Fault Detection of Permanent Magnet Synchronous Machines: An Overview

1. Introduction

Permanent magnet synchronous machines (PMSMs) become growingly popular in industries due to their high efficiency and high torque density [1]. However, reliability has been one of the major concerns that preclude the further promotion of PMSMs in many safety critical applications, such as aerospace, electrical vehicles, and wind power generation. To improve the reliability of PMSMs, a lot of efforts have been put into this subject from the perspective of the topology and design of PMSMs, the condition monitoring, as well as the fault mitigation and fault tolerant control. It has been found that condition monitoring could play a vital role in protecting the PMSM drive system from the catastrophic consequences of the faults [2]. In the area of condition monitoring, fault detection is an important subject, which aims at detecting the faults after their occurrence, thus providing information about the type, location, and scale of faults for subsequent emergency operations. On most occasions of online fault detection, the rapidness of detection is also required to reserve enough reaction time.

Extensive literature has been published about fault detection of PMSMs, covering a wide range of faults. In general, the faults in PMSMs could be classified into 3 categories based on the failure parts, including electrical faults, mechanical faults, and magnet faults [3], as shown in Figure 1. Electrical faults can be further classified according to the faulty components as winding faults, inverter faults, and sensor faults. Mechanical faults include rotor eccentricity, rotor misalignment, bearing faults, etc. Magnet faults consist of magnet mechanical damage, uniform and partial demagnetization, etc. Winding faults can be further classified according to the failure modes, including inter-turn short-circuit (ITSC) faults, phase-to-phase and phase-to-ground faults, open circuit faults, high resistance connection (HRC) faults, etc.

Several review papers about fault detection have been published. A very comprehensive review is presented in [3] about the fault detection methods. The detection methods are generally classified into 3 categories, signal-based, model-based, and artificial intelligence (AI)-based methods. Diverse kinds of faults are covered in [3] including ITSC fault, magnet faults, and several kinds of mechanical faults. Special attention is paid to the application of advanced signal processing and AI algorithms, nevertheless, lack of analysis about the signal sources. This could be vital because the available signals might vary dramatically among different applications, and thus, the corresponding fault signatures and the analysis methods might be completely different from each other. In [4], the focus is laid on the detection methods, whereas the corresponding fault signatures are not summarized. Several detailed surveys are accomplished about the detection of ITSC [5], demagnetization [6], and eccentricity [7] faults. However, these papers concentrate more on one fault, thus failing to reveal the common features and differences among these faults, which is critical for distinguishment among these faults. In summary, existing review papers are not systematic and detailed enough. Hence, the target of this paper is to provide a more comprehensive review about fault detection of PMSMs in the perspective of the signals and their signatures.

In this paper, the detection methods of three typical faults in PMSMs are reviewed, i.e., the ITSC fault, PM partial demagnetization (PD) fault, and eccentricity fault, because they are very common while difficult to detect. The detection methods of each fault are classified into 3 major categories, i.e., the signal-based, model-based, and data-based methods [3,4,8], as illustrated in Figure 2. Signal-based methods mainly concentrate on extracting the fault-related features of signals directly or indirectly. Model-based methods treat the fault as a variation of the model of motor, hence using various ways to observe the variation of model. Data-based methods utilize the algorithms of machine learning, trying to extract high dimensional hidden information about fault features from the data. Figure 3 concludes the basic principles of signal-, model-, and data-based methods, from the perspective of the general fault detection procedure, i.e., feature extraction and decision-making process.

Furthermore, this paper specifically classifies the papers in literature according to their signal sources and fault signatures, intending to provide a unique insight into the state-of-art techniques for fault detection of PMSMs and their general issues, as shown in Figure 2.

This paper is organized as follows. Section 2, Section 3 and Section 4 discuss the detection methods of ITSC fault, PD fault, and eccentricity fault, respectively. Section 5 discusses some differences in the detection methods of these three faults. Section 6 concludes the paper and presents the future work.

2. Inter-Turn Short-Circuit Fault Detection

2.1. Background

The ITSC fault is the short-circuit between one or a few turns of a winding due to the inter-turn insulation degradation and breakdown. The ITSC fault may be the result of various factors including high winding temperature, mechanical stress, transient overvoltage, etc. When the ITSC fault occurs, the broken insulation part forms a short-circuit path shown in Figure 4 (a). The residual resistance on the short-circuit path can be modelled as a resistance R_f parallel to the short-circuit part of winding. The equivalent circuit of PMSMs with the ITSC fault is shown in Figure 4 (b), where the ITSC fault is assumed in phase A without losing generality.

As shown in Figure 4 (b), a large circulating current i_f will flow through the short-circuit resistance R_f as well as the shorted coil L_af, causing an extremely high temperature in the local area. High temperature can further accelerate insulation degradation, which may lead to the development of ITSC fault in the manner of avalanche, resulting in more serious fault such as phase-to-phase or phase-to-ground short-circuit [2]. Hence, it is vital to detect the ITSC fault as soon as possible.

However, ITSC fault detection is not easy. The high circulating current does not flow through the regular phase current sensors so that phase currents only contain insignificant fault-related features. Thus, numerous methods are developed to extract and magnify the fault-related features from all types of signals in electrical machines.

2.2. Signal-Based Methods

The signal-based methods refer to those concentrating on extracting explicit signatures from specific signals with time or frequency analysis techniques. The signals adopted for fault detection can be either common electrical signals available in a PMSM drive system, or signals obtained with extra sensors such as Hall sensors or search coils. Generally speaking, the signal-based methods can be grouped as electrical-signal-based and magnetic-signal-based. Obviously, the signatures in electrical signals and magnetic signals are different. Thus, they have different SNR, advantages and disadvantages. The detailed classification is shown in Figure 5.

2.2.1. Electrical Signals

Electrical signals generally include voltage and current signals. Sampled current signals are commonly available in PMSM drive system due to the need of close-loop control. As for voltage signals, usually the reference voltage signals are used, while extra voltage sensors are usually required to sample the zero sequence voltage component (ZSVC). In general, the methods based on electrical signals can be grouped according to the type of signatures:

a) Symmetrical component

Symmetrical component analysis is one of the common ways to detect asymmetry in machines [2,9], where negative sequence component (NSC) and zero sequence component (ZSC) emerge in addition to positive sequence component (PSC). Obviously, ITSC fault introduces asymmetry in stator winding, and thus, ZSC and NSC can be found in either machine voltages or currents. A method based on the ZSC in voltage (for wye-connect machines) and current (for delta-connect machines) is proposed in [10]. The location of ITSC fault can be decided by examining the phase angle information of ZSC, which is also found able to discriminate HRC and ITSC faults [11]. These methods are further integrated with mitigation methods in [12,13]. The similar idea is applied in brushless DC (BLDC) machine with a simple fault severity evaluation method [14], reducing the impact of working condition on fault detection. Laadjal et al. [15] propose a fault index defined as the ratio between ZSC, NSC and PSC so that the ITSC fault can be distinguished from unbalanced supply voltage. The algorithm is improved by introducing short-time least square Prony’s algorithm to track the fundamental frequency component [16,17]. In [18], the ZSC in voltage is found has better signal-to-noise ratio (SNR) and sensitivity to fault than the 2^nd harmonics in q-axis current and the speed harmonics.

Similar to ZSC, NSC will emerge in the voltage or current of machines [19]. Jeong et al. [20] successfully utilize the NSC in machine voltages to detect the ITSC fault of PMSMs. It is also discovered that NSC in voltage is more sensitive to ITSC fault and more general, compared to ZSC in voltage, because the amplitude of ZSC in voltage is dependent on the leakage inductance of the machine. Meanwhile, NSC in current can also be utilized to detect ITSC fault [21]. A fault indicator is further proposed in [22], by combining the NSC and PSC in current. The phase angle between NSC and PSC in current is used to locate the ITSC fault. Furthermore, NSC in voltage and current can be utilized at the same time. A novel fault indicator is proposed in [23], which is the phase shift between NSC in the machine voltage and current. The indicator can discriminate the unbalanced load of the PM synchronous generator (PMSG) and the ITSC fault, meanwhile the location of ITSC fault is revealed by the phase angle of the NSC in voltage. Similarly, the NSC in voltage can be also combined with ZSC and PSC to detect ITSC fault [15,24].

b) Low frequency (LF) pattern

On the other hand, frequency pattern of signals usually also contains effective signatures for ITSC detection. From the perspective of the way to obtain frequency pattern, diverse frequency analysis techniques are adopted in existing literature, such as Fast Fourier Transform (FFT) [25,26,27,28], orthogonal phase-lock-loop (PLL) [29,30], empirical model decomposition (EMD) [31,32], and wavelet transform [33] , etc.

Meanwhile, existing methods also cover a wide range of fault signatures. For example, 3^rd harmonics in phase currents are found feasible for detection [31], thus being tracked with quadratic time–frequency techniques, targeting at fault detection at non-stationary situations. Dogan et al. [32] extract 3^rd harmonics in phase currents using Gabor transform and Ensemble EMD. Lee et al. [27] made a conclusion that the negative frequency part of 3^rd harmonics is free from supply imbalance and inherent structural asymmetry, leading to its high robustness. The ITSC fault signatures are selected as (2k-1)^th harmonics in [26], which is a more general fault signature. Naderi et al. [25] obtain the fault related frequency signatures in the currents of PMSM through comprehensive theoretical analysis, concluding the capability of each frequency component to distinguish among different faults.

The frequency pattern In dq-axis quantities is also widely applied. It is discovered that ITSC fault introduces 2^nd harmonics in dq-axis voltage and currents [34], which can be treated as an alternate perspective of the same fault signatures with NSC [2]. However, the variation in controller bandwidth will cause re-distribution of the 2^nd harmonics between voltage and current [35]. Huang et al. [30] introduce a fault index called Rayleigh Quotient so that the 2^nd harmonics in voltages and currents can be spontaneously synthesized to reduce the influence of current controller bandwidth. The 2^nd harmonics extraction method can be improved by applying a short-time Adaline neural network [36], variational mode decomposition (VMD) [37], and wavelet transform [38]. The transient false alarm is mitigated in [39] through estimated compensation signals calculated according to the mechanical close-loop transfer function. There are also methods using the 2^nd harmonics in Extended Park’s vector approach (EPVA) [40].

Power, which is essentially the multiple of voltage quantities and current quantities, combines both the signatures from voltage and current. The 2^nd harmonic is found appearing in instantaneous reactive power (IRP) when ITSC fault occurs in a PMSM working in motoring mode, while appears in instantaneous active power (IAP) in generating mode [28]. The analysis results show the amplitude of the 2^nd harmonics in IRP or IAP is sensitive to working conditions. Consequently, the look-up-table (LUT) is adopted to make fault decision. Similar detection method is proposed in [41], where the 2^nd harmonics is observed in instantaneous power of a synchronous generator.

Information in the amplitude and phase angle of the fundamental frequency component in phase voltages or currents, i.e., the DC components in dq-axis quantities, are utilized in some papers for ITSC fault detection and discrimination from other faults. The voltage vector variation in dq-axis is used as the fault indicator in [42]. The static eccentricity fault, PM PD fault and ITSC fault cause the voltage vector deviating from healthy vector in different direction. Meanwhile the deviation of current angle is found increasing when ITSC fault occurs while demagnetization causes current angle decreasing [43]. Similarly, the torque angle is utilized in [44], which is essentially the angle between voltage vector and the d-axis. The amplitude and phase differences among αβ fundamental frequency currents are considered together in [45]. It is worth mentioned that recently, the fundamental frequency components in xy subspace of dual-three-phase PMSM (DTPPMSM) is found feasible to detect and locate the ITSC fault [29,46,47,48].

c) Highfrequency(HF)pattern

Previously mentioned papers mainly concentrate on frequency pattern in LF range. As for pattern in HF range, two types of methods are developed, i.e., methods based on the switching frequency components and those based on the HF injection.

Switching frequency components exists in switching-inverter based drive system. It is a naturally embedded excitation source providing useful information for fault detection. Sen et al. [49] observe that the admittance of phase winding increases at the HF region after ITSC fault happens. Hence, switching frequency ripple is adopted as the HF test signal to obtain the HF impedance of winding. Similar idea is also investigated in [50]. Hu et al. [51] model and analyse the pulse-width modulation (PWM) signals in a PMSM and obtain the analytical expression of PWM ripple current. The analytical expression as well as simulation shows that the amplitude of PWM ripple current of the faulty phase increases when ITSC fault occurs. A fault index is further proposed defined as the ratio of the 2f_switching component in adjacent phases. Same principle is adopted in [52], while its capability to distinguish between high resistance connection (HRC) and ITSC fault is discovered. HRC fault only causes resistance asymmetry, hence insensitive to HF voltage, while ITSC causes inductance variation, hereby discrimination of two kinds of fault. The method is improved by combined the ZSVC and switching frequency component analysis [53]. The fundamental component of ZSVC is used to detect asymmetry in stator, while the HF component in ZSVC is used to distinguish between HRC and ITSC fault. Meanwhile, GAO et al. [54] combined the NSVC and switching frequency component analysis. Wang et al. [55] compare interior PMSMs (IPMSMs) and surface-mounted PMSMs (SPMSMs), where the saliency ratio is proved to be insignificant to the switching frequency and its sideband in current spectrum.

Switching frequency components in voltage rely on the modulation index, thus being influenced by working conditions. On the other hand, by injecting HF voltage or current, the excitation source becomes independent from working conditions, making fault signatures much more consistent across different working conditions. A method based on HF rotating current is proposed to distinguish between HRC and ITSC fault [56]. The injected frequency component is extracted from ZSVC. Xu et al. [57] inject specially designed HF current so that injected current only flow through two phases at one time. The line voltage difference when injecting different pairs of phases is selected as the fault indicator. Other than current injection, rotating square wave voltage is injected in [58,59] and the differences of the HF response in three phases are selected as the fault features. Different kinds of signal injection methods, including rotational voltage, static pulsating voltage, rotational pulsating voltage, and rotational current are systematically compared in [60]. The rotational voltage injection with constant amplitude is found with more significant fault features while simpler injection structure than the others. HF injection method is improved in [61] by replacing the zero voltage vector in the output of space vector PWM (SVPWM), which results in a spatial asymmetrical distributed rotating HF voltage. The capability of locating ITSC and HRC faults with high frequency injection is further extended in [62].

Impedance can be calculated based on signal injection. It can be seen as another perspective to evaluate the ITSC fault. Qi et al. [63] find that ITSC fault and eccentricity have different influence on the impedance over the whole region of magnetic working points, which can be used to distinguish these two faults. Another different way of utilizing impedance information is presented in [64], where the resistance of phase winding is used as the fault indicator. When ITSC fault occurs, the resistance of corresponding phase reduces, leading to reduction of d-axis resistance, which is estimated by injecting DC voltage at standstill condition.

d) Others

Apart from the aforementioned methods, there are various ways of extracting fault signatures from current and voltage. Hang et al. [65] provide a novel method that uses the differences among three phase current amplitude as the fault indicator. Obeid et al. [66] manage to detect the distortion in phase current or voltage waveforms caused by intermittent ITSC fault using adaptive wavelet transform. Cost function of the model predictive control system is utilized as a fault signal source in [67]. Discrete wavelet transform (DWT) is adopted to analyse the cost function signal, since it is non-stationary signal. ITSC fault is detected by monitoring the normalized energy-related feature vector calculated from the wavelet transform coefficients. Skewness of currents [68], current envelope [69], model signal of phase currents [70], are also found effective for detection of ITSC fault.

2.2.2. Magnetic Signals

Unlike electrical-signal-based methods, magnetic-signal-based fault detection methods extract fault signatures through the magnetic field. These methods provide a different view of the ITSC fault. As shown in Figure 6, different mounting places of sensors require different level of modification to the machine. According to the significancy of modification, magnetic-signal-based methods can be classified into two kinds, i.e., invasive methods and less-invasive methods. Since the flux distribution varies in space, the signals captured in different places are certainly different, and thus, different fault features are discovered with these two kinds of methods.

a) Invasive methods

Invasive methods place the sensors on the main flux path of the machine, as shown in Figure 6 (a). Thus, sensors are usually placed in the slots [71,72], stator yoke [73], around the tooth coils [74], etc., meaning that significant modification of the machine is required. Da et al. [71] utilized the search coils on all tooth for the detection of various faults. The polar diagram is then drawn to see the distortion of airgap flux linkage due to ITSC fault. Similarly, Zeng et al. [75] also use search coils wound on tooth. The frequency components (2h_k±1)f_e (h_k is the carrier wave ratio) are found having high SNR, and a fault index based on these two components is proposed.

On the other hand, only 6 search coils are required in [72] by setting the coil pitch of search coils the same as the coil pitch of phase windings. The NSC with frequency of 2f_s±f_e component is extracted as the fault indicator, where is f_s the switching frequency. Furthermore, a universal design method for search coil structure is developed in [76] considering the reliability, sensitivity, SNR, and positioning capability of the induced voltage.

b) Less-invasive methods

Compared with the invasive methods, less-invasive methods make less significant modifications to the machine. The sensors are usually placed at the end region or outer side of the stator back iron, as shown in Figure 6 (b). Liu et al. [77] build and analyse the magnetic equivalent circuit model of a PMSM and found that the stray flux at stator back side changing with the magnetomotive force (MMF) of windings. Thus, tunnelling magneto-resistive (TMR) sensors are placed at the stator back side to obtain the stray flux density. Polar diagram is drawn to discover the distortion in spatial distribution. A similar method using only two search coils at symmetrical position of stator back side is proposed in [78]. Pearson correlation coefficient is adopted to analyse the induced voltage in two search coils. Gurusamy et al. [79] specifically use the 3^rd harmonics in stray flux to detect ITSC fault in PMSMs because the fundamental frequency component may lead to false negative error. Eldeeb et al. [80] propose a different method by utilizing the switching frequency components in the stray flux, where an antenna placed at relatively far away from the machine is used to extract the switching frequency and its odd number multiple components.

The housing of the machine may affect the stray flux distribution at the stator back side. In such case, placing sensors at the end region is an alternative way. Assaf et al. [81] analyse the spectrum of the axial stray flux as well as the sensitivity of different frequency components to ITSC fault and unbalanced supply. Lamim et al. [82] place two stray flux sensors at the end region, which are orthogonal to each other. Orbit diagram is used to illustrate the distortion of leakage flux linkage. Kumar et al. [83] combine temperature and the leakage flux density together for ITSC fault.

2.2.3. Other Signals

Other signals are also utilized for ITSC fault detection, such as mechanical signals [84], and temperature signals [83,85]. For example, since the ITSC fault leads to harmonics in currents, it can be expected that these will excite corresponding mechanical response. Hence, the 4^th harmonics in speed signals [37,86,87] and vibration signals [84] are investigated for ITSC fault detection.

2.2.4. Comparison

Table 1 concludes the general advantages and disadvantages of different kinds of signal-methods discussed in this section.

2.3. Model-Based Methods

Instead of explicitly extracting certain time or frequency domain features from signals, model-based methods utilize the mathematical or finite element method (FEM) models of PMSMs to estimate fault-related quantities. Compared with signal-based methods, model-based methods rely more on the machine parameters, but are usually more suitable for transient conditions.

To better reveal the essential differences between these methods, in this paper model-based methods are classified according to their observed signatures. For ITSC fault detection, the commonly used signatures include estimation residual, estimated shorted turn ratio, etc., as shown in Figure 7.

2.3.1. Estimation Residual

Healthy PMSM models are commonly used in this category of detection methods. When the ITSC fault occurs, the model of real machine becomes different from the assumed healthy model, causing errors in the estimation results. Hence, the estimation errors can be used as a fault indicator.

Three phase current residuals are estimated with open-loop calculation in [88]. Meanwhile, a αβ current observer with feedback is adopted in [89] to calculate current residuals considering the non-linearity of inverter model and unbalanced inductance. Mazzoletti et al. [90] further consider the influence of parameter errors. It is proved that NSC in current estimation residual is independent to parameter errors, hence being used as fault indicator to reduce the influence of the parameter variation. The NSC of the residual is also selected as the fault indicator in [91], but the current estimation is achieved with the assistance of FEM so that saturation effect can be considered. The FEM model provides the relationship between dq-axis flux linkages and currents. Similar principle is adopted in [92] while the influence of working conditions is mitigated with a LUT. Mahmoudi et al. [93] use the Luenberger observer to estimate the dq-axis currents. The NSC in the estimation residual is still used as the fault indicator. A full-order Luenberger observer [94] and the extended Kalman Filter [95] have also been investigated for residual calculation. To better account for parameter uncertainty and nonlinearity, digital twin models are used to obtain the current residuals [96].

Similar to current residuals, voltage residuals are also utilized. Hang et al. [97] model the PMSM under the ITSC fault condition with two terms of voltage disturbances at dq-axis, , and use the Luenberger observer to estimate the voltage disturbances. The sum of two voltage disturbances is taken as the fault characteristics signal, and the 2^nd harmonics in it is used as the fault indicator. Similarly, Du et al. [98] estimate the disturbance in back-EMF with an extended state observer and extract the 2^nd PSC in the disturbance as the fault indicator. A five-phase PMSM is investigated in [99], where the voltage disturbances in all phases are estimated.

Other than currents and voltages, various physical quantities can also be estimated to monitor the deviation of machine model due to occurrence of fault. Sarikhani et al. [SAR13] estimate the back-EMF of PMSM, and similarly, take the residual of estimation as the fault indicator. However, the fault index is set to the linear average of residual normalized against speed. Cui et al. [100] estimate the electromagnetic torque using torque equation at healthy condition. The estimation error is obtained by subtracting estimation value with the real value measured with torque transducer. Through theoretical analysis, the DC component in residual is found only exist when the ITSC fault occurs, thus being used as the fault indicator. Xu et al. [101] analyse the model of sensorless SPMSM with ITSC fault considering the inverter nonlinearity and the current measurement error. The residual voltages in estimated dq-axis contain various harmonics when ITSC fault happens. With proper harmonics elimination algorithm, the 8^th harmonics in voltage estimation residual is selected for ITSC fault detection.

Upadhyay et al. [102] estimate the flux linkage in dq-axis and draw the XY diagram to observe the trace of flux linkage. By extracting the DC component in the flux linkage trace variation, the fault can be detected and located.

For DTPPMSM, Yang et al. [103] utilize the difference of voltage vectors of two sets windings to detect the ITSC fault.

Furthermore, the model-based method can be combined with high frequency injection method for better SNR, by estimating the high frequency injection residual [104,105].

2.3.2. Estimated Shorted Turn Ratio

This kind of methods utilize the PMSM model with ITSC fault. The faulty model contains the shorted turn ratio as a parameter. Thus, the short-circuit ratio can be estimated with the information in machine voltages or currents, either through open-loop expressions [106], or close-loop observers [107,108].

Aubert et al. [107] build a PMSM model consisting of the standard PMSM voltage equations and the short-circuit loop voltage equation. Kalman filter is adopted to estimate the shorted-turn ratio of each phase. Sayed et al. [108] take similar way, but also compare the performance of extended Kalman filter and the unscented Kalman filter. Furthermore, Hidden Markov model is adopted in [109] to estimate the range of shorted turn ratio and short-circuit resistance.

2.3.3. Others

Except for the estimation residual and the estimated shorted turn ratio, several other fault indicators have also been investigated, such as estimated torque [110], the probability distribution function of estimated machine parameters [111], speed estimation residual [112], etc.

2.3.4. Comparison

Table 2 concludes the general advantages and disadvantages of different kinds of model-based methods discussed in this section.

2.4. Data-Based Methods

Compared with signal- and model- based methods, data-based methods generally use a significantly larger amount of machine operating data. Fault signatures in the data are extracted and analyzed implicitly through machine learning process, while the machine model and explicit fault signatures are usually less important.

Data-based methods can be classified according to the signals they used: electrical signals, magnetic signals, and other signals, as shown in Figure 8.

2.4.1. Electrical Signals

Machine learning algorithms put no limit on the input data forms. Some algorithms such as CNN or recurrent neural network (RNN) can directly process the time-series data, and the implicit fault features are automatically learnt from the training data. In contrast, fault signatures in the time series data can also be processed with signal analysis tools such as FFT, and the fault-related information in the results can be subsequently extracted with machine learning algorithms. Thus, the methods are further classified according to the forms of input data of machine learning algorithms.

a) Time series data

Data in the form of time sequence can be directly fed into the machine learning algorithms. Thus, the fault signatures are automatically determined and extracted with machine learning process, and then being analysed and classified. In [113], three phase voltage and current signals are directly input into the proposed framework. Multiple feature extraction methods, as well as multiple classification are used in parallel and synergised with the Fisher’s ratio. Wang et al. [114] manage to detect ITSC fault with two of three phase currents and the deep autoencoder. Furthermore, only one phase current is required in [115].

A wide range of existing methods apply the convolutional neural network (CNN) for fault detection. The convolution layer usually plays a role as the feature extraction methods in the machine learning structure. Compared with traditional spectrum analysis tools such as FFT, the corresponding features are identified and extracted with convolution layer self-adaptively according to the training data, because the convolution core is obtained through training process. In [116], a method based on CNN is proposed, where time series of three-phase currents compose a 3×n array as the input of CNN. Shih et al. [117] compare the performance of SVM and CNN on ITSC fault detection. Time series of q-axis voltage and current are used as the input of SVM, while the waveforms of them are converted into 2D image and fed into CNN. A conclusion is made that SVM which is assisted with mathematical model of PMSM requires much less data. On-Bayesian-optimization-based residual CNN is applied in [118] to reduce network depth and avoid degradation effect when feeding time series data into CNN. A residual dilated CNN is applied in [119] combined with transfer learning techniques. Some other applied methods such as deep Q-network [120] and stacked autoencoder [121], also have one or multiple convolutional layers.

Apart from CNN, transformer neural network [122], RNN [123], etc., can also learn the features from time-series data.

b) Symmetricalcomponents

Instead of directly feeding three phase current signals, NSC and PSC are used as input for a neural network in [124]. Similarly, an attention-based RNN is adopted in [123] to analyze the NSC, PSC and speed. Meanwhile, Pietrzak et al. [125] take NSC and PSC as the input of the K-Nearest Neighbour (KNN) algorithm.

In [126], an efficient method based on stacked sparse autoencoders and Siamese networks is proposed to reduce the amount of data required for training. Seven kinds of signals and features are integrated into the dataset, including three-phase current, NSC and PSC in current, NSC impedance, and electronic torque.

c) Frequency pattern

Machine learning algorithm can also process the spectrum of three phase currents [127,128] or dq currents [129]. Specifically, Pietrzak et al. [130] use Bisepctrum analysis to pre-process three-phase currents. The result of Bispectrum analysis is a 2D image, which is then fed into a CNN for further feature extraction. Meanwhile, other time-frequency analysis techniques such as the wavelet transform [131] are also adopted.

d) Other features

The image of Park’s vector trajectory is used as the input of neural network in [132]. In [133], 15 features are extracted from current signals to build up the dataset, including mean value, maximum value, root-mean-square (RMS) value, etc. In [134], the difference between two of three phase currents are obtained to enhance the fault signatures.

The estimation residual of q-axis current based on Luenberger observer is used in [135] so that the information of PMSM model can be incorporated into machine learning process.

2.4.2. Magnetic Signals

Search coils are placed on tooth to capture the abnormality in airgap flux density distribution due to the ITSC , eccentricity and partial demagnetization fault in [136], where the induced voltage is processed into image by short-time Fourier transform. Meanwhile, the stray flux density in end region [137] and at the stator back side [138] are also investigated.

2.4.3. Other Signals

Apart from the electrical and magnetic signals, many other signals are also used for ITSC detection. For example, torque signal is combined with currents in [120]. Speed signal is combined dq voltage and currents in [129] to reduce the influence of controller bandwidth. Furthermore, the input current on the power side [139] and the vibration signals [140] are also investigated.

2.4.4. Comparison

Table 3 concludes the general advantages and disadvantages of different kinds of model-based methods discussed in this section.

3. Partial Demagnetization Detection

3.1. Background

Demagnetization of a magnet means the magnetic property of this magnet degrades. Once the applied field strength exceeds the knee-point of the B-H curve of a PM, as shown in Figure 9, the working point of the PM will not recover following the original B-H curve but will follow the recoil line. This means that the remanence is reduced from B_r1 to B_r2, in other words, the PM is demagnetized. Generally, high temperature and large armature field are the main cause of demagnetization fault [141]. The demagnetization severity of a PM is usually reflected in its remanence B_r, which decreases as the severity of demagnetization increases.

In the area of fault detection, PM demagnetization faults are usually classified into two categories: uniform demagnetization (UD) and PD. UD means all PMs in a PMSM are demagnetized uniformly, while PD means only one or a few PMs are demagnetized, and the severity of demagnetization can be non-uniform. In recent research, PD detection gains more attention due to its commonness and difficulties in detection. Hence, this paper concentrates most on PD detection rather than UD detection.

PD causes unwanted harmonics in the back-EMF and currents, as well as torque ripple, unbalanced magnetic pull, and vibration [3], leading to deterioration in performance and efficiency of the PMSM drive system.

Detection of PD has been widely investigated in the past few decades. The proposed methods can be classified into signal-based, model-based, and data-based, similar to the ITSC detection methods.

3.2. Signal-Based Methods

Classification of signal-based methods for PD detection is shown in Figure 10.

3.2.1. Electrical Signals

a) Symmetrical components

Similar to ITSC fault detection, symmetrical component analysis can also be applied in PD detection. The mostly used symmetrical component in PD detection is ZSC. Urresty et al. [143] discuss the influence of winding configurations on the spectra of currents and ZSVC of a PMSM with demagnetization fault. It is proved that harmonics appear in both current spectrum and ZSVC spectrum in fractional-slot PMSM in the case of PD. But in integral-slot PMSM, new harmonics may not emerge in current or ZSVC spectrum, depending on the specific winding configuration. It is also discovered in [144] that fractional harmonics in current spectrum related to PD might be cancelled in phase current with certain winding configurations. However, the ZSVC will not be eliminated in such winding configuration, hence eligible for PD detection. Furthermore, Zhan et al. [145] extend the ZSC-based methods to DTPPMSM. The drop of the 3^rd harmonics in the difference of ZSVC in two set of windings is chosen as the fault indicator for an integer-slot machine, while for other types of machines, the fractional harmonics are used.

b) LF pattern

Fractional harmonics are very frequently used in demagnetization fault detection. When PD occurs, the frequencies of harmonics emerge in three phase currents can be expressed as (1) [146]:

$$f_{PD}=f_e\left(1\pm {\displaystyle \frac{k}{p}}\right),k=\mathrm{1,2},3,\cdots ,$$

(1)

where f_e is the electrical frequency, p is the pole-pair number, and k is any positive integer.

According to the pole number of a certain machine, the corresponding fault harmonics can be obtained FFT [147]. However, at transient condition, the fundamental frequency components interact with fault-related harmonics [148], causing false alarm. Thus, advanced signal processing techniques are adopted to track and analysed the harmonics such as wavelet transform [149], box-counting algorithm [150], Vold-Kalman filter [151], T-f decomposition [148,152], HHT [153], etc. Other than harmonics in the phase currents, harmonics in dq currents are also evaluated for PD detection in [154].

The capability of a certain harmonics to distinguish PD from eccentricity is paid special attention. In [147], the 2/3^rd and 4/3^rd harmonics emerge in stator spectrum of a 6-poles SPMSM at both PD and dynamic eccentricity fault condition, thus unable to distinguish them. On the other hand, the 1/4^th and 1/2^nd harmonics are found able to distinguish PD from static eccentricity [155] in a 9-slot 8-pole SPMSM. Amplitude and phase angle of the (1-1/p)f_e harmonics are used together to distinguish PD and eccentricity fault. Furthermore, Naderi et al. [25] systematically analyse the homopolar frequency components in stator current and conclude the components with the capability to distinguish between PD, eccentricity, and ITSC fault.

The PD-related frequency components are studied comprehensively in [156,146] considering pole-slot combination, double-layer and single-layer winding, as well as skew angle. It is found that the harmonics at triple multiple of mechanical frequency may disappear due to winding configuration. Specifically, in the case when the following criteria is satisfied

$$n_{layer}\times {\displaystyle \frac{n_{slot}}{2p}}=3k,k=\mathrm{1,2},3,\cdots ,$$

PD does not create harmonics in stator currents, where n_layer, n_slot, and p is the number of winding layers, slots, and pole-pairs, respectively. A systematic analysis is presented in [157] about the harmonics in back-EMF at PD and eccentricity conditions.

Apart from harmonics, the variation of current angle [43] and torque angle [44] are found feasible to distinguish PD and ITSC faults, and furthermore, the torque angle can be combined with the voltage vector amplitude to further discriminate PD, SE, and ITSC faults [42].

Methods based on current spectrum, Park’s vector approach (PVA), extended Park’s vector approach (EPVA) are compared in [158] and the EPVA is found superior in terms of high sensitivity and robustness.

Furthermore, the signatures in the 5^th harmonics subspace voltage vector of the DTPPMSM are investigated [159].

c) HF pattern

Generally, very few papers utilize the HF pattern in voltages or currents for PD detection. However, it is found in [160] that the demagnetization can change the saturation level of the machine, and thus, it can be detected by inspecting the d-axis incremental inductance. It is also claimed that this method can detect PD fault in the case where machine current signature analysis (MCSA) cannot.

d) Others

Instead of using spectrum analysis, Hong et al. [161] use the waveform of currents to detect PD at standstill condition, essentially utilizing the impact of PD on the saturation level of iron core. It is also worth mentioned that Fernandez et al. [162] and Reigosa et al. [163] managed to utilize the magnetoresistance effect of PM to estimate the magnetization state of PM.

3.2.2. Magnetic Signals

a) Invasive methods

As discussed previously [146], some PD signatures in currents or voltages may not emerge depending on the topology of PMSM. Magnetic-signal based methods can get around that limitation through extracting the fault signatures directly from the flux or flux density.

The radial airgap flux density is analysed at different working conditions in [164], and Hall sensors are placed in the airgap to obtain the corresponding signatures. Search coils wound on each tooth are adopted in [71], and the first-order harmonics amplitudes in each search coil are used. This method is improved in [165] by introducing a more robust fault index, which subtracts the average value of induced voltage in the range of 120° mechanical angle from the original induced voltage waveforms. Multiple harmonics are used to evaluate the severity of PD in [166]. The abovementioned methods require search coils with only one tooth pitch wound on each tooth, which might be too complicated for PMSM with many slots.

A method [167] is proposed by using two search coils placed on the tooth and facing the airgap. Four search coils are placed at the slot opening in [168]. In [169], six search coils with one pole pitch are placed in stator slots of a 36-slot 4-pole machine. The induced voltage in the search coils is found containing (2k+1±2/p)^th order harmonics when PD occurs. Rafaq et al. [170] find out that for PD detection, only one search coil on the tooth is essentially enough, because the demagnetized PM can be continually scanned by the search coils when the rotor is rotating. Skarmoutsos et al. [171] place a search coil in slots with coil-pitch being exactly four pole pitch. Consequently, the induced voltage will be zero at healthy condition, resulting in a better signal-to-noise ratio (SNR). When PD occurs, the peak-to-peak value of induced voltage is used to calculate the fault index.

Besides in [172], the search coils are wound at the stator yoke. Two search coils are placed at the bottom of two slots, whose distance is exactly one pole-pitch and connect with each other in a way to makes sure the ideal output is zero at healthy condition. Chen et al. [173] improved this method by adding another search coil which is also one pole-pitch away. Totally 8 search coils on the stator yoke are used in [73] to achieve discrimination between the PD and ITSC faults. Furthermore, the search coil on the rotor is also investigated in [174].

In [175], a special type of demagnetization was discussed, where the trailing edge of PM is demagnetized. It is demonstrated that monitoring airgap flux has the detectability of this type of demagnetization.

b) Less invasive methods

The stray flux at the stator back side contains similar harmonics and signatures as airgap flux [79,176]. Goktas et al. [177] apply two flux-gate sensors located behind the tooth and behind the slots respectively. A circumferentially equidistant array of eight TMR sensors are placed outside the stator in [178]. Time domain fault index based on the envelope and the average of stray flux density is calculated, which shows better performance than methods based on machine current signatures.

The stray flux in the end region can also be utilized. Hall sensors placed at the end region close to the PM are used to detect demagnetization [179]. The method is improved in [180] using the ZSC flux in the end region, which presents higher sensitivity. Park et al. [181,182] reduces the amount of Hall sensors to one, and further discuss the feasibility of fault detection with digital Hall sensors. The basic principle is that if the rotor PMs are not symmetrical due to local demagnetization or damage, the flux measurement of the Hall sensor will be smaller when the demagnetized PM passes the Hall sensor.

3.2.3. Other Signals

PD causes harmonics in the spectrum of currents and voltages, consequently causing harmonics in torque, i.e. torque ripple. In [183], the amplitude of (λ±ξ/p)^th harmonic in torque is chosen as the fault index, where λ and ξ are integers, p is the pole-pair number. Furthermore, the 0.25^th harmonic in vibration signal is used in [184].

3.2.4. Comparison

Table 4 concludes the general advantages and disadvantages of different kinds of signal-based methods discussed in this section.

3.3. Model-Based Methods

Classification of model-based methods for PD detection is shown in Figure 11.

3.3.1. Estimation Residual

In [185], the airgap flux density is estimated in real-time through the analytical equations assisted with offline FEM model and compared with the measured signal obtained with Hall sensors. The deviation is then processed with wavelet transform to extract the target frequency, followed by a classifier. Bisschop et al. [186,187] managed to build up the analytical expression of the induced voltage in search coils. Then, the predicted induced voltage is compared with measured value, and the deviation is used to evaluate the severity of PD.

3.3.2. Estimated Rotor Flux

Roux et al. [188] estimate d-axis flux linkage with the voltage equation of PMSM and monitored the decreases in the estimated value. Moon et al. [189] present a method to estimate the PM flux and the L_d, L_q at the same time. The variation in PM flux is combined with deviation of L_d and L_q to detect PD. Zhu et al. [190] estimate the rotor flux by using torque ripple as the input of rotor flux estimation algorithm. Liu et al. [191] also estimate the rotor flux, while simultaneously monitoring the HF d-axis inductance to discriminate the eccentricity and the PD faults. Han et al. [192] further take the variation of parameter into consideration, and successfully separated the estimated quantity irrelevant to parameter error. Rotor flux linkage can also be estimated with PWM voltage measurement [193] to exclude the influence of inverter nonlinearity.

For DTPPMSM, PD detection can be achieved by monitoring the flux linkage in the 5^th harmonics subspace [194].

3.3.3. Comparison

Table 5 concludes the general advantages and disadvantages of different kinds of model-based methods discussed in this section.

3.4. Data-Based Methods

Classification of data-based methods for PD detection is shown in Figure 12.

3.4.1. Electrical Signals

Skowron et al. [195] target at detect the ITSC fault and the PD fault at the same time. The raw current signals are constructed as a 20×25×3 array and fed into a CNN. Similar principle is adopted in [196], while transfer learning technique is integrated into the diagnosis framework. Chen et al. [135] also adopted CNN to distinguish ITSC and PD, but processed the current and voltage signals with model-based method, and fed the CNN with q-axis current residual. Other than CNN, linear discriminant analysis and principle component analysis are combined with variational autoencoder in [73] to extract implicit features in current signals.

The method proposed in [197] convert both voltage and current signals into images, and then process them with local binary pattern texture analysis. The extracted features are then classified by KNN algorithm. Bispectrum is adopted to convert the three phase currents into images in [198] to detect eccentricity and PD faults. Continuous wavelet transform (CWT) is combined with ResNet in [115], and the short-time Fourier transform (STFT) is combined with KNN and multilayer perceptron in [199].

It is worth noting that because of the strong capability of machine learning algorithm, a lot of papers investigate the discrimination among different fault types based on voltage and current signals, such as ITSC and PD, [115,131,135,200,201], eccentricity and PD [198,200,202,203], etc.

Since demagnetization of PM is sensitive to temperature, the influence of PD on machine temperature is analyzed in [204], and the temperature of the shell and winding are used in addition to current, torque, and speed signals.

3.4.2. Magnetic Signals

In [205,206,207], airgap flux density is measured with three gauss meter probes located in different position in airgap of a double side linear PMSM. In [205], complex CWT and Teager-Kaiser energy operator are used to extract the fault feature, while in [206], time-time-transform is used. Random forest and extreme learning machine are adopted as classifier respectively. Other than using flux density measuring sensors, search coils wound on tooth [136] and yoke [208] are also adopted to capture the distortion in airgap flux density.

In [209], stray flux signal sampled outside the stator back-iron is converted into symmetric dot pattern image and analysed by wavelet scattering convolution network and semi-supervised deep rule-based classifier. Similar stray flux sensors are placed in [210], while the sampled signals are processed with the local outlier filter and deep Q-network.

3.4.3. Comparison

Table 6 concludes the general advantages and disadvantages of different kinds of data-based methods discussed in this section.

4. Eccentricity Detection

4.1. Background

Eccentricity refers to the misalignment of the geometrical center of rotor and stator, resulting in an unevenly distributed airgap. Many factors can cause eccentricity, such as manufacture tolerance, bearing faults, rotor deformation, etc. Generally, there are 3 types of eccentricity, including static eccentricity (SE), dynamic eccentricity (DE), and mix eccentricity (ME).

SE occurs when the rotating center of rotor is not aligned with the center of stator O_s, while being aligned with the geometric center of rotor O_r. As shown in Figure 13 (a), SE results in an unevenly distributed but time-invariant airgap.

On the other hand, DE means the rotating center of rotor is aligned with the center of stator O_s but misaligned with the geometric center of rotor O_r. It can be seen from Figure 13 (b) that the position of minimum airgap rotates with rotor, but the average airgap length is still the nominal airgap length.

ME is essentially the mixture of SE and DE.

With an uneven airgap, eccentricity causes harmonics in back-EMF [211] and currents, torque ripple, the unbalanced magnetic pull (UMP) [212], and vibration [213]. It is also reported in [214] that eccentricity has very significant influence on cogging torque in machines having 2p=N_s±1, where p is the pole-pair number and N_s is the number of stator slots.

4.2. Signal-Based Methods

Classification of signal-based methods for eccentricity detection is shown in Figure 14.

4.2.1. Electrical Signals

a) LF pattern

Fractional harmonics are commonly observed in stator currents and voltages when eccentricity fault occurs. The harmonics caused by DE can be expressed as (2) [147,215]:

$$f_{de}=f_e\left(1\pm {\displaystyle \frac{2k-1}{p}}\right),k=\mathrm{1,2},3,\cdots .$$

(2)

Meanwhile, the (1+2k/p)^th harmonics are found related to SE in [215] for the studied machine.

For DE detection, the 2/3^rd and 4/3^rd harmonics of a 6-poles BLDC are analysed and tracked with windowed Fourier transform [216], analytical wavelet transform [217], and quadratic time-frequency representation methods [218]. Meanwhile, frequency components with even multiple of mechanical frequency harmonics are selected as fault indicator for DE in [26], and found able to distinguish DE from ITSC faults. Discrimination of DE from PD is accomplished in [219] based on the amplitudes and phase angles of monitored harmonics. IPMSMs and SPMSMs are compared in [220] and found no impact on the harmonics existence. Similar frequency components are obtained in [221] through synchronous resampling to get better consistency at various operation conditions.

For SE detection, the 1/4^th harmonics is selected of a 9-slot 8-poleSPMSM in [155]. Skarmoutsos et al. [222] propose an algorithm to analytically calculate the fractional harmonics that exist in phase voltage and point out that the coil and pole number can influence the harmonic existence expressed in (2). It is also discovered in [211] the back-EMF is not affected by eccentricity in rotational symmetrical machines, leading to a reduction in fault harmonics.

Integer harmonics also emerge when eccentricity fault occurs. In [188], the NSC of the 7^th harmonic in currents is chosen as fault indicator for SE. The fundamental frequency component in ZSVC difference between two sets of windings in a DTPPMSM is used for SE detection in [223], while the sideband (1±1/p)^th harmonics are used for DE detection.

As mentioned in the methods of ITSC and PD faults, the variation of fundamental frequency components are used in [42] to discriminate ITSC, PD, and SE. Back-EMF at various speed and speed-fluctuation conditions is collected and used for DE detection in [224].

b) HF pattern

HF pattern can reflect the changes of inductance due to the fault. It is pointed out that variation in PM flux results in the saturation degree in d-axis, and thus, changes the d-axis inductance [225]. A pulsating voltage at d-axis is injected at standstill condition to calculate the d-axis inductance, and consequently detect SE. The method is further extended [160] to non-standstill conditions and enabled to discriminate demagnetization and DE. A DC voltage in d-axis is further injected superimposed with a pulsating HF voltage, and incremental inductance is calculated with the response current. The increase and decrease of inductance compared with healthy machine are used as features for demagnetization and eccentricity faults. Furthermore, Liu et al. [191] utilize the mechanical frequency fluctuation of d-axis inductance to detect DE. Similarly, Aggarwal et al. [226] propose an off-line test method based on incremental inductance, while another criterion is added, that is the hump height of inductance curve versus current.

4.2.2. Magnetic Signals

Similar to the PD fault, the eccentricity fault features may be influenced by the machine topology [211,222]. In comparison, the magnetic signals directly monitor the field distribution, while without relying on the signals filtered by the phase windings. Thus, magnetic signals have a significant advantage of universality in the detection of eccentricity.

a) Invasive methods

Search coils wound on each tooth are feasible to detect both SE and DE, and also able to distinguish them from each other [71,165,170]. In [165], two specially designed fault indicators are calculated from the induced voltages to separately distinguish PD and eccentricity faults. Meanwhile, the nominal peak values of the induced voltages are used as fault indicators in [170]. Furthermore, only one search coil in [227] is placed in slots, whose coil pitch is about the even number of pole-pitch to eliminate induced voltage at healthy condition. In [228], Hall sensors are placed at the slot opening. The NSC and the single-side-band components in flux density are used as fault signatures of SE and DE, respectively.

b) Less invasive methods

Sensors are mounted onto the stator back side of a synchronous generator to capture the stray field [229,230] and manage to distinguish SE and DE. An analytical method is developed in [231] to predict the stray flux at the stator back side at SE condition. Hall sensors are mounted at the end region in [182,232] to capture the distortion due to eccentricity. In [232], the RMS of induced voltages is compared among the four search coils. It is proved that digital Hall sensor is also feasible as a cheaper alternative of analog Hall sensor [182].

4.2.3. Other Signals

Cogging torque is found to contain signatures of eccentricity in [233,234]. It is also stated in [235] by measuring the UMP, discrimination can be achieved between SE and DE. Furthermore, the vibration and acoustic noise caused by eccentricity is investigated in [236] and [237], respectively.

In [238], the temperature asymmetry of the whole machine is utilized for SE detection. The asymmetry in iron loss distribution is discovered when SE exists, resulting in the asymmetry of temperature.

4.2.4. Comparison

Table 7 concludes the general advantages and disadvantages of different kinds of signal-based methods discussed in this section.

4.3. Model-Based Methods

According to the criteria adopted in this paper, very few methods for eccentricity detection are classified as model-based methods. Thus, this category of methods is omitted here.

4.4. Data-Based Methods

Classification of data-based methods for eccentricity detection is shown in Figure 15.

4.4.1. Electrical Signals

Ebrahimi et al. [215] extract the sideband frequency components from stator currents, and KNN classifier is cascaded with an artificial neural network (ANN) to distinguish SE, DE and ME. Furthermore in [239], the fault features of SE and DE are extracted from the harmonics around the fundamental frequency component and classified with SVM. Similar harmonics are used in [240], while the harmonics are obtained with wavelet analysis. The classification is achieved with KNN, and the evaluation of severity is achieved with SVM. In [198], three phase currents are processed with Bispectrum analysis and converted into images. Specifically, the information in different frequency ranges is separately extracted and transformed into images to minimize the invalid information for ME detection.

In [200], the time and frequency indices are calculated with dq currents and fed into different outlier detection methods, including Isolation Forest, SVM, and Robust Covariance Ellipse. The results of these methods are ensembled by Majority Voting Ensemble techniques.

The eccentricity fault in DTPPMSM is analysed in [203]. The voltage angle increases at SE conditions compared with healthy conditions. Thus, the voltage angle is then processed with multivariate regression analysis and ANNs for fault classification.

A method is proposed in [241] using generative adversarial network (GAN) to solve the problem of data lacking for data-based eccentricity detection methods. GAN is employed to generate back-EMF data similar to the actual data, based on the data calculated from the analytical model. Furthermore, in [242], a new fault diagnosis framework is proposed where only healthy data is required. The GAN is used to combine the data provided by ideal mathematical and FEM model of PMSM with eccentricity and the actual data, and eventually provides accurate prediction of back-EMF at different eccentricity degrees.

4.4.2. Magnetic Signals

In [136], the ITSC, PD, SE, and DE can be successfully detected and separated based on the search coils wound on tooth. The SqueezeNet is adopted to capture those fault features in the induced voltages that can distinguish these faults from each other.

In [202], six search coils placed in slots are used as the signal sources, and the amplitudes of the fractional harmonics in the induced voltage are classified based on Random Forest, achieving an accurate detection of DE and discrimination from PD.

In [243], the stray flux in the end region of a linear PMSM is collected with TMR sensors. Gramian Angular Field and Markov Transition Field are used to convert 1-D data into 2-D images, which is then processed with fusion feature extraction algorithms and neural networks.

4.4.3. Comparison

Table 8 concludes the general advantages and disadvantages of different kinds of data-based methods discussed in this section.

5. Discussion of Detection of Three Types of Faults

5.1. Signal-Based Methods

The ITSC, PD, SE, and DE all introduce unwanted harmonics in currents but with different frequencies. However, the PD, SE, and DE fault-related harmonics are much more sensitive to the machine topology compared to the ITSC fault. This is because the windings act as filters to the harmonics introduced by rotor faults. In contrast, it is more difficult for the windings to filter through the harmonics caused by the faults in themselves. Consequently, it can be observed that magnetic signals have higher SNR and better universality for rotor faults detection.

5.2. Model-Based Methods

The model-based methods are much more popular in the area of ITSC fault detection than the other two. This could be due to the phenomenon stated before, that is, in some cases the information in the stator currents and voltages are not enough.

5.3. Data-Based Methods

Data-based methods have gained more and more attention in recent years. It can be observed that the recent trend in data-based methods is integrating multiple types of signals, i.e., sensor fusion, and distinguishing different kinds of faults. These tasks are very difficult for traditional techniques due to the highly non-linear and uncertain relationship between the fault signatures and the faults, especially in practical applications. However, since machine learning algorithms, especially those deep learning algorithms, have very strong representation capability, this relationship can possibly be learnt with enough data and appropriate algorithms.

6. Conclusion and Future Work

The state-of-the-art techniques of fault detection methods of PMSMs are comprehensively reviewed in this paper. Three major faults are covered, i.e., the ITSC fault, the PD fault, and the eccentricity fault. The existing methods are classified into signal-, model-, and data-based methods, and further categorized according to the signal types they used. Then, the existing methods are discussed in detail, and special attention is paid to the fault signatures they use. Subsequently, comparison is conducted between methods with different signal sources and fault signatures, as well as the methods for three types of faults. Generally speaking, signal-based methods are relatively simple and low computational burden, while difficult to suppress transient false alarm. Model-based methods are more suitable for transient working conditions, but at the cost of higher computational burden and lower SNR. Data-based methods are the hottest topic in recent years. They have advantages of high SNR and strong capability of distinguishing different faults. But the computational burden is usually very high.

According to the discussions and analyses in this paper, future research can be initiated in the following directions:

(1)

Improving the capability of distinguishing different faults. It has been widely investigated how to distinguish different faults, while very few methods with good universality are developed.

(2)

Reduction in the number of sensors. Much effort has been made to reduce the number of sensors used for magnetic signal and ZSVC collection. Further investigation can follow this direction and try to find a balance point between the detection capability and complexity.

(3)

Detection of faults in DTPPMSM. Compared with traditional three phase PMSM, DTPPMSM has more control degrees, and also more sampled current signals. Thus, potentially higher SNR can be achieved.