Simulation and Fault Detection in PMSM under Dynamic

Conditions

J. Rosero, J. Romeral,

Motion Control and Industrial Applications Group, Technical University of Catalonia

C/ Colom. 08222 Terrassa. Catalonia. Spain, e-mail: rosero@eel.upc.edu

Abstract: This paper presents a study of PMSM under short circuit in turns at different speed and torque values.
PMSM in a healthy state and under fault conditions is simulated at different working points using a Flux 2D, two-
dimensional (2-D) finite element analysis (FEA). Currents and flux density are presented and their harmonic
content is obtained. In addition, harmonics content is compared to one obtained from a healthy motor. This paper
applies the appropriate tools for determining the fault condition from the current analysis as Empirical mode
decomposition (EMD) with Wigner Ville Distribution (WVD). The EMD and WVD are used to fault detection in
the operation range of PMSM. Simulated and experimental stator current spectra from healthy and faulty
machines are obtained and compared.

I. Introduction
The permanent magnet synchronous motor (PMSM) is becoming popular in high performance
applications as compared to other types of ac motors. High speed operation and precise torque control
are some of advantageous features, which include high torque to current ratio as well as high power to
weight ratio, high efficiency, low noise and robustness.
In such applications it is advantageous to use a drive capable of continuing operation even with the
presence of any single point failure. Such a drive is termed fault tolerant and in order to develop such
drive it is necessary to know and understand the faults in motor, which is the aim of the research
presented here [1, 2].
A short circuit between turns is the most critical fault in a machine, is quite difficult to detect and
almost impossible to remove. Stator winding faults might have a destructive effect on the stator coils.
[3]. Stator faults are usually related to insulation failure. There are a number of techniques used to
detect turn-to-turn faults, the majority of them based on stator voltages and currents, axial flux and d-q
current and voltage component analysis [4, 5]. The best way to analyze the content frequency of non-
stationary stator current is with a joint time-frequency analysis. Therefore, a number of time-frequency
analysis technologies have been developed for analyzing non-stationary stator current [6, 7] as can be
see Fig. 1. Apart from their theoretical advantages, the method proposed for motor fault detection
should be really accurate, easy to perform and simple to implement on a Digital Signal Processing
(DSP) development tool.
The EMD method was motivated by computation of instantaneous frequency defined in terms of
Hilbert transform. The local energy and the instantaneous frequency derived from the intrinsic mode
function (IMF) [8, 9]. IMF is the decomposition of signal at different frequency ranges.
The Wigner-Ville distribution (WVD) is a time-frequency representation [7]. If the analyzed signal
contains more than one component, the WVD method suffers from cross term interference as the
Pseudo Wigner Ville Distribution (SPWVD) exist for applying the weighting function to the
instantaneous correlation.
Flux 2D is proposed for simulation of the system PMSM-Control-Drive because involves non-linear
external circuit parts, such as solid state switching devices; also, the electric motor comprises non-
linear magnetic materials. Flux 2D allows the user to consider local phenomena such as saturation,
high circulating currents, field current dynamics, Eddy currents and hysteresis losses, non-sinusoidal
voltage induced (EMF), etc., whose effects are not important in normal behaviour, but are significative
during a fault transient [10, 11].

Fig. 1. Scheme of fault detection techniques analysis in PMSM

In this paper, a PMSM with short circuit is simulated at different speed values using a Flux 2D, two-
dimensional (2-D) Finite Element Analysis (FEA). Currents and flux density are presented and their
harmonics contents are obtained. In addition, the PMSM running at speed variation are analyzed by
means of (EMD) with (WVD). Simulations are also compared with experimental results.

IV. Workbench of simulation and experimental

The fault analysis for a PMSM of 6000 rpm, 2.3 Nm, 3 pair of poles is based on simulation and
experimental tests at different speed values. Simulations and experiments have been carried out for
motors with 4, 8 and 12 short turns of stator phase winding.
The spectrum has been normalized to a rotor frequency for every case. Numerical simulations were
developed with the combination of Flux 2D for the motor model and electric circuit, and
Matlab-Simulink for power electronics and control as is shows in Fig. 2.

Fig. 2. Schematic of the finite element analysis (FEA) for a PMSM

The workbench development used to apply and improve motor current signal analysis (MCSA) start
with current, a filter that keeps only those components of particular interest; followed by selection of
frequency components and finally the software classifies.

III. Simulation of Short Circuit

The way of Modeling shorted turn faults by means of Flux 2D can be see in Fig. 3. The fault of short
turns of stator phase winding leads to two main effects on the machine flux. The first is that the large
current in the shorted turns, leads to an increase in the local leakage flux, particularly slot leakage. This
changes the saturation conditions from of the teeth locally. Secondly, the currents induced in the
shorted coils oppose the establishment of the main, air-gap flux. They thus reduce that flux and the
corresponding main flux path saturation along the winding axis of the shorted coil.

Fig. 3. Circuit scheme for short circuit simulation

Fig. 4. Flux density of a PMSM with short circuit

Therefore, the new frequency component appears in the stator current spectrum as a result of a fault
in the stator windings. Only a rise in the rotor slot harmonic frequencies can be expected because
under fault conditions a greater number of flux density waves exist in the machine and all of these flux
densities make a contribution at the same frequencies. Thus, there is a greater probability of flux
density with the basic number of pole pairs now existing [12] as shown in Fig. 4 and Fig. 5.

Fig. 5. Flux density distribution of a PMSM with short circuit

The simulations of the PMSM in stationary conditions are shown in Fig. 6 through the stator current
harmonic. The amplitudes of the current harmonics 1st, 7th, 9th are higher for a faulty motor than for a
healthy one. However, at a low speeds the the discrimination of current harmonics between both
motors is more difficult to be seen than when motors are running at nominal speed. The experimental
results are similar for all speed range.
0
w=1500 rpm
M Healthy
-10 M 4 Short turns
M 8 Short turns
M 12 Short turns
-20
Amplitude ( dB)

-30

-40

-50

-60
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
Harmonics

Fig. 6. Stator current harmonics for a PMSM with short circuit, Simulation at 1500 rpm
V. Result
The short circuit is analyzed under different speed conditions. Five intrinsic mode functions (IMF) are
obtained from the stator current by means of EMD and afterwards WVD is calculated. The results
obtained from a machine with short circuit are compared with those obtained from a healthy machine.
IMF 3 contains the stator main current harmonic and IMF 1 and 2 together contain the failure
harmonic. In this way, the EMD allows you to obtain the IMF that contain the short circuit fault
harmonic and thus concentrate our analysis on them, obtaining better resolution and accuracy with
fault detection. Now, the biggest harmonic is the 9th fault harmonic (900 Hz in this case); the main
harmonic corresponding to power supply, that is always higher to the other ones when it is analyzed
with standard MCSA [13], has been eliminated.
The simulation with a linear speed change from 1500 rpm to 1000 rpm, starting at 0.2 s and ending at
0.29 s can been seen in Fig. 7 for healthy PMSM. The stator currents decomposition by means of
EMD allows to obtain 5 IMF and each one contains its own characteristic frequency range. In this
case, the spectrum for healthy machine results clean or empty; on the contrary, for short circuit fault
we could locate the fault frequency in a more easy and evident way.
 Reassigned smoothed pseudo Wigner-Ville distribution TFRSPWV
600
60
550

500 50

450

Frequency (Hz)
40

400
30
350

300 20

250
10
200

150
0 0.05 0.1 0.15 0.2 0.25 0.3
Time (S)

Fig. 7. SPWVD of IMF 1&2 in healthy PMSM. Speed change from 1500 to 1000 rpm
600
60
550

500 50

450
Frequency (Hz)

40

400
30
350

300
20

250 Failure Harmonic

10
200

150
0.05 0.1 0.15 0.2 0.25 0.3
Time (S)

Fig. 8. SPWVD of IMF 1&2 in PMSM with 12 short circuit turns. Speed change from 1500 to 1000 rpm
600
60
550

500 50

450
Frequency (Hz)

40

400
30
350

300
20
Failure Harmonic
250
10
200

150
0.05 0.1 0.15 0.2 0.25 0.3
Time (S)

Fig. 9. SPWVD of IMF 1&2 in PMSM with 12 short circuit turns. Speed change from 1500 to 1000 rpm, Experimental results

Fig. 8 shows SPWVD of IMF 1 and 2 of the stator currents of a PMSM with 12 short circuit turns in
comparison with Fig. 7. Some experimental results are shown in Fig. 9. These figures represent
SPWVD of IMF 1 and 2 PMSM stator currents with a 12 turn short circuited. These are the results for
low speed where the fault detection is more critical. The results for the whole speed range are also
very satisfactory. The method allows remarking the analysis in the characteristic failure and decreasing
the computational burden; besides, it maximizes the failure relative value. The quick filter by means of
EMD allows eliminating the undesired frequencies for the fault detection and the system can be more
accurate. Now, it presents the short circuit analysis with stator currents of PMSM by means of EMD
and IMF. Next, WVD has been applied to IMF 1 and 2.
VI. Conclusion
Flux 2D model allows PMSM behaviour to be determined under different fault conditions. Flux 2D is
an excellent tool for simulation of the PMSM under fault and for all speed range. This software allows
the user to know the behaviour of the motor in different conditions and to analyze what has happened
after a failure.
Joint EMD and SPWVD show the ability of fault identify within a short time. These transforms do not
need many samples, and they can work independently of the speed variations. The featured signature
allows differentiating between healthy and faulty conditions, as well as between different degrees of
fault. The results can be used for fault detection and diagnosis of such internal faults at low speed and
dynamic conditions. This method is suitable to be implemented in embedded on-line applications. On
the whole, their precision and reliability is very high. The results show an increase in the spectral
resolution and also reliability for the fault diagnosis.
VII. References
[1] J. A. Haylock, B. C. Mecrow, A. G. Jack, and D. J. Atkinson, "Operation of a fault tolerant PM drive for an
aerospace fuel pump application," in Eighth International Conference on Electrical Machines and Drives (Conf. Publ. No.
444), 1997, pp. 133-137.
[2] N. Ertugrul, W. L. Soong, S. Valtenbergs, and H. Chye, "Investigation of a fault tolerant and high performance
motor drive for critical applications," in TENCON. Proceedings of IEEE Region 10 International Conference on Electrical
and Electronic Technology, 2001, pp. 542-548 vol.2.
[3] M. A. Awadallah and M. M. Morcos, "Application of AI tools in fault diagnosis of electrical machines and drives-
an overview," IEEE Transactions on Energy Conversion, vol. 18, pp. 245-251, 2003.
[4] H. A. Toliyat, S. P. Waikar, and T. A. Lipo, "Analysis and simulation of five-phase synchronous reluctance
machines including third harmonic of airgap MMF," IEEE Transactions on Industry Applications, vol. 34, pp. 332-339,
1998.
[5] H. A. Toliyat and T. A. Lipo, "Transient analysis of cage induction machines under stator, rotor bar and end ring
faults," IEEE Transactions on Energy Conversion, vol. 10, pp. 241-247, 1995.
[6] N. Arthur and J. Penman, "Condition monitoring with non-linear signal processing," in IEE Colloquium on Non-
Linear Signal and Image Processing (Ref. No. 1998/284), , 1998, pp. 4/1-4/5.
[7] J. Rosero, J. Cusido, A. Garcia, J. A. Ortega, and L. Romeral, "Broken Bearings Fault Detection for a Permanent
Magnet Synchronous Motor under non-constant working conditions by means of a Joint Time Frequency Analysis," in
IEEE International Symposium on Industrial Electronics, ISIE 2007, Vigo, Spain, 2007, pp. 3415-3419.
[8] B. Liu, S. Riemenschneider, and Y. Xu, "Gearbox fault diagnosis using empirical mode decomposition and Hilbert
spectrum," Mechanical Systems and Signal Processing, vol. 20, pp. 718-734, 2006.
[9] N. E. Huang, Z. Shen, S. R. Long, M. C. Wu, H. H. Shih, Q. Zheng, N. C. Yen, C. C. Tung, and H. H. Liu, "The
empirical mode decomposition and the Hilbert spectrum for nonlinear and non-stationary time series analysis," Proceedings
of the Royal Society A: Mathematical, Physical and Engineering Sciences, vol. 454, pp. 903-995, 1998.
[10] A. A. Arkadan and B. W. Kielgas, "The coupled problem in switched reluctance motor drive systems during fault
conditions," IEEE Transactions on Magnetics, vol. 30, pp. 3256-3259, 1994.
[11] J. Rosero, J. Cusido, A. Garcia, J. A. Ortega, and L. Romeral, "Study on the Permanent Magnet Demagnetization
Fault in Permanent Magnet Synchronous Machines," in The 32nd Annual Conference of the IEEE Industrial Electronics
Society, IECON06, Paris - FRANCE, 2006, pp. 879-884.
[12] G. M. Joksimovic and J. Penman, "The detection of inter-turn short circuits in the stator windings of operating
motors," IEEE Transactions on Industrial Electronics, vol. 47, pp. 1078-1084, 2000.
[13] J. Rosero, O. Almonacid, M. Amaya, and L. Romeral, "Simulation and Fault Detection of Short Circuit Winding in
a Permanent Magnet Synchronous Machines (PMSM)," in ISEF 2007 - XIII International Symposium on Electromagnetic
Fields in Mechatronics, Electrical and Electronic Engineering, Prague, Czech Republic, 2007.