A Survey of Broken Rotor Bar Detection Using PT and

2015 IEEE Student Conference on Research and Development (SCOReD)

A Survey of Broken Rotor Bar Detection Using PT and HT in Squirrel Cage Electrical Machine Mohammad Rezazadeh Mehrjou1,2, Norman Mariun1,2, Norhisam Misron2, Mohd Amran Mohd Radzi1,2 1

Centre for Advanced Power and Energy Research (CAPER), Faculty of Engineering, Universiti Putra Malaysia 2 Department of Electrical and Electronic Engineering, Faculty of Engineering, Universiti Putra Malaysia Serdang, Malaysia [email protected],[email protected] by broken rotor bar fault, such as low-frequency torsional oscillations in drive trains with gearboxes, voltage fluctuations, or bearing faults that cause an incorrect detection [3]. Another drawback for this method is the effect of side band leakage from fundamental frequency to fault frequency. The frequency feature of fault “ ” during transient period or motor start-up varies between zero and fundamental frequency and therefore a fault detection algorithm cannot be misinterpreted [3]. To overcome these problems, some researcher proposed to use demodulation methods to extract the fault related feature in low frequency from raw signal. This paper presents a survey of broken rotor bar detection using low harmonic stator current signal in Squirrel cage electrical machine.

Abstract—Early detection of faults in electrical machinesare imperative because of their diversity of use in different fields.A suitable fault monitoring scheme helps to stop propagation of the failure or limit its escalation to severe degrees and thus prevents unscheduled downtimes that causeloss of production and financial income. In this study, asurvey of methods based on the Park transform and Hilbert transform for broken rotor bar fault monitoring in Squirrel cage electrical machine is presented. Keywords—squirrel cage electrical machine; broken rotor bar; motor current signature analysis; Park transform; Hilbert transform

I.

INTRODUCTION

Three-phase squirrel-cage electrical machines are utilized in the most industrial applications for conversion of electrical power to mechanical power. These machines are usually rugged and reliable those requirelow maintenance. However, similar to otherdevices, they finally decline and be stopped. In industry, health monitoring of an electrical device and detecting its malfunction before it becomes catastrophic is vital to prevent severe damage of the device. In this regards, the need of cost-effective preventive maintenance procedures is raised. These procedures are generally based on condition monitoring of the device that includes sensing and analysing the real-time signals obtained from it. Squirrel-cage electrical machines are susceptible to different types of failure. Among these failures, broken rotor bar fault has been considered as the most important one [1] and detectionof this fault has been widely regarded among researchers since 1970 [2]. In this respect, a wide range of monitoring techniques has been proposed and developed for efficient and reliable detection of broken rotor bar in electrical machines [1]. Among all condition monitoring techniques, motor current signature analysis (MCSA) is the most widely used method for broken rotor bar detection, because it is noninvasive and easy to use. The MCSA methods utilized inliteratures for broken rotor bar detection can becategorized in two types based on measurement of stator current, which are stator current during steady-state operation and stator current during the start-up transient. Both of these methods have their own benefits and limitations [1].

II.

LOW FREQUENCY FAULT FEATURE

A. Theoretical background New frequency components are added to the electromagnetic field profile with presence of broken rotor bar failure in squirrel cage electrical machine [4]. Bellini et al., based on the backward filed theory, explained the flux density in electrical machines in two states of motor, it is faulty or fault-free [5]. Theoretically, when broken rotor bar causes rotor asymmetry, the stator current can be written as [6,7]: ∑ ∑ (1) where: : Stator current : Amplitude of fundamental frequency. : Fundamental frequency. : Main phase shift angle of stator current. : Amplitude for harmonic component : Amplitude for harmonic component : Phase shift angle of harmonic component : Phase shift angle of harmonic component k: Integer number (1, 2, 3,…). Above (1) can be rephrased as:

The limitations of stator current during steadystateoperation is frequencies similar to frequency that caused

. . . .

(2)

552

2015 IEEE Student Conference on Research and Development (SCOReD)

The space-vector components of currentis given by and . These components can be computed as a function of the three-phase stator currents:

∑ ( (

) )

√

(3)

√

√

√

√

∑ ( (

) )

√ (5) (6)

(11)

In 1995, Cruz et al.[15] introduced a new approach for broken rotor bar detection in three-phase squirrel cage electrical machine using Park’s Vector of motor currentin stationary regime. They proved when any bar is broken, the spectrum for the space vector modulus in the stationary reference frame indicates a dc level, which is generated by the fundamental component of the supply current, plus two additional components at frequencies ( ). They mentioned that the later components ( ) in the spectrum that is obtained through extended Park’s vector approach can be easily separated from the fundamental component. The reason is they are far apart from fundamental component and also the amplitude of fundamental component is automatically reduced by Park’s transformation. Later on the absolute valuemagnitude of positive-sequence component was used to detect the fault [8,16]. Noted that Jaksch mentioned if the Park transform is used for demodulation, all amplitudes of the modulating signals are magnified by 1.5, therefore to find out the real amplitudes of modulating signal they need to be multiplied by 2/3, or be subtracted approximately 3.5 dB in the case of dB scale [16]. Kia et al. used park transform to compute squared space vector magnitude as an input for calculation of energy feature in DiscreteWavelet analysis of the signal [6].The advantage of their research is the energy feature is used to detect the rotor fault as well as to evaluate its severity without using any slip estimation.

(7) From (5), the current envelope and its phase shift depend on broken rotor bar fault can be extracted [5]. The broken rotor bar faults in squirrel cage electrical machine induced the stator current amplitude at frequency components of , as shown in equation and . The amplitude of these frequency components can be used as indication of fault related feature [8,9]. In [10], it was shown the amplitude increases upon when the number of broken bars increases and on the other hand, the term rises due to increase of the slip [11].The modulation index for this fault frequency, and estimated fault frequency amplitude can be thus referred as [10,12,13]: (8)

III.

(10)

(12)

With:

Where is the number of rotor bars and consecutive broken rotor bars.

(9)

The signal magnitude “ ” and phase “ ” of positive-sequence component returned by Park transform are calculated as:

(4)

The equation (2) can be rewritten as follows:

√

√

is the number of

LOW FREQUENCY SIGNAL EXTRACTING METHOD

This section presents the methods that are used for broken rotor bars fault diagnosis in squirrel cage electrical machine based on the Park Transform and Hilbert transform. This two methods used as extraction of low frequency fault related feature as mentioned in previous section.

However, Park's Vector pattern has some vague issues that limit its application for fault diagnosis. These issues include: first, it is not clear whether patterns for different faults are distinctive or not. Since similar deviation in the current Park’s vector can be generated from different faults, it would not be practical to isolate different faults. Second, if noise or any practical problems are considered how it will affect the pattern [17]. Third, for a specific fault, the information inside onephase current and space vector current are equal. Forth, Park's Vector approach ignores the non-idealities in the squirrel-cage electrical machine and inherent unbalance of the supply voltages. Moreover, Zhang et al. [18] stated that when the load is light or nonexistent, extended Park’s vector approach fails to detect the characteristic fault components. Since, more

A. Park transform Park Transform (PT) is an approach relied on identifying a specific pattern of the current, which istaken from transformation of three-phase stator currents to an equivalent two-phase system. When there is no failure in a squirrel-cage electrical machine, the sum of stator current in three phases is equal to zero. In this condition, Park’s vector transform of the stator currents is a circular pattern centered at the origin of coordinates. If any failure presents in the motor, Park’s vector pattern of the stator currents becomes elliptic. The pattern ellipticity can be proportional to the fault severity, and orientation of its major axis depends on the faulty phase [14].

553

2015 IEEE Student Conference on Research and Development (SCOReD)

|

computational effort is required to acquire the space vectors spectrum; this procedure may be adopted as supplementary method in diagnostic systems [5].

|

(14)

IV.

A new complex signal, called analytic signal, is created by adding a real signal and its Hilbert transform defined as (15) .

also can be defined as (16)

Where and are the instantaneous amplitude and the phase of , respectively. These parameters can be computed as follows. √

|

(21)

CONCLUTION

Motor current signature analysis is a great technique for condition monitoring and fault diagnosis in squirrel-cage electrical machine. The measurement of stator current is sufficient to detect fault of the machine without disturbing the normal operation.This paper has presented a motor fault diagnosis method forsquirrel-cage electrical machine based on MCSA approach for brokenrotor bars detection using Park Transform and Hilbert transform. Table I shows the summary of published papers with the aim of using Park vector transform and Hilbert transform of the stator current for broken rotor bar detection.

(15) The analytic signal

|

Spectral analysis of the phase current based on Hilbert transform has some significant advantages over traditional methods that used for fault detection and it is mentioned by [21]:First, In the spectrum of the signal only one fault-related frequency feature is appeared, instead of two sidebands frequency features , which are directly positioned at the characteristic frequency of the failure. Second, prevent the effect of frequency leakage due to the fundamental frequency . Third, linear scale is used instead of a logarithmic because of removing the value of fundamental frequency and also the legibility of graph improves. Fourth, need to use low sampling rate because of detection in low frequency.

(13)

filters all the negative frequencies of

|

Hilbert transformation provide an alternative MSCA approach, which enables detection of frequency component associated to the broken bar fault even at low slips [11,21].

Equation 14 relies on the fact that positive frequencies of the spectrum are shifted by degree of –90◦, while the negative frequencies are shifted by 90◦ .Then, theHilbert transform of a signal can be viewed as a filter of amplitude unity and phase ±90◦ depends on the sign of the frequency in the spectrum of the input signal.

The signal

(19)

In [11,21,22] the following equation was used to calculate the alternating component of the modulus of the analytic signal for diagnose of broken bar from the values of the phase current, :

The Fourier Transform of the function ( ) is defined as: {

|

(20)

The first step of this approach is to extract the analytic signal from one phase current signal by applying the Hilbert transformto the current signal. The Hilbert transform of a signal is a convolution between the original signal and the function ⁄ . Considering a real time current signal , Hilbert transform for this signal, , is expressed as [19]:

( )

|

Different types of Hilbert transform approach may be used for detection of bar breakage. Liu et al.described a new method, based on the spectral analysis of the current Hilbert Modulus (absolute value), to detect faults in rotor cage of three-phase squirrel cage electrical machine [20]. The Hilbert Modulus was defined as the square of a signal and its conjugation.

B. Hilbert transform Hilbert transform (HT), which is used to extract the instantaneous magnitude (envelope) of current signal, is an ideal phase shifting tool in signal processing techniques. The important advantage of Hilbert transform is it increases resolution, both in amplitude and frequency regions [8]. The complete explanation on application of Hilbert transform for current analysis of electrical machines, both in faultyand faultfree conditions, is discussed in [11]. Hilbert transform is usually a preprocessing step when using other methods of signal analysis, like Fast Fourier Transform, Wavelet Transform (WT), Empirical Mode Decomposition (EMD) and etc., to extract the fault-related features.

∫

|

As future work, it is necessary to perform studies on the effects of load variation on stationary operation of machine in order to evaluate the proposed methods. The effects of starting load onthe motor start-upoperation(transient) is another point for furtherstudy to improve the diagnosis of incipient faults using these methods.

(17) (18)

When the analytic signal is obtained using (15) or (16), the envelope of this complex, defined as the absolute value of the signal, is comprised using (19):

554

2015 IEEE Student Conference on Research and Development (SCOReD)

TABLE I.Summary of published paper with the aim of using Park Transform and Hilbert transform for broken rotor bar fault detection Year

Fault

Phase

Signal

Load Condition

Methodology

Feature

Reference

1995 2000

1-10 BRB 1,2,3,4BRB

3 3

Stationary Stationary

-

PT PT

Park's Vector pattern 2sf

15 23

2001

BRB

3

Stationary

Varying Loads

PT

Park's Vector pattern

24

1,3

Stationary

PT and HT

Three different Magnitude

8

1,3

Stationary

PM and HM HT,PT and B&K demodulation techniques HT , EMD and FFT

2skf

20

Three different Magnitude

16

2sf

18

2004

Stator fault, Unbalancing, 2,4 BRB 1,2BRB

2005

BRB and Eccentricity

1,3

Stationary

25%, 50% ,75% and 85% of rated Load

2007

1 BRB

1

Stationary

Unload and Full Load

2007

1 BRB

1

Stationary

-

HT and WT

Statistical parameters (mean value estimate)

19

2007

Stator fault, Unbalancing, 2,4 BRB

1,3

Stationary

25%, 50% and 85% of rated Load

PT and HT

Three different Magnitude

25

2009

1,3 BRB

1,3

Stationary

various load

(HT and PT) with DWT

Energy

6

2009

Stator fault, Unbalancing, 2,4 BRB

1,3

Stationary

PT and HT

Three different Magnitude

26

2009

1BRB

1

Stationary

DHT and FFT

2sf

11

2009 2011

1BRB 1,2 BRB

1 1

Stationary Start-up

25%, 50% and 85% of rated Load Low, medium and full rated Load Low rated Load Unload and Load

2sf Energy

21 27

2011

1 BRB

1

Stationary

-

HT and FFT DWT and HT HT and ensemble EMD

2sf , 4sf , 6sf

28

2003

2012

1 BRB

1

Stationary

10%, 50% and 100% of rated Load

HT and FFT

1

Stationary

Static load and variable load

HT and FFT

1,3

Stationary

3

2013

3 BRB, Bowed rotor,Unbalanced rotor, Statorfault, Bearing Fault Stator fault, Unbalancing, 2,4 BRB 2,4 BRB

2013

2 BRB

1

2013

BRB

1

Start-up Start-up and Stationary Stationary

2014

1,2 BRB

1

Stationary

2015

1 BRB

1

Stationary

2012 2012

25%, 50% and 85% of rated Load -

PT and HT

Three different Magnitude

30

2sf

31

-

HT and WT

Statistical parameters (Skewness, Kurtosis)

32

Low load 0%, 50% and 100% of rated Load Unload and 70% of the rated load

HT and ESPRIT

2sf , 4sf

33

HT

DC and 2sf

7

HT and Zero crossings

2sf

34

[4]

[5]

REFERENCES

[3]

29

PT and CWT

The authors would like to express their gratitude to Ministry of Education Malaysia for financial support through grant number FRGS-5524356 and Universiti Putra Malaysia for the facilities provided during this research work.

[2]

22

25%, 50% and 85% of rated Load Unload

ACKNOWLEDGMENT

[1]

Statistical parameters (Peak, Average, Variance, Skewness and Kurtosis) Statistical parameters, Frequency features and Autoregressive model coefficients

S.S. Kalsi, B. Adkins, & D.D. Stephen, “Calculation of system-fault currents due to induction motors,” In Proceedings of the Institution of Electrical Engineers, vol. 118, pp. 201-215, 1971. M.R. Mehrjou, N. Mariun, M. Hamiruce Marhaban, and N. Misron, “Rotor fault condition monitoring techniques for squirrel-cage induction machine-A review,” Mechanical Systems and Signal Processing, vol. 25, no. 8, pp. 2827-2848, 2011 J. Antonino-Daviu, M. Riera-Guasp, J.R. Folch, & M. Palomares, “Validation of a new method for the diagnosis of rotor bar failures via

[6]

[7]

555

wavelet transform in industrial induction machines,” IEEE Transactions on Industry Applications, vol. 42, no. 4, pp. 990-996, 2006. V. Ghorbanian, & J. Faiz, “A survey on time and frequency characteristics of induction motors with broken rotor bars in line-start and inverter-fed modes,” Mechanical Systems and Signal Processing, vol. 54, pp. 427-456, 2015. A. Bellini, F. Filippetti, G. Franceschini, C. Tassoni, G.B. Kliman, “Quantitative evaluation of induction motor broken bars by means of electrical signature analysis,” IEEE Trans. Ind. Appl., Vol. 37, no 5, pp. 1248–1255, 2001. S.H.Kia, H. Henao,G.A. Capolino, “Diagnosis of broken bar fault in induction machines using discrete wavelet transform without slip estimation,” IEEE Trans Appl, vol. 45, pp. 1395–1404, 2009. W. Laala, S.E. Zouzou, & S. Guedidi, “ Induction motor broken rotor bars detection using fuzzy logic: experimental research,” International Journal of System Assurance Engineering and Management, vol. 5, no. 3, pp. 329-336, 2014.

2015 IEEE Student Conference on Research and Development (SCOReD)

[8]

[9]

[10]

[11]

[12]

[13]

[14]

[15]

[16]

[17]

[18]

[19]

[20]

I. Jaksch, “Faults diagnosis of three-phase induction motors using envelope analysis, ” 4th IEEE International Symposium on SDEMPED, 2003, pp. 289-293. A.M.da Silva, R.J. Povinelli, and N.A.O. Demerdash, “Induction machine broken bar and stator short-circuit fault diagnostics based on three-phase stator current envelopes,” IEEE Trans. Ind. Electron., vol. 55, no. 3, pp. 1310–1318, Mar. 2008. F. Filippetti, M. Martelli, G. Franceschini, & C. Tassoni, “Development of expert system knowledge base to on-line diagnosis of rotor electrical faults of induction motors,” IEEE In Industry Applications Society Annual Meeting, 1992, pp. 92-99. R. Puche-Panadero, M. Pineda-Sanchez, M. Riera-Guasp, J. Roger-Folch, E. Hurtado-Perez, & J. Perez-Cruz, “Improved resolution of the MCSA method via Hilbert transform, enabling the diagnosis of rotor asymmetries at very low slip,” IEEE Transactions on Energy Conversion, vol. 24, no. 1, pp. 52-59, 2009. C.Hargis, B. G. Gaydon, K. Kamash, “The detection of rotor defects in induction motors, “ Proc. IEE Conf. on Electrical Machines, pp.216-220, London, Sept. 1982 W. T. Thomson, D. Rankin, “Case Histories of on Line Rotor Cage Fault Diagnosis,” in Proc. Int. Conf: Condition Monitoring (Swansea, UK), Pineridge Press, 1987, pp. 798-819. M.A. Awadallah, M.M. Morcos, “Application of AI Tools in Fault Diagnosis of Electrical Machines and Drives-An Overview,” IEEE Transactions on Energy Conversion, vol. 18, no. 2, pp. 245-251, 2003. S. M. A. Cruz, J. F. S. Carvalho, & E. S. Saraiva, “Rotor cage fault diagnosis in three-phase induction motors, by Park's vector approach,” Thirtieth IAS Annual Meeting, 1995, pp. 642-646. I. Jaksch, & J. Bazant, “Demodulation methods for exact induction motor rotor fault diagnostic” . 5th IEEE International Symposium on SDEMPED, 2005, pp. 1-5. M. Haji, & H. Toliyat, “Pattern recognition-a technique for induction machines rotor broken bar detection,” IEEE Transactions on Energy Conversion, vol. 16, no. 4, pp. 312-317, 2001. J.W. Zhang, N.H. Zhu, L. Yang, Q. Yao, Q. LU, “A Fault Diagnosis Approach for Broken Rotor Bars Based on EMD and Envelope Analysis,” Journal of China University of Mining and Technology, vol. 17, no. 2, pp. 205-209, 2007. G.A. Jimenez, A. O. Munoz, and M. A. Duarte-Mermoud, “Fault detection in induction motors using Hilbert and Wavelet transforms,” Electrical Engineering, vol. 89, no. 3, pp. 205-220, 2007. Liu, Z., Zhang, X., Yin, X., & Zhang, Z. “Rotor cage fault diagnosis in induction motors based on spectral analysis of current Hilbert modulus,” In Power Engineering Society General Meeting, June, 2004.

[21] R. Puche-Panadero, J. Pons-Llinares, V. Climente-Alarcon, M. PinedaSanchez, “Review diagnosis methods of induction electrical machines based on steady state current,” In XI Portuguese–Spanish conference in electrical engineering, 2009. [22] D. Matić, F. Kulić, M. Pineda-Sánchez, & I. Kamenko, “Support vector machine classifier for diagnosis in electrical machines: Application to broken bar,” Expert Systems with Applications, vol. 39, no. 10, pp. 8681-8689, 2012. [23] S.M.A. Cruz, & A.J. Marques Cardoso, “Rotor cage fault diagnosis in three-phase induction motors by extended Park's vector approach,” Electric Machines & Power Systems, vol. 28, no. 4 ,pp.289-299, 2000. [24] S.M.A. Cruz, & A.J. Marques Cardoso, “Discriminating between rotor asymmetries and time-varying loads in three-phase induction motors,” InProc. COMADEM, September 2001, pp. 319-327. [25] I. Jaksch, & P. Fuchs, “Rotor cage faults detection in induction motors by motor current demodulation analysis,” IEEE International Symposium on SDEMPED, September 2007, pp. 247-252. [26] I. Jaksch, & P. Fuchs, ”Demodulation analysis for exact rotor faults detection under changing parameters,” IEEE International Symposium on SDEMPED, 2009, pp. 1-7. [27] J. Antonino-Daviu, S. Aviyente, E. Strangas, & M. Riera-Guasp, “A scale invariant algorithm for the automatic diagnosis of rotor bar failures in induction motors,” IEEE International Symposium on Industrial Electronics, June 2011, pp. 496-501. [28] J. Wang, R.X. Gao, & R. Yan, “Broken-rotor-bar diagnosis for induction motors,” In Journal of Physics: Conference Series, vol. 305, no. 1, 2011. [29] J. Wang, S. Liu, R.X. Gao, & R. Yan, “Current envelope analysis for defect identification and diagnosis in induction motors,” Journal of Manufacturing Systems, vol. 31 ,no. 4, pp. 380-387, 2012. [30] I. Jaksch, Rotor Cage Fault Detection in Induction Motors by Motor Current Demodulation Analysis, Intech Open Access Publisher, 2012. [31] M.U. Petushkov, “Analysis of defects in the rotor asynchronous motor during start,” International Journal Of Applied And Fundamental Research, 2013 [32] D. Matic, Z. Kanovic, D. Reljic, F. Kulic, D. Oros, & V. Vasic, “Broken bar detection using current analysis-A case study” 9th IEEE International Symposium on SDEMPED, August 2013, pp. 407-411. [33] B. Xu, L. Sun, L. Xu, & G. Xu, “Improvement of the Hilbert method via ESPRIT for detecting rotor fault in induction motors at low slip,” IEEE Transactions on Energy Conversion, vol.28, no. 1, pp. 225-233, 2013. [34] A. Sapena-Bano, M. Pineda-Sanchez, R. Puche-Panadero, J. MartinezRoman, & Z. Kanovic, “Low-Cost Diagnosis of Rotor Asymmetries in Induction Machines Working at a Very Low Slip Using the Reduced Envelope of the Stator Current,” IEEE Transactions on Energy Conversion, pp. 1-11, 2015

556