research during the past five years ‘on the application of current monitoring to diagnose, for example, bearing problems and shafveoupling misalignment [41,43,44.45}. ‘There is evidence, via laboratory tests, that faults in roller clement beatings or shafveoupling ristlignment ‘cause changes in the current spectrum, This means there is the potential that relatively inexpensive current ‘wonsformers can be used to detect mechanical faults in Induction motor drives in comparison to more expensive vibration transducers and associated pre-amplifies, However, there is a distinct lack of case histories of current monitoring being applied in industry to detect mechanical faults, although it has been stated that Beta site testing is underway on one particular system for detecting bearing faults [44]. Before current monitoring, can replace vibration monitoring for detecting, mechanical problems on critical induction motor drives application in industry. If it is proven to be reliable it ‘could be used on a wide range of induction motor drives ‘of different ratings as a first line surveillance technique for detecting mechanical abnormalities. Vibration ‘monitoring and advanced analysis could then be used to identify the specific mechanical problera. It is highly unlikely that current monitoring will replace vibration monitoring on large (MW) induction motors that use plain bearings. IV. CONCLUSIONS - FUTURE DEVELOPMENTS, ‘On-line monitoring and diagnostics are now being effectively applied for the early detection of broken rotor bars, airgap eccentricity problems and insulation degradation of HV (¢.1kV and above ) stator windings. Case histories have verified that these problems can be reliably detected in industrial induction motor drives and commercial systems are available, ‘The reliable, on-line diagnosis of shorted tums (e.g. ‘one/two tums in a coil) in LV stator windings has stil 10 ‘obtain credibility in industry but it i likely this will be achieved in due course. Al/Nueral Nets/Fuzzy Logic have been researched to improve the diagnosis of problems in induction motors and some interesting publications on this topic are available (25-27, 45-53]. ‘The reliability of such schemes has still 10 be verified via successful on-site case histories ‘The monitoring and diagnostic strategies previously discussed consider the analysis of signals ftom induction motors operating under steady state conditions. Research has also been done on the analysis of the transient current to diagnose problems (54-56). When ‘the motor is not connected to a mechanical load the problem of detecting broken rotor bars or end rings is not possible since the current in the rotor is negligible (5=0) and the twice stip frequency sidebands cannot be separated from the supply frequency component. In certain installations large motors are not permitted to be removed from the site, Repair and maintenance is therefore carried out in the plant. This often means there js not a loading facility to test arepaired motor, Jn these ‘circumstances it is the transient current that has to be analysed to determine the rotor winding condition. Results have shown that rotor winding asymmetry can be diagnosed via transient current analysis and Finite Element simulations have velidated the predictions (54- ST. In general, the focus has been on the sensing and analysis of signals to diagnose only the problem or fault. ‘The identification of the fundamental cause of the problem has not featured in the monitoring strategy. In ‘the future signals should also be analysed that would provide knowledge on the possible causes of the problem that has been diagnosed. For example vibration should be analysed to determine whether itis the cause of rotor winding problems or mechanical degradation of the stator winding insulation. This would be in addition to monitoring current and PDs to detect rotor and stator ‘winding problems. It is also known that shafv/coupling, misalignment causes high mechanical forces and these ean cause electrical problems. The concept of diagnostics systems that include problem and cause ‘detection is the way forward, ‘This leads on to the prediction of remaining lifetime after a problem has been identified. This is slways asked by the user and is extremely difficulvimpossible to awnser due to number of variables that can effect remaining lifetime. The fundamental cause of the problem has to be determined at the same time a the diagnosis of the inception of the problem has been identified for there to be any chance of predicting the remaining life of the machine. The operational characteristics of the drive system have (0 also be assessed etc, At present only very broad guidelines can bbe given based on case histories and problems in similar machines, ‘The main challenge in the furure is to create intelligent diagnostic systems that include: “FAULT DIAGNOSIS, CAUSE DIAGNOSIS AND PROGNOSIS. OF ‘REMAINING LIFETIME” V. REFERENCES LU EPRI: “improved Motors (or Utility Applications and Improve Motors for Usity Applications, Industry Assessment Stuy" Vol I, EPRUEL-2678, Vol | 1763+, il roport and EPRI EL 2614, Vol2, 1763-1 final eport Oaober 1982 2} TREE: Motor Retisbility Working Group, “Report of Lage ‘Moor Reliniity Survey of Industrial Comercial Insalltons”, Par I, Transactions on Industry Application, Vol LA-21, No 4, {ly August, 1985, Pa | and, pp 853-872 D1 OV Thorsen and M_ Dalve: “Condition Montoriog Methods, Fallue Identfeson and Analysis for High Voltage Motors in Petoshemical Industy", Proc 8 IEE Int Conf, EMD'97, Univesity of Cambridge, No 444, pp 109-113 8) GC Stone, B A Floyd, SR Campbell and HG Seddig: “Development of Automatic, Continuous Partial Discharge ‘Monitoring Sysieme to Detect Motor and Generator Partial Dischacges, Proc IEEE, IEMDC', 1997, pp MA2-3.1 - MA2- 33 {8} DG Edwards: “Planned Maintesance of High Voltage Rotating ‘Machine Iosulaon Based Upon fnformaion Derived fom On= line Discharge Measurements", Proe TEE Int Caaf on “Life “Managerocnt of Power Plans", Vol 401, pp 101-107 16) MM (Thys) Botha: “Electrical Machine Failures, Causes and Cures", Pro IEE, EMD'97, University of Cambridge, No 464, |= 3.1997, pp 1417 4 (7) AH Bonnet and G C Soukup: “Cause and Analysis of Stator and ‘Rotor Faloes ia Three-Phase Squiel-Cage Industion Motor", IEEE Transpetions on Industry Applications, Vol No 28, No 4, Joly/August 1992, pp 921-537 18) AH Bonnet: “Analyse ofthe Impact of Pule-Wikth Module! Toverter Voltage Wavetorms 00 AC Induction Motors IEEE Transactions om Indosty Applications, Vol 37, No 2, MareW Apel 1996, pp 385-392 [PLM Kauthol, G Borner, M Ehertar and J Speck: “Palle ‘Mechanism of the Interim lesulation of Low Voltage Electric ‘Machines Fed by Puse-Convolled inverters, IEEE. Eleesical Inslction Magazin, Seplmbe/October 1996, Vol 17, No, 5 6 [N0} AM Baye, F Grigonescu, T Lebey and Bai Ai: “Existence of arial Discharges in Low-Voltage Induction Machines Supplied ‘by PWM Drives", {EEE Transactions on Dieleeien end letra! Inulin, Vol3, No‘, August 1996, gp 584-560 [11] WT Thomson: “Diagnosing Faults in Induction Motor ~ Engineering Meas”, Electrical Review, Vol 215, No 11, Nov ist (12) WT Thomson, ND Deans, RA Leonard and A 3 Miloe: “Monitring Stalegy for Discriinating Between Difiereat “Types of Rotor Cage Feuls", Proc 18* UPEC, University of Surrey, April 1985, {13} W-T Thomson and D Rankin: “Case Histories of Rotor Winding Fault Diagnosis in Induction Motors", 2" Int Cont Proe on ‘Condition Monitoring, Unversity College Swanse, March 1987 {14) GB Xiiman and} Stein: “Induction Motor Fault Detection Via Passive Curren Monitoring”, Prec Int Conf (ICEM'90), MIT, ‘Boston, USA, 1950, pp 13-7 [15] WT Thomson, § J Chalmers and D Rankin: “On-line Curren Monitoring and Fault Diagncsis In High Volige Tadusion Motor Case Histories and Cost Savings in. Offhore Thstallaions", Offshore Europe 87, Conf Pros SPE September 1987, Aberdeen, SPE 16577/1~ SPE 1657710 106) 1 R Cameron, WT Thomson and A B Dow: “Vibration and CCareat Monitoring for Detecting Aitgap Becengisity in Large duction Mora, Proc IEE Joural, Part B, Vol 133, No 3, May 1986. {17] WT Thomson, D Rankin and DG Donel: “Ondine Curent “Monitoring to Diagnose Airgap Eccentricity ~ An Industrie Case istry of Large HV, 3-Phase Induction Motors", Pro, of TEEE Int Conf FEMDC'S7, May 1997, Milwaukee, USA (Also due (0 ‘be published in Transactions of IEEE on Energy Conversion) 8] § Fruchenecht, E Pius and H Seinsch: "A Diagnosis Sytem for Three Phase Asynchronous Machines", Proc TEE Conf, EMDA'S9, Vol 310, IEE Savoy Place, London, 1989, pp 163+ m [u9) P-Tavner aod J Penman; “Condition Monitoring of Electrical Machines”, Research Studies Li, Joho Wiley & Sons, °, England, 1987 {20} C Hirgis, B G Gaydon and K Kamish: “The Dereon of Rotor Defect in Indoction Mol", Proc TEE EMDA Conf, Londos. 1982, pp 216-220 {21} F Thollon, G Geliet, A Jamal: “Asynchronous Motor Cage auf Detection Through Eleczomagneic Torque Mesremest”, EER, Vol 3, No 5, SeprembeOetaber, 1993 {22} H Yahout and @ Grelet: “Analysis of Harmonie Components of the Elecromagnetc Torque of an Asynchronous Motor With an End-Ring Fault, Proc ICEM96, Vigo, Spain, Vol Il, 1998. PP 396398 0) WT Thomson: “Online Curent Monitoring ~ The Influence of Mechanical Losds/Utique Rowr Designs oo the Detection of Broken Rotor Bart in SCIMS",ICEM92, UMIST, 1992 [24] RR Sohoen and F G Haheter: “Eos of Time-Varying Loads fon Rotor Fautt Detection in Indvtion Machines", IEEE, 1AS ‘Rana Coaf Pros, October 1993, p 324-330 5) F Filippi, G Fancheschiok and C Tassool: "Neural Networks ‘Aided Online Dingrasties of Toducéon Motor Rotor Faults” IEEE, 1AS Annael Conf Pree, October 1993, pp 346-523 Po) F Filipe, G Franshesshin| end C Tassoni: “Newal Nesworks Aided Ortine Diagoostis of Induction Machines’, Transactions TEEE-IAS,Jay/Augus, Vel 31, No 4, 1995, pp 897-899 [21 RIRSchoco, B K Lin, FG Hubeter, HJShlog and $ Farag: “An Untupervied Online System for Induction Motor Fault election Using Sisior Curent Monitoring”, IEEE-IAS ‘Transactions, November/December, Vol 31, No 6, 1995, pp 1240-1286 (28) A} M Cardoso, § MA Cruz, JF § Caraiho and E § Sea ‘eoior Cage Fault Dlagnsis in Thre-Phase Induction Motos by Park's Vector Approtsh, Proc Coat TEEE-IAS, 1995, pp 642- 646 [29] AJM Cardoso and E 5 Serva: "Onne Diagnostics of Three- Phase Induction Motors by Pu’s Vector” Proc. ICEM'BS, Piss, oll pp 231-234 [Go] WT Thorton: “On-ine Curent Moritoing to Detect Blecricl td Mechanical Fauhs in Three-Phase Induction Motor Drives", Proc IEE and IMECHE (London), Int Conf Proc on Life Management of Power Plans, Heriot-Watt Universiy, ainburgh, December 1994 [21] WT Thomson and § J Chakners: “A New Ondine Computer Based Current Monitoring, Sysiem for Expert System Fault Dingnosis of Induction Motos", Proe UPEC'SA, Trent Polyiecnic, 1985 152} J Peaman, JG Hiadwick and B Barbour; “Detection of Faulty in Eleetrieal Machines by Examioaton of the Axilly Directed Flakes", Proceedings ICEM'78, Brscels [D3] J Penman, HG Sedding,B A Lloyd und W 1 Fink: “Detetion ‘and Location oferta Short Cults inthe Stator Windings of ‘Operating Mixon, TEEE Transactions on Energy Conversion, ‘Vol 9, No, December 1994 [24] D E Knights: “Laboratory Assessment of Vibration and Stuy Field Condition Monitoring”, ERA Sport No #2-0031, Jantary 1962, ERA, Surey, KT2275A (115 page epor) 15] WT Thomson and C MMcRee: “Online Current Monitoring 1 Detect ImerTur Winding Faults in Induction Motors", Proc UUPEC'S9, Queens University of Belfi, 1989 15 » [36] D G Béwards: “Ondige Diagnosis of Defects in the Stator ‘Wining Structures of High Voltage Rotating Machines" Proc of ICEM'92, September 1992, UMIST, England [37] @.C Stone, HG Sedding and MJ Costello: “Apslication of Paral Diccharge Testing to Motor and Generator Stator Winding Maintenance", IBEE Transactions on Industy Applications, Vo $32, No 2, March/April 1996, pp 459-164 [581 DG Dorrell, W T Thomson snd S Roach: "Analysis of Airgap Flux, Curent and Vibration Signals as a Function of the Combination of Site and Dynamic Airap Ecearcty in 3- Phase Induction Motos", IEEE Transactions on Indusry ‘Applionions, Vol33, No I January/February, 1997, pp 24-34 189} WT Thomson, A Bartour, C Tasonl and F Filipe: “An Appraisal of the MME-Permeance Method and Finite Element Models o Sandy Sule Airgap Eoceatcty and its Diagnosis in Induction Machines’, Proc ICEM'98, Intanbul, 1998 {40} G Kron: “Equivalent Ciruits of Elvtrie Machines", New York Power Publication, 1967 141] WT Thomason: “Ondine Curent Monitoring to Diagnose Shaft ‘Misalignment in 3-phase Induction Motor Drive Systems’, roe ICEM'94, Pars, 1994, pp 238-243 [42] WT Thomson and A Barbour: “Ontine Curent Monitoring and Application of a Finke Elerent Method to Predict the Level of ‘Attap: Eccentricity ia Phase Induction Motors’. IEEE ‘Trannctions on Energy Conversion, Vol 13, No 4, December 1998, pp 347-357 (includes discussion and closure) [HS] RR Schoco, FG Haber, F Kampon and KG Barthelé: “Motor Hearing Dumage Detection Using Stator Cament Monitoring”, Proc 29" IEEE-IAS Meeting, October 1994, pp 110-116 44] B Yuri, GB Kliman, W J Premedini, RA Foes, G B Robinson and A Abdel-Malek: “An Adaptive, On-line Sisal Method fo Bewsing Fault Detection Using Stor Curent” Proc 32! IEEE-IAS Meeting, New Orleans, October 1997, pp 213+ 20 [45] MF Cansbas, M G Melero, 1G Alexandre and J Solar: “Sha Misalignment of Induction Motoe Using Current Special ‘Anaijs: A. Theorleal Approach”, Prot, ICEM'96, Vio, ‘Spain, pp 384-589 [A6} F Filippet, G Frenceshini and C Tasso: “AI Techalques io Induction Machines Diagnosis Including the Speed Ripple Effect", IEEE Transsctions on indusuy Aplications, Vol 34, Nol, January/February 1998, pp 98-107 [49] F Filippo, © Franceshii and C Tastoni: “NN Aled On-ne Diagnaies of Induction Motor Rotor Fouts”, IEEE ‘Transactions ind App, Vol 31, Joly/August 1995, 9p 892-894, {8} F ips, Franceshini,C Tassonl and P Vas: “A Fuzzy Logic ‘Approach to Online Induction Motor Diagnostics Based on Stator Curent Montaciag”, IEEE Proc. Stockholm Power Tech Cont, Sweden, June 1995, Vol Eleeical Machines & Drives, pp S6-161 (49) D Leith, ND Desns und 1D Stewart: “Condition Mositoring of leaical Machines Using Real Time Exper Systems", Proc. ICEM'R8, Pisa, September 1988, pp 297-302 [50] C Mo-Yoen und C Manyum: “An NN Approach to Real Time Condition” Monitoring of Induction Machines"” IEEE “Transactions Ind Electron Vol 38, December 1991, pp 448-483, [34] J Penman and C M Yi “The Application of Artscial NN to denity Faults In Induction Machines", Proc ICEM"92, ‘Manchester, UK, 197, pp 1256-1260, {52} E Richie et a: “Diagnosis of Rotor Faults in Squinel-Cage Indection Motor: Using Fuzzy Lagic Approsch, Pros ICEM™94, Pai, Vo 2, 1994, pp 348-352 53] P. Vas: "Parameter Estimation, Condition Monitoring and Diagnosis of Hlecuieal Machines, Landon, UK, Oxford University Press 1992. [4] © Bléer, 1 F Watson and W T Thomson: "Fault Detection in Induction Motors as a Result of Transient Analysis” Proc. IEE [EMDA Conf London, Vol 310, 1989 pp182-186. (55) R Bumett and J F Watson: “The Application of Modem Signal Processing Techniques for use in Rotor Peult Deletion and Location within 3-Phate Induction Motos”, European Joual Vol 49,10. 1 Apel, 1996 [56] 1 F Watson andN Paterson: "The use of Finite Element Methods fo Improve Teshniques for Eaiy Detection of Faults in 3-Phast Indoction Motots,” Proe. IEEE Inlemetional Conference on Eecirieal Machies and Drives IEMDC'97, Milwaukee, 1997 (Accepted for publication in the IEEE ransactions on Energy Conversion (devalue to be confirmed) 16

You might also like