Demagnetization fault diagnosis in permanent magnet synchronous motors: A review of the state-of-the-art

Journal of Magnetism and Magnetic Materials 391 (2015) 203–212

Contents lists available at ScienceDirect

Journal of Magnetism and Magnetic Materials journal homepage: www.elsevier.com/locate/jmmm

Demagnetization fault diagnosis in permanent magnet synchronous motors: A review of the state-of-the-art S.S. Moosavi a,d,n, A. Djerdir a, Y.Ait. Amirat b, D.A. Khaburi c a

University of Technology Belfort Montbeliard (UTBM), Laboratory of IRTES-SET, Belfort, France Laboratory of Femto-ST, University of Franche-Comte, France c Center of Excellence for Power System Automation and Operation, Iran University of Science and Technology (IUST), Tehran, Iran d Engineering Department, Amol University of Special Modern Technology, Amol, Iran b

art ic l e i nf o

a b s t r a c t

Article history: Received 1 July 2014 Received in revised form 22 January 2015 Accepted 15 April 2015 Available online 27 April 2015

There are a lot of research activities on developing techniques to detect permanent magnet (PM) demagnetization faults (DF). These faults decrease the performance, the reliability and the efficiency of permanent magnet synchronous motor (PMSM) drive systems. In this work, we draw a broad perspective on the status of these studies. The advantages, disadvantages of each method, a deeper view investigated and a comprehensive list of references are reported. & 2015 Elsevier B.V. All rights reserved.

Keywords: Permanent magnet machines (PMSM) Demagnetization Fault diagnosis Fast Fourier Transforms (FFT) Time–frequency transform Analytical models

1. Introduction The positive specific characteristics of permanent magnet motors make them highly attractive candidates for several classes of drive applications, such as: servo-drives containing motors with a low to mid power range, robotic applications, motion control system, aerospace actuators and specially in air, sea and land transportation [1–3]. Some of the most common advantages of permanent magnet synchronous motors (PMSMs) over other electric motors available on the market are: high dynamic response performance, high efficiency, long lifetime, low acoustic noise, high power factor, high power to weight ratio, high torque to inertia and volume ratio, high flux density and high speed ranges [1–17]. Permanent magnet (PM) motors also have some inherent disadvantages just like any other electrical machine. Some of them are included in the following [14,18–29]: 1. Magnet cost: rare-earth magnets such as samarium-cobalt and neodymium boron iron are especially costly. 2. Very large opposing magneto motive forces (MMF) and high temperature can demagnetize the magnets. n

Corresponding author. Fax: þ 33(0)384582032. E-mail address: [email protected] (S.S. Moosavi).

http://dx.doi.org/10.1016/j.jmmm.2015.04.062 0304-8853/& 2015 Elsevier B.V. All rights reserved.

3. For surface-mounted permanent magnet (SPM/SMPM) motors, high speed operation is limited or not possible because of the mechanical construction of the rotor. 4. There is a limitation in the range of the constant power region, especially for SMPM motors. 5. Because there is a constant energy on the rotor due to the permanent magnets, motors present a major risk in the case of short-circuit failures in the inverter. 6. The interior permanent magnet motor (IPM) generates a high mechanical vibration and the noise by electromagnetic vibration sources such as variation of radial force, cogging torque and commutation torque ripple compared to a SMPM. In several applications such as electric vehicles, the levels of operating temperature and the MMF from the stator winding are severely higher than those of the conventional ones. The effect on demagnetization of the permanent magnet must be considered as a main design and control parameter. Because, when a partial demagnetization takes place, the same load torque is generated by a higher current than the one rated in the safety case. Consequently, the thermal level of the operating point is more increased due to the Joule effect [8,30]. Progress and usage of new signal processing methods and also some analytical and modeling methods lead to new and useful application methods for fault diagnosis, especially for demagnetization fault diagnosis (DFD) region.

204

S.S. Moosavi et al. / Journal of Magnetism and Magnetic Materials 391 (2015) 203–212

Several papers about demagnetization faults (DF) and its diagnosis methods have been presented so far. Different methods of fault diagnosis (FD) are continuing to be expanded and used more effectively for electrical fault detection in the initial steps of its occurrence by measuring different values such as current, voltage, temperature, vibration, magnetic flux, torque, speed, etc. Nonetheless, choosing an appropriate method for fast, on time and suitable fault diagnosis has become the main concern of every user to achieve the desired goal. Also the difference between various proposed methods in terms of advantages, disadvantages and its optimization in regard of their application are considered as a difficult and determinative cases to achieve the main goal. Many papers deal with the problem of fault diagnosis on electrical machine [14,31–36]. Most of them are concerned with the induction motors. In this work, we focus on the last presented method in the recent years by looking back to advantages and disadvantages when applied to PMSMs. Also critical discussion and comparisons of different references have been presented. So, the readers can choose the right method according to each situation. Most of the investigated references address particularly the demagnetization fault.

2. Specification of demagnetization fault Faults in PMSMs are classified into three parts: electrical such as stator windings short circuits, magnetic such as demagnetization, and mechanical faults such as rotor eccentricities and bearing damages [37–39]. Magnet faults include microscopic fissures, chips, disintegration due to corrosion, complete or even partial demagnetization. Among these, demagnetization faults hold an important place in magnet failure [40–42]. Demagnetization can be complete, that is, all over the pole, or partial, on a certain region of the pole [12,13,39]. Depending on the severity of fault, demagnetization can be reversible or irreversible [17,39,43]. However it has been verified that irreversible demagnetization does not arise in the PMs under the steady states. Instead, it arises under transient states [44]. The demagnetization phenomenon is due to armature reaction, especially under conditions of operation requiring strong torque, for example, at high loads, during sharp transients or even at high temperature. Such demagnetization limit is considered to be depending on the operating temperature and the machine size. Furthermore, the comparison between the continuous load and demagnetization conditions shows that low and medium size machines can be stiffer against demagnetization, in comparison with larger machines, and have ability for transient overload. Nevertheless, the leakage permeance and the peak MMF are not much influenced by the magnet thickness: Thin magnets result in a bit more leakage field to the rotor yoke and in consequence, a somewhat higher demagnetization risk in thin magnets. For the considered combinations of number of poles and number of stator slots, the combinations with low numbers of poles and slots seem to be a bit more sensitive to demagnetization, and the risk is not much dependent on the magnet thickness [13,17,46-49,76]. In high performance applications, the rotor magnets are usually made of sintered rare earth materials such as samarium-cobalt (SmCo) and neodymium–boron iron (NdFeB). Such materials are easy to crack, brittle and easy to erode owing to high humidity or dew. During its installation, the permanent magnets are exposed to mechanical pressure which may cause small cracks that can lead to disintegration at high speed [6]. In addition, metallurgical changes in the magnet material, at high temperatures and/or due to corrosion/oxidation, can result in irreversible demagnetization

fault too. A direct impact on the motor may also damage the magnets, leading to partial demagnetization. Additionally, under certain circumstances, the magnets may be exposed to different types of contaminants, including dust pollution, salt and cooling lubricants and aging of magnet among others, which also may lead to disintegration [12,17,48]. Normally, the thickness of the magnets is designed to tolerate the current due to the maximum rated torque or to the short circuit torque according to the following equation [10]:

⎛ ⎞ 3 N 2I g 1 lm⟩ ⎜⎜ + Br ⎟⎟ −g 2 ⎝ 2p μ 0 ⎠ Hci

(1)

where lm is the magnet thickness, N is the number of the conductors in series per phase, p is the pole pairs number, Br is the residual flux density and Hci is the intrinsic coercive force of the magnetic material employed, g is the air-gap, and I is the RMS of the maximum current among the maximum torque current and the short circuit current. However even if the magnet thickness is well designed, the MMF due to the high current in the stator can lead to a demagnetization on magnet trailing edges when the rotor is overheated [49]. The Influence of the temperature on the magnetic remanence is approximately linear below the Curie temperature [12] expressed in the following equation:

Br (T ) = Br (T0 )[1 + ΔB (T − T0 )]

(2)

where T is magnet's operation temperature, T0 is the preferred temperature, Br(T0) is the remanence at the temperature T0, and ΔB is the reversible temperature coefficient, which is a negative number. The moving of operating point due to the increasing temperature is illustrated in Fig. 1. Magnet's permeance coefficient Pc is a function of magnet length, air gap length and armature current. It is usually greater than one to keep the operation point far away from the knee point because operation around the knee area will cause irreversible demagnetization too. However, temperature change along with demagnetization fault lead to displacement of operation point. If the failure causes the operating point to “fall off" the lower end of recoil line, there will be an irreversible flux loss [4,11,15]. According to the demagnetization characteristics, permanent magnets can be divided into several groups, the three main ones are as follows: – Alinco (Alinco5, Alinco5-7, Alinco9, etc.) – Ferrites (barium ferrite, strontium ferrite, etc.) – Rare earths (Sarrium cobalt (SmCo), neodymium–iron–boron (Nd–Fe–B))

Fig. 1. Effect of increasing temperature on the operating point [11,15].

S.S. Moosavi et al. / Journal of Magnetism and Magnetic Materials 391 (2015) 203–212

Fig. 2. Demagnetization curve of magnetic materials [49].

The demagnetization curves of these materials are presented in Fig. 2. It is clear from this figure that the Alinco have the highest residual induction, a very small coercive field and a nonlinear curve. The Alinco can be magnetized and demagnetized easily. They were widely used in DC motors with permanent magnets until the ferrite magnets have become commercially available. Ferrites are better than the Alinco, in terms of the coercive field but their residual induction is lower. The ferrite permanent magnets are the cheapest on the market. They are commonly used for PM machines with low power. Rare earth materials like samariumcobalt (SmCo) and neodymium iron boron (Nd–Fe–B) have almost a linear demagnetization curve. They have a remanent flux density and a high coercivity. However, the cost of SmCo is much higher than other PMs. Although the cost of Nd–Fe–B is higher than that of ferrites, it is more appropriate for machines with high performance, because of superior magnetic properties. The disadvantage of Nd–Fe–B lies in their low corrosion resistance [50]. The approximate maximum operation temperature of some commercial magnets is illustrated in Fig. 3 [12]. A major fault condition of permanent-magnet machines is the stator fault in the opening or shorting of one or more of stator phase windings [5,18,51,52]. Normally the synchronous machines should be proof against the sudden short-circuit current after the rated load condition. It should be considered in inter turn short circuit faults when a small number of turns are short-circuited, the additional flux is not enough to demagnetize the PM. In opposite, when a large number of turns are involved in a stator turn fault, the additional flux resulting from the turn fault can demagnetize the PMs, and consequently, can result in an irreversible damage to the motor too [18]. Two-phase short circuit is more common and more dangerous from the demagnetization of permanent magnets point of view than three-phase and one-phase short circuits. Such condition will drastically increase the demagnetizing effect of the

Fig. 3. Maximum operating temperature of different magnet materials.

205

external magnetic field, enlarging the magnetic loop response of the permanent magnets. These effects can bring the permanent operating point below the knee region on the magnetic characteristic, i.e., out of linear region. Linear assumption of magnet behavior seems to be very simplified regarding the real material behavior in the electric machine operation [1]. In such applications, the permanent magnets have the undesirable quality of potential demagnetization. The demagnetization states associated with the macroscopic behavior of permanent magnets are strongly dependent on the external magnetic field. It is necessary to consider both: linear and nonlinear behavior of magnet if someone wants to use a finite element method (FEM) or an analytical analysis method for the simulation of the demagnetization fault. A suitable model can be obtained for accurate representation of the reversible and irreversible magnetic processes [1,39]. In addition to be irreversible, the flux losses from demagnetization fault which worsen the motor efficiency and can also lead to the magnetic force harmonics that can provoke noises, vibrations, and increase the copper losses. In addition, because of the power factor changes, the air-gap magnetic flux distributions deteriorate will lead to decrease of the electromagnetic torque in the machine [3,5,11,14,18,39,43,45,53-55].

3. Diagnosis methods, investigation and comparison he history of fault diagnosis, state supervision and protection is as old as electrical appliances. In general, monitoring and diagnosis require the detection and analysis of signals that contain specific information (symptoms or signatures) in order to characterize some degradation of the machine [56]. Condition monitoring and fault diagnosis of PMSMs problems are essential to guarantee its availability with high motor performance, efficiency, and reliability [2]. In order to carry out a fault diagnosis scheme, it is highly desirable to use an easy-to-calculate fault severity index with low computational burden [6]. So far, several reviews have been presented on different methods of fault detection and diagnosis, especially, on induction motors. A summary of these works is presented in Table 1 [14,31,33–36]. According to the purpose of this work, new methods are developed so far and are studied in the continuation. Demagnetization fault investigation can be divided into two main parts. The first one is a software review on the simulation (analytical or finite element) based methods. The second one is a hardware review based on the magnetic design changes and practical design as mentioned in Fig. 4. Details of each part have been explained futher. 3.1. Software review Today, finite element method (FEM) becomes a powerful and reliable tool for comparison between theoretical methods and the experimental results. There are different methods to make data analysis for fault diagnosis. In this study, we consider four types of the most popular methods to complete this task. 3.1.1. Frequency domain analysis Any observed signals, as acquired from measuring sensors in raw form, are in the time-domain. Historically, Fourier spectral analysis has provided a general method for examining the global energy-frequency distributions and has been used for damage detection. In [54] the results obtained are based on analyzing the harmonics of stator and zero sequence currents. These harmonics are obtained from the Fast Fourier Transform (FFT). The fault frequencies due to the rotor demagnetization are localized as in

206

S.S. Moosavi et al. / Journal of Magnetism and Magnetic Materials 391 (2015) 203–212

Table 1 A synthesis of different parameters measurement and measurement devices aiming fault detection. Parameter

Measurement device

Potential information richness

Intrusive to electrical machines

On/ off line

Operator skill required

Measurement frequency

Measured as part of control strategy

References

Current Voltage Flux

Hall effect transducer DVM Hall effect device Search coil Dynamometer Accelerometer Microphone Hand-held probe Thermal paint Thermocouple Infrared camera

Average Average Very high

No No Yes and no

On On On

High High High

Continuous Continuous Hourly

Yes Yes No

[32,36,37,54] [32,37] [13,32,36,37,56]

Very high High High Low

On On On Off

High Expert Expert Low

[32,37] [32,37] [32,37] [32,37]

On

Average Expert

Encoder

Average

No

On

High

Continuous Hourly Hourly Monthly or on suspected deterioration Continuous Monthly or on suspected deterioration Continuous

No No No No

Average High

No Yes and no no Yes and no Yes Yes No

Yes

[37]

No

On

Expert

Continuous

Yes and no

[32,36,72]

Force Vibration Acoustics Temperature

Instantaneous angular speed Torque

Torque (magneto elas- High tic, piezoelectric, strain gauge)

Yes No

– relatively time consuming to some certain extent; – it is too hard to determine the source of specific harmonics since we cannot discriminate faults that have the same signature frequencies, like partial demagnetization, dynamic eccentricity (DE) and unbalanced load (UL) [33,62].

Fig. 4. The scope of demagnetization fault investigation.

[2,4,6,15,17,39,57–59]

⎛ k⎞ ff = fe ⎜1 ± ⎟ p⎠ ⎝

(3)

where ff is the fault frequency, fe is the electrical fundamental frequency, k is an integer, and p is the number of pole pairs. The results clearly show that odd harmonics (1/3, 5/3 among others) appear when dealing with a 3-pole-pair PMSM with a partially demagnetized pair of poles. It is very essential to note that in some PMSMs, by using the specific stator windings configuration like the sample one mentioned in [6], the fault harmonics predicted by Eq. (3) are not present in the current spectrum. Since it is the asymmetry in the flux pattern that induces the current component in Eq. (3), uniform demagnetization cannot be detected with motor current signature analysis (MCSA). Uniform demagnetization is a very common type of PM defect that can occur when the motor operates at high temperature. Another critical limitation of the MCSA is the case of load torque oscillating (e.g., reciprocating compressors) which can produce sidebands at the same frequency component as Eq. (3). Its magnitude is larger than the one due to the demagnetization fault and there is no practical means to separate them [2,58]. Similar to current frequencies, the indications of several faults are also found in other spectrums, such as noises, vibrations and torque [60]. However, due to the high cost of the sensors (accelerometer or torque-meter), they are implemented in relatively larger machines. Notice that there are some limitations of these frequency analysis based algorithms such as:

Authors in [10] have proposed a new non-invasive method for magnet faults detection that includes local and uniform demagnetization by means of a Fourier transform of the space vector of back-EMF. The proposed approach is then validated for three kind of permanent magnet synchronous motors with different winding configurations (as mentioned before, different winding configurations may lead to canceling of some harmonic ranges used for fault detection, see [3,6,64–66]). The results show how the chains of frequency are univocally related to the considered winding configuration for a local detection on the rotor. That makes the fault easier to detect. On the other hand, for uniform demagnetization, only the variation of harmonics (in respect to the fundamental one) which are already present in the spectrum of the back-EMF can be used in signature analysis approach. Also, by surveying of article [40], this reality is unveiled that between two different stator winding configurations (series and parallel connected windings), and by fault detection strategy based on the information contained in winding-current spectrum and electromagnetic torque, current harmonic analysis has not any valuable effect for fault detection in series winding connection. It should be taken into account that the different ranges of load are effective on reduction of frequency component amplitudes that make the fault detection difficult. Refs. [3,6,64,66] present a zero-sequence voltage component (ZSVC) measure based method. It is mentioned that if the back-EMF in a single pole of a phase winding contains a fractional harmonic, this does not appear when considering an entire phase winding. This means that for the analyzed SPMSM, it is not feasible to detect this particular demagnetization fault by analyzing the fractional harmonics in the back-EMF voltage or in the stator currents spectrum. An accessible neutral point and three external resistances for each phase are needed and for critical systems, this disadvantage may be persecutor. The ZSVC is based on the difference between fundamental harmonic amplitude of the measured voltage between resistance and star point of motor against the first harmonic amplitude of the current in a healthy and a faulty SPMSM.

S.S. Moosavi et al. / Journal of Magnetism and Magnetic Materials 391 (2015) 203–212

Although FFT allows detecting demagnetization faults by analyzing the amplitude of harmonics in stationary signals (compliance with limitations expressed), it fails when dealing with non-stationary signals. Thus, when dealing with signals that contain changes of speed and torque, this tool is not well-suited for detecting the demagnetization fault in PMSM [4,11,12,15,61]. 3.1.2. Time–frequency (TF) domain analysis Frequently, the motor operates in a non-stationary environment and the motor's state variables such as currents, torque, and speed change over time. When applying FFT, information about time is lost. Therefore, it is not convenient to deal with FFT when the motor is operating under non-stationary conditions, under continuous changes of speed and torque. For these reasons, processing tools able to feature extraction are used, based on the TF transform. Notice that TF can show what frequencies (notes) are present in a signal, just like Fourier transform does, and can also show when they appear in the signal and how long they last. That is what “time localization of frequency” implies. As the conventional representations in the time domain or frequency domain are inadequate in these situations where the time localization of frequencies is required, an obvious solution is to seek a representation of the signal as a two-variable function (or distribution) whose domain is the two-dimensional (time, frequency) space. So far, more than 20 different types of TF methods (such as: Gabor expansion coefficient function, Cohen's Nonnegative Distribution, Active Unterberger Distribution (AUD), Rihaczek Distribution (RD), Bertrand Distribution (BD), Born–Jordan Distribution (BJD), Generalized Wigner Distribution (GWD), Levin Distribution (LD), etc.) were explored in three major groups as: linear, quadratic and non-linear/non-quadratic. Only in five of these methods demagnetization fault detection has been taken into account: Wavelet Transform (WT), Wavelet Packet Transform (WPT), Short-Time Fourier Transform (STFT), Wigner–Ville Distribution (WVD), Hilbert–Huang Transform (HHT) and Choi–Williams Distribution (CWD) have been taken into account for studying the demagnetization fault detection [4,11,15,58,67]. These methods are classified in Fig. 5 with regard to the linearity [68,69]. In [4,15], time–frequency wavelet-based methods have been successfully applied to detect demagnetization faults in PMSM motors under non-stationary conditions. The effect of demagnetization faults that are reflected on the stator current spectrum of a PMSM motor has been analyzed. A wavelet method has been considered in [4]. It associates the original signal with a shortduration wave named the mother wavelet which has a finite length and defined frequency. Wavelets are continuous functions that have zero mean and are in both time and frequency space. WT divide a continuous-time function into wavelets. Thus, from the information collected from the wavelets it will be possible to detect local features in signals. The usage of a variablesized-regions windowing technique is known as main advantage of the WT against Fourier transform (FT) and STFT. Large scales s—

Fig. 5. A classification of TF methods used in demagnetization fault detection and diagnosis.

207

windows—are introduced in order to obtain the slow variations of the signal, i.e., low frequency components, whereas small scales lead to possibility of the detection of signal transients, i.e., highfrequency components. On the one hand, basic functions obtained by shifting and scaling a particular function, uniform resolution, non-stationary data analysis and feature extraction possibility are well-known strength of WT, but on the other hand, leakage generated by the limited length of the basic wavelet function, nonadaptive nature, and high-frequency range observation to define local events can be considered as its weakness. Two different methods based on Continuous Wavelet Transform (CWT) and Discrete Wavelet Transform (DWT) have been presented for demagnetization fault diagnosis (DFD) [4,11]. The uniform resolution for all the scales has been considered as the most appealing feature of the CWT, but due to generation a huge volume of data that are not independent from each other and also produce a high redundancy, CWT results in a non orthogonal decomposition. For this reason, DWT has been employed to cover this weakness of CWT. The severe cross terms as indicated by the existence of negative power for some frequency ranges has been introduced as the difficulty of Wigner–Ville Distribution (WVD). In [15,16,39,70] other types of time frequency methods namely Hilbert–Huang Transform (HHT) and Choi–Williams Distribution which are a part of the Cohen class of distribution are used for demagnetization fault diagnosis. This method can overcome the FFT drawbacks since it allows analyzing the stator current obtained from experimental data for both steady state and dynamic conditions where speed changes at high, medium, and low velocities. The Hilbert– Huang transform (HHT) is based on the instantaneous frequencies resulting from the intrinsic-mode functions (IMFs) of the signal being analyzed. Thus, it is not constrained by the uncertainty limitations with respect to the time and frequency resolutions to which other time–frequency techniques are subjected. The end effects due to spline fitting that leads to inability of Hilbert transform to separate signals with very close frequencies and no physical meaning of some IMFs have been considered as a disadvantage of HHT but also some significant advantages as high time–frequency resolution, generalized Fourier analysis with variable amplitudes and frequencies and etc. led to increasing use of this type of TF method. Considering different properties of the time frequency representation (TFR) introduced in this section, one can realize there is no optimal TFR for every application. Different methods provide different advantages and have their own drawbacks. However, for a wise choice of method among different TFRs one should be familiar with basics of each method [13]. 3.1.3. Analytical model based analysis The analytical model is based on well-known vector potential formulation by Maxwell's equations. A lot of earlier research has already been carried out to model accurately different machine parameters, where the analytical method is a two-dimensional model in polar coordinates and solves the Laplacian/quasi-Poissonian field equations in the air-gap and magnet regions [43,7174]. In [43], a novel analytic model approach to detect magnet faults such as local demagnetization in brushless permanent-magnet motors has been presented. What adds in this work is that with the magnet subdivided into elements, equations are written for each element region and a new form of air-gap vector potential equation has been developed which is a function of remanent induction (Br) of all magnet elements. To reach this important result in [43], a new form of analytical model that solves the Laplacian/quasi-Poissonian field equations in the machine's air-gap and magnet element regions has been developed. The model is

208

S.S. Moosavi et al. / Journal of Magnetism and Magnetic Materials 391 (2015) 203–212

Table 2 Number of required search coils for faults.

Fig. 6. Induced EMF due to fault.

verified by using finite-element software in which the demagnetization faults were simulated and the corresponding electromotive force is calculated as a function of the rotor position. Then the numerical data of electromotive force is introduced into a gradient-based algorithm that uses the analytical model to locate demagnetized regions in the magnet as simulated in the finiteelement package. The induced EMF under the fault condition and the analytic FEM are shown in Fig. 6. Finally, the results show that this method is able to find the faulty magnet and approximate location of the fault. The simulation time is just 15–20 s and the algorithm needs only 11 iterations to converge. Notice that the analytic model can be applied to dynamic systems for diagnostic purposes. 3.1.4. Modeling method design based analysis The modeling design-based method analysis is divided to four main subsections as shown in Fig. 7. 3.1.4.1. Field reconstruction method (FRM). In [5,7,63], because the conventional finite element analysis (FEA) is time consuming, a field reconstruction method has been used to a partial demagnetization fault in PMSM. In this method a comparison of average torque, torque ripple and emitted noise for both healthy and faulty conditions are presented. FRM is known as a new method that improves the computational time used to determine magnetic field distribution within the electromechanical energy converters. Considering the accuracy of the FRM, the model created in these investigations can substitute the FEA to calculate field and magnetic force (torque) quantities. The difference in computational time is significant. The FRM model can run three times quicker than FEA with the same level of precision. 3.1.4.2. Search coil design by help of FEM and field component analysis. In [13], an alternative multi-faults detection method using search coils is proposed. These invasive coils are wound around armature teeth, so they typically need to be installed during manufacturing. But its immunity to high frequency harmonics

Fault case

Number of search coils required

Eccentricity Demagnetization Phase failure Inter-turn fault

3 Number of poles Number of phases Number of solenoids

makes it suitable for inverter/rectifier fed motors or generators, such as wind turbines and automotive systems. In addition, this method does not require the knowledge of machine parameters. Since the air gap flux is directly measured in this method, it provides much more diagnosis reliability. The number of required search coils for different fault cases is given in Table 2. To analyze all the given four kinds of fault cases, maximum number of search coils, 12, is chosen. The drawback of this method is that it is invasive, so it might not be very economical for the machines that have already been manufactured, but holds potential for emerging applications. 3.1.4.3. Inverter-embedded technique. In [2,17], an inverter-embedded technique for automated detection and classification of PMSM rotor faults is proposed as an alternative. The main concept is to excite the machine with a pulsating field at different angular positions and observe the stator current pattern to detect local or uniform PM demagnetization at motor standstill and to detect rotor faults independent of operating conditions or load torque oscillations which is not possible with MCSA. The proposed method is a simple technique that can be implemented in an inverter without additional hardware, test equipment, or manpower as a built-in diagnostics feature or used as a stand-alone diagnostics equipment for quality assurance or machine inspection as shown in Fig. 8. The d-axis is excited with a (direct-current)þ (alternating-current) signal, and the variation in the inductance pattern due to the change in the degree of magnetic saturation caused by demagnetization or eccentricity is observed for fault detection. It is well shown that in the presence of demagnetization fault, the rate of X ′d changes and increases by the fault rate. The rate of X ′d is shown in Fig. 9(a) uniform demagnetization and (b) partial demagnetization will be increased. It is an interested result when compared to inductance change in presence of eccentricity faults wherever the eccentricity faults cause X ′d to decrease from its nominal rate. However presented classification of eccentricity and demagnetization faults in this reference show a success of the presented method to separate introduced faults but other faults that have similar effect on inductance change, like unbalanced voltage, voltage dips, inter-turn short circuit, should be considered for using of inductance changes aiming demagnetization fault diagnosis. 3.1.4.4. Permeance network approach. In [45] the permeance

Fig. 7. Modeling method design-based analysis subsections flowcharts.

Fig. 8. Main concept of proposed technique for PMSM rotor fault detection; stator ac and dc field excitation in the d-axis is used for extraction of L′d .

S.S. Moosavi et al. / Journal of Magnetism and Magnetic Materials 391 (2015) 203–212

Fig. 9. The results for PM demagnetization (a) X ′d versus Id for uniform PM demagnetization (b) ΔX ‵d versus Id for partial demagnetization.

network (PN) approach and superposition theorem have been used to check out the semi-analytical model. The permeance network is based on magnetic network analogous to electrical network which used machine's geometry. It saves information on geometry within PN. The connection scheme of PN can either be in parallel or in series. It is possible to represent a certain part of

209

machine's geometry by one common permeance. The PN method has future potential as a research tool, in particular, in the dynamic simulations of an electrical machine. As compared to an accurate finite element method (FEM) analysis, a simple permeance model can solve the problem much more quickly with a little compromise to computational error. In other words, it is a good compromise between precise results and small simulation times. The corresponding equations for PN which depend on the machine's dimensions and different materials have been evaluated in [45]. Equivalent PN of the air-gap region have been defined as a function of the angular position of the rotor regarding to the stator shown in Fig. 10. This figure presents permeance of the yoke that is include the rotor steel permeance, flux source of an element of magnet include the north and South Pole, remmant MMFs, fault MMFs and PM permeance. Also leakage permeance between magnets and tooth, permeance of airgap and tooth and permeance of staror steel have been shown in Fig. 10 in detail. Ref. [45] explains how to use these parameters for modeling of demagnetization phenomenon thoroughly. The main advantage of the proposed model is its simplicity and its fast execution time, regarding to the FE method. Furthermore, it is possible to incorporate in the model, the control scheme by power converters, the effects of the magnetic saturation motor, and the temperature variation on the machine. 3.2. Investigations based on magnet design changes – practical design In [18,29,30,47], for minimization of mechanical vibration caused by demagnetization fault, the optimal PM shape is designed. The robust shape model of the permanent magnet for the outer rotor PMSM related to partial demagnetization, cogging torque, load angle curve, and back EMF is suggested. In Fig. 11 samples of different shapes of magnet due to optimizing process

Fig. 10. Equivalent PN.

210

S.S. Moosavi et al. / Journal of Magnetism and Magnetic Materials 391 (2015) 203–212

4. Synthesis study

Fig. 11. Different shapes of magnet due to optimizing process against demagnetization fault [31] (finally, the author introduces the shape model 1 as a robust permanent magnet type of outer rotor PMSM for HEV).

against demagnetization fault, are shown. Also in [30] a structure of high power density PM assisted synchronous reluctance motor employing the ferrite PM, that does not cause irreversible demagnetization even in condition of the heavy flux-weakening excitation or inverter's fault present. A permanent-magnet-assisted salient-pole synchronous machine (PMa-SM) under sudden short circuits has been investigated in [44]. Based on this, an additional damper bars is used to reduce the demagnetization in the PMs. As a result, it has been found that making the PM shape a rectangle and adding the damper bars on the outer surfaces of the PMs is effective to reduce the PM demagnetization. Adding the damper bars on the inner surfaces is not effective. Increasing the conventional damper bars is also not effective. Causes of the effectiveness and ineffectiveness of the additional outer and inner damper bars are also shown. Four different rotor configurations with copper sheets, dual squirrel cages, magnetic barriers and bridges, and with magnetic barriers and dual cages for a Line start permanent magnet synchronous motors (LSPMSM) has been Investigated in [47]. The results prove that the comprehensive configuration with both dual squirrel cages and magnetic barriers on the rotor is rather effective to protect the magnets, but hardly deteriorates the motor operation performance under the rated condition. Parameter estimation techniques are used in previous years for fault detection and diagnosis of permanent magnet motors [75]. The newest work in this area is presented in [8]. A method based on torque changes is presented to determine magnet strength. The average of the magnet flux linkage derivative terms yields to a value that corresponds to twice time the mean value of the flat part of the phase back-EMF. The symbol used here for the esti^ mation of this mean value is E that is calculated according to the following equation:

^ E = V − 2rs Idc

As it is obvious from existed references, in recent years this kind of fault has been noted by researchers. Because of its importance, each year different methods are proposed for detecting it more precisely on time. Regarding the vast scope of references that have been studied in this paper, besides past methods shown in Table 1, the latest achievements of researchers in recent years were investigated and analyzed. Advantages and disadvantages of each were shortly discussed. A summary of the newest methods that were proposed is brought below. 1. Fast Fourier Transform (FFT) of stator components (zero sequence currents, voltage and current) – [2,4,6,15,17,39,55,5759] – 2006–2010 and 2012. 2. Time frequency transform of stator components – [4,11,12,15,16,39,61,71] – 2007–2010. 3. Noise, vibration and torque – [32,39,62,47] – 2006, 08, 10, and 2013. 4. Optimization in PM shape – [18,28,29,47] – 2006, 10, and 2011. 5. Field reconstruction method (sound power radiated) – [5,7,63] – 2010–2012. 6. Measure of the zero sequence voltage components (the spectrum of the back-EMF) – [3,6,64] – 2011 and 2012. 7. Additional damper bars – [44] – 2012. 8. Inverter embedded technique – [2,17] – 2010 and 2012. 9. Estimation of torque mean value – [8] – 2007. 10. Search coils – [13] – 2010. 11. Analytical models – [43,71-74] – 2000, 02, 03, 06, and 2008. 12. Semi-analytical model – [45] – 2006.

5. Conclusions Today, PMSM have significant attraction for different industries because of their considerable advantages that were discussed in this paper in detail. By continuously using these machines, normally they would be exposed to damages that necessitate the need for monitoring, detection of faults and prevention of fault propagation. This paper has concentrated its goal uniquely on inclusive and precise investigation into demagnetization fault in PMSM. Critical points in fault diagnosis that lead to correct results and the increase of the safety and reduction of the cost are mentioned in the following:

(4)

 The exact knowledge of system under operation will be effec-

where V is the average supply voltage and Idc is the average of the DC current. These calculations are simple to perform since the Hall sensors in the BLDC motor indicate the six regions of opera^ tion. When the parameter E estimated in Eq. (4) is divided by the electrical rotor speed (rad/s), the result obtained is the average value of kt , which measure the magnet strength. Thus, the estimation of the magnet strength is given by the following expression:

tive in avoidance of mistake in fault diagnosis and control of motor. As mentioned in some PMSMs by using the specific stator windings configuration, the fault harmonics predicted by Eq. (3) are not presented in the current spectrum. So the kind of stator windings configuration must be considered. The uniform demagnetization cannot be detected with motor current signature analysis. Another critical limitation of the MCSA is the similar effect under oscillating load torque that produces side-bands at the same frequency component as Eq. (3). So considering other faults by the same effect is urgent. It must be taken into attention that, it is too hard to determine the source of specific harmonics in the faults that have the same signature frequencies, like partial demagnetization, dynamic eccentricity and unbalanced load. These faults have the same effects on some of motor parameters too as mentioned in [2,3,6,17,32,37,62,64], so the selection of the best signal for processing in view of the diagnosis of the demagnetization fault will be very important for avoiding mistakes. As it is mentioned earlier in this paper, although a short circuit fault, in an early stage when a small number of turns is

kt =

V − 2rs Idc ωr

(5)





This value gives a good indication on health of the magnet and can be used to detect the broken or the demagnetized rotor magnets. The estimation would give a more accurate result when the back-EMF (E ) is used directly as the following equation:

kt =

E ωr

(6)



S.S. Moosavi et al. / Journal of Magnetism and Magnetic Materials 391 (2015) 203–212



concerned, cannot lead to the demagnetization, no timely diagnosis will lead to additional flux from turn to turn and demagnetization in the PMSM. It mentions the relation between different faults and their effect together. This would be true about open phasing fault and two phases short-circuit fault too. So a complete diagnosis system must be able to detect all major faults and their discrimination. The quick response and high accuracy of analytical and some of modeling method design based analysis for demagnetization fault diagnosis must be taken into account by the researchers that want to present new methods other than the powerful FEM method.

Even though, the use of FFT in non-stationary conditions does not let us to detect the fault, it is still used as an applicable and attractive method in stationary conditions by researchers. Today, time frequency method is used by a great number of researchers in stationary and non-stationary conditions as a compensation for FFT method. However, considering the appropriate diagnosis of this method and other similar methods when facing different faults, the existence of a supplementary mean like neural network for classification of faults seems to be a necessary factor.

References [1] M. Rosu1, J. Saitz, A. Arkkio, Hysteresis model for finite-element analysis of permanent-magnet demagnetization in a large synchronous motor under a fault condition, IEEE Trans. Magn. 41 (6) (2005) 2118–2123. [2] J. Hong, S. Park, D. Hyun, T. j Kang, S.B. Lee, Ch Kral, A. Haumer, Detection and classification of rotor demagnetization and eccentricity faults for PM synchronous motors, IEEE Trans. Ind. Appl. 48 (3) (2012) 923–932. [3] J. C´esar Urresty, J.R. Riba, M. Delgado, L. Romeral, Detection of demagnetization faults in surface-mounted permanent magnet synchronous motors by means of the zero-sequence voltage component, IEEE Trans. Energy Convers. 27 (1) (2012) 42–51. [4] J.R.R. Ruiz, J.A. Rosero, A.G. Espinosa, L. Romeral, Detection of demagnetization faults in permanent-magnet synchronous motors under nonstationary conditions, IEEE Trans. Magn. 45 (7) (2009) 2961–2969. [5] D. Torregrossa, A. Khoobroo, B. Fahimi, Prediction of acoustic noise and torque pulsation in PM synchronous machines with static eccentricity and partial demagnetization using field reconstruction method, IEEE Trans. Ind. Electron. 59 (2) (2012) 934–944. [6] J. Urresty, R.J. Riba, A. Romeral, Back-emf based method to detect magnet failures in PMSMs magnetics, IEEE Trans. Magn. 49 (1) (2013) 591–598. [7] A. Khoobroo, B. Fahimi, A novel method for permanent magnet demagnetization fault detection and treatment in permanent magnet synchronous machines, in: Proceedings of the 25th Annual IEEE Conference on Applied Power Electronics Conference and Exposition (APEC), Palm Springs, California, February, 2010. [8] S. Rajagopalan, W. Roux, T.G. Habetler, R.G. Harley, Dynamic eccentricity and demagnetized rotor magnet detection in trapezoidal flux (brushless DC) motors operating under different load conditions, IEEE Trans. Power Electron. 22 (5) (2007) 2061–2069. [9] K. Kim, S.B. Lim, J. Lee, The effect on partial demagnetization of permanent magnet synchronous motor for hybrid electric vehicle, in: Proceedings of the 12th Biennial IEEE Conference on Electromagnetic Field Computation, Miami, Florida, 2006. [10] D. Casadei, F. Filippetti, C. Rossi, A. Stefani, Magnets faults characterization for permanent magnet synchronous motors, in: Proceedings of the IEEE International Symposium on Diagnostics for Electric Machines, Power Electronics and Drives, 31 August–3 September 2009, pp. 1–6. [11] J. Rosero, L. Romeral, J. Cusido, J.A. Ortega, Fault detection by means of wavelet transform in a PMSMW under demagnetization, in: Proceedings of the 33rd Annual Conference of the IEEE, IECON, Industrial Electronics Society, Taipei, China, 2007. [12] Y. Da X. Shi M. Krishnamurthy. Health monitoring, fault diagnosis and failure prognosis techniques for brushless permanent magnet machines, in: Proceedings of the IEEE Vehicle Power and Propulsion Conference (VPPC), Chicago, September 2011, pp. 1–7. [13] Y. Da, X. Shi, M. Krishnamurthy, A new approach to fault diagnostics for permanent magnet synchronous machines using electromagnetic signature analysis, IEEE Trans. Power Electron. 28 (8) (2013) 4104–4112. [14] M.E.H. Benbouzid, A review of induction motor signature analysis as a medium for faults detection, IEEE Trans. Ind. Electron. 47 (5) (2009) 984–993. [15] A.G. Espinosa, J.A. Rosero, J. Cusid´o, L. Romeral, J.A. Ortega, Fault detection by means of Hilbert–Huang transform of the stator current in a PMSM with demagnetization, IEEE Trans. Energy Convers. 25 (2) (2010) 312–318.

211

[16] J. Rosero, A. Garcia, J. Cusido, L. Romeral, J.A. Ortega, Fault detection by means of Hilbert Huang Transform of the stator current in a PMSM with demagnetization, in: Proceedings of the IEEE International Symposium on Intelligent Signal Processing, WISP, October 2007, pp. 1–6. [17] J. Hong, D. Hyun, S.B. Lee, J. Yoo, K. Lee, Automated monitoring of magnet quality for permanent magnet synchronous motors at standstill, IEEE Trans. Ind. Appl. 46 (4) (2010) 1397–1405. [18] H.K. Kim, J. Hur, B.W. Kim, G.H. Kang, Characteristic analysis of IPM type BLDC motor considering the demagnetization of PM by stator turn fault, in: Proceedings of the Energy Conversion Congress and Exposition (ECCE) Atlanta, Georgia, September 2010, pp. 3063–3070. [19] H.S. Ko, K.J. Kim, Characterization of noise and vibration sources in interior permanent-magnet brushless DC motors, IEEE Trans. Magn. 40 (6) (2004) 3482–3489. [20] J.A. Rosero, L. Romeral, J.A. Ortega, E. Rosero, Short-Circuit detection by means of empirical mode decomposition and Wigner–Ville distribution for PMSM running under dynamic condition, IEEE Trans. Ind. Electron. 56 (11) (2009) 4534–4547. [21] B.M. Ebrahimi, J. Faiz, M. Javan-Roshtkhari, A. Zargham Nejhad, Static eccentricity fault diagnosis in permanent magnet synchronous motor using time stepping finite element method, IEEE Trans. Magn. 44 (11) (2008) 4297–4300. [22] A. Djerdir, J.-A. Farooq, A. Rezig, A. Miraoui, Faults in permanent magnet traction motors: state of the art and modelling approaches, in: Proceedings of the Power and Energy Society General Meeting, Minneapolis, July 2010. pp.1– 6. [23] M. Hadef, M.R. Mekideche, A. Djerdir, Vector controlled permanent magnet synchronous motor (PMSM) drive with stator turn fault, in: Proceedings of the 19th International Conference on Electrical Machines-ICEM, Rome, 2010, pp. 1–6. [24] J. Rosero, J.A. Ortega, J. Urresty, J. Cárdenas, L. Romeral, Stator short circuits detection in PMSM by means of higher order spectral analysis (HOSA), in: Proceedings of the Applied Power Electronics Conference and Exposition, Washington, February, 2009, pp. 964–969. [25] J. Rosero, L. Romeral, J.A. Ortega, E. Rosero, Short-circuit detection by means of empirical mode decomposition and Wigner–Ville distribution for PMSM running under dynamic condition, IEEE Trans. Ind. Electron. 56 (11) (2009) 4534–4547. [26] J.A. Farooq, T. Raminosoa, A. Djerdir, A. Miraoui, Modelling and simulation of stator winding inter-turn faults in permanent magnet synchronous motors, Int. J. Comput. Math. Electr. Electron. Eng. 27 (4) (2008) 887–896. [27] J. Luukko, Direct torque control of permanent magnet synchronous machines – analysis and implementation (Ph.D. dissertation), Lappeenranta University of Technology, Lappeenranta, Finland, 2000. [28] IEC, Rotating Electrical Machines, Part 4: Methods for Determining Synchronous Machine Quantities from Tests, IEC Standard 60034-4, 2008. [29] S. Ooi, S. Morimoto, M. Sanada, Y. Inoue, Performance evaluation of a highpower-density PMASynRM with ferrite magnets, IEEE Trans. Ind. Appl. 49 (3) (2013) 1308–1315. [30] K. Kim, S. Lim, D. Koo, J. Lee, The shape design of permanent magnet for permanent magnet synchronous motor considering partial demagnetization, IEEE Trans. Magn. 42 (10) (2006) 3485–3487. [31] S. Nandi, H. Toliyat, X. Li, Condition monitoring and fault diagnosis of electrical motors – a review, IEEE Trans. Energy Convers. 20 (4) (2005) 719–729. [32] D. Basak, A. Tiwari, S. Das, Fault diagnosis and condition monitoring of electrical machines – a review, in: Proceedings of the IEEE International Conference on Industrial Technology (ICIT), Mumbai, December 2006, pp. 3061– 3066. [33] G.K. Singh, S. Ahmed Saleh, Al Kazzaz, Induction machine drive condition monitoring and diagnostic research – a survey, Electr. Power Syst. 64 (2) (2003) 145–158. [34] A.H. Bonnett, G.C. Soukup, Cause and analysis of stator and rotor failures in three-phase squirrel-cage induction motor, IEEE Trans. Ind. Appl. 28 (4) (1992) 921–937. [35] A. Bellini, F. Filippetti, C. Tasoni, G.A. Capolino, Advances in diagnostic techniques for induction machines, IEEE Trans. Ind. Electron. 55 (12) (2008) 4109–4126. [36] A. Siddique, G.S. Yadava, B. Singh, A review of stator fault monitoring techniques of induction motor, IEEE Trans. Energy Convers. 20 (1) (2005) 106–114. [37] B.M. Ebrahimi, J. Faiz, M. Javan Roshtkhari, Static-, dynamic-, and mixed-eccentricity fault diagnoses in permanent-magnet synchronous motors, IEEE Trans. Ind. Electron. 56 (11) (2009) 4727–4739. [38] M.A.S.K. Khan, M. Azizur Rahman, Development and implementation of a novel fault diagnostic and protection technique for IPM motor drives, IEEE Trans. Ind. Electron. 56 (1) (2009) 85–92. [39] M. Delgado, A. Garcia, J. Riba, C. Urresty, J.A. Ortega, Feature extraction of demagnetization faults in permanent- magnet synchronous motors based on box-counting fractal dimension, IEEE Trans. Ind. Electron. 58 (5) (2011) 1594–1605. [40] C. Ruschetti, G. Bossio, C. De Angelo, C. Verucchi, Effects of partial rotor demagnetization on permanent magnet synchronous machines, in: Proceedings of the IEEE International Conference on Industrial Technology (ICIT) Vi a del Mar, March 2010, pp. 1233–1238. [41] W. Le Roux, R. Harley, T. Habetler, Rotor fault analysis of a permanent magnet synchronous machine, in: Proceedings of the International Conference on Electrical Machines (ICEM'02), Brugge, Belgium, 2002. [42] S. Rajagopalan, W. le Roux, T.G. Habetler, R.G. Harley, Diagnosis of potential

212

[43]

[44]

[45]

[46]

[47]

[48] [49]

[50]

[51]

[52]

[53]

[54]

[55]

[56]

[57]

[58]

[59]

S.S. Moosavi et al. / Journal of Magnetism and Magnetic Materials 391 (2015) 203–212

rotor faults in brushless dc machines, in: 2nd International Conference on Power Electronics, Machines and Drives (PEMD), Edinburgh, 31 March–2 April 2004, pp. 668–673. J.A. Farooq, A. Djerdir, A. Miraoui, Analytical modeling approach to detect magnet defects in permanent-magnet brushless motors, IEEE Trans. Magn. 44 (12) (2008) 4599–4604. T. Hosoi, H. Watanabe, K. Shima, T. Fukami, R. Hanaoka, S. Takata, Demagnetization analysis of additional permanent magnets in salient-pole synchronous machines with damper bars under sudden short circuits, IEEE Trans. Ind. Electron. 59 (6) (2012) 2448–2456. J. Farooq, S. Srairi, A. Djerdir, A. Miraoui, Use of permeance network method in the demagnetization phenomenon modeling in a permanent magnet motor, IEEE Trans. Magnet. 42 (4) (2006) 1295–1298. A. Vagati, B. Boazzo, P. Guglielmi, G. Pellegrino, Design of ferrite assisted synchronous reluctance machines robust towards demagnetization, IEEE Trans. Ind. Appl. 50 (3) (2014) 1768–1779. J. Shen, P. Li, M. Jin, G. Yang, Investigation and countermeasures for demagnetization in line start permanent magnet synchronous motors, IEEE Trans. Magn. 49 (7) (2013) 4068–4071. W. Li, A. Li, H. Wang, Anisotropic fracture behavior of sintered rareearth permanent magnets, IEEE Trans. Magn. 41 (8) (2005) 2339–2342. S. Ruoho, A. Arkkio, Partial demagnetization of permanent magnets in electrical machines caused by an inclined field, IEEE Trans. Magn. 44 (7) (2008) 1773–1778. J.A. Farooq, Study and detection of demagnetization and short-circuit in permanent magnet synchronous machines (Ph.D. thesis), University of Technology Belfort Montbéliard, France, 2008. G.H. Kang, J. Hur, H. Nam, J.P. Hong, G.T. Kim, Analysis of irreversible magnet demagnetization in line-start motors based on the finite-element method, IEEE Trans. Magn. 39 (4) (2003) 1488–1491. S. Ruoho, J. Kolehmainen, J. Ikaheimo, A. Arkkio, Interdependence of demagnetization, loading, and temperature rise in a permanent- magnet synchronous motor, IEEE Trans. Magn. 46 (3) (2010) 949–953. S. Yu, R. Tang, Electromagnetic and mechanical characterization of noise and vibration in permanent magnet synchronous machines, IEEE Trans. Magn. 42 (4) (2006) 1335–1338. K.C. Kim, S.B. Lim, D.H. Koo, J. Lee, The shape optimization of permanent magnet for permanent magnet synchronous motor considering partial demagnetization, IEEE Trans. Magn. 42 (10) (2006) 3485–3487. J.A. Rosero, J. Cusido, A. Garcia, J.A. Ortega, L. Romeral, Study on the permanent magnet demagnetization fault in permanent magnet synchronous machines, in: proceedings of the 32nd Annual Conference on IEEE Industrial Electronics, IECON, Paris, November, 2006, pp. 879–884. J.A. Farooq, A. Djerdir, A. Miraoui, Analytical modeling approach to detect magnet defects in permanent-magnet brushless motors, IEEE Trans. Magn. 44 (12) (2008) 4599–4604. W. le Roux, R.G. Harley, T.G. Habetler, Detecting rotor faults in low power permanent magnet synchronous machines, IEEE Trans. Power Electron. 22 (1) (2007) 322–328. S. Rajagopalan, J.M. Aller, J.A. Restrepo, T.G. Habetler, R.G. Harley, Detection of rotor faults in brushless DC motors operating under non-stationary conditions, IEEE Trans. Ind. Appl. 42 (6) (2006) 1464–1477. J. Rosero, J.L. Romeral, J. Cusido, J.A. Ortega, A. Garcia, Fault detection of eccentricity and bearing damage by means of wavelet transforms decomposition of the stator current, in: Proceedings of the 23rd Annual IEEE on Applied Power Electronics Conference and Exposition, IEEE APEC, Austin, February, 2008, pp. 111–116.

[60] B.M. Ebrahimi, J. Faiz, Demagnetization fault diagnosis in surface mounted permanent magnet synchronous motors, IEEE Trans. Magn. 49 (3) (2013) 1185–1192. [61] J.A. Rosero, J. Cusido, A. Garcia, J.A. Ortega, L. Romeral, Broken bearings and eccentricity fault detection for a permanent magnet synchronous motor, in: Proceedings of the 32nd Annual Conference on IEEE industrial electronics, IECON, Paris, November, 2006, pp. 964–969. [62] G.H. Kang, Y.D. Son, G.T. Kim, The noise and vibration analysis of BLDC motor due to asymmetrical permanent-magnet overhang effects, IEEE Trans. Ind. Appl. 44 (5) (2008) 1569–1577. [63] M. Kiani, D. Torregrossa, B. Fahimi, F. Peyraut, A. Miraoui, Detection of faults in PMSM using field reconstruction method and mechanical impulse response, in: Proceedings of the 26th Annual IEEE on Applied Power Electronics Conference and Exposition (APEC), Fort Worth, March, 2011, pp. 1896–1901. [64] J. Urresty, J. Riba, H. Saavedra, J. Romeral, Analysis of demagnetization faults in surface-mounted permanent magnet synchronous motors with symmetric windings, in: Proceedings of the IEEE International Symposium on Diagnostics for Electric Machines, Power Electronics & Drives (SDEMPED), Bologna, September 2011, pp. 240–245. [65] J. Urresty, R. Atashkhooei, J. Riba, L. Romeral, S. Royo, Shaft trajectory analysis in a partially demagnetized permanent-magnet synchronous motor, IEEE Trans. Ind. Electron. 60 (8) (2013) 3454–3461. [66] J. Urresty, J. Riba, L. Romeral, Influence of the stator windings configuration in the currents and zero-sequence voltage harmonics in permanent magnet synchronous motors with demagnetization faults, IEEE Trans. Magn. 49 (8) (2013) 4885–4893. [67] L. Cohen, Time–frequency distributions – a review, Proc. IEEE 77 (7) (1989) 941–981. [68] R. Carmona, H.L. Hwang, B. Torresani, Practical Time–Frequency Analysis: Gabor and Wavelet Transforms With an Implementation in S, Academic Press, New York, 1998. [69] A. Papandreou- Suppappola, Applications in Time–Frequency Signal Processing, CRC Press, Boca Raton, FL, 2003. [70] J. Rosero, L. Romeral, J.A. Ortega, J.C. Urresty, Demagnetization fault detection by means of Hilbert Huang transform of the stator current decomposition, in: Proceedings of the PMSM IEEE International Symposium on Industrial Cambridge, 30 June–2 July 2008, pp. 172–177. [71] Z.Q. Zhu, D. Howe, C.C. Chan, Improved analytical model for predicting the magnetic field distribution in brushless permanent magnet machines, IEEE Trans. Magn. 38 (1) (2002) 229–238. [72] X. Wang, Q. Li, S. Wang, Q. Li, Analytical calculation of air-gap magnetic field distribution and instantaneous characteristics of brushless DC motors, IEEE Trans. Energy Convers. 18 (3) (2003) 424–432. [73] K.F. Rasmussen, J.H. Davies, T.J.E. Miller, M.I. McGlip, M. Olaru, Analytical and numerical computation of air-gap magnetic fields in brushless motors with surface permanent magnets, Ind. Appl. 36 (5) (2000) 1547–1554. [74] A.B. Proca, A. Keyhani, A. El-Antably, W. Lu, M. Dai, Analytical model for permanent magnet motors with surface mounted magnets, IEEE Trans. Energy Convers. 18 (3) (2003) 386–391. [75] X.Q. Liu, H.Y. Zhang, J. Liu, J. Yang, Fault detection and diagnosis of permanentmagnet DC motor based on parameter estimation and neural network, IEEE Trans. Ind. Electron. 47 (5) (2010) 1021–1030. [76] P. Sergeant, A.P.M. Van den Bossche, Influence of the amount of permanentmagnet material in fractional-slot permanent-magnet synchronous machines, IEEE Trans. Ind. Electron. 61 (9) (2014) 4979–4989.