Condition Monitoring of Mechanical Faults in Induction Machines from Electrical Signatures: Review of Different Techniques Y. Gritli, Member, IEEE, A. Bellini, Member, IEEE, C. Rossi, Member, IEEE, D. Casadei, Fellow, IEEE, F. Filippetti, Member, IEEE and G-A. Capolino, Fellow, IEEE. Abstract— Condition monitoring of electric machines is a procedure of increasing importance, as fault tolerant systems are becoming mandatory in many applications. Early diagnosis of faults is a basic pillar in order to achieve full tolerance by the maintenance of faulty parts in electrical machines. As far as electrical drives are concerned, the share of mechanical faults (imbalances, gears and bearings) is very high. The mechanical fault detection is typically based on vibration signals, a robust and effective technique, that is quite invasive and with high latency. Recently, many theoretical and signal– based methods have been investigated for early diagnosis of mechanical faults by electrical machine signals. This paper will review methods for condition monitoring of mechanical faults, with special reference on those based on electrical signals and quite effective also for an early detection: generalized roughness and realistic incipient faults. Index Terms — Mechanical faults, Induction Machine, Fault Diagnosis, Bearings, Shaft, Gearbox, Eccentricity, Motor Current Signature Analysis, Torsional vibration.
I. INTRODUCTION Early stage fault diagnosis of Induction Machines (IMs) is an important research topic for cost and maintenance savings. IMs are being widely used in industry applications, mainly because their low price, ruggedness, efficiency and reliability. Recently, electrical drives are replacing hydraulic and pneumatic actuation in many processes, for their flexibility and performances. Large number of papers related to condition monitoring of IMs can be found in the literature [1]–[2], where different failure modes are analysed. Here, the focus is on mechanical faults detection which is a key item as they are the most frequent failures in IMs. The failures distribution within the machine subassemblies is reported in several reliability survey papers Y. Gritli, A. Bellini, C. Rossi, D. Casadei and F. Filippetti are with the Department of Electrical, Electronic and Information Engineering "Guglielmo Marconi", University of Bologna, 40126 Bologna, Italy (email:
[email protected];
[email protected];
[email protected];
[email protected];
[email protected]). Y. Gritli is also with the Department of Electrical Engineering, University of Tunis El Manar, LA.R.A., National Engineering School of Tunis BP 37, 1002 Tunis Belvedére, Tunisia. (e-mail:
[email protected]). G.-A. Capolino is with the University of Picardie “Jules Verne,” 80000 Amiens, France (e-mail:
[email protected]).
[3]–[4]. Four main classes can be identified: mechanical faults, stator related faults, rotor related faults, other faults (cooling, connection, terminal boxes). Mechanical faults can be roughly classified in bearing faults, shaft–load connection faults, gear faults and mechanical imbalances. Depending on the type and size of the machine, the bearing faults distribution varies from about 40% to about 90% from large to small machines. Recent studies considering the whole drive and including the mechanical load have revealed that the primary problems in IMs are: 1) rotor eccentricities; 2) rotor bars; 3) bearings damage [5]–[6]. Signal processing techniques are widely used for condition monitoring to overcome common issues [7]–[10]. Actually, condition monitoring of electrical faults based on space–vector (SV) of electrical signals is now well–known and established approach. This approach was used for diagnosing voltage supply unbalance [11], voltage source inverter and power switch faults [12]–[13], stator faults windings [7], [9], [14]–[19], rotor faults windings [7], [9], [15]–[16], [18]-[20], airgap eccentricity faults [21], or rotor cage faults [22]-[24]. Vibration signals are commonly adopted for mechanical faults detection in IMs and/or related mechanical loads. Generally, diagnosis techniques based on the analysis of the mechanical signals have shown good performances [25]– [32]. However, in harsh industrial environments mechanical signals cannot be acquired. Moreover, the typical latency of both sensing and condition monitoring of mechanical faults by vibration signals is quite high. Hence, condition monitoring based on electric signals would be preferable [26]–[29], [31]–[38]. A perfect diagnostic procedure should consider the minimum data acquisition from an electrical machine and establish a clear indication of incipient failures in a minimum time. Motor current signature analysis (MCSA) is the first option by the frequency analysis of current signals and a suitable signal processing leading to efficient fault detection in steady state at constant speed. The replacement of hydraulic and pneumatic actuators by electrical drive fosters the development of variable speed electrical machines. In time–varying operating conditions,
,(((
the MCSA fails as both speed and slip vary, and thus the sideband frequency components are spread proportionally to the speed variations. Under the above conditions, the time– frequency analysis is necessary, and more specifically, wavelet analysis can be tailored for efficient condition monitoring of mechanical faults under any time-varying operating conditions. This paper is organized as follows. Section II reviews the mechanical behaviour of radial bearings, eccentricity, and gear faults, and the corresponding diagnosis techniques based on vibration signature analysis. Section III reviews the use of MCSA for the detection of the same mechanical failures. Section IV reviews time–frequency signal processing methods for mechanical fault detection under time–varying conditions, whereas section V reports selected experimental results with realistic faults. The main conclusions are reported in section VI.
Similarly, faults in the inner raceway, outer raceway or rolling elements produce effects in vibration signals. Specific signatures appear in the vibration signal spectrum: (1)-(3), that are related to the ball diameter db, the pitch diameter dc, the number of rolling elements Nb, and the ball contact angle ȕ. The signature related the inner or outer raceway faults fi+,o- is expressed as (1), cage faults fca in (2), and rolling elements faults in (3) [39]:
II. VIBRATION SIGNATURE ANALYSIS FOR MECHANICAL FAULT DETECTION
On the other hand, small or manufacturing tolerance defects produce slight torsional vibrations. Similar conditions occur for surface chemical corrosion, early pitting fault. These conditions are usually referred to as “generalized roughness”, that is the most common damage in rolling bearings [21]. It occurs in normal operating conditions and no specific fault signature is associated to it. In fact, it produces a frequency spreading of characteristics fault frequencies, that cannot be detected easily with spectral or envelope analysis. Many papers are dealing with condition monitoring for mechanical faults with different methods, most of them are referring to intense defects. Large mechanical imbalances appear as rotor asymmetry faults [40]–[41] which are commonly classified as eccentricity–related faults. In fact, eccentricity faults are very frequent failures in IMs. This category of faults can be classified as static, dynamic or mixed eccentricity [27]–[28], [42]–[50]. A static eccentricity occurs when the center of rotation is displaced. On the other hand, dynamic eccentricity occurs when the rotation cylinder is displaced, while the center of rotation is kept at its original position. In case of mixed eccentricity, both the rotation cylinder and the center of rotation are displaced from their original position. Any kind of eccentricity increases largely the bearing wear. Furthermore, the radial magnetic force, owing to the eccentricity, acts on the stator core and exposes the stator windings to excessive vibrations. Main papers [39]–[43] report that all types of eccentricity are related to both torque and speed oscillations. They also cause specific signatures in the vibration signal at frequency [44]:
Mechanical faults as bearing faults, shaft-load connection faults, gear faults and mechanical imbalances are usually monitored by vibration signals. The vibration monitoring is wide spread and relies on common standards, such as ISO 10816. Vibration signals are obtained by sensor placed on the external part of machine and they are quite reliable and mature. Vibration spectra can be analysed and some “signature” can be isolated that are related to specific mechanical problems: imbalances, misalignment, looseness, bent shaft and bearing problems, the latter being the most critical and likely. Any change in the mechanical structure of a rotating machine generates a periodical force variation that produces a “signature” in the vibration signal spectrum. In any type of mechanical faults, this process starts with very low intensity: cracks or pitting, and later degenerates in higher effects. The main challenge is to detect early phenomena, that might be below noise thresholds or hidden. As an example, radial bearings are running with periodic changes in pressure and friction, leading to failures. Specifically, radial bearings are made by two concentric rings, separated by rollers. Rolling elements are bound by a cage that keeps a constant angular pitch and prevents any contacts. Under normal operating conditions, bearing defects often are caused by material fatigue. Typically, small cracks appear on tracks or on rolling elements. This initial degradations typically degenerates because of pitting and tearing off of the material and because of repetitive impacts of moving components. Specific vibration signal signatures are associated to intense defect. As an example, in case of pitting defects on the surface of a bearing component, local stiffness reduces and consequently, periodical impulses in the radial directions appear that change the vibration pattern.
fi +,o − =
§ d · fr N b ¨1 ± B cos(θ ) ¸ , 2 © dC ¹
(1)
f ca =
· fr § d B cos(θ ) ¸ , ¨1 − 2 © dC ¹
(2)
f ro =
§ § d ·2 · dC f r ¨ 1 − ¨ B ¸ (cos(θ )) 2 ¸ . ¸ 2d B ¨ © d C ¹ © ¹
(3)
f ecc = 2 f s ± f r .
(4)
Usually, faults are combined and there are interactions between the phenomena. Eccentricity can be caused by many problems such as bad bearings alignment during the motor
assembly, worn bearings, bent rotor shaft or large torque variations and torsional oscillations. A recent investigation on wind turbines reliability has shown that 70% of gearbox failures are directly related to bearings failures and 26% of failures are caused by gear faults leading to long downtimes [5]. The most common faults which can affect gear teeth are: abrasive wear caused by particle contamination or worn gear teeth particles; fatigue wear under normal stresses which eventually degenerates in cracks or fractures. A specific signature in the vibration signals is associated to this type of imbalances. Hence, a periodical assessment of vibration signal is a nice sensor of the degree of degradation of gear tooth, and of the gearbox and of the whole electromechanical system [51]. Among different techniques, vibration–based condition monitoring of gears is a well–known tool and it has proved its efficiency for fault detection, diagnosis, and prognosis of gears up to now [31], [52]–[55]. For healthy gearboxes, vibrations spectra show the main shaft frequencies, corresponding mesh frequencies fm and associated sideband frequency components given by [56]:
f side = f m ± f r ,
(5)
where:
f m = N teeth . f r .
(6)
is the teeth number and fr is the rotational frequency of the shaft being analyzed. Under faulty conditions, an increase of magnitude for fm and fside and their corresponding harmonics is expected. Eventually, condition monitoring of mechanical faults based on vibration signals suffer from the effects of noise caused by external mechanical excitations, and from measurement tolerances [56]. On the other hand, condition monitoring of mechanical faults by electrical signals (i.e. MCSA) uses noninvasive sensors already installed in the drive for control purposes. Hence, condition monitoring of electrical and mechanical parts can be realized at a very competitive cost. The feasibility of MCSA techniques will be detailed in the next section. III. ELECTRICAL/CSA FOR MECHANICAL FAULT DETECTION Condition monitoring of bearings based on electrical quantities is a widely investigated topic [8], [38]–[40], [44], [56]–[59]. However, while it is straightforward the physical link between bearings faults and vibration, the link between bearings faults and machine currents is still an open issue. In the literature, two different approaches were proposed. In the first approach, the physical chain is vibration, torque ripple, and eventually speed ripple [62]-[63]. Vibration produces additional torque components, which affect current spectrum, where a specific signature appears at frequencies:
f bec = f s ± k f r ,
(7)
According to the second approach, the vibration caused by bearing faults produces a static eccentricity [64]. The latter approach best fits to local defects and to severe bearing faults, while the former best fits to incipient bearing faults. According to the first approach, the fault can be modelled as an additional torque component with magnitude īload at of the mechanical characteristic frequencies. This component is additional to the load torque TL. Hence, the instantaneous torque applied to the motor is [7]: Tload ( t ) = TL + ΔTload ( t ) = TL + Γ load cos(ωcar t ),
(8)
This additional torque component produces a current and flux modulation at the same frequency. Hence, a relationship exists between current and torque ripples. With a suitable computation technique, this leads to a relationship between the current modulation and mechanical imbalances:
ΔI = I cos φ
Γload , K + ω J TL KT
2 T
2 car
2
(9)
The relationship (9) states that a signature appears in the machine current, that is related to a specific bearing defect, depending on the characteristic frequency. However, for real and incipient defects the magnitude of torque oscillations is very small and is negligible with respect to the load torque. Though the parameters KT and inertia vary with machine size and are difficult to estimate, the current modulation is typically lower than torque oscillation. Hence, it is hard to realize an effective condition monitoring of bearing faults based on current signals. On the other hand, it could be based on torque signals, provided that a torque meter with suitable sensitivity is available. In the literature this approach was investigated in a few papers [64], [65]. Condition monitoring of rolling bearings based on either vibration or torque signals is reviewed in [65]. Both techniques have shown interesting results for detecting faults at incipient stage. The ratio between current modulation and torque ripple (9) is inversely proportional to the characteristics frequency. Hence, current signal can be effectively used for condition monitoring of bearing faults on for those defects featuring low mechanical frequencies. The same achievements are reported in [64]. The use of current signals for condition monitoring of mechanical faults was extended to the diagnosis of gear faults with different approaches [38], [66]. In this framework, a one-stage gearbox was considered, for which specific signatures can be retrieved in the stator current spectrum around the supply frequency [66]. This method was successfully extended to condition monitoring of tooth breakage fault in multistage gearboxes. In this case, specific signatures appear in the stator current spectrum and are
related to the gear mechanical frequency components [67]– [69]. Moreover, a theoretical analysis was made to state the link between current signatures and gear mechanical fault frequency components [70]. Recently, condition monitoring of gear tooth surface damage (usually referred to as gear tooth localized fault) based on stator current SV instantaneous frequency has been proposed [71]–[72]. IV. REVIEW OF EXPERIMENTAL RESULTS A large number of papers include experiments, in order to assess the methods reviewed in the previous sections. These are mainly torque and current signatures analysis in the context of mechanical faults monitoring. For the sake of simplicity, two examples with experimental validation have been reported. The former is related to a case study of bearings faults and the latter to a case of faulty gears. Experiments are detailed in papers [8], [71], [73]. A. Bearings faults Figure 1 reports the torque spectrum for a small power induction machine (1.5 kW, 380 V, 50Hz) with a local defect
on the outer raceway introduced by chemical etching. A torque meter has been used in order to measure the torque ripple created by the bearing fault. The chosen torque meter uses the strain gauge technology, and a contactless torque signal transmission from the rotating shaft using a frequency modulation (accuracy is ±2 mNm). The healthy bearing was used as a reference compared to the faulty one. Supplying the machine at 50 Hz grid frequency, the torque ripple varies from about 1.5 mNm to 6 mNm in healthy and faulty conditions, respectively. Near rated load, the torque ripple caused by the bearing fault is independent of the load torque, and the current signature is within the range 0.2–0.5 mA at 50 Hz [8]. Hence, the fault signature is not visible as it would require uncommon resolution (less than 1mA) or either very expensive for industrial drives. Figure 2 shows the current SV spectra under healthy and faulty cases, with a supply frequency of 50 Hz. Spectra are normalized to the magnitude of the fundamental component, and the amplitude of negative component at –f is far lower than the positive components. Moreover, the amplitudes of direct rotating field components are similar (±3f), while negative components in counter rotating fields are higher (±3f). Hence, for single defects motor current signature
Fig. 1. Spectra of the torque signal for an IM supplied at 50 Hz, under healthy (top), and faulty bearing (bottom) cases [26].
Fig. 3. Spectra of the measured torque ac part: a) Healthy gear - b) Faulty gear [71].
Fig. 2. Spectra of the currents SV of an IM supplied at 50 Hz under healthy (top), and faulty bearing (bottom) cases [26].
Fig. 4. Spectra of the stator current space vector instantaneous frequency ac part: a) Healthy gear - b) Faulty gear [71].
analysis is not an effective fault detection method, since signatures are buried in the noise coming from the data acquisition. B. Gears faults Recently, the spectral content of the current SV has been used for the diagnosis of gear tooth localized fault (pinion fault) [71]-[72]. It has been shown by analytical computation that the current SV signature analysis (equivalent to the MCSA) contains fault signature information similar to the mechanical torque experimented by the driven electrical machine. Fig. 3 reports the torque spectrum for a small power three–phase squirrel–cage induction machine (250 W, 400 V) connected to a digital controllable brake through a one–stage gear with a number of teeth at the input Nr1=25 and at the output Nr2=75. fr1 and fr2 represent rotation frequencies at both input and output stages of the gear respectively. A torque meter with a 5 kHz frequency bandwidth is placed between the shaft of the IMs and the input stage of the gearbox and it has been used to measure the torque ripple created by the pinion fault. The rotation frequencies are fr1=23.1 Hz and fr2=7.7 Hz. In faulty conditions, components at k1=5,…,34 feature higher magnitudes in the alternative part of the torque signal with k1 being the harmonics order of the frequency fr1. Fig. 4 reports the spectra of stator current SV in healthy and pinion–faulty conditions. In faulty conditions, the higher magnitudes appear at nsv=6,...,10, being nsv the harmonic order of fr1=23.1 Hz (at rated load). It has been also shown that the application of the profile reconstruction method to the instantaneous frequency of the current SV leads to an effective tool for fault detection [71]. More specifically, after computing the discrete Fourier transform
(DFT) of the current SV, frequencies of interest kffp are filtered. The fault profile in the time domain is reconstructed using the inverse Fourier transform of the filtered current. In fact, the kffp frequency components exist also for healthy cases. Hence, the fault profile reconstruction can be realized exclusively using the most sensitive harmonics to the fault. Obviously, this procedure can be applied to each frequency range to track simultaneously different fault signatures. This leads to an effective tool for multiple fault identification in gearboxes. An example of a profile reconstruction for both pinion and wheel faults has been reported (Fig. 5) showing an effective discrimination of faults [71]. V. TIME–FREQUENCY METHOD FOR MECHANICAL FAULT DETECTION
As already mentioned, one of the most critical issues is the condition monitoring in time–varying conditions. Any mechanical fault detection by techniques recalled in the previous section can fail for two major reasons. Typically, mechanical faults degenerate quickly into generalized roughness. Thus, models detailed in the previous section are not valid as they assume that a single defect occurs. Then, for variable speed drives, fault signatures are spread in a frequency proportionally to the speed variation. Therefore, the MCSA will become inefficient.
Fig. 6. Spectra of axial vibration signals under large speed transient: Healthy (red) - Unbalanced mechanical load (blue) [30]. -a-
-b0.3
3000
a8
Speed (rpm)
3050
2950
0
2900
2850
0
5
10
15
-0.3
20
0
5
-c-
20
15
20
3000
a8
Speed (rpm)
15
0.3
2950
0
2900
2850
-0.3 0
5
10
Time (s)
Fig. 5. Computation of a fault profile from current–space vectors under 80% of the rated load: (a) Healthy gear - (b) Faulty gear (pinion) - (c) Faulty gear (wheel) [71].
10
-d-
3050
15
20
0
5
10
Time (s)
Fig. 7. Speed transient (a,c), approximation signal “a8” (b,d) resulting from WT of the axial vibration, for healthy and unbalanced mechanical load respectively [30].
The time–frequency analysis can solve this critical issue by adding the capability of fixing transient periods of time and recognizing them. As an example, the wavelet analysis can be used for the signal decomposition by using successive combination of approximation and detail signals [72]. The operation is successively repeated until the original signal is decomposed to a pre–defined number (J) of level decomposition. Frequency bands for each decomposition level are directly related to the sampling rate established by the well–known dyadic down sampling process. Hence, these bands cannot be changed unless a new acquisition with a new sampling frequency. Then, the fault detection based on the discrete wavelet transformation (DWT) is intrinsically inefficient for time–varying conditions. In [74], an efficient solution to overcome this limitation has been proposed after being effectively applied to the detection of electrical faults [75]. With a fixed sampling frequency (a few kHz), the 8th decomposition level has been adopted to cover the frequency bands in which the frequency component characteristics of the fault have been localised. Under constant speed operation and for healthy or faulty cases, the mechanical characteristic frequency is fixed in the spectrum. However, in time varying conditions (Fig. 6), the rotational component related to mechanical characteristic frequency is spread in a wide frequency range. A simple processing of the motor vibration signal vAX (t), allows shifting the fmec to the frequency bandwidth corresponding to a single approximation signal (eight). By doing so, the fault related information is insulated and confined in a single frequency bandwidth. Specifically, the motor vibration signal vAX(t) is demodulated by a variable carrier at the frequency fsl as detailed in [30]. The choice of fsl is made in each time slice according to the speed range of the transient. Then, the signal vsl(t) is analysed by the DWT where a specific fault signature can be retrieved. Fig. 7 shows that the difference between healthy and faulty conditions can be clearly stated leading to a useful tool for identifying specific fault signatures. It is also possible to insulate it from other misleading phenomena. This example use mechanical fault signals, but the methodology can be extended also for electrical signal based diagnostic systems.
mechanical frequency rate and it is effective quite only for severe faults. Time–varying conditions have been also investigated since they are very important in industrial cases. Recently, variable speed drives are becoming the standard option for many applications. Then, condition monitoring must be tailored for time–varying conditions. Here, time–frequency analysis methods have been reviewed and a solution based on wavelet transforms was reviewed. Assuming that the speed variation is known, the axial vibration signal can be operated in order to obtain a fixed signature strictly related to mechanical imbalances. Gearboxes faults were investigated also, though they are highly dependent on the applications and on the chosen kinematic chain. These faults can be grossly classified into two categories: uniform wear of the gear, which results in backlash; wearing of teeth of single gear part that results in imbalances. Moreover, signal processing method to detect the torque ripple generated by pinion–gear faults are reviewed. Signal processing is based on vibration or electrical signals and are based on spectral or time-frequency analysis. Space vector was successfully adopted for condition monitoring of electrical faults. Recently, it was extended to mechanical faults, leading to an effective condition monitoring under specific conditions. A general method for condition monitoring of mechanical faults based on space vector analysis has yet to be developed. Despite the large number of publication and research activities, there are still open issues on this topic. Effectively at this stage it is not possible to state a general purpose method which can be applied to a large number of fault types and machines. Each technique can be applied usefully in a suitable context, depending on operating conditions, the possible accessible fault signals and their sensitivity. REFERENCES [1]
[2]
[3]
VI. CONCLUSIONS The condition monitoring of mechanical faults is a critical issue for economical and safety reasons. This paper reviews recent advances in research for mechanical faults: bearings, gears and load. Mechanical faults can be modeled as eccentricities or torque ripples. Here, both methods have been reviewed. The former is more effective for large imbalances. Current signal or vibrations are commonly used for fault detection and the vibration analysis is more robust and effective while more invasive. Current signals can be used only for faults characterized by a relatively low
[4]
[5]
[6] [7]
A. Bellini, F. Filippetti, C. Tassoni, G.-A. Capolino, “Advances in diagnostic techniques for induction machines,” IEEE Trans. Ind. Elect., vol. 55, no. 12, pp. 4109–4126, Dec. 2008. ͒ G-A. Capolino, J-A. Antonino-Daviu, M. Riera-Guasp, “Modern Diagnostics Techniques for Electrical Machines, Power Electronics, and Drives”, IEEE Trans. Ind. Elec., vol. 62, no. 3, pp. 1738 – 1745, March 2015. ͒ O. V. Thorsen and M. Dalva, “A survey of faults on induction motors in offshore oil industry,petrochemical industry, gas terminals, and oil refineries,” IEEE Trans. Ind. Appl., vol. 31, no. 5, pp. 1186–1196, Sep./Oct. 1995. A. H. Bonnett, “Root cause AC motor failure analysis with a focus on shaft failures,” IEEE Trans. Ind. Appl., vol. 36, no. 5, pp. 1435–1448, Sep./Oct. 2000. S. Sheng, “Report on wind turbine subsystem reliability—A survey of various databases,” Nat. Renew. Energy Lab., Washington, DC, USA, Tech. Rep. REL/PR-5000-59111, Jun. 2013. P. Tavner, L. Ran, J. Penman, H. Sedding, Condition Monitoring of Rotating Electrical Machines, 2nd ed. Stevenage, UK, IET, 2008. H. Henao et al., “Trends in fault diagnosis for electrical machines: A review of diagnostic techniques,” IEEE Ind. Electron. Mag., vol. 8, no. 2, pp. 31–42, Jun 2014.
Z. Gao, C. Cecati, S. X. Ding “A Survey of Fault Diagnosis and Fault-Tolerant Techniques—Part I: Fault Diagnosis With ModelBased and Signal-Based Approaches”, IEEE Trans. Ind. Elec., vol. 62, no. 6, pp. 3757–3767, June 2015.
[23] A. Aboubou, M. Sahraoui, S. E. Zouzou, H. Razik, A. Rezzoug,
M. Riera-Guasp, J. Antonino-Daviu, G.-A. Capolino, “Advances in Electrical Machine, Power Electronic and Drive Condition Monitoring and Fault Detection: State of the Art,” IEEE Trans. Ind. Elec., vol. 62, no. 3, pp 1746 - 1759, March 2015.
[24] R. Salehi Arashloo, J. Luis Romeral Martinez, M. Salehifar, "A
[10] V. Fernandez-Cavero, D. Morinigo-Sotelo, O. Duque-Perez, J.
[25] J. R. Stack, T. G. Habetler, R. G. Harley, “Fault-signature
Pons-Llinares “A Comparison of Techniques for Fault Detection in Inverter-Fed Induction Motors in Transient Regime”, IEEE Access, vol. 5, pp. 8048 - 8063, May 2017.
modeling and detection of inner-race bearing faults,” IEEE Trans. Ind. Appl., vol. 42, no. 1, pp. 61–68, Jan./Feb. 2006. F. Immovilli, A. Bellini, R. Rubini, C. Tassoni, “Diagnosis of Bearing Faults in Induction Machines by Vibration or Current Signals: A Critical Comparison”, IEEE Trans. on Ind. Appl., vol. 46, no. 4, 2010, pp. 1350 – 1359. D. G. Dorrell, W. T. Thomson, S. Roach, “Analysis of airgap flux, current, and vibration signals as a function of the combination of static and dynamic airgap eccentricity in 3-phase induction motors”, IEEE Trans. Ind. Appl., vol. 33, no. 1, 1997, pp. 24 – 34. E. Tarkesh Esfahani, S. Wang, V. Sundararajan, “Multisensor Wireless System for Eccentricity and Bearing Fault Detection in Induction Motors”, IEEE/ASME Trans. on Mechatronics, vol. 19, no. 3, pp. 818 – 826, June 2014. P. Henriquez, J. B. Alonso, M. A. Ferrer, C. M. Travieso, “Review of Automatic Fault Diagnosis Systems Using Audio and Vibration Signals”, IEEE Trans. Sys., Man, Cyb. Sys., vol. 44, no. 5, pp. 642 – 652, May 2014. Y. Gritli, A. O. Di Tommaso, R. Miceli, C. Rossi, F. Filippetti, "Diagnosis of mechanical unbalance for double cage induction motor load in time-varying conditions based on motor vibration signature analysis," in Proc. IEEE-ICRERA, Madrid, 2013, pp. 1157-1162. H. Henao, S. Hedayati Kia, G-A. Capolino, “Torsional-Vibration Assessment and Gear-Fault Diagnosis in Railway Traction System IEEE Trans. Ind. Elect., vol. 58, no. 5, pp. 1707 – 1717, May 2011. T.G. Habetler, F. Kamran, R.G. Bartfield, “Motor bearing damage detection using stator current monitoring”, IEEE Trans. Ind. Appl., vol. 31, no. 6, Nov/Dec 1995, pp. 1274 – 1279. W. Zhou, T. Habetler, R. Harley, “Stator current-based bearing fault detection techniques: A general review,” in Proc. IEEESDEMPED, Sept. 2007, pp. 17–22. R. R. Shoen et al., “Motor bearing damage detection using stator current monitoring”, IEEE Trans. Ind. Appl., vol. 31, no. 6, Dec. 1995, pp. 1274-1279. M. Blödt, P. Granjon, B. Raison, J. Regnier. Mechanical fault detection in induction motor drives through stator current monitoring- Theory and application examples, Chapter in Wei Zhang, Fault Detection, INTECH, pp.451-488, 2010. S. Singh, A. Kumar, N. Kumar, “Motor Current Signature Analysis for Bearing Fault Detection in Mechanical Systems”, Procedia Materials Science, vol. 6, 2014, pp. 171-177. N. Feki, G. Clerc, P. Velex. "An Integrated Electromechanical Model of Motor Gear Units - Applications to Tooth Fault Detection by Electric Measurements," Mech. Syst. & Sig. Process., 2012, pp. 377–390. J.R. Ottewill and M. Orkisz. "Condition Monitoring of Gearboxes Using Synchronously Averaged Electric Motor Signals," Mech. Syst. & Sig. Process, 2013. R. R. Schoen, T. G. Habetler, F. Kamran, R. G. Bartfield, “Motor bearing damage detection using stator current monitoring,” IEEE Trans. Ind. Appl., vol. 31, no. 6, pp. 1274–1279, Nov./Dec. 1995.
[8]
[9]
“Broken bars and/or end rings detection in three-phase induction motors by the extended Park's vector approach”, in Proc. IEEECIEP’2004, pp. 128 – 133. novel broken rotor bar fault detection method using park's transform and wavelet decomposition”, in Proc. of the 9th IEEESDEMPED, Valencia, Spain, Sept. 2013, pp. 412 – 419.
[26]
[11] S. M. A. Cruz and A. J. M. Cardoso, “Rotor cage fault diagnosis
in three-phase induction motors, by Extended Park’s Vector Approach,” Elec. Mach. Power Sys., vol. 28, no. 4, pp. 289–299, Apr. 2000.
[27]
[12] A. M. S. Mendes and A. J. M. Cardoso, “Voltage source inverter
fault diagnosis in variable speed AC drives, by the Average Current Park’s Vector Approach”, in Proc. IEEE-IEMDC, pp. 704-706, May 1999. [13] V. Smet, F. Forest, J. Huselstein, F. Richardeau, Z. Khatir, S.
Lefebvre, M. Berkani, “Ageing and failure modes of IGBT modules in high temperature power cycling”, IEEE Trans. Ind. Elec., vol. 58, no. 10, pp. 4931-4941, October 2011 [14] S. M. A. Cruz and A. J. M. Cardoso, "Stator winding fault
diagnosis in three-phase synchronous and asynchronous motors, by the Extended Park’s Vector Approach", IEEE Trans. on Ind. Appl., vol. 37, no. 5, pp. 1227–1233, Sept./Oct. 2001. [15] Y. Gritli et al., "Experimental Validation of Doubly Fed
[28]
[29]
[30]
Induction Machine Electrical Faults Diagnosis Under TimeVarying Conditions", Elect. Power Syst. Res., vol. 81, no. 3, pp. 751-766, March 2011. [16] F, Vedreño-Santos et al.,
”Diagnosis of Rotor and Stator Asymmetries in Wound-Rotor Induction Machines Under Nonstationary Operation Through the Instantaneous Frequency”, IEEE Trans. Ind. Electron., vol. 61, no. 9, pp.4247–4259, Sep. 2014.
[17] Y. Gritli, L. Zarri, C. Rossi, F. Filippetti et al., “Advanced
diagnosis of electrical faults in wound rotor induction machines,” IEEE Trans. Ind. Electron., vol. 60, no. 9, pp. 4012–4024, Sep. 2013.
[31]
[32]
[33]
[18] J. O. Estima, Nuno M. A. Freire, A. J. Marques Cardoso, “Recent
advances in fault diagnosis by Park's vector approach”, in Proc. IEEE-WEMDCD, March 2013, Paris, France, pp. 279-288.
[34]
[19] Y. Gritli, C. Rossi, D. Casadei, F. Filippetti, G-A. Capolino, “A
Diagnostic Space Vector-Based Index for Rotor Electrical Fault Detection in Wound-Rotor Induction Machines Under Speed Transient”, IEEE Trans. Ind. Elect., vol. 64, no. 5, pp. 3892 3902, May 2017.
[35]
[20] J. Antonino-Daviu, A. Quijano-López, V. Climente-Alarcon, C.
[36]
Garín-Abellán, “Reliable Detection of Rotor Winding Asymmetries in Wound Rotor Induction Motors via Integral Current Analysis”, IEEE Trans. Ind. Appl., vol. 53, no. 3, pp. 2040–2048, Feb. 2017.
[37]
[21] A. J. M. Cardoso and E. S. Saraiva, "Computer-aided detection of
airgap eccentricity in operating three-phase induction motors by Park’s Vector Approach", IEEE Trans. on Ind. Appl., vol. 29, no. 5, pp. 897-901, Sept./Oct. 1993.
[38]
[22] S. M. A. Cruz and A. J. M. Cardoso, “Rotor cage fault diagnosis
in three-phase induction motors, by Extended Park’s Vector Approach,” Electric Machines and Power Systems, vol. 28, no. 4, pp. 289–299, Apr. 2000.
[39]
[40] IEEE and Standard, IEEE draft guide for induction machinery maintenance testing and failure analysis, Jul. 2006. [41] G. B. Kliman and J. Stein, “Induction motor fault detection via passive current monitoring,” in Proc. ICEM, Cambridge, MA, USA, Aug. 1990, pp. 13–17. [42] H. A. Toliyat, M. S. Arefeen, A. G. Parlos, “A method for dynamic simulation of air-gap eccentricity in induction machines,” IEEE Trans. Ind. Appl., vol. 32, no. 4, pp. 910–918, Jul./Aug. 1996. [43] W. T. Thomson, D. Rankin, D. G. Dorrell, “On-line current monitoring to diagnose airgap eccentricity in large three-phase induction motors—Industrial case histories verify the predictions,” IEEE Trans. Energy Conv., vol. 14, no. 4, pp. 1372–1378, Dec. 1999. [44] S. Nandi, H. Toliyat, X. Li, “Condition monitoring and fault diagnosis of electrical motors—A review,” IEEE Trans. Energy Conv.., vol. 20, no. 4, pp. 719–729, Dec. 2005 [45] R.N. Andriamalala, H. Razik, L. Baghli, F.M. Sargos, “Eccentricity fault diagnosis of a dual-stator winding induction machine drive considering the slotting effects”, IEEE Trans. Ind. Electron., vol. 55, no. 12, pp. 4238–4251, Dec. 2008. [46] M. Akar, “Detection of a static eccentricity fault in a closed loop driven induction motor by using the angular domain order tracking analysis method”, Mech. Sys. Sig. Proc., vol. 34, no. 1-2, pp. 173– 182, Jan. 2013. [47] M. Yazid Kaikaa, M. Hadjami, A. Khezzar, “Effects of the simultaneous presence of static eccentricity and broken rotor bars on the stator current of induction machine” IEEE Trans. Ind. Elec., vol. 61, no. 5, pp. 2452 – 2463, May 2014. [48] D. Dorrell, J. Shek, M.-F. Hsieh, M. Mueller, “Unbalanced magnetic pull in cage induction machines for fixed-speed renewable energy generators,” IEEE Trans. Magn., vol. 47, no. 10, pp. 4096–4099, Oct. 2011
[58] C. Piantsop Mbo'o and K. Hameyer, “Fault Diagnosis of Bearing Damage by Means of the Linear Discriminant Analysis of Stator Current Features From the Frequency Selection”, IEEE Trans. Ind. Appl., vol. 52, no. 5, Sept.-Oct. 2016, pp. 3861 – 3868. [59] J. Jung, Y. Park, S. Bin Lee, C-H. Cho, K. Kim, E. J. Wiedenbrug, M. Teska, “Monitoring Journal-Bearing Faults: Making Use of Motor Current Signature Analysis for Induction Motors”, IEEE Industry Applications Magazine, vol. 23, no. 4, pp. 12-21, April 2017. [60] F. Immovilli and M. Cocconcelli,”Experimental Investigation of Shaft Radial Load Effect on Bearing Fault Signatures Detection”, IEEE Trans. Ind. Appl., vol. 53, no. 3, pp. 2721-2729, Nov. 2016. [61] S. Singh and N. Kumar, “Detection of Bearing Faults in Mechanical Systems Using Stator Current Monitoring”, IEEE Trans. Ind. Inf., vol. 13, no. 3, pp. 1341 - 1349, June 2017. [62] G. Salles, F. Filippetti, C. Tassoni, G. Crellet, G. Franceschini, “Monitoring of induction motor load by neural network techniques,” IEEE Trans. on Power Electr., vol. 15, no. 4, pp. 762–768, Jul. 2000. [63] M. Blodt, P. Granjon, B. Raison, G. Rostaing, “Models for bearing damage detection in induction motors using stator current monitoring,” IEEE Trans. Ind. Elect., vol. 55, no. 4, pp. 1813–1822, Apr. 2008. [64] C. M. Riley, B. K. Lin, T. G. Habetler, R. R. Schoen, “A method for sensorless on-line vibration monitoring of induction machines,” IEEE Trans. Ind. Appl., vol. 34, no. 6, pp. 1240–1245, Nov./Dec. 1998. [65] H. Smith, E. Wiedenbrug, and M. Lind, “Rotating element bearing diagnostics in a nuclear power plant: Comparing vibration and torque techniques,” in Proc. IEEE-SDEMPED, Sept. 2007, pp. 17–22. ͒ [66] M. Fenger, B. A. Llyod, W. T. Thomson, “Development of a tool to detect faults in induction motors via current signature analysis,” in Proc. IEEE IAS/PCA Cement Ind. Tech. Conf., 2003, pp. 37–46.
[49] S. Haroun, A. Nait Seghir, S. Touati, S. Hamdani, “Misalignment fault detection and diagnosis using AR model of torque signal”, in Proc. IEEE-SDEMPED, pp. 322 - 326, Sept. 2015.
[67] A. R. Mohanty and C. Kar, “Fault detection in a multistage gearbox by demodulation of motor current waveform,” IEEE Trans. Ind. Electron., vol. 53, no. 4, pp. 1285–1297, Jun. 2006.
[50] J. Sobra, T. Vaimann, A. Belahcen, “Mechanical vibration analysis of induction machine under dynamic rotor eccentricity”, in Proc. International Scientific Conference on Electric Power Engineering, Karlsruhe, Germany, September 2016, pp. 1-4.
[68] C. Kar and A. R. Mohanty, “Vibration and current transient monitoring for gearbox fault detection using multiresolution Fourier transform,” J. Sound Vib., vol. 311, no. 1/2, pp. 109–132, Mar. 2008.
[51] C. Yoo and I.O. Park, "Analysis of structural vibration characteristic of harmonic drive, " in Proc. Int. Conf. on Ubiquitous Robots and Amb. Intel., July 2014. [52] C. M. Harris and A. G. Piersol, Harris’s Shock and Vibration Handbook, New York, NY, USA: McGraw-Hill, 2002. [53] C. W. de Silva, Vibration and Shock Handbook, Boca Raton, FL, USA: CRC Press, 2005. [54] L. Feng and D. Kong, “Fault diagnosis of tooth root crack in helical gear”, in Proc. International Conference on Ubiquitous Robots and Ambient Intelligence, pp. 686 - 691, June 2016 [55] C. Verucchi, G. Bossio, J. Bossio, G. Acosta, “ Fault detection in gear box with induction motors: an experimental study”, IEEE Latin America Trans., vol. 14, no. 6, pp. 2726 - 2731, June 2016. [56] J. Jose Saucedo-Dorantes, M. Delgado-Prieto, J. Antonio OrtegaRedondo, R. Alfredo Osornio-Rios, R. de Jesus Romero-Troncoso, “Multiple-Fault Detection Methodology Based on Vibration and Current Analysis Applied to Bearings in Induction Motors and Gearboxes on the Kinematic Chain”, Shock and Vibration, article ID 5467643, vol. 2016. [57] V. C. M. N. Leite, J. Guedes Borges da Silva, G. Francimeire Cintra Veloso, L. Eduardo Borges da Silva, G. Lambert-Torres, E. Leandro Bonaldi, L. Ely de Lacerda de Oliveira, “Detection of Localized Bearing Faults in Induction Machines by Spectral Kurtosis and Envelope Analysis of Stator Current”, IEEE Trans. Ind. Elect., vol. 62, no. 3, March 2015, pp. 1855 – 1865.
[69] F. Cheng, Y. Peng, L. Qu, W. Qiao, “Current-Based Fault Detection and Identification for Wind Turbine Drivetrain Gearboxes”, IEEE Trans. Ind. Appl., vol. PP, no. 99, pp. 1 – 1, Nov. 2016. [70] S. Hedayati Kia, H. Henao, G.-A. Capolino, “Analytical and experimental study of gearbox mechanical effect on the induction machine stator current signature,” IEEE Trans. Ind. Appl., vol. 45, no. 4, pp. 1405–1415, Jul./Aug. 2009. [71] S. Hedayati Kia, H. Henao, G-A. Capolino, “Gear Tooth Surface Damage Fault Detection Using Induction Machine Stator Current Space Vector Analysis”, IEEE Trans. Ind. Elect., vol. 62, no. 3, pp. 1866 - 1878, March 2015. [72] S. Hedayati Kia, H. Henao, G-A. Capolino, “Fault Index Statistical Study for Gear Fault Detection Using Stator Current Space Vector Analysis”, IEEE Trans. Ind. Appl., vol. 52, no. 6, pp. 4781 – 4788, Nov.-Dec. 2016. [73] S. Hedayati Kia, H. Henao, G.-A. Capolino, “Gear tooth surface damage fault profile identification using stator current space vector instantaneous frequency,” in Proc. IEEE IECON, Vienna, Austria, Nov. 10–13, 2013, pp. 5482–5488. [74] S.G. Mallat, "A theory for multiresolution signal decomposition: the wavelet representation", IEEE Trans. Pattern Anal. Mach. Intell., Vol. 2, N°7, pp. 674–693, 1989. [75] Y. Gritli, L. Zarri, C. Rossi, F. Filippetti, G.-A. Capolino, D. Casadei, “Advanced diagnosis of electrical faults in wound rotor induction machines,” IEEE Trans. Ind. Elec., vol. 60, no. 9, pp. 4012–4024, Sep. 2013.