Simulative and experimental investigation on stator winding turn and unbalanced supply voltage fault diagnosis in induction motors using Artificial Neural Networks

ISA Transactions ∎ (∎∎∎∎) ∎∎∎–∎∎∎

Contents lists available at ScienceDirect

ISA Transactions journal homepage: www.elsevier.com/locate/isatrans

Simulative and experimental investigation on stator winding turn and unbalanced supply voltage fault diagnosis in induction motors using Artificial Neural Networks Negin Lashkari, Javad Poshtan n, Hamid Fekri Azgomi Iran University of Science and Technology, Iran

art ic l e i nf o

a b s t r a c t

Article history: Received 10 March 2015 Received in revised form 23 May 2015 Accepted 3 August 2015 This paper was recommended for publication by Steven Ding

The three-phase shift between line current and phase voltage of induction motors can be used as an efficient fault indicator to detect and locate inter-turn stator short-circuit (ITSC) fault. However, unbalanced supply voltage is one of the contributing factors that inevitably affect stator currents and therefore the three-phase shift. Thus, it is necessary to propose a method that is able to identify whether the unbalance of three currents is caused by ITSC or supply voltage fault. This paper presents a feedforward multilayer-perceptron Neural Network (NN) trained by back propagation, based on monitoring negative sequence voltage and the three-phase shift. The data which are required for training and test NN are generated using simulated model of stator. The experimental results are presented to verify the superior accuracy of the proposed method. & 2015 Published by Elsevier Ltd. on behalf of ISA.

Keywords: Fault diagnosis Induction motor Interturn stator short circuit Voltage unbalance

1. Introduction Induction motors are widely used in industry. In fact, fair selfstarting capability, rugged construction, easy maintenance, low cost and reliability are contributing factors that lead induction motors to be extensively applicable [1].The condition monitoring of induction motors should be done in order to guarantee their reliability, efficiency and safety. Stator winding failures are considered as one of the most serious faults an induction motor may encounter, since they are highly probable and their damage are inextricably associated with high fault currents and high cost of maintenance [2]. Detecting stator ITSC fault is of great importance, since it probably causes a large circulating current to flow and afterward generates excessive heat in the shorted turns. Moreover, it may result in partial discharges between turns in the stator and therefore erodes the magnet wire insulation. Thus, if the diagnostic system fails to detect ITSC fault in appropriate time, it undoubtedly results in subsequent failures. Unfortunately, it is difficult to detect ITSC faults at early stages. For this reason, considerable interest has been shown in the literature to solve the difficulty in detecting this kind of fault [3–6].

n

Corresponding author. E-mail addresses: [email protected] (N. Lashkari), [email protected] (J. Poshtan), [email protected] (H.F. Azgomi).

Furthermore, occurrence of voltage unbalance at the motor stator terminals leads the life span of the machine to be shortened and its performance to be degraded due to increased losses, unbalanced line currents and excessive heating [7]. In [8] the most common causes of voltage unbalance are introduced as: unbalanced supply voltage, faulty operation of power factor equipment, unevenly distributed single-phase loads on the same power system, an open circuit on the primary distribution system and unidentified single phase to ground faults, to name a few. For induction motor ITSC fault diagnosis, different kinds of fault indicators have been used in several literatures. Among various ITSC fault indicators, stator current is widely used for diagnostics purposes, for instance: current and speed [9], line currents and phase voltages [10], current and vibration signals [11], slip and symmetrical components of stator currents [12] etc. In fact, availability of the needed sensors in the existing drive system and possessing informative instinct are the main reasons that make the stator current more preferable than other fault indicators. In another seminal work, Bouzid et al. developed a Neural Network to detect ITSC fault by using phase shifts between the line currents and phase voltages [13]. Regrettably, in [13], the study is constrained by the diagnosis of ITSC fault under balanced source voltage conditions. In fact, a very important factor that is missing in [13] is that both ITSC and voltage unbalance fault lead the stator currents to lose its balance and therefore alter the phase shift between the current and voltage of each phases similarly. Thus, another fault indicator should be utilized alongside stator current,

http://dx.doi.org/10.1016/j.isatra.2015.08.001 0019-0578/& 2015 Published by Elsevier Ltd. on behalf of ISA.

Please cite this article as: Lashkari N, et al. Simulative and experimental investigation on stator winding turn and unbalanced supply voltage fault diagnosis in induction motors using Artificial Neural.... ISA Transactions (2015), http://dx.doi.org/10.1016/j.isatra.2015.08.001i

N. Lashkari et al. / ISA Transactions ∎ (∎∎∎∎) ∎∎∎–∎∎∎

2

in order to distinct between ITSC and unbalanced supply voltage fault. Extensive research efforts have been put forth to detect unbalanced supply voltage on electric motors using negative sequence current [14–18]. Although negative sequence current has attractive features to reveal unbalanced supply voltage and stator faults, it is highly sensitive to inherent machine asymmetry. Hence, this paper considers fault indicators which are able to diagnose ITSC fault and to distinguish it from the supply voltage unbalance while they are immune from inherent machine asymmetry. Further, unlike previous model-based methods [19] in which the model of stator was always needed during the fault diagnosis procedure, this research proposes a less model dependent methodology that uses the stator model only to collect data for training and testing the Neural Network under various faulty conditions and afterwards the model would not be required anymore. The present paper is organized as follows. The model of induction motor including ITSC fault that is used to get simulated database for training and test procedure of NN is described in Section 2. In Section 3, the behavior of selected fault indicators is investigated. The description of the proposed diagnostic system for identifying ITSC fault and unbalanced voltage based on NN are detailed in Section 4. In Section 5, efficiency of the proposed method is demonstrated by the use of experimental data's coming from 3 kW squirrel-cage induction motor. Finally, the conclusion is provided in Section 6.

2. Modeling of three-phase induction motor Studying the behavior of induction motors under different fault conditions by creating real faults into the motor and monitoring its evolution does not seems sensible since firstly created faults can be dangerous for the motor and might lead to the destruction of the motor [20]; And secondly, presence of other uncertainties in real systems, like inherent machine asymmetry, non-ideal sensors, noise and disturbances, may cause the achieved results not to be reliable. Thus, an accurate model of faulty induction motor can be useful in this regard.

2.2. Model under ITSC fault The schematic of an induction motor with ITSC fault on a single phase is shown in Fig. 1. An accurate analytical model to describe ITSC fault has been presented in [22]. Assuming that motor operates under an ITSC fault in phase A, the equations of faulty model can be expressed as: Equations of stator winding flux in dq frame dλqs 2 ¼ vqs  r s iqs þ ωλds þ μr s if cos θ 3 dt

ð6Þ

dλds 2 ¼ vds r s ids þ ωλqs þ μr s if sin θ 3 dt

ð7Þ

Equations of short circuit winding flux dλas2 ¼ r f if  μr s ðids cos θ þ iqs sin θ  if Þ dt

ð8Þ

The stator and winding currents in dq frame iqs ¼ λqs a1  λqr a2 þ ðLr a3 Lm a4 Þif cos θ

ð9Þ

ids ¼ λds a1  λqr a2 þðLr a3  Lm a4 Þif sin θ

ð10Þ

if ¼ ð  λas2 þ ða5 iqs þ a6 iqr Þ cos θ þ ða5 iqs þ a6 iqr Þ sin θÞ=a4

ð11Þ

Note that the constant coefficients ai are shown in Table 1. Also i, v, r, Ll and Lm are representing currents, voltages, resistance, leakage and mutual inductance respectively. Having simulated the model discussed above by MATLAB, the stator current of the induction motor under various ITSC fault is obtained. The simulated motor is 3 kW induction motor having 250 turns per phase winding on the stator. The characteristics of the motor used to obtain the simulated results are shown in Table 2. Fig. 2 shows the simulated three-line's currents (a) in healthy condition, (b) under a stator fault of 50-shorted turns on one of the three phases and (c) under unbalanced supply voltage that affects only one phase voltage magnitude with 5% of the rated phase voltage. This figure clearly shows that both ITSC and voltage unbalance fault lead the stator currents to lose their balance and therefore change the three-phase shifts.

2.1. Healthy model In order to model the stator of a healthy induction motor, it is better to use the basic equations of induction motor in dq stationary reference frame. Dynamic equations of a three-phase healthy induction motor stand as follows [21]: Stator (subscript with s) voltage equations vas ¼ Rs ias þ

dλas dt

ð1Þ

vbs ¼ Rs ibs þ

dλbs dt

ð2Þ

dλcs dt

ð3Þ

vcs ¼ Rs ics þ

Fig. 1. Schematic of three-phase winding with ITSC fault on phase A.

where winding flux, λ, given by dλqs ¼ vqs r s iqs  ωλds dt dλds ¼ vds  r s ids þ ωλqs dt

ð4Þ

ð5Þ

Table 1 Constant coefficients of currents' equations. a0

a1

a2

a3

a4

a5

a6

Ls Lr  L2m

1 a0 Ls

Lm a1 Lr

2 μLs 3 a0

2 μLm 3 a0

μLs

μLm

Please cite this article as: Lashkari N, et al. Simulative and experimental investigation on stator winding turn and unbalanced supply voltage fault diagnosis in induction motors using Artificial Neural.... ISA Transactions (2015), http://dx.doi.org/10.1016/j.isatra.2015.08.001i

N. Lashkari et al. / ISA Transactions ∎ (∎∎∎∎) ∎∎∎–∎∎∎

3. Extraction of the fault indicators In 1918, Fortescue developed a transformation that decomposes any unbalanced system on three balanced system (negative, positive and zero). Therefore, this transformation is used to map unbalanced voltage into three sets of symmetrical balanced phases which are positive, negative and zero sequence components. The complex Fortescue transformation which used three unbalanced set of supply voltage (Va, Vb,Vc) to obtain symmetrical components (Vp, Vn, Vo) is 2 32 2 3 3 Va Vp 1 ej2π =3 ej4π =3 6 76 V 7 6V 7 j4 π =3 j2 π =3 ð12Þ e 54 b 5 4 n 5 ¼ 1=34 1 e Vo Vc 1 1 1 Since the balanced supply voltage is symmetric, only positive sequence voltage exists and other symmetrical components (negative and zero sequence) remain zero. When unbalanced supply voltage occurs, the magnitude of negative sequence voltage increases. Thus, specific attention should be paid to study the behavior of negative sequence voltage in order to identify whether power supply imbalance has occurred. The block diagram of computing symmetrical components and phase shifts is shown in Fig. 3. Table 2 Characteristics of the induction motor. Lm

Ls

Rs

Poles

f

460 V

538.68 mH

13.97 mH

4.05 Ω

4

60 Hz

4 ia ib ic

(a)

0 -2 -4 2.5

2.505

2.51

2.515

2.52

2.525

2.53

2.535

2.54

2.545

In what follows, the behavior of the phase shift and symmetrical component in presence of ITSC fault and unbalanced supply voltage will be discussed. 3.1. Study of phase shift and symmetrical component characteristics under ITSC fault Healthy operation of the induction motor results in phase voltages and line currents which are equal in magnitude and shifted by 1201 electrical. However, under faulty operation, the magnitude and therefore the phase shifts altered (see Fig. 2b). In order to study the behavior of three-phase shift under various ITSC fault and unbalanced voltage, the model of faulty stator introduced in Section 2 has been simulated. Fig. 4 shows the characteristic of the three-phase shifts as function of the number of short-circuited turns (under the load torque of 5 N m). It can be understood from Fig. 4 that the difference of the three-phase shift values is related to importance of the fault. Moreover, the smallest value of the three-phase shifts usually devotes to the phase where the fault has occurred. In fact, in case ITSC fault is in phase A, the phase shifts' increasing order is: (Phia-Phib-Phic) [13]. As stated before, since ITSC fault does not affect supply voltage asymmetry, it would not alter the negative sequence voltage. Therefore, the magnitude of negative sequence voltage remains “zero” in this case. 3.2. Study of phase shift and symmetrical component characteristics under unbalanced supply voltage

Supply voltage

2

3

2.55

10

Fig. 5 shows the behavior of three-phase shift under unbalanced supply voltage affecting phase A, B or C source voltage magnitude with %1–5 of the rated phase. Furthermore, for occurrence of unbalanced magnitude on phase A of supply voltage, the phase shifts' ascending order is: (Phib-Phia-Phic). The magnitude of negative sequence supply voltage linearly depends on the level of supply voltage imbalance, as shown in Fig. 6. Thus, this feature can be used in order to distinguish as to whether the change in the three-phase shift is the result of ITSC fault or unbalanced supply voltage.

(b)

5 0

4. Design of the Neural Network

-5 -10 2.5

2.505

2.51

2.515

2.52

2.525

2.53

2.535

2.54

2.545

2.55

2.505

2.51

2.515

2.52

2.525 time (s)

2.53

2.535

2.54

2.545

2.55

4

(c)

2 0 -2 -4 2.5

Fig. 2. Simulated stator currents (a) Healthy condition, (b) under 20% ITSC fault on phase A, (c) under 5% unbalanced supply voltage.

This study used feedforward multi-layer perceptron (MLP) Neural Network trained by Levenberg–Marquardt back propagation (BP) algorithm in order to automatically detect and locate ITSC and voltage fault. The input number of NN is fixed by the number of the fault indicators, which are Phia, Phib, Phic and |Vn|. The output number of NN is four, which are corresponding to the three phases of the induction motor where the ITSC fault may occur, as well as unbalanced supply voltage occurrence. The used NN consists of only one hidden layer including 5 neurons (see Fig. 3). In fact, the number of neurons in the hidden layer is selected by trial and error. Neural Network

Induction Motor

Vs ///

Phase Shifts

/// FFT

/// is

///

Magnitude and phase angle

Symmetrical Components

Phia

O1

ITSC fault on as

Phib

O2

ITSC fault on bs

Phic

O3

ITSC fault on cs

|Vn|

O4

Unbalanced voltage fault

Fig. 3. Block diagram of the proposed method.

Please cite this article as: Lashkari N, et al. Simulative and experimental investigation on stator winding turn and unbalanced supply voltage fault diagnosis in induction motors using Artificial Neural.... ISA Transactions (2015), http://dx.doi.org/10.1016/j.isatra.2015.08.001i

4

N. Lashkari et al. / ISA Transactions ∎ (∎∎∎∎) ∎∎∎–∎∎∎

Fig. 4. Phase shift characteristic for ITSC fault on (a) phase A, (b) phase B, (c) phase C.

Fig. 5. Phase shift characteristic for unbalance on (a) phase A, (b) phase B, (c) phase C of supply voltage.

In order to find the optimal number of hidden neurons, we start with a few neurons and train the NN; then we add other ones until we reach a number of hidden neurons that provides the lowest

mean square error. Also the activation functions of the neurons in the hidden layer are “tansig” and in the output layer are “logsig”.

Please cite this article as: Lashkari N, et al. Simulative and experimental investigation on stator winding turn and unbalanced supply voltage fault diagnosis in induction motors using Artificial Neural.... ISA Transactions (2015), http://dx.doi.org/10.1016/j.isatra.2015.08.001i

N. Lashkari et al. / ISA Transactions ∎ (∎∎∎∎) ∎∎∎–∎∎∎

5

The training procedure of NN is performed offline. In order to achieve a NN which is efficient in detecting ITSC and unbalanced voltage fault, the designed network should have been more generalized using training data that cover the complete range of the operating conditions, including all possible fault occurrences and healthy operation with various load torques. To pursue this purpose, a training input data set consists of different operating cases of the induction motor: “healthy (one point), fault of an odd

number of shorted turns (n ¼1, 3, 5, 7, 9, 11, 13, 15) on each stator phase (24 points) and fault of voltage unbalance (1%, 2%, 3%, 4%, 5%) affecting each phase of supply voltage (15 points)” is composed. The training data set have been constructed by simulating faulty model of induction motor in MATLAB environment (Fig. 7). Having structure of the network constructed, learning parameter should be chosen by trial and error, when using Levenberg–Marquardt backpropagation algorithm. In order to choose learning parameter, numbers between 0.05 and 0.9 in increments of 0.05 had been tried and the best performance achieved with 0.15 for training parameter. Thus, after four thousands epochs, the NN reaches a low mean square error (mse). Although the consumed time for training procedure is very short, online identification of faults would not be affected by the time it takes for iterations since the training procedure is offline. The performance function which is mse between the NN's output and the target output is shown in Fig. 8. Also the consumed time for training procedure is very short. Fig. 9 shows the desired training outputs and NN's training errors for ITSC fault on phase A, B and C as well as voltage unbalance fault. The training results are very conclusive and it is clear that the NN has well learned the training data and therefore reproduced the desired outputs with very low training errors.

Fig. 6. Evolution of magnitude of negative sequence voltage as function of the percent of voltage unbalance on phase A or B or C.

Fig. 8. Training performance of the NN.

It is worth mentioning that the NN's desired outputs which indicate the state of each phase as well as the condition of supply voltage are as follows: [0;0;0;0] [1;0;0;0] [0;1;0;0] [0;0;1;0] [0;0;0;1]

Healthy operation; ITSC fault occurred on phase A of stator; ITSC fault occurred on phase B of stator; ITSC fault occurred on phase C of stator; voltage unbalance fault occurred.

4.1. Training results

Fig. 7. Simulated training input data base of the NN.

Please cite this article as: Lashkari N, et al. Simulative and experimental investigation on stator winding turn and unbalanced supply voltage fault diagnosis in induction motors using Artificial Neural.... ISA Transactions (2015), http://dx.doi.org/10.1016/j.isatra.2015.08.001i

N. Lashkari et al. / ISA Transactions ∎ (∎∎∎∎) ∎∎∎–∎∎∎

6

4.2. Test results The even numbers of shorted turns (n¼2, 4, 6, 8, 10, 12 and 14) on each stator phase as well as 1.5%, 2.5%, 3.5% and 4.5% unbalanced voltage fault affecting each phase of supply voltage were reserved for test data set. Using the aforementioned data set for test procedure, the generalization capacity of the trained NN would be assessed. The results of test set are shown in Fig. 10. According to the test results, it can be concluded that the NN based method is able detect ITSC fault on each of the three phases as well as supply voltage unbalance. 5. Experimental results In order to validate the proposed fault diagnosis scheme, several experiments were conducted for diagnosis of ITSC in

the stator windings of motor and unbalanced supply voltage fault. The three-phase, 50 Hz, 4-pole, 3 kW, squirrel cage induction motor used in laboratory tests has the same parameters as those of the simulated motor (Table 2). The experimental test bed is shown in Fig. 11. The stator current and voltage signals are sampled at a rate of 10 kHz using a PCI-1711 data acquisition card. Given that the Neural Network has successfully learned the behavior of motor under ITSC and voltage unbalance fault, and that the simulated motor and laboratory motor have identical features and behave the same way in case of a fault, the Neural Network trained earlier using simulated data can be utilized to detect faults of experimental motor. The following sections discuss the results of the proposed fault diagnosis system under healthy operation as well as in presence of ITSC fault and supply voltage unbalance.

Fig. 9. Training outputs (left) and training errors (right) of the NN.

Fig. 10. Test outputs (left) and test errors (right) of the NN.

Please cite this article as: Lashkari N, et al. Simulative and experimental investigation on stator winding turn and unbalanced supply voltage fault diagnosis in induction motors using Artificial Neural.... ISA Transactions (2015), http://dx.doi.org/10.1016/j.isatra.2015.08.001i

N. Lashkari et al. / ISA Transactions ∎ (∎∎∎∎) ∎∎∎–∎∎∎

5.1. Experimental results with healthy machine Experimental line currents and phase voltages under healthy operation is shown in Fig. 12. The current and voltage signals are utilized to compute the fault indicators (Fig. 13). It is clearly shown in Fig. 13 that in term of healthy operation, the three-phase shifts are equal and the magnitude of negative sequence component is approximately zero. Then, the fault indicators are applied to the NN. The output signal of the fault detector system is shown in

7

Fig. 14. The result of experimental test on healthy machine supplied by a balanced voltage demonstrates that the NN is capable of detecting healthy operation. 5.2. Experimental results in presence of unbalanced supply voltage Fig. 15 shows the stator current and voltage signals of the aforementioned machine working under unbalanced voltage 1.5

ANN Response

1 0.5 No fault 0 -0.5

ITSC fault on A ITSC fault on B ITSC fault on C voltage fault

-1 -1.5 0.35 Fig. 11. The experimental setup.

0.4

0.45 time [s]

0.5

0.55

Fig. 14. Experimental output result of healthy machine.

Fig. 12. Experimental line currents and phase voltages under no fault.

Fig. 15. Experimental line currents and phase voltages under unbalanced supply voltage fault.

Fig. 13. Fault indicators of healthy machine.

Please cite this article as: Lashkari N, et al. Simulative and experimental investigation on stator winding turn and unbalanced supply voltage fault diagnosis in induction motors using Artificial Neural.... ISA Transactions (2015), http://dx.doi.org/10.1016/j.isatra.2015.08.001i

N. Lashkari et al. / ISA Transactions ∎ (∎∎∎∎) ∎∎∎–∎∎∎

8

Fig. 16. Fault indicators of machine under unbalanced supply voltage fault.

1.5 unbalanced supply voltage fault

ANN Response

1 0.5 0 -0.5

ITSC fault on A ITSC fault on B ITSC fault on C voltage fault

-1 -1.5 0.35

0.4

0.45 time [s]

0.5

0.55

Fig. 17. Experimental output result under unbalanced supply voltage fault.

5.3. Experimental results with faulty machine Stator currents and voltages in presence of ITSC faults on phase B of stator is shown in Fig. 18. The corresponding fault indicators are also demonstrated in Fig. 19. As it can be observed in Fig. 20, the scheme is able to detect the presence of the ITSC fault in addition to locate the phase in which ITSC fault has occurred even for lower percentage of faults.

6. Conclusion

Fig. 18. Experimental line currents and phase voltages under 13 (5%) shorted turns in phase B.

supply. There are obvious asymmetries in Fig. 15, in comparison to Fig. 12. The fault indicators and response of NN is also shown in Figs. 16 and 17 respectively. As shown in Fig. 17, the forth output of the NN would change into “one” by the time unbalanced supply voltage occurs in the system. So this proves the capability of the designed NN to detect the operation condition when line voltages applied to the motor are not exactly the same .

In this paper, a major drawback of the existing fault diagnosis technique, based on three-phase shifts between the line current and phase voltage, to detect and locate short circuit on the stator windings in induction motors has been investigated. Confining the diagnosis of stator fault to utilizing only the three-phase shifts may cause serious ambiguity, since the three-phase shifts are found to be sensitive to the supply voltage unbalance. This study described an Artificial Neural Networks based scheme for diagnosis and discrimination between the effects of stator interturn fault and those due to unbalanced supply voltages in three-phase induction motors. The NN fault diagnosis is based on monitoring the magnitude of negative sequence voltage and the three-phase shifts between the line current and phase voltage. The proposed NN is a simple MLP feedforward network having one hidden layer composed of five neurons, with a very short training time. Various parameters used in the backpropagation training algorithm are also presented. The data collected from the simulated faulty model of stator were used for off-line training and test of proposed NN. The training and test data set are

Please cite this article as: Lashkari N, et al. Simulative and experimental investigation on stator winding turn and unbalanced supply voltage fault diagnosis in induction motors using Artificial Neural.... ISA Transactions (2015), http://dx.doi.org/10.1016/j.isatra.2015.08.001i

N. Lashkari et al. / ISA Transactions ∎ (∎∎∎∎) ∎∎∎–∎∎∎

9

Fig. 19. Fault indicators of machine under ITSC fault.

1.5 ITSC fault on B

ANN Response

1 0.5 0 -0.5 ITSC fault on A ITSC fault on B ITSC fault on C voltage fault

-1 -1.5 0.4

0.45

0.5 time [s]

0.55

0.6

Fig. 20. Experimental output result under ITSC fault.

collected from different fault possibilities. The performance of the NN is found to be accurate for the fault diagnosis field. The main merit of the proposed fault diagnosis system is the fact that an induction motor model during the fault diagnosis process is not needed anymore. In the proposed method, no additional sensors are required to detect supply unbalance and stator faults since the terminal voltage and stator currents data would be enough. The validation of the proposed scheme experimentally confirms the accuracy and efficiency of the method. There are multiple causes of voltage unbalance commonly existing at the motor stator terminals. In future work, it will be interesting to identify the root causes of voltage unbalance during the diagnosis procedure. References [1] Tag Eldin EM, Emara HR, Aboul-Zahab EM, Refaat SS. Monitoring and diagnosis of external faults in three phase induction motors using artificial neural network. In: Proceedings of the IEEE power engineering society general meeting; 2007. p. 1–7. [2] Refaat SS, Abu-Rub H, Saad MS, Aboul-Zahab EM, Iqbal A. Detection, diagnoses and discrimination of stator turn to turn fault and unbalanced supply voltage fault for three phase induction motors. In: Proceedings of the IEEE international conference on power and energy (PECon); 2012. p. 910–5. [3] Mo-Yuen C, Yee SO. Methodology for on-line incipient fault detection in single-phase squirrel-cage induction motors using artificial neural networks. IEEE Trans Energy Convers 1991;6:536–45.

[4] Benbouzid MEH, Nejjari H. A simple fuzzy logic approach for induction motors stator condition monitoring.In: Proceedings of the IEEE international conference on electric machines and drives; 2001. p. 634–9. [5] Siddique A, Yadava GS, Singh B. A review of stator fault monitoring techniques of induction motors. IEEE Trans Energy Convers 2005;20:106–14. [6] Jee-Hoon J, Lee J-J, Bong-Hwan K. Online Diagnosis of Induction Motors Using MCSA. IEEE Trans Ind Electron 2006;53:1842–52. [7] Ahmed SM, Texas A, Qatar MUa, Abu-Rub H, Texas A, Qatar MUa, et al. Diagnosis of stator turntoturn fault and stator voltage unbalance fault using ANFIS. Int J Electr Comput Eng 2013;3:129–35. [8] Williams JE. Operation of 3-phase induction motors on unbalanced voltages [includes discussion]. Power Appar Syst, Part III. Trans Am Inst Electr Eng 1954;73. [9] Mo-Yuen C, Sharpe RN, Hung JC. On the application and design of artificial neural networks for motor fault detection. IEEE Trans Ind Electron 1993;40:181–8. [10] Kolla SR, Altman SD. Artificial neural network based fault identification scheme implementation for a three-phase induction motor. ISA Trans 2007;46:261–6. [11] Yang B-S, Kim KJ. Application of Dempster–Shafer theory in fault diagnosis of induction motors using vibration and current signals. Mech Syst Signal Process 2006;20:403–20. [12] Filippetti F, Franceschini G, Tassoni C, Vas P. Recent developments of induction motor drives fault diagnosis using AI techniques. IEEE Trans Ind Electron 2000;47:994–1004. [13] Bouzid M, Champenois G, Bellaaj NM, Signac L, Jelassi K. An effective neural approach for the automatic location of stator interturn faults in induction motor. IEEE Trans Ind Electron 2008;55:4277–89. [14] Arkan M, Perović DK, Unsworth P. Online stator fault diagnosis in induction motors. Electric Power Applications, IEE Proceedings 2001;148(6) IEE. [15] Ballal MS, Suryawanshi HM, Mishra MK. Stator winding inter-turn insulation fault detection in induction motors by symmetrical components method. Electr Power Compon Syst 2008;36:741–53. [16] Kowalski CT, Wolkiewicz M. Stator faults diagnosis of the converter-fed induction motor using symmetrical components and neural networks. In: Proceedings of the IEEE 13th european conference on power electronics and applications; 2009. p. 1–6. [17] Kliman G, Premerlani W, Koegl R, Hoeweler D. A new approach to on-line turn fault detection in AC motors. In: Proceedings of the 31st IAS annual meetingindustry applications conference. Conference record of the 1996 IEEE; 1996. p. 687–93. [18] Bouzid M, Champenois G. Accurate stator fault detection insensitive to the unbalanced voltage in induction motor. In: Proceedings of the XXth international conference onelectrical machines (ICEM); 2012. p. 1545–51. [19] Isermann R. Model-based fault-detection and diagnosis–status and applications. Ann Rev Control 2005;29:71–85. [20] Bouzid M, Champenois G. An efficient, simplified multiple-coupled circuit model of the induction motor aimed to simulate different types of stator faults. Math Comput Simul 2013;90:98–115. [21] Krause Paul C. Analysis of electric machinery and drive systems, 75. John Wiley & Sons; 2013. [22] Azgomi HF, Poshtan J, Poshtan M. Experimental validation on stator fault detection via fuzzy logic. In: Proceedings of the 3rd international conference on electric power and energy conversion systems (EPECS); 2013. p. 1–6.

Please cite this article as: Lashkari N, et al. Simulative and experimental investigation on stator winding turn and unbalanced supply voltage fault diagnosis in induction motors using Artificial Neural.... ISA Transactions (2015), http://dx.doi.org/10.1016/j.isatra.2015.08.001i