55_

Page 1

Proc. of Int. Conf. on Control, Communication and Power Engineering 2010

Fault Detection of Induction Motor using Energy and Wavelets Biju K, Jagadanand G (Member, IEEE), Saly George Department of Electrical Engineering College of Engineering Munnar National Institute of Technology Calicut, Kerala, India Email: bijuk1978@gmail.com, jagadanand@nitc.ac.in, saly@nitc.ac.in Abstract— Induction motors are widely used in industries for the conversion of electrical energy into mechanical energy. The failure of induction motors leads to plant shutdown, production loss etc in industries. So a reliable fault detection scheme which can be used for detecting the faults in the beginning stage itself is necessary for the protection of induction motors. Fault diagnosis in induction motor based on stator current is conventionally conducted by spectral analysis. But classical spectral analysis doesn’t provide good results for transient and continuously varying load torque conditions. In this work, energy contained in the signal is used for the fault detection of induction motors. The experimental data from the induction motor are acquired under healthy and faulty conditions of the motor. Wavelets have been used for the feature extraction from the acquired data. Using the extracted features the data is classified into healthy and faulty classes by computing the energy.

primarily vibration sensors for the detection of the faults. These vibration sensors are delicate and expensive. But the stator current monitoring can provide the same indications without requiring access to the motor [3], [4]. This technique mostly utilizes the results of spectral analysis of the stator current of an induction motor to spot the failure of the induction motor [9], [15]. Nowadays diagnosis of induction motors uses the modern measurement techniques, data processing techniques and spectral analysis techniques [10]. The most commonly used fault related feature extraction techniques are Fourier spectral analysis by FFT, Instantaneous power FFT [20], Bi-spectrum, High resolution spectral analysis [27], wavelet time-frequency analysis and Artificial intelligence techniques [2], [12]. II.

Keywords— Induction motor, fault detection, feature extraction, feature selection, energy, Wavelets.

I.

A. Problem Statement The current signature analysis of induction motors is conducted conventionally by spectral analysis [7]. But the method is not accurate under transient conditions and varying load torques [29]. The frequency components depend on the value of slip and therefore the spectral analysis gives good results only under near full load conditions [21].

INTRODUCTION

Induction motors are used as the work force in most of the industrial applications. They are very robust and highly reliable machines. However they may be subjected to different types of faults [18]. Failure of such induction motors may cause plant shut down, reduced production, accidents etc. in the production line. Prevention of induction motor failure is thus a major issue in industries [1]. The main reason for the motor faults is mechanical and electrical stresses [19]. Mechanical stresses are caused by overloads and abrupt load changes, which may cause bearing faults and rotor bar breakage. The electrical stresses may produce stator winding short circuits and result in a complete motor failure [17]. The major faults in induction motor include stator winding faults, rotor faults, bearing faults, eccentricity related faults etc. [11]. Bearing failures are responsible for approximately two-fifths of all faults. Inter-turn short circuits in stator windings stand for nearly one-third of the reported faults [13]. Broken rotor bars and end ring faults represent around ten percent of the induction motor faults [5], [14]. Condition monitoring of the performance of induction motors received considerable attention in recent years [7], [16]. Many condition monitoring methods have been proposed for different type of induction motor fault detection and localization [8]. Large electromechanical systems are usually equipped with mechanical sensors,

B. Proposed Approach In classical Fourier analysis, the power of a signal can be computed by taking the square of the absolute value of the Fourier transform coefficients. Similarly with the wavelet transform the power of a signal can be computed by adding the square of all the coefficients of the details and the final approximation [23]. The wavelet analysis gives both time and frequency information simultaneously [24], [28]. The energy contained in the signal can be used as a good tool for the fault detection of induction motors. The energy for every detail has been calculated by adding the squared coefficients of the details and final approximation [26]. So the proposed approach uses a fault detection method based on wavelet decomposition and energy computation [25]. The wavelet decomposition and energy computation is done using MATLAB 7.1. C. Experimental Work Data are acquired from the induction motor using current sensors, Data acquisition card (DAC) and LABVIEW. The current sensor used is LA 55-P of LEM make. The Data acquisition card used is PCI 6521 of NI make. Number of samples acquired is 1024 so that it is a 210

© 2009 ACEEE

METHODOLOGY


Proc. of Int. Conf. on Control, Communication and Power Engineering 2010

turn short in the same phase and turn to turn short between different phases. Turn to turn short in same phase is created in R phase. Here 6% of the total turns in the R phase are shorted through different resistances such as 1 k Ohm, 500 Ohm and 100 Ohm. Turn to turn short between different phases is created between R and Y phase. This is done by interconnecting different turns in R and Y phases by resistances such as 50 K, 20 K and 10 K.

power of 2. The sampling frequency selected is 5.12 kHz so as to get an integer number of cycles. The data from the induction motor in the form of stator current is sensed using sensors and is sent to the CPU using DAC. Data are then collected in the form of instantaneous voltage values and is saved as text files using NI-LABVIEW software. The ratings of the induction motor selected are Make Crompton Greaves Rated power 3 hp/2.2 kW Rated voltage 415 V Rated current 4.5 A Rated speed 1440 rpm The Data acquisition system is shown in Figure 1

F. Feature Extraction The data acquired doesn’t directly reveal any information usable for the detection of the fault. So feature extraction is done. Here energy contained in the signal is calculated to extract the features. The feature extraction is done with the help of wavelets. In this case, first wavelet coefficients are obtained and then energy is calculated using it. The wavelet coefficients are obtained by wavelet decomposition. The wavelet selected for analysis is DB10 of level 8 [22]. The frequency bands for the 8 level decomposition of the signal using DB10 are shown in table 1. The sampling frequency selected for data acquisition is 5.12 kHz. TABLE 1. WAVELET DECOMPOSITION DETAILS

Figure 1. Data acquisition system

Data under healthy condition of the motor were acquired at first. Then different faults were then created in the motor and data was obtained for the faulty conditions. D. Good Condition Data acquired for the induction motor under different load conditions such as 1. No load condition 2. Light load condition(60% load) 3. Medium load condition(80% load) 4. Full load condition

Frequency bands (Hz)

1 2 3 4 5 6 7 8

2560-1280 1280-640 640-320 320-160 160-80 80-40 40-20 20-10

G. Feature Selection On analyzing the wavelet coefficients of different signals, it is observed that significant variations exist in the D4, D5, D6 and D7 levels for faulty machines compared to that of healthy one. From the energy calculated for various levels, energy of the wavelet coefficients in the D5 level is selected as the feature for fault detection.

E. Faulty Conditions Data acquired for the induction motor under different fault conditions such as 1. External faults 2. Internal faults 1) External Faults: Unbalance in the supply voltage and single phasing are the two types of external faults of an induction motor that are considered in this work. Unbalance in supply voltage is created by connecting three single phase transformers in three phases of the supply. Here supply voltage in B phase is reduced to 10% less than the rated voltages in the other two phases to create unbalance. Single phasing is created by connecting a knife switch in B phase. Here one phase is opened to create single phasing. 2) Internal Faults: The two types of internal faults considered for induction motor in this work are turn to

III. RESULT ANALYSIS In each case the energy of the wavelet coefficients for every detail is calculated and tabulated. The energy for every detail has been calculated by adding the squared coefficients of the details and final approximation. The energy for each detail is calculated as a percentage of the total energy. A. Healthy Condition Energy of wavelet coefficients for 8 different frequency levels is calculated in all the three phases. From the different levels it is seen that there are significant variations in the wavelet details D4, D5, D6 and D7 compared to faulty conditions. They are tabulated as shown in table 2, 3 and 4. 211

Š 2009 ACEEE

Decomposition details


Proc. of Int. Conf. on Control, Communication and Power Engineering 2010 TABLE 2. ENERGY FOR WAVELET DETAILS (R PHASE)

TABLE 6. ENERGY FOR WAVELET DETAILS (Y PHASE)

Load

D4

D5

D6

D7

No load

0.0631

0.3206

20.297

4.944

Light load

0.0921

0.2542

21.107

2.0152

Medium load Full load

0.0411

0.2637

19.230

3.8348

0.0306

0.3306

17.866

3.8636

Unbalance 10 % unbalance 20 % unbalance Single phasing

D4

D5

D6

D7

No load

0.063

0.3206

20.299

4.942

Light load

0.0923

0.2548

21.105

2.015

Medium load Full load

0.0419

0.2631

19.236

3.834

0.0307

0.3304

17.866

3.8632

Unbalance 10 % unbalance 20 % unbalance

D4

D5

D6

D7

No load

0.0639

0.321

20.292

4.947

Light load

0.0926

0.2545

21.10

2.0159

Medium load Full load

0.0415

0.2639

19.234

3.8349

0.0301

0.3309

17.862

3.8638

D6 20.1625

D7 1.1913

0.0452

0.3626

17.093

3.0166

0.0602

0.3932

20.374

3.1381

D4 0.1906

D5 0.2443

D6 20.611

D7 5.0651

0.2843

0.2119

20.529

3.1297

From the energy analysis, it was seen that energy for D5 level was increased for R and Y phases and it was decreased for B phase with an increase in voltage unbalance. From the results, it was concluded that the voltage in B phase was reduced to create the unbalance. This could be used as a good indicator of unbalance in the supply voltage.

TABLE 4. ENERGY FOR WAVELET DETAILS (B PHASE) Load

D5 .3499

TABLE 7. ENERGY FOR WAVELET DETAILS (B PHASE)

TABLE 3. ENERGY FOR WAVELET DETAILS (Y PHASE) Load

D4 0.2858

C. Internal Faults Different types of internal faults considered are turn to turn fault between two phases and turn to turn fault in the same phase. 1) Turn to turn fault between different phases: Energy of wavelet coefficients for different levels were calculated. From the different levels it was seen that there are significant variations in D4, D5, D6 and D7. They were tabulated as shown in table 8, 9 and 10. TABLE 8.

B. External Faults

ENERGY FOR WAVELET DETAILS (R PHASE)

The different types of external faults considered here include unbalance in supply voltage and single phasing (most severe case of unbalancing). Energy of wavelet coefficients for different levels was calculated. From the different levels it is seen that there are significant variations in D4, D5, D6 and D7. They are tabulated as shown in table 5, 6 and 7.

Fault level (Ohm) 50 k

D4

D5

D6

D7

0.1048

0.3143

22.7344

4.7477

20 k

0.0355

0.3655

18.6597

4.3909

10 k

0.0283

0.3697

18.598

3.6572

TABLE 5.

TABLE 9.

ENERGY FOR WAVELET DETAILS (R PHASE)

ENERGY FOR WAVELET DETAILS (Y PHASE) D4

D5

D6

D7

3.1804

Fault level (Ohm) 50 k

0.0688

0.3184

18.5188

1.6101

18.371

3.1257

20 k

0.0764

0.3442

20.5825

1.9409

20.568

3.1579

10 k

0.0654

0.3545

20.9324

1.9653

Unbalance

D4

D5

D6

D7

10 % unbalance 20 % unbalance Single phasing

0.0374

0.3163

17.754

0.0404

0.3258

0.0583

0.3692

212 Š 2009 ACEEE


Proc. of Int. Conf. on Control, Communication and Power Engineering 2010 TABLE 10.

IV. CONCLUSION

ENERGY FOR WAVELET DETAILS (B PHASE) Fault level (Ohm) 50 k

D4

D5

D6

D7

0.1033

0.2718

25.3213

4.7324

20 k

0.0851

0.2581

22.6845

4.7638

10 k

0.0608

0.236

21.3042

4.5256

A fault detection method using energy contained in the signal is proposed in this paper. From the results of energy analysis, it was seen that the proposed method could be used for the detection of faults in the induction motor. Wavelets were used for feature extraction and it was seen that good feature could be extracted using wavelets. Thus the experiments performed and the results obtained showed that the energy analysis using the wavelets achieved good results in the field of fault diagnosis of induction motor.

From the energy analysis it was seen that energy in D5 level has been increasing with increase in level of fault for both R and Y phase. So from the results, it was identified that the fault was made between R and Y phases. This could be used as a good indicator for turn to turn fault between different phases.

REFERENCES [1] Mohamed El Hachemi Benbouzid, “A Review of Induction Motor Signature Analysis as a Medium for Fault Detection”, IEEE Transactions on Industrial Electronics, Vol. 47, No. 5, October 2000. [2] Fiorenzo Filippetti, Giovanni Franceschini, Carla Tassoni, Peter Vas, “Recent Developments of Induction Motor Drives Fault Diagnosis Using AI Techniques”, IEEE Transactions on IndustrialElectronics,vol47,no5,Oct 2000. [3] R. R. Schoen et al., “An unsupervised, on-line system for induction motor fault detection using stator current monitoring,” IEEE Trans. Industry Applications, Vol. 31, pp. 1280–1286, Nov./Dec. 1995. [4] Randy. R. Schoen, Thomas G.Habetler, Farrukh Kamran and Robert G.Bartheld, “Motor bearing damage detection using stator current monitoring,” IEEE Transactions on Industry Applications, vol. 31, pp. 1274–1279, Nov.ember/December 1995. [5] N. M. Elkasabgy et al., “Detection of broken bars in the cage rotor on an induction machine,” IEEE Transactions on Industry Applications, vol. 28, pp. 165–171, Jan. /Feb. 1992. [6] Andreas stavrou, Howard G. Sedding and James Penman, “Current monitoring for detecting inter turn short circuits in induction motors”, IEEE Transactions on Energy conversion vol.16, No.1, March 2001. [7] S. Chen et al., “A new approach to motor condition monitoring in induction motor drives,” IEEE Transactions on Industry Applications, vol. 30, pp. 905–911, July/Aug. 1994. [8] R. Hirvonen, “On-line condition monitoring of defects in squirrel cage motors,” in Proc. 1994 Int. Conf. Electrical Machines, vol. 2, Paris, France, pp. 267–272. [9] M. E. H. Benbouzid et al., “Induction motor faults detection using advanced spectral analysis technique,” in Proc. 1998 Int. Conf. Electrical Machines, vol. 3, Istanbul, Turkey, pp. 1849– 1854. [10] Mohamed El Hachemi Benbouzid, Michelle Vieira and Celine Theys, “Induction motors faults detection and localization using stator current advanced signal processing techniques,” IEEE Trans. Power Electronics, vol. 14, pp. 14–22, Jan. 1999. [11] Arfat Siddique, G. S. Yadava, and Bhim Singh,,” A Review of Stator Fault Monitoring Techniques of Induction Motors”, IEEE Transactions on Energy Conversion, vol. 20, No. 1, March 2005. [12]. M. E. H. Benbouzid et al., “Monitoring and diagnosis of induction motors electrical faults using a current Park’s vector pattern approach,” in Proc. 1999 IEEE Int. Electric Machines and Drives Conf., Seattle, WA, pp. 275–277. [13] Randy R. Schoen, Thomas G. Habetler, Farrukh Kamran, and Robert G. Bartheld, “ Motor bearing damage detection using stator current monitoring”, IEEE Transactions on Industry Applications, vol. 31, No 6, November/December1995

2) Turn to turn fault in the same phase: Energy of wavelet coefficients for different levels was calculated. From the different levels it was seen that there were significant variations in D4, D5, D6 and D7. They were tabulated as shown in table 11, 12 and 13. TABLE 11. ENERGY FOR WAVELET DETAILS (R PHASE) Fault level (Ohm) 1k

D4

D5

D6

D7

0.0821

0.3275

18.3833

1.8717

500 ohm

0.1095

0.3335

23.441

5.1421

100 ohm

0.0306

0.3443

19.1094

4.5948

TABLE 12. ENERGY FOR WAVELET DETAILS (Y PHASE) Fault level (Ohm) 1k

D4

D5

D6

D7

0.1032

0.3003

24.495

5.079

500 ohm

0.1122

0.3245

25.98

4.96

100 ohm

0.0571

0.3157

21.2326

4.5274

TABLE 13. ENERGY FOR WAVELET DETAILS (B PHASE) Fault level (Ohm) 1k

D4

D5

D6

D7

0.0798

0.2426

20.662

1.8647

500 ohm

0.0679

0.2264

19.277

1.6759

100 ohm

0.0663

0.2543

21.1379

2.0922

From the energy analysis it was found that the energy in the D5 level has been increasing with increase in level of fault for R phase. From the results, it was identified that the turn to turn fault was developed in R phase. This could be used as a good feature for the detection of turn to turn fault in any phase.

213 © 2009 ACEEE


Proc. of Int. Conf. on Control, Communication and Power Engineering 2010 Induction Machines Using Power Spectral Density in Wavelet Decomposition’, IEEE Transactions on Industrial Electronics, vol 55, No2, Feb2008. [24] Jose A.Antonino, Martin Riera, José Roger, M.Pilar ,“Validation of a new method for the diagnosis of Rotor bar failures via Wavelet Transform in induction machines”, IEEE Transactions on IndustryApplications,vol.42,No.4, July/August 2006. [25] Hugh Douglas, Pragasen pillay, Alireza K. Ziarani, “A new algorithm for transient motor current signature analysis using wavelets”, IEEE Transactions on Industry Applications, vol.40, No.5, September/October 2004. [26] Zhongming Ye,Bin Wu and Alireza Sadeghian, “Current signature analysis of induction motor mechanical faults by wavelet packet decomposition”, IEEE Transactions on Industrial Electronics,vol.50,No.6, December 2003. [27] Neil Arthur and Jim Penman, “Induction machine condition monitoring with higher order spectra”, IEEE Transactions on Industrial Electronics, vol.47, No.5, October 2000. [28] Martin Blodt, Marve Chabert, Jeremi Regnier and Jean Faucher, “Mechanical load fault detection in induction motors by stator current time frequency analysis”, IEEE Transactions on Industry Applications, vol.42, No.6, November/December 2006. [29] Fernando Briz, Michael W. Degner, Pablo Garcia and David Bragado, “Broken rotor bar detection in line fed induction machines using complex wavelet analysis of startup transient”, IEEE Transactions on Industery Applications,vol.44,No.3, May/June 2008.

[14] A. H. Bonnett et al., “Cause and analysis of stator and rotor failures in three-phase squirrel-cage induction motors,” IEEE Trans. Ind. Applications., Vol. 28, pp. 921–937, July/Aug.1992 [15] Behrooz Mirafzal and Nabeel A.O. Demeradash, “On innovative methods of induction motor interturn and broken bar fault diagnostics”, IEEE Transactions on Industry Applications,vol.42,No.2,March/April 2006. [16] Subhasis Nandi, “Detection of stator faults in induction machines using residual saturation harmonics”, IEEE Transactions on Industry Applications, vol.42, No.5, September/October 2006. [17] G. B. Kliman et al., “Noninvasive detection of broken bars in operating induction motors,” IEEE Trans. Energy Conversion, vol. 3, pp. 873–879, Dec. 1988. [18] Neelam Mehala and Ratna Dahiya, “Motor current signature analysis and its applications in induction motor fault diagnosis”, International journal of Systems applications, Engineering and development, vol.2, issue 1, 2007. [19] R. R. Schoen et al., “Effects of time-varying loads on rotor fault detection in induction machines,” IEEE Trans. Ind. Applicat, vol. 31, pp. 900–906, July/Aug. 1995. [20] S. F. Legowski et al., “Instantaneous power as a medium for the signature analysis of induction motors,” IEEE Trans. Ind, Applicat, vol. 32, pp. 904–909, July/Aug. 1996. [21] Douglas. H.Pillay, P. and Ziarani.A, “Detection of broken rotor bars in induction motors using wavelet analysis”, IEEE Transactions, pp.923-928, 2003. [22] Douglas, H.Pillay, P., “The impact of wavelet selection on transient motor current signature analysis”, IEEE Transactions [23] Jordi Cusidó, Luis Romeral, Juan A. Ortega,Javier A. Rosero, and Antonio García Espinosa,“Fault Detection in

214 © 2009 ACEEE


Turn static files into dynamic content formats.

Create a flipbook
Issuu converts static files into: digital portfolios, online yearbooks, online catalogs, digital photo albums and more. Sign up and create your flipbook.