Successfully reported this slideshow.
We use your LinkedIn profile and activity data to personalize ads and to show you more relevant ads. You can change your ad preferences anytime.

Development of a low cost test rig for standalone wecs subject to electrical faults


Published on

In this paper, a contribution to the development of low-cost wind turbine (WT) test rig for stator fault diagnosis of wind turbine generator is proposed. The test rig is developed using a 2.5 kW, 1750 RPM DC motor coupled to a 1.5 kW, 1500 RPM self-excited induction generator interfaced with a WT mathematical model in LabVIEW. The performance of the test rig is benchmarked with already proven wind turbine test rigs. In order to detect the stator faults using non-stationary signals in self-excited induction generator, an online fault diagnostic technique of DWT-based multi-resolution analysis is proposed. It has been experimentally proven that for varying wind conditions wavelet decomposition allows good differentiation between faulty and healthy conditions leading to an effective diagnostic procedure for wind turbine condition monitoring.

Published in: Engineering
  • Be the first to comment

  • Be the first to like this

Development of a low cost test rig for standalone wecs subject to electrical faults

  1. 1. Development of a low cost test rig for standalone WECS subject to electrical faults Himani n , Ratna Dahiya Department of Electrical Engineering National Institute of Technology Kurukshetra, India a r t i c l e i n f o Article history: Received 30 October 2015 Received in revised form 24 July 2016 Accepted 15 August 2016 Available online 21 September 2016 This paper was recommended for publica- tion by Jeff Pieper Keywords: Wind turbine generator (WTG) Wind Turbine Emulator Non-stationary signals Self-excited induction generator (SEIG) Condition monitoring Short circuit fault a b s t r a c t In this paper, a contribution to the development of low-cost wind turbine (WT) test rig for stator fault diagnosis of wind turbine generator is proposed. The test rig is developed using a 2.5 kW, 1750 RPM DC motor coupled to a 1.5 kW, 1500 RPM self-excited induction generator interfaced with a WT mathe- matical model in LabVIEW. The performance of the test rig is benchmarked with already proven wind turbine test rigs. In order to detect the stator faults using non-stationary signals in self-excited induction generator, an online fault diagnostic technique of DWT-based multi-resolution analysis is proposed. It has been experimentally proven that for varying wind conditions wavelet decomposition allows good differentiation between faulty and healthy conditions leading to an effective diagnostic procedure for wind turbine condition monitoring. & 2016 ISA. Published by Elsevier Ltd. All rights reserved. 1. Introduction The recent technological developments of the wind turbine (WT) systems focus on O&M cost reduction [1] and operational reliability. Condition monitoring of wind turbine generator (WTG) is important as defects in generators have shown to be a major reason for WT downtime [2,3]. Among all the possible faults occurring in WTG, faults related to stator comprise a significant percentage [4,5]. Stator short circuit fault is quite common in electrical machines [2]. In most cases, it begins as an inter-turn fault and eventually grows in to major one's such as coil to coil, phase to phase, phase open circuit and phase to ground, that may lead to system break down. It can cause catastrophic damage to the WTG in a very short time, making any fault compensation impossible. This demands better fault-detection and remediation strategies [5]. The evaluation of fault diagnosis method on the real physical system by creating a fault can be dangerous as this may lead to the destruction of wind turbine [6]. Therefore, for CM evaluation, adequate models of the test rigs are needed [6]. The test rig should drive the WTG in a similar way as a wind turbine imitating the non-stationary conditions and the torque developed for a given wind velocity under laboratory conditions [7]. The widely used and studied standalone wind turbine systems are based on induction generators due to their advantages over synchronous generators, such as low cost, small size, and low maintenance requirements. Therefore, induction generators are the suitable option as a WTG for isolated applications [8]. Though many researchers have worked on the stator fault both for induction motors as well as for induction generator, but a very few have done the research using the dynamic characteristic of the WT. WTs are of variable speed; variable load machines as a result, standard analysis techniques like Fourier analysis cannot be directly applied simply to the monitoring of non-stationary signals produced by WT [9,10]. Non-stationary signal faults detection by using current, voltage [11], power [12,13] has been performed using advanced signal processing techniques such as instanta- neous frequency [10,14], wavelet-based techniques [5,15], sequence network analysis[16], power signature analysis [12] and Short-Time Fourier transform (STFT) [6]. The quantification of the fault has been done using DWT [24,25], artificial neural networks [26,27]. Both these technique has been used successfully to model complex nonlinear dynamic systems. In most of the earlier research papers for CM, the various experimental setups are used for the non-stationary signal gen- eration at the generator output. These setups include test rig controlled by aerodynamics forces [12,17], wind tunnel [18], speed fluctuations by PLC [14] and analog inputs [10]. Contents lists available at ScienceDirect journal homepage: ISA Transactions 0019-0578/& 2016 ISA. Published by Elsevier Ltd. All rights reserved. n Corresponding author. E-mail addresses: (Himani), (R. Dahiya). ISA Transactions 65 (2016) 537–546
  2. 2. The main objective of this paper is modeling of the low-cost test rig and fault diagnosis of stator faults in SEIG (self-excited induction generator) in non-stationary conditions. The novelties of this research paper are summarized as follows: For CM of WT, a low-cost setup is designed and developed. The real-time interface for monitoring and control of WTE is designed in LabVIEW. It controls the prime mover according to the WT principle and follows the torque speed characteristics of WT. Hardware part of the emulator is developed to imitate the WT characteristics. Each component has been designed keeping in consideration flexibility and cost. The setup can be reconfigured for given specifications as each component is designed inde- pendently. The total cost of the setup is approx. $1200. The designed WT test rig follows the power curves of WT characteristics. The results show a very good operation in the whole operating range of the SCIG and due to its simplicity, real-time interfacing and control for monitoring; – it is suitable for practical use. Benchmarking of stator current signatures with already proven setup – available in the literature [9,13,17] (University of Manchester and Durham University). An experimental study shows that the proposed approach using the DWT quantitative analysis offers a good diagnostic cap- ability for interpreting the non-stationary WT CM signals. The method is in real time, it detects the fault at the moment of its existence in WT, which allows easy maintenance without damaging other parts of the system. The organization of this paper is as follows. The development of a wind turbine test rig is discussed in Section 2. Section 3 presents the experimental test results. The stator short circuit fault results are discussed in Section 4. The conclusion is given in Section 5. 2. Development of wind turbine test rig 2.1. Wind turbine modeling The aerodynamic power output of wind turbines (Po) can be modeled as [7,19]: Po ¼ 1=2ρCpAv3 ð1Þ where ρ¼air density (kg/m3 ), A¼WT swept area (m2 ), v¼wind speed (m/s) The fraction of the power extracted by the turbine is called the power coefficient (Cp), which depends on wind speed, shaft speed and the mechanical parameters such as shape and pitch angle (β) of the blades. The maximum value of Cp, theoretically given by the Betz limit [7,19] is 0.593. It is a function of tip speed ratio (TSR) (λ), which is defined as the ratio between the linear velocity of blade tip and the wind velocity. The TSR is given as: λ ¼ ωwR v ð2Þ where ωw ¼angular speed of WT (rad/s), R¼blade radius (m) The mechanical torque generated can be calculated from the WT power and the shaft speed as given by [19]: Tw ¼ Po ωw ¼ 1 2λ ρCpπR3 v2 ð3Þ 2.2. DC motor control strategy The DC motor control is based on the armature current reg- ulation. The schematic representation of a separately excited DC motor is in Fig. 1 [7]. Ve ¼ IeRe þLe dIe dt : For the excitation ð4Þ Va ¼ IaRa þLa dIa dt þLmIeωm: For the excitation ð5Þ ζem ¼ ζr þfωm þJ dωm dt ð6Þ ζem ¼ KcIaKe ¼ LmIeKc ¼ Ke ð7Þ ζr: Resistive Torque; ζem: Electromagnetic Torque; Lm: Mutual inductance excitation armature; f: Coefficient of friction; J: Inertia moment; Ke: Torque constant; ωm ¼angular speed of DC motor; Va is the voltage applied to the motor armature, Ra ,Re are the resis- tance; La ,Le are the inductance of the armature circuit and field respectively. The DC motor characteristic is such that, the mechanical power output of the machine is a function of the armature voltage Va and angular speed ωm under steady state. When a constant Va is applied to the motor the output power is a conic function of speed ωm. 2.3. Self-excited induction generator (SEIG) modeling The generated stator voltage and current are derived from d–q axis values using the equations below. Fig. 2(i) and (ii) shows the equivalent d–q axis circuit of SEIG. Across the stator windings of the generator a capacitor is con- nected. Using Kirchhoff’s voltage the loop Eqs. (8)–(11) of the SEIG equivalent circuit are written as [214]: Rsiqs þLIs diqs dt þLm diqr dt þ 1 C diqs dt ¼ Vcq ð8Þ Rriqr þLIr diqr dt þLm diqs dt þ 1 C diqr dt ¼ ωrλdr ð9Þ Rsids þLIs dids dt þLm didr dt þ 1 C dids dt ¼ ÀVcd ð10Þ Rridr þLIr didr dt þLm dids dt þ 1 C didr dt ¼ Àωrλqr ð11Þ At each step of integration, the magnetizing current has to be updated. Therefore, by using the Eq. (12), the new magnitude of the magnetizing current is obtained jim j ¼ ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ðiqs þiqrÞ2 þðiqs þiqrÞ2 q ð12Þ Fig. 1. The electric model of the DC motor. Himani, R. Dahiya / ISA Transactions 65 (2016) 537–546538
  3. 3. Hence, in the steady state the magnitude of the generated air gap voltage of SEIG is Eg ¼ ωLm imj ð13Þ Lmis not a constant but depends on magnetizing current imj . This dependency is determined by a synchronous impedance test and can be expressed as Lm ¼ f m imj ð14Þ The developed electromagnetic torque Te and torque balance equations are Te ¼ 3 2 P 2 Lmðidriqs ÀiqridsÞ ð15Þ 2.4. Capacitor bank model Current drawn by each capacitor [22]: Icap ¼ Q1 Ell ¼ ¼ 0:849 A; Capacitive reactance: Xc ¼ Ell Icap ¼ 1 2ÂπÂf ÂC ¼ 488:38Ω: So, the capacitance (connected in delta) required for excitation in full load condition is: C ¼ 1 2ÂπÂf ÂXc ¼ 6:52μF: 2.5. Stator modeling taking into account stator fault The dynamic model for the stator windings of the three-phase squirrel cage induction generator (Fig. 3) is developed and the relevant volt–ampere equations are: ÀVs ¼ RsIs þ d∅s dt ð16Þ Vs ¼ Vsa; Vsb; Vsc½ ŠT is the stator voltage vector; Is ¼ Isa; Isb; Isc½ ŠT is the stator current vector; Rs ¼ diag Rsa; Rsb; Rsc½ Š is a 3 by 3 stator resistance matrix;. ∅s ¼ ∅sa; ∅sb; ∅Vsc½ ŠT is a stator flux vector. With the Park transformation ðPsÞ, the voltage equations for the stator windings can be written as: ÀVsdq ¼ PsRsIsdqPÀ1 s þPs dðPÀ1 s ∅sdqÞ dt ¼ RSDQ Isdq þ d∅sdq dt þωs 0 À1 1 0 ! ∅sdq ð17Þ where RSDQ is the equivalent resistance matrix and is given by: RSDQ ¼ PsRsPÀ 1 s ¼ Rds Rsdq Rsdq Rqs # ð18Þ PS ¼ ffiffiffi 2 3 r sin ðθÞ sin θÀ2π 3 À Á sin θþ2π 3 À Á cos θ À Á cos θÀ2π 3 À Á cos θþ2π 3 À Á # ð19Þ is not a constant but depends on magnetizing current. 2.6. Wind Turbine Emulator (WTE) The emulation scheme is shown in Fig. 4. Steady-state char- acteristics of a given WT at various wind velocity are studied. The wind speed, turbine radius and angular speed of motor are con- sidered as inputs of this model. To calculate the reference torque, the control program reads wind velocity from an input file, shaft speed from tacho-generator and supply current from the current transducer [7]. The wind data file can be generated in different ways, depending on the desired test conditions, or even being real data from an anemometer. The electromagnetic torque of a DC motor should be equal to the reference torque of WT. From the mathematical model, the reference armature current is obtained, which is used to control the armature current of the DC motor by using PI controller and semi converter rectifier. It gives controlled rectified voltage depending upon the triggering pulses to thyr- istors for the regulation of DC motor armature voltage. The var- iation in armature voltage controls the motor current in accor- dance with the reference current and the system reaches a steady state. Here, the DC motor current is directly controlled so that the required torque is directly proportional to the current. Fig. 2. d–q axes equivalent circuit of SEIG. Fig. 3. SEIG with a capacitor excitation system driven by the wind. Fig. 4. Schematic of Wind Turbine Emulator. Himani, R. Dahiya / ISA Transactions 65 (2016) 537–546 539
  4. 4. 2.7. Integration of system instrumentation A low-cost test bench equipped with 220 V, 1.5 kW squirrel cage induction machine with the prime mover, a 2.5 kW sepa- rately excited DC machine has been designed [20,21]. Based on required specifications and cost, Advantech 4704 based data acquisition card is selected. The model of a wind turbine and the PI controller are developed using LabVIEW software. The control panel designed in LabVIEW (Fig. 5) allows real time communica- tion between the setup and the user. Turbine parameters such as the wind velocity and turbine radius can be set by the user through the software. Various off grid-connected wind turbine system parameters as stator voltage, generated current and speed of machine are sampled at 2 kHz sampling rate using DAQ. A low- cost interfacing circuitry including the current transducer (CTs) and voltage transducer (VTs) are used to monitor the generator currents and terminal voltages. Signal conditioning is done to provide the interface the signals/sensors and the data acquisition system. The coordinate transformations, controllers and further calculations are carried out and current reference values are given to the PI controller. The block diagram of test rig is shown Fig. 6 and parameters are reported in Appendix. 3. Experimental results of wind turbine test rig The experimental results of the designed test rig are verified with the test rigs made at University of Manchester, UK and Dur- ham University UK [12,13,17]. The specifications of these test rigs are tabled in Appendix. The equations describing the stator current spectral content for the healthy conditions is [12] f k ind ¼ 6kð1ÀsÞ7l f ð20Þ Fig. 5. Control panel for Wind Turbine Emulation. Fig. 6. Block diagram of the system instrumentation. Himani, R. Dahiya / ISA Transactions 65 (2016) 537–546540
  5. 5. where f is the fundamental frequency, s is the induction generator fractional slip, l ¼ 1; 2; 3…, k ¼ 1; 2; 3…: Constants k and l relate respectively to air-gap field space harmonics resulting from the layout of the machine and supply time harmonics in the current. For benchmarking [12], reduced set of steady state frequencies are listed in Table 1. The test rig was run up to the required super-synchronous speed for a constant wind speed of 7 m/s. The generated current is acquired using the current transducers and DAQ. The healthy current spectra in Figs. 7 [12] and Fig. 8 indicates a comparable spectral content despite different levels of rotor resistance. Frequencies ‘c’ and ‘d’ are related directly to the fundamental frequency and 3rd harmonics, consistent for all the test rigs. Machine dependent frequencies ‘a’ and ‘b’ are also present in all the machines. Frequencies ‘e’ and ‘f ’ also show the comparable results despite different operating conditions. 4. Short circuit fault In order to analyze the phase-to-phase faults, a short-circuit between two phases is considered. This fault can produce specified frequency components f short in different signals [15,23] as: f short ¼ f s n 1Àsð Þ p 7k k ¼ 1; 3; …:: n ¼ 1; 2; 3……::Þ ð21Þ where p is the number of pole-pairs, s is the per-unit slip and f s is the supply frequency. 4.1. Stator winding fault-experimentation Fault conditions were emulated by introducing phase to phase short circuit fault through a resistor. Terminal voltage contains unique fault frequency components that can be used for the stator winding fault detection. The only difficulty with this technique is that the data needs to be collected very quickly. With the appli- cation of short circuit fault, the large amount of current flows through the shortest path causing the machine to heat up very quickly. So the fault was applied for a very short duration of 0.2 s. Application of this method allowed multiple tests of the WTG to be performed without permanent damage. A terminal voltage measured from the test rig under various cases (Table 2) has been processed for CM of WTG for the stator winding fault diagnosis. The first set of experiments was performed under static conditions with no load and a constant wind speed of 7 m/s as shown in Fig. 9; the shaft speed almost remains constant at 1480 RPM, generated voltage is 220 V and the generated current is almost negligible. The next set of experiments was performed with short circuit fault at no load, with constant and then variable wind conditions applied through the WT model as shown in Fig. 10. At full load, the generated current is max 0.5 A and terminal voltage is 220 V as shown in Fig. 11. Similarly Figs 12 and 13 show the generated current and terminal voltage at constant wind speed and variable wind speed respectively. 4.2. Signal processing using frequency domain analysis The power spectrum of voltage in healthy conditions at no load is shown in Fig. 14(a) and under full load is given in Fig. 14(b). The odd harmonics (i.e. 1st, 3rd, 5th), at 50 Hz, 143 Hz and 241 Hz are most evident under no load conditions and 50 Hz, 143 Hz and 239 Hz in full load conditions. In faulty condition and at no load, with n¼3, k¼1, and slip¼0.22, side bands at 20.75 Hz, 79.25 Hz, 120.75 Hz and 179.25 Hz around first and third harmonics are quite evident as shown in Fig. 15(a). Similarly, side bands at 22.5 Hz, 77.5 Hz, 118.5 Hz and 177.5 Hz are evident at full load Table 1 Constants for healthy current spectra. Frequency label Constant Line current j k L a, b (healthy) 1 1 1 f k ind ¼ 6kð1ÀsÞ8lj f c, d 1 4 1 f k ind ¼ k pð1ÀsÞ8l f e 1 8 1 f k ind ¼ k pð1ÀsÞ8l f 1 16 1 f k ind ¼ k pð1ÀsÞ8l Fig. 7. Current spectra of test rig made at (a) University of Manchester (b) Durham University. Fig. 8. Current spectra of test rig made at NITK. Table 2 Experimental conditions for short winding fault detection. Case no. Load Wind speed WTG condition Observed output 1 No-load Constant Healthy Fig. 9 2 No-load Constant Faulty Fig. 10(a) 3 No-load Variable Faulty Fig. 10(b) 4 Full-load Constant Healthy Fig. 11 5 Full-load Constant Faulty Fig. 12 6 Full-load Variable Faulty Fig. 13 Himani, R. Dahiya / ISA Transactions 65 (2016) 537–546 541
  6. 6. conditions as shown in Fig. 15(b). Under variable wind speed conditions, the slip varies; the magnitudes of the above compo- nents of the fault are spread in a bandwidth proportional to the variation of speed as shown in Fig. 16(a). There are no clear side- bands available in case of no load under variable wind conditions. The same phenomena also exist under full load conditions as shown in Fig. 16(b). The results are summarized in Table 3. As a result of varying slip under transit conditions, the use of Fourier analysis may result in an erroneous diagnosis. For pro- viding a reliable diagnosis procedure, a more appropriate techni- que is required. 4.3. Stator winding fault diagnosis using discrete wavelet transformation An effective fault diagnosis requires the measurements of a quantity-sensitive to the faults and a suitable method to obtain a diagnostic index and a threshold stating the edge between faulty and healthy condition. Under variable-speed conditions, the fault fre- quency components, whose amplitude is usually monitored for fault detection, are spread in frequency with width which is related to the load, speed and slip variations. The wavelet analysis allows the pro- cessing of the stator terminal voltage in such a manner that the fault Fig. 9. At constant wind speed, no load, healthy conditions- (a) terminal voltage (b) shaft speed (c) line current. Fig. 10. Terminal voltage with short circuit fault-at no-load (a) constant wind speed (b) variable wind speed. Fig. 11. Terminal voltage in healthy conditions- on load and constant wind speed (a) terminal voltage (b) line current. Himani, R. Dahiya / ISA Transactions 65 (2016) 537–546542
  7. 7. Fig. 12. Terminal voltage with short circuit fault on load and constant wind speed (a) terminal voltage (b) line current . Fig. 13. Terminal voltage with short circuit fault on load and variable wind speed (a) terminal voltage (b) line current. Fig. 14. Power spectrum of generated voltage in healthy conditions at constant wind speed (a) at no load (b) at full load. Fig. 15. Power spectrum of voltage in faulty conditions at constant wind speed (a) at no load (b) at full load. Himani, R. Dahiya / ISA Transactions 65 (2016) 537–546 543
  8. 8. frequency components can be shifted to a dedicated frequency band. In this way, the information related to the fault can be isolated and confined to obtaining a diagnostic index and a threshold stating the edge between faulty and healthy condition. As illustrated by Fig. 17, the logarithmically spaced frequency bands are obtained by dividing the frequency content of the original signal using DWT. A high-order mother wavelet is suitable for carrying out the DWT. If a low-order wavelet is used, there is an increase in the overlap between adjacent frequency bands and the frequency response gets worse.In this paper, a 10th order mother wavelet of Daubechies family has been chosen [24,25]. In order to cover the frequency band with a sampling frequency of fs¼2000 Hz, an eight level decomposition (J¼8) is chosen [5]. In this range, we can track the characteristics of default frequency components. For fault analysis, a diagnosis index based on multiresolution mean power indicator is introduced. The signal is decomposed at different frequency levels by MRA analysis. To reduce the computational time, only detail level dj is considered. In case any fault occurs, the distribution of the energy in signal changes at the resolution levels associated with the characteristic frequency bands of the default. Therefore, the increase in the energy constrained to certain details is measured as the appearance of an irregularity. The mean power concentrated in each detail dj generated by the MRA of the generated voltage signal is defined as below [24,25]: mPdj ¼ 1 N XN n ¼ 1 djðnÞ 2 ð22Þ Here N¼number of samples j¼level decomposition. The data flow of the MRA feature vector extraction is shown Fig. 18. The mean power of the details (d1 Àd8Þ resulting from wavelet decomposition at no-load (Case 1, 2, 3 Table 2) is shown in Fig. 19 (a) and under load conditions (Case 4, 5, 6 Table 2) is shown in Fig. 19(b). Fig. 16. Power spectrum of voltage in faulty conditions at variable wind speed (a) at no load (b) at full load. Table 3 Power spectrum analysis for short circuited winding fault. Case no. Condition Lower Side Band Upper Side Band Observation FF Mag. FF Mag. Case 1 Fig. 14(a) – No Visible side bands in healthy conditions Case 2 Fig. 14(a) 20.75 8 79.25 8 Visible Case 3 Fig. 16(a) – – – – No Visible side bands because of variable wind speed. Case 4 Fig. 14(b) – No Visible side bands in healthy conditions Case 5 Fig. 15(b) 22.5 15 77.5 15 Visible Case 6 Fig. 16(b) – – – – No Visible side bands because of variable wind speed. Fig. 17. Wavelet tree decomposition with three details levels. Fig. 18. Multiresolution feature vector extraction. Himani, R. Dahiya / ISA Transactions 65 (2016) 537–546544
  9. 9. There is a significant discrimination between healthy and faulty conditions. The fault discrimination is significantly increased at load in varying wind conditions especially at d5 (around funda- mental frequencies). By comparing the detail signal d5 calculated in fault conditions (Fig. 19(a) and (b)), it is clearly visible that there is an amplitude evolution of the faulty component for loaded conditions. 5. Conclusion The paper presents the development of a low-cost wind tur- bine test rig for detection and diagnosis of the stator winding fault. Various tests have been carried out on the test rig and the results are benchmarked with already proven test rigs developed at the Durham University and the University of Manchester. The measured current spectra are in good agreement and shows comparable results despite different operating conditions. The findings of research presented in this paper can be summarized as follows. The designed wind turbine test rig follows the power curves of wind turbine characteristics. The result shows a very good operation in the whole operating range of the SCIG and due to its simplicity, real-time interfacing, and control for monitoring – it is suitable for practical use. Stator winding fault can be successfully detected in normal IM using FFT analysis whereas in the case of WTG, due to variations in wind speed and hence slip variations, fault frequency com- ponents are not visible in FFT spectrum. Quantitative analysis using wavelet transformation has been performed to discriminate the healthy from faulty. The analysis shows the occurrence of characteristic patterns through the energy of the involved wavelet signals and oscillations appear- ing in the wavelet signals. The experimental results show that the proposed approach offers good diagnostic capability com- pared to the existing techniques. This work further can be extended to the following topics could be studied: Detection of the fault subtypes. There is the need to study the effect of electric drives, use dif- ferent generator types-DFIG, WRIG as these may change the electrical signature's. The influence of gear box components needs to be investigated. Appendix A See Tables A1–A4. Fig. 19. Mean power of the details d1–d8 resulting from the wavelet decomposition (a) at no load (b) at full load. Table A1 Wind turbine parameters. Rated power 500 W Rated wind speed 7.5 m/s Radius of WT 1.4 m Power coefficient 0.48 Table A2 DC motor parameters. Rated power 2.5 kW Nominal speed 1750 rpm Armature resistance (Ra) 1.8711 Ω Filed resistance (Rf) 470 Ω Armature inductance (La) 98 mH Filed inductance (Lf) 14.160 H Table A3 Induction generator parameters. Rated power 1.5 kW Nominal speed 1500 rpm Table A4 Comparison of test rigs. Manchester Durham NITK Generator Type DFIG or WRIG 30 KW WRIG 30 KW SCIG 1.5 KW No. of poles 4 4 4 Converter Back-to-back, 8 kHz switching None None DC motor Driving motor (DC) 40 kW constant speed 54 kW variable speed 2.5 KW Gearbox None 5:1 Helical None Data acquisition Hardware Precision oscilloscope NI LabVIEW Advantech Sampling fre- quency, kHz 2 5 2 MATLAB FFT analysis 0–500 Hz 0–500 Hz Himani, R. Dahiya / ISA Transactions 65 (2016) 537–546 545
  10. 10. References [1] Kamel RM. Effect of wind generation system types on Micro-Grid (MG) fault performance during both standalone and grid connected modes. Energy Convers Manag 2014;79:232–45. [2] Tavner PJ, Faulstich S, Hahn B, van Bussel GJW. Reliability availability of wind turbine electrical electronic components. Eur Power Electron J 2011;20(4). [3] Ribrant J, Bertling L. Survey of failures in wind power systems with focus on Swedish wind power plants during 1997–2005. IEEE Power Eng Soc General Meet 2007;22:167–73. [4] Gandhi A, Corrigan T, Parsa L. Recent advances in modeling and online detection of stator interturn faults in electrical motors. IEEE Trans Ind Electron 2011;58(5):1564–75. [5] Seshadrinath J, Singh B, Panigrahi BK. Single-turn fault detection in induction machine using complex-wavelet-based method. Ind Appl IEEE Trans 2012;28 (6):1846–54. [6] Attoui I, Omeiri A. Modeling, control and fault diagnosis of an isolated wind energy conversion system with a self-excited induction generator subject to electrical faults. Energy Convers Manag 2014;82:11–26. [7] Garg H, Dahiya R. Modeling and development of wind turbine emulator for the condition monitoring of wind turbine. Int J Renew Energy Res 2015;5(2):591–7. [8] Zou Y, Elbuluk M, Sorez Y. Simulation comparisons and implementation of induction generator wind power systems. IEEE Trans Ind Appl 2013;49 (3):1119–28. [9] Yang W, Tavner PJ, Crabtree CJ, Feng Y, Qiu Y. Wind turbine condition monitoring: technical and commercial challenges. Wind Energy 2014;17(5):673–93. [10] Alarcon VC, Daviu JA, Haavisto A, Arkkio A. Particle filter-based estimation of instantaneous frequency for the diagnosis of electrical asymmetries in induction machines. IEEE Trans Instrum 2014;63(10):2454–63. [11] Shah D, Nandi S, Neti P. Stator-interturn-fault detection of doubly fed induc- tion generators using rotor-current and search-coil-voltage signature analysis. IEEE Trans Ind Appl 2009;45:1831–42. [12] Djurović S, Crabtree CJ, Tavner PJ, Smith AC. Condition monitoring of wind turbine induction generators with rotor electrical asymmetry. IET Renew Power Gener 2012;6(4):207–16. [13] Yang W, Tavner PJ, Wilkinson MR. Condition monitoring and fault diagnosis of a wind turbine synchronous generator drive. IET Renew Power Gener 2009;3 (1):1–11. [14] Santos FV, Guasp MR, Henao H, Sanchez MP. Diagnosis of rotor and stator asymmetries in wound-rotor induction machines under nonstationary operation throughthe instantaneous frequency. IEEE Trans Ind Electron 2014;61(9):4947–59. [15] Kia MY, Khedri M, N. H.R, Nejad MAS. Hybrid modelling of doubly fed induction generators with inter-turn stator fault and its detection method using wavelet analysis. Gener, Transm Distrib IET 2013;7(9):982–90. [16] Howard DF, Habetle TG, Harley RG. Improved sequence network model of wind turbine generators for short-circuit studies. Energy Convers IEEE Trans 2012;27(4):968–77. [17] Yang W, Tavner PJ, Crabtree CJ, Wlikinson M. Cost-effective condition mon- itoring for wind turbines. IEEE Trans Ind Electron 2010;57(1):263–71. [18] Gong X, Qiao W. Imbalance fault detection of direct-drive wind turbines using generator current signals. IEEE Trans Energy Convers 2012;27(2):468–2766. [19] Satpathy, AS, Kishore, NK, Sahoo, NC. Emulation of WT characteristics based on separately excited DC motor using LabVIEW. In: Proceedings of CCEE IISc Bangalore; 2011. p. 235–240. [20] Himani, Ratna DAHIYA. Condition monitoring of a wind turbine generator using a standalone wind turbine emulator. Front. Energy. 2016;10(3):286–97. [21] Himani Garg, Ratna Dahiya. Current signature analysis and its application in the condition monitoring of wind turbine for rotor faults. Energy Syst 2016. [22] Tiwari A, Murthy S, Singh B, Shridhar L. Design-based performance evaluation of two-winding capacitor self-excited single-phase induction generator. Electr Power Syst Res 2003;vol. 67(no. 2):89–97. [23] Wu Y, Li Y. Diagnosis of rotor winding interturn short-circuit in turbine generators using virtual power. Energy Convers IEEE Trans 2015;30(1):183–8. [24] Y. Gritli, et al. Double frequency sliding and wavelet analysis for rotor fault diagnosis in induction motors under time-varying operating condition. In: Proceedings of the IEEE International Symposium on Diagnostics for Electric Machines, Power Electronics and Drives. Bologna, Italy; Sep. 2011. p. 676–683. [25] Gritli Y, Stefani A, Rossi C, Filippetti F, Chatti A. Experimental validation of doubly fed induction machine electrical faults diagnosis under time-varying conditions. Elect Power Syst Res 2011;81(3):751–66. [26] Xiaosong H, Feng-chun S, Sheng-bo L, et al. NARX modelling of a lithium iron phosphate battery used for electrified vehicle simulation. Int J Model Identif Control 2013;20(2):181–9. [27] Amel Adouni, Dhia Chariag, Demba Diallo, Mouna Ben Hamed, Lassaâd Sbita. FDI based on artificial neural network for low-voltage-ride-through in DFIG- based wind turbine. In: ISA Transactions, 2016.05.009%200019-0578. Himani, R. Dahiya / ISA Transactions 65 (2016) 537–546546