Origins of Stator Current Spectra in DFIGs with Winding Faults and Excitation Asymmetries

Transcription

1 Origins of Stator Current Sectra in DFIGs with Wing Faults and Excitation Asymmetries S. Williamson * and S. Djurović * University of Surrey, Guildford, Surrey GU2 7XH, United Kingdom School of Electrical and Electronic Engineering, The University of Manchester, P.O. Box 88, Sackville Street, Manchester M6 1QD, United Kingdom Abstract- In this aer the authors derive simle exressions for the frequencies of the harmonic comonents in the steady state stator line current of a DFIG oerating under various conditions of suly and/or wing unbalance. The underlying urose of the work is to identify signature frequencies which may be used in condition monitoring instrumentation to both identify and discriminate between a range of likely faults that may arise in ractice. The analytical exressions are verified first by comarison with the sectra calculated using an advanced timesteed circuit model, and then by comarison with measurements made on an exerimental test rig. I. INTRODUCTION Doubly-fed uction generators (DFIGs) are the referred choice for contemorary large variable seed w turbines [1,2] due to the advantages they offer over other conventional generator toologies, rimarily in terms of ower flow control, electromechanical daming and constant outut frequency. Maintenance of these machines is raidly gaining imortance in the w ustry. The results of various studies have icated that wing faults reresent one of the most significant causes of uction machine failure [3,4]. The commonly emloyed techniques for detecting the asymmetry arising from wing faults are for the most art based on exloring the steady state frequency sectra of certain measurable quantities and searching for fault-secific harmonic comonents. In this resect, line current is the referred medium [5,6]. Consequently, the analysis of the oeration of a DFIG with wing faults has recently received increased attention. Several authors have examined DFIG current frequency sectra in an attemt to identify an efficient means for condition monitoring [7-12]. The work ublished to date, however, is mostly based on the analysis of exerimental results [7,9]. It does not, for the most art, resent an exlanation of the origins of the DFIG current sectra nor does it rovide exressions for the defined harmonic signatures [7-9]. Additionally some of the methods emloyed may rove to be unreliable as they are sensitive to inherent machine asymmetry and DFIG unbalanced suly oeration, which is a common occurrence during regular generator oeration [7,1,11]. Furthermore the modeling aroach taken in [1,11] does not aint a comlete icture of the current sectra as it accounts for fundamental fields only. In [12] a more advanced analysis of the rotor current sectra is resented with a theoretical exlanation of the uction of stator fault related harmonics in the rotor circuit. The work is confirmed exerimentally by identifying harmonic comonents in the rotor current sectra. However, these are shown to be difficult to extract without invasive monitoring tools due to their small magnitude. In addition the theoretical aroach taken in [12] is limited to fault secific harmonic content and offers no exlanation of the source of other comonents found in the sectrum, i.e. it does not discriminate between the frequencies of the harmonic comonents resent in the healthy DFIG current sectrum and those of the fault-uced comonents. The available literature icates that there is a distinct lack of information on the DFIG stator current harmonic content. A detailed analysis of the origins of the sectral content of the stator current would therefore rove beneficial for the design of an efficient DFIG condition monitoring system. This work aims to resent a comrehensive theoretical analysis of the DFIG stator current sectrum content for the machine oerating in steady state, both with and without suly and/or wing asymmetries. The urose of this analysis is to establish an understanding of the henomena that give rise to harmonic comonents found in the DFIG current sectrum, as well as to investigate the ossibility of using relatively simle exressions for the frequencies to seek when erforming current sectral analysis. The method yields a series of comact exressions that identify the stator current harmonic comonents that are roduced by any of the oerating conditions examined. These analytical exressions are first verified using a time-steed couled-circuit model and then by exerimental measurement on a secially-constructed test rig. II. OUTLINE DESCRIPTION OF THE TIME-STEPPED MODEL AND THE EXPERIMENTAL APPARATUS The time-steed DFIG model used in this work is based on the established concets of couled-circuit theory and comlex conductor distributions [13]. These methods account for higher order air-ga field sace harmonics when evaluating wing uctances, and also take into consideration recise wing conductor distributions. This technique is conveniently incororated into a mathematical model comrising of a standard set of DFIG electromagnetic and mechanical equations, which are solved in a time steing numerical rocedure. The develoed model has been extensively verified against measurements obtained from a urose designed test /9/$ IEEE 563 Authorized licensed use limited to: The University of Manchester. Downloaded on July 16, 29 at 6:25 from IEEE Xlore. Restrictions aly.

2 rig that enables the exerimental simulation of various DFIG oerating conditions. The laboratory rig emulates a tyical w ower DFIG drive. It comrises of a wound rotor uction machine mechanically couled by a common shaft to a DC motor that acts as a rime mover, i.e. mimics the role of the turbine in a wmill. The wound rotor uction machine rotor circuit is excited by an active front-end converter caable of accommodating bi-directional ower flow. The DFIG electrical quantities are measured on the rig using ower analyzers installed in the stator and rotor circuit and seed and torque transducers mounted on the shaft; the currents are recorded using a recision digital oscilloscoe. The stator coils are brought out to a atch-anel, which facilitates their interconnection, allowing a range of stator wing configurations, both balanced and unbalanced, to be examined exerimentally. This setu also enables the exerimental investigation of rotor wing unbalance, which is achieved by electromagnetically reversing the roles of rimary and secondary wings. Detailed model and rig descritions can be found in [14,15], along with comarisons between measurement and timesteed model redictions. III. ANALYSIS A. Air-ga Fields and Induced EMFs The analysis in this section is based on a three-hase 2-ole DFIG. The stator wing of this machine is excited at frequency ω, whilst its rotor rotates at seed ω r and is excited at frequency sω, where ω ω = 1 (1) r ( The rotor of a DFIG is cylrical, and in this work will be assumed to be mounted concentrically in the stator bore. The stator-driven field will therefore uce only suly-frequency emfs in the stator wings. Non suly-frequency emfs uced in the stator must be roduced by fields driven by the currents flowing in the rotor, assuming that the effects of saturation can be ignored. There are two sources of rotor current; the rotor excitation and the emf uced in the rotor by the stator-driven air-ga field. These will be considered in turn. A.1. Rotor-driven fields roduced by the rotor excitation current The air-ga field roduced by the rotor excitation will have the general form bgr ( θ, t) e 2B e (2) j( sωt θ ) =R gr θ is an angular co-ordinate measured in a co-ordinate system fixed to the rotor. At this stage we wish to lace no restriction on the ole numbers,, other than that they are non-zero signed integers. The transformation between the rotor coordinate system and that fixed to the stator (θ), is given by θ = θ + ω t r (3) The rotor driven air-ga field may now be exressed in the stator reference frame by substituting from (1) and (3) into (2) bgr ( θ, t) =Re 2B e j ( s+ (1 s)) ωt θ gr (4) Each of these rotating flux waves will uce an emf in the stator conductors at the corresonding frequency, ω s + ( 1 s) ω = (5) For currents not to flow in the wings at this frequency, the ividual conductor emfs must sum to zero across the wing. A.2. Rotor-driven fields roduced by currents uced in the rotor Currents may also be uced in the rotor as a consequence of the air-ga fields roduced by the stator excitation. Such fields have the general form bgs ( θ, t) e 2B e =R j( ωt θ) gs (6) In the rotor reference frame these become, bgs ( θ, t) =Re 2B e j (1 (1 s)) ωt θ gs (7) Each of these rotating flux waves will uce emfs in the rotor conductors. The magnitudes of the resulting rotor wing emfs will deend on the configuration of the rotor wings, and the rotor currents they drive will deend on the imedance of the rotor suly. However, at this stage in the develoment we will assume that all of the fields in (7) are able to uce currents in the rotor wing. The distribution of the air-ga fields roduced by these uced currents deends on the configuration and interconnection of the rotor wings. Each harmonic comonent of the uced rotor current can set u a sectrum of air-ga fields with different numbers of oles. The rotor-driven flux density roduced by the uced rotor currents therefore has the general form 564 Authorized licensed use limited to: The University of Manchester. Downloaded on July 16, 29 at 6:25 from IEEE Xlore. Restrictions aly.

3 j (1 (1 ) ωt μθ, μ bgr, ( θ, t) =Re 2Bgr, e μ (8) Again, no restriction is laced on the ole-air number, μ, other than to state that they are non-zero integers. Equation (8) exresses the flux density wave roduced by the uced rotor currents, in the rotor reference frame. Transferring this into the stator reference frame roduces, ( μ) j 1 (1 s) ωt μθ, μ bgr, ( θ, t) =Re 2Bgr, e μ (9) Equation (9) shows that the emfs uced in the stator wings due to the currents uced in the rotor wings, will have frequencies given by,, μ μ ω = 1 + (1 ω (1) Equations (5) and (1) may be used to derive simle exressions for the emfs uced in the stator wings under a range of oerating conditions. B. Machine oerating normally For a machine with no wing faults and balanced excitation on the stator and rotor, and μ have the same form, μ = (1 6 m), m =, ±1, ±2, ±3 (11) Substituting these exressions into either (5) or (1), and simlifying gives k ω = 6 k(1 ± 1ω, k =, 1, 2, 3 (12) k= corresonds to the stator excitation frequency. Non-zero values o relate to seed-deendant high-frequency comonents of emf which are uced in the stator of a 'healthy' DFIG oerating under ideal balanced suly conditions. C. Machine oerating with asymmetry on the rotor side C.1. Rotor wings balanced. Rotor excitation unbalanced An un-balanced rotor excitation will roduce air-ga fields with ole-air numbers given by = ± ( 1 6m) m =, ±1, ±2, ±3 (13) The ositive sign in (13) gives values of corresonding to (11), and will therefore result in frequencies of uced emf given by (12). The negative sign, on the other hand, after substitution into (5), results in frequencies given by k ω 6k(1 ± (1 2s)ω =, k =, 1, 2 (14) Additional field comonents will be roduced by the rotor if a zero-sequence voltage, roduced by the suly unbalance, is able to drive zero-sequence currents through the rotor wings. This would require the rotor to be star-connected, with the star oint connected to neutral. The authors do not believe that such a configuration is commonly used, and this otion will be neglected. C.2. Rotor wings unbalanced. Rotor excitation balanced A rotor wing fault has the effect of introducing asymmetry into the rotor mmf attern. As a consequence, the resulting rotor-driven air-ga field may now contain subharmonic ole numbers, i.e. = 1 2m, m =, ±1, ±2, ±3 (15) If the stator wings are series-connected, the stator will be able to resond only to those harmonic fields which have the same ole numbers as those roduced by the stator wing itself (i.e. those ole numbers icated by (13)). Under such circumstances the emfs uced in the stator wings will have the frequencies given by (12) and (14). If the stator wings are arallel-connected, however, the stator may be able to resond to some of the sub-harmonic fields. For examle, with two arallel aths in the stator, the wings may be caable of having emfs uced by fields with any odd number of ole airs. Substituting (15) into (5) gives, ω k 2k 1 s = (1 ± s ω + (16) D. Machine oerating with asymmetry on the stator side D.1. Stator wing balanced. Stator excitation unbalanced. Under these circumstances the stator roduces fields with ole-air numbers given by (13). The resonse of the rotor to these fields is to roduce the same ole-air numbers, with the same sign. That is, when has the form +(1-6m), μ has the form +(1-6n), where m and n are signed integers. And when has the form -(1-6m), μ has the form -(1-6n). Substitution of these two alternatives into (1) roduces the frequencies given by (12). That is, stator excitation asymmetry does not introduce any comonents into the stator current sectrum that are not resent in a 'healthy' machine with a balanced suly. D.2. Stator wings unbalanced. Stator excitation balanced. A stator wing fault has the effect of introducing asymmetry into the stator mmf attern. As a consequence, the 565 Authorized licensed use limited to: The University of Manchester. Downloaded on July 16, 29 at 6:25 from IEEE Xlore. Restrictions aly.

4 resulting stator-driven air-ga fields may now contain all ole numbers. The resonse of the rotor to this harmonic-rich airga field will deend on the layout of the rotor wings. If the rotor wings are series-connected, only those fields which have ole-air numbers given by = ± (1-6m) will be able to uce non-zero sequence emfs in the rotor. This is analogous to the situation discussed in D.1. where it is argued that the frequencies of the stator uced emf sectra will be given by (12). That is, an oen-circuit or short-circuit fault on the stator will not result in additional comonents of uced emf in the stator. If, on other hand, the rotor wings contain (say) two arallel aths, those stator-driven fields which have odd numbers of ole airs will also coule with the rotor. That is, = 2m + 1, where m=, ±1, ±2, ±3 The rotor's resonse to these uced emfs will be to introduce fields with the same range of ole numbers, i.e. μ=2n+1, where n=, ±1, ±2, ±3 Substituting these two exressions into (7), and simlifying, k 2k ω = s 1 ( 1 ) ± ω k=, 1, 2, 3 (17) a) Balanced wings b) Unbalanced wings Figure 1. Healthy and faulty wing configurations oerating scenarios examined in the revious section. These are: 1. Balanced stator and rotor wings; Balanced stator and rotor suly. 2. Balanced stator and rotor wings; Unbalanced rotor suly, balanced stator suly. 3. Unbalanced rotor wings, balanced stator wings; Balanced stator and rotor suly. 4. Balanced stator and rotor wings; Unbalanced stator suly, balanced rotor suly. 5. Unbalanced stator wings, balanced rotor wings; Balanced stator and rotor suly. A summary of the analytical exressions derived in this section is given in Table I. IV. MODEL STUDY The first ste in the verification of the exressions given in Table 1 was accomlished by means of the DFIG time steing model, which was used to simulate each of the five conditions listed in the table. The urose of this work was to verify the develoed equations through comarison with the redictions from a verified analytical model. DFIG oeration for all ossible combinations of suly and wing unbalance was investigated here, equivalent to the The model study was carried out for a suer-synchronous oerating seed of 1619 rm and a load torque of 24.9 Nm, which corresonded to test rig measurements erformed for the wing configurations shown in Fig. 1. Balanced wings in this work comrise of two wing grous connected in arallel er hase. Wing unbalance is achieved by oen circuiting one of the arallel legs in a hase wing. The values of stator and rotor excitation used in the time-steed simulation calculations were set to be equal to those measured during the exeriments. The only excetion to this was when balanced suly conditions were under examination, because comlete suly balance could not be achieved in the laboratory. TABLE I STEADY STATE STATOR CURRENT HARMONIC FREQUENCIES FOR VARIOUS BALANCED/UNBALANCED SUPPLY AND WINDINGS OPERATING CONDITIONS WINDINGS SUPPLY STATOR ROTOR STATOR ROTOR BALANCED BALANCED BALANCED BALANCED BALANCED BALANCED BALANCED UNBALANCED BALANCED UNBALANCED BALANCED BALANCED BALANCED BALANCED UNBALANCED BALANCED UNBALANCED BALANCED BALANCED BALANCED INDUCED DFIG STATOR CURRENT FREQUENCIES = 6 k(1 ± 1 f = 6 k(1 ± 1 f = 6k(1 ± (1 2s) f 2k 1 s = (1 ± s + = 6 k(1 ± 1 f 2k = (1 ± 1 f f 566 Authorized licensed use limited to: The University of Manchester. Downloaded on July 16, 29 at 6:25 from IEEE Xlore. Restrictions aly.

5 57.9 Hz Hz Hz a) Balanced stator and rotor wings, Balanced stator and rotor suly & & Hz & Hz 57.9 Hz & 75.5 Hz & Hz 14 & Hz & Hz -6 c) Balanced stator and unbalanced rotor wings, Balanced stator and rotor suly e) Balanced stator and rotor wings, Unbalanced stator and balanced rotor suly Figure 2.Time-steed model redictions for DFIG current sectrum, 1619rm The excitation voltage used in the time-steed simulation was reresented as consisting of fundamental comonent only. The suly unbalance introduced on either the stator or the rotor b) Balanced stator and rotor wings, Balanced stator and unbalanced rotor suly Hz Hz Hz Hz -6 d) Unbalanced stator and balanced rotor wings, Balanced stator and rotor suly side is equal to that measured in the laboratory for the exerimentally achieved suer-synchronous oerating oint. The steady state mechanical seed value was ket at constant 1619 rm in the time-steed simulations for the urose of this study in accordance with the assumtions made in the analytical work in section III. Study results for steady state current sectrum are given in Fig. 2 for all of the oerating conditions investigated. The magnitudes are normalized with resect to the fundamental harmonic. The bandwidth of data resented is limited to a maximum frequency of 8 Hz as this is the region that the authors found to be most attractive for analysis, due to the relatively high magnitude of sectral comonents. Frequencies at which the most dominant harmonic comonents aear in the resective current sectra are labeled in the grahs. The current sectrum in Fig. 2a for a DFIG oerating with balanced wings and balanced excitation is shown to contain four higher order harmonic comonents, i.e those shown in the grah at,, and. Fig. 2b gives the stator line current sectrum when the machine wings and the stator excitation are balanced, but the rotor 567 Authorized licensed use limited to: The University of Manchester. Downloaded on July 16, 29 at 6:25 from IEEE Xlore. Restrictions aly.

6 TABLE II CALCULATED NUMERIC VALUES OF INDUCED STEADY STATE STATOR CURRENT HARMONIC FREQUENCIES FOR VARIOUS BALANCED/UNBALANCED SUPPLY AND WINDINGS OPERATING CONDITIONS WINDINGS SUPPLY CALCULATED DFIG STATOR STATOR ROTOR STATOR ROTOR CURRENT FREQUENCIES [Hz] BALANCED BALANCED BALANCED BALANCED BALANCED BALANCED BALANCED UNBALANCED BALANCED UNBALANCED BALANCED BALANCED BALANCED BALANCED UNBALANCED BALANCED UNBALANCED BALANCED BALANCED BALANCED suly is not. The sectrum not only contains all of the harmonic comonents resent in Fig. 2a, but also exhibits the resence of additional sideband comonents at 57.9 Hz, Hz and Hz. These are clearly roduced by the rotor suly unbalance. In the case of a DFIG oerating with rotor wing unbalance, balanced stator wings, and balanced excitation on both, Fig. 2c icates that additional harmonic comonents of stator current are uced at 57.9 Hz, 14 Hz, Hz, Hz, Hz, Hz, 381.7, Hz, Hz, Hz, Hz, Hz and 75.5 Hz. Again, this sectrum also exhibits the four harmonic comonents resent in a machine oerating with balanced wings and excitation. The current sectral content for a DFIG oerating with balanced suly and rotor wings and unbalanced stator wings is shown in Fig. 2d to contain additional harmonic comonents at Hz, Hz, Hz and Hz. The four harmonic comonents observed for machine oeration with balanced suly and wings are again seen to be resent. Finally, the time-steed model results shown in Fig. 2e for DFIG oeration with unbalanced stator and balanced rotor excitation and with balanced wings, icate that, the resence of stator excitation asymmetry does not introduce new comonents in the stator current sectrum. Comarison of Fig. 2e with Fig. 2a shows that no new harmonic comonents have been introduced. The stator current harmonic sectra resented in figures 2a-e all exhibit the harmonic comonents that are resent in a machine with balanced wings and excitation. Rotor excitation unbalance, or rotor wing asymmetry are seen to introduce additional frequency comonents in the stator line current, as does stator wing unbalance. These observations suggest that the resence of certain harmonic comonents may eed be used to identify different tyes of wing or suly asymmetry. The frequencies of harmonic comonents uced in the current sectra at oerating seed of 1619 rm that are redicted by the equations in Table I are listed in Table II. The derived exressions yield a series of ossible stator current harmonic frequencies. The magnitudes of these comonents deend on a number of arameters, including oerating seed, wing configuration, and rotor skew. The numeric values listed in Table II are those that match the ones found in the analytical exressions given in Table I. Comarison of the corresonding data in Table II and Fig. 2 demonstrates that the time-steing model roduces harmonic comonents which accord with the analytical exressions derived in section III. Although the time-steed model has been validated against measurement, this is still the use of one model to confirm another, and so an exerimental investigation was also carried out. V. MEASUREMENTS AND PREDICTIONS In this section we demonstrate the validity of the timesteed model for the rediction of the frequencies of harmonic currents, by comaring measured and redicted sectra for a range of DFIG wing configurations. The measurements were taken for a machine oerating at 1619 rm with both balanced wings and with an asymmetry resent in stator or rotor wings, as shown in Fig. 1. The measured excitation voltages are tyically unbalanced and corresond to the asymmetry that exists in the laboratory suly, which was used to generate data in Fig. 2. The excitation and load torque values recorded in the exeriments are used as model data inuts in order to credibly reresent the oerating conditions observed in the laboratory tests. As before, suly voltages are modeled in calculations as fundamental sine waves for the urose of this analysis. It is also imortant to oint out that the measured current data contains seed rile effects, as minute seed fluctuations arise due to the electromagnetic 568 Authorized licensed use limited to: The University of Manchester. Downloaded on July 16, 29 at 6:25 from IEEE Xlore. Restrictions aly.

7 unbalance that is inevitably resent in any DFIG drive [16,17]. This henomenon was reroduced in the model by allowing the mechanical seed to vary in the simulations, i.e. by resolving the model steady state mechanical characteristic for angular seed in each numerical iteration. As both the -6 a) Balanced stator and rotor wings, model redictions 42.1 Hz 57.9 Hz 65.9 Hz Hz Hz & Hz Hz Hz Hz Hz b) Balanced stator and unbalanced rotor wings, model redictions theoretical analysis and the model study were erformed on the assumtion of constant seed the results resented here are also an efficient demonstration of the seed rile effects on healthy and faulty DFIG line current sectrum. The redicted and measured current sectra for the wing 15 Hz 45 Hz 25 Hz 35 Hz -6 a) Balanced stator and rotor wings, test rig measurement 57.9 Hz Hz 65.9 Hz Hz 35 Hz & Hz Hz Hz 42 Hz 15 Hz Hz 25 Hz Hz 5 b) Balanced stator and unbalanced rotor wings, test rig measurement Hz Hz Hz 15 Hz Hz 212 Hz Hz Hz c) Unbalanced stator and balanced rotor wings, model redictions Figure 3. Model redictions for DFIG current sectrum, variable seed, 1619 rm 112 Hz 25 Hz 35 Hz c) Unbalanced stator and balanced rotor wings, test rig measurement Figure 4. Test rig measurements for DFIG current sectrum, 1619 rm 569 Authorized licensed use limited to: The University of Manchester. Downloaded on July 16, 29 at 6:25 from IEEE Xlore. Restrictions aly.

8 and excitation conditions that have been reviously considered are shown in Fig. 3 and Fig. 4, resectively. The corresonding calculated and exerimental data sets are seen to be in good agreement when sectral content is concerned. The most dominant frequency comonents are labeled in the grahs and are clearly identifiable in redictions and measurements alike. Minor discreancies between some of the measured and redicted harmonic frequency values are due to sectral resolution difference between the simulated and exerimental data. In addition, harmonic comonents that exist in the measured current sectrum at multiles of line frequency (15 Hz, 25 Hz, 35 Hz ) tyically originate from the higher order harmonic content in the ower suly but may also arise from magnetic saturation within the machine. These effects are not accounted for in the model and are thus not resent in the calculated sectra. Furthermore the absence of these comonents in model redictions is exected, since model considerations [14] and the resented theoretical analysis are focused on investigation of the current sectra harmonic comonents that originate from machine based electromagnetic henomena. The measured data also exhibit an aarent higher noise level when comared to model calculations. This is uced rimarily by the rotor converter switching and is not accounted for in the analytical considerations. The sectra resented in Figs 3 and 4 amly demonstrate the validity of the time-steed model, which itself has been used to confirm the frequency exressions given in Table I. It is also noticeable, however, that some of the harmonic comonents of small magnitudes in the redicted sectra are buried in the noise in the exerimental data. Furthermore, where results from Fig. 2 and Table II differ from the measured data is in redicting the existence of sideband comonents, which arise as a result of the seed rile. It will be recalled by the reader that the seed is assumed to be constant throughout the analysis in sections II and III. It will be shown in a searate aer that seed rile gives rise to additional sideband comonents throughout the current sectra in Fig. 3 and Fig. 4, and most rominently to those around fundamental frequency. Other sideband comonents of lesser magnitude can also be observed at higher frequencies in the sectra. This is esecially evident for machine oeration with rotor wing unbalance. VI. CONCLUSIONS In this aer the authors have develoed a set of simle exressions for the stator line current harmonic frequency comonents which may be resent in a DFIG under a range of oerating conditions. The magnitudes of these harmonic comonents will deend largely on the design of the machine under investigation. Design artifacts such as coil itch and rotor skew will have an influence on harmonic magnitudes. However, not all will be attenuated, so that the identification of one of the characteristic frequency comonents may be sufficient to uniquely identify a fault. The exressions have been verified by comarison with the redictions of a timesteed couled-circuit model, which is itself validated using exerimental measurements made on a test rig, for a range of oerating conditions. The authors believe that work resented in this aer is an imortant ste towards the develoment of condition-monitoring techniques for DFIGs, and rovides a better understanding of the mechanisms that generate articular harmonic frequencies in the stator line current. REFERENCES [1] Y. Amirat, M.E.H. Benbouzid, B.Bensaker, R.Wamkeue, Generators for W Energy Conversion Systems: State of the art and Coming Attractions, Journal of Electrical Systems 3-1, 26-38, 27. [2] A.D. Hansen, F.Iov, F. Blaabjerg, L.H.Hansen, Review of Contemorary W Turbine Concets and their Market Penetration, W Engineering, vol. 28, Issue 3, , 24. [3] Motor Reliability Working Grou, Reort of large motor reliability survey of ustrial and commercial installations, Part I, IEEE Transactions on Industry Alications, Vol IA-21, No 4, July/August [4] O.V. Thorsen, M. Dalva, Failure identification and analysis for highvoltage uction motors in the etrochemical ustry, IEEE Transactions on Industry Alications, Vol 35, No 4, July/August [5] A. Siddique, G.S. Yadava, B. Singh, A Review of Stator Fault Monitoring Techniques for Induction Motors, IEEE Transactions on Energy Conversion, Vol 2, No 1, March 25. [6] M.E.H. Benbouzid, A review of uction motors signature analysis as a medium for faults detection, IEEE Transactions on Industrial Electronics, Vol 47, No 5, October 2. [7] I. Boldea, L.M. Poa, B. Jensen, E. Ritchie, Condition monitoring of w generators, Industry Alications Conference, Conference record of the 38th IAS Annual Meeting, Vol 3, , October 23. [8] Q.F. Lu, Z.T. Cao, E. Ritchie, Model of stator inter-turn short circuit fault in doubly-fed uction generators for w turbine, 35th Annual IEEE Power Electronics Secialist Conference, Vol 2, , June 24. [9] H. Douglas, P. Pillay, P. Barendse, The detection of inter-turn stator faults in doubly-fed uction generators, IEEE Industry Alications Conference, Vol 2, , October 22. [1] A. Stefani, A. Yazidi, C. Rossi, F. Fillieti, D. Casadei, G.A. Caolino, Doubly fed uction machines diagnosis based on signature analysis of rotor modulating signals, IEEE Transactions on ustry alications, Vol 44, No 6, November/December 28. [11] D. Casadei, F. Fillieti, A. Stefani, C. Rossi, A. Yazidi, G.A. Caolino, Exerimental fault characterization of doubly fed uction machines for w ower geenration, International Symosium on Power Electronics, Electrical Drives, Automation and Motion, SPEEDAM, , May 26. [12] D. Shah, S. Nandi, P. Neti, Stator inter-turn fault detection of doublyfed uction generators using rotor current and search coil voltage signature analysis, Conference record of the 27 IEEE Industry Alications Conference, , Setember 27. [13] S. Williamson, E.R. Laithwaite, Generalized harmonic analysis for the steady-state erformance of sinusoidally-excited cage uction motors, IEE Proc., Vol. 133, Pt. B, No 3, May [14] S. Djurovic, S. Williamson, A. Renfrew, Dynamic model for doubly-fed uction generators with unbalanced excitation, both with and without wing faults, IET Electric Power Alications, March/Aril 29. [15] S. Djurovic, S. Williamson, A couled-circuit model for s DFIG oerating under unbalanced conditions, Proceedings of IEEE International Conference on Electrical Machines, ICEM 28, Portugal, Setember 28. [16] F. Fillieti, G. Francheschini, C. Tassoni, P. Vas, AI Techniques in uction machines diagnosis including the seed rile effect, IEEE Transactions on Industry Alications, Vol 34, No 1, January/February [17] Y. Liao, L. Ran, G.A. Putrus, K.S. Smith, Evaluation of the effects of rotor harmonics in a doubly fed uction generator with harmonic uced seed rile, IEEE Transactions on Energy Conversion, Vol 18, No 4, December Authorized licensed use limited to: The University of Manchester. Downloaded on July 16, 29 at 6:25 from IEEE Xlore. Restrictions aly.