Sie sind auf Seite 1von 12

1318

IEEE TRANSACTIONS ON INDUSTRY APPLICATIONS, VOL. 36, NO. 5, SEPTEMBER/OCTOBER 2000

Closed-Loop Control Impact on the Diagnosis of Induction Motors Faults


Alberto Bellini, Associate Member, IEEE, Fiorenzo Filippetti, Giovanni Franceschini, Associate Member, IEEE, and Carla Tassoni, Senior Member, IEEE

AbstractIn this paper, the impact of control on faulted induction machine behavior is presented. The diagnostic indexes usually used for open-loop operation are no longer effective. Simulation and experimental results show that the spectrum of the field in a field-oriented controlled machine has current component suitable features that can lead to an effective diagnostic procedure. Specifically, in the case of stator and rotor faults the spectrum components at frequencies 2 and 2 , respectively, are quite independent of control parameters and dependent on the fault extent. Index TermsFault diagnosis, induction motor drives.

I. INTRODUCTION N the supervision of electrical equipment, the task of diagnostic systems is to detect an upcoming machine fault as soon as possible, in order to save expensive manufacturing processes or to replace faulty parts. This task is often not trivial, as the impact of many faults is small, and can be masked by the noise of electrical quantities or by changes in environmental conditions, therefore, the diagnostic techniques have to be very sensitive. As far as open-loop induction machines are concerned usually the input current [1][4], the stray flux [5] or the electromagnetic (EM) torque [6] is monitored in quasi-steady-state conditions, in order to detect machine faults. In the literature, the issue of faults injected by a voltage-source inverter (VSI) in the motor supply was also tackled in order to diagnose further anomalous operating conditions [7], [8]. Addressing diagnostic techniques based on current signature analysis, several procedures for the detection and severity classification of both stator short circuits and rotor bar breakage have been proposed for machines supplied at fixed frequency or at fixed-voltage hertz ratio. Usually, the anomalous components introduced by a specific fault in the current spectrum are investigated and the diagnostic procedure is able to correlate the amplitude of these components to the fault extent [9][14]. As for closed-loop drives, the control itself modifies the behavior of supply variables and more sophisticated procedures
Paper IPCSD 00005, presented at the 1999 Industry Applications Society Annual Meeting, Phoenix, AZ, October 37, and approved for publication in the IEEE TRANSACTIONS ON INDUSTRY APPLICATIONS by the Electric Machines Committee of the IEEE Industry Applications Society. Manuscript submitted for review June 15, 1999 and released for publication March 14, 2000. A. Bellini, G. Franceschini, and C. Tassoni are with the Dipartimento di Ingegneria dellInformazione, University of Parma, I-43100 Parma, Italy (e-mail: a_bellini@ee.unipr.it; g_franceschini@ee.unipr.it; c_tassoni@ee.unipr.it). F. Filippetti is with the Dipartimento di Ingegneria Elettrica, University of Bologna, I-40136 Bologna, Italy (e-mail: fiorenzo.filippetti@mail.ing.unibo.it). Publisher Item Identifier S 0093-9994(00)07609-X.

must be adopted in order to assess machine conditions. A recent proposal has been presented in [15][17], where the difference of torque amplitude computed in different ways, as a function of rotor position, is sensed to detect the presence of rotor faults. The proposed torque computation method aims at producing the independence of torque ripple produced by the fault of operating conditions and control parameters. With the aim of extending the diagnostic procedures developed for open-loop faulted machines to the closed-loop operations, the authors have started in [18] a systematic analysis of the behavior of controlled induction machines with stator or rotor electric faults. In this first approach, the faults were introduced as one-side asymmetries, caused by an additional resistance in the stator or in the rotor windings. Simulation and experiments have shown that, as expected, the typical spectrum lines produced by asymmetries in the machine input currents are present in the voltage spectrum as well. The amplitude of these lines depends on the control structure and on its parameters, therefore, the usual diagnostic indexes are no longer effective. However, it resulted that the flux control current, i.e. the current in the field-oriented control structure considered, is quite independent of the control parameters and depends on asymmetry degree. In this paper, real electric faults, i.e., stator short circuits and rotor bar breakage, are considered (Fig. 1). For the machine with broken rotor bars, the authors use a simplified dynamic model, developed starting from more complex models, which provides sufficiently good results and that is here recalled [19][21]. For the machine with a stator short circuited, the extension to the dynamic behavior of a simple steady state model is here presented [22], [23]. These models allow for the consideration of the fault effects in the dynamic simulation of the whole drive. By simulations, a map of the amplitude of the specific lines introduced by the faults in the spectrum of different variables can be drawn versus control parameters for some fault degrees. and Besides machine current and voltages, field current torque current are investigated. The experiments performed with a digitally controlled drive confirm the effectiveness of the simulation results and, thus, the effectiveness of the proposed faulted machine model. The amplitude of machine current spectrum lines is compared with that of the open-loop machine in steady-state conditions in order to address the different behavior. The features of the other electric variables are monitored, too, and, as expected, the amplitude of the specific component introduced in the current spectrum by the faults is practically independent of the control parameters, namely, the proportional speed loop gain, and depends on fault degree. Therefore, it should be possible to de-

00939994/00$10.00 2000 IEEE

BELLINI et al.: DIAGNOSIS OF INDUCTION MOTORS FAULTS

1319

Fig. 1. Stator short circuit and rotor bar breakage.

Fig. 2. Two-axes simplified model of stator short circuits and rotor bar breakage.

velop new diagnostic indexes, based on spectrum line amplitude. However, the authors are not able to state the relationship between the diagnostic index and the fault extent, since the the parameters of other drive components, beyond the machine parameters, affect the control variables. II. FAULTED MACHINE MODELS Several models for a machine with broken bars were proposed in the literature [24], [25] while more recently, stator short-circuited machine models have been introduced, too [26], [27]. These models are steady-state models and, thus, are not suitable for dynamic simulation. In the last few years, dynamic models were developed both for rotor breakage [19] and stator short circuit [28]. However, the complexity of these models leads to unacceptable run time of controlled drive simulation. Therefore, simpler models are needed in order to allow several simulations in different conditions and to predict the outstanding effects produced by the faults. Specifically, the stator fault model must proof the mavide the negative-sequence current at frequency chine input variables, and the rotor fault model must provide the . These models can current spectrum line at frequency be easily commuted to steady-state models, using the transforto symmetrical components [29][31]. Adding mation from further simplifications, direct diagnostic indexes can be determined which still allow an effective first-level diagnosis of the faults entity [32]. In the modelization of the controlled machine, the rotor speed is obviously one of the variables, therefore, the faulted machine models allow the consideration of the effect of speed ripple due and . If to the torque ripple, respectively, at frequencies mains supplied machines are considered, the diagnostic procefrequency. dures usually neglect the effect of phenomena at

On the contrary, the speed ripple at frequency causes a sequence of new lines in the current spectrum that cannot be neglected [33]. Obviously, in the frequency-controlled machine frequency must be considered, too, if the the speed ripple at frequency is low, because it is no more damped by inertia. In the following, the models for a stator shorted machine and rotor bar breakage are presented. A. Stator Shorted Circuit Model The shorted circuit presence in a stator winding requires the analysis of a two-poles fundamental flux density wave in the machine air gap. Therefore, no variable transformation could be applied and computational methods, accounting for all the time and spatial harmonics of current and flux density distribution, should be used [28]. However, the short-circuit occurrence allows the machine to operate only if this event affects a restricted number of turns and, thus, when a simplified assumption can be adopted. Namely, a machine with shorted turns in one phase with turns can be replaced with a machine with a healthy stator (three windings with turns) and a dummy stator winding with shorted turns in the same axis of the shorted winding. The dynamic behavior of such machine can be analyzed with reference to a two-axes model in a stationary frame with the shorted and, therefore, the dummy winding in the axis (see Fig. 2). Specifically, the transformation that maintains the current and voltage amplitude and not the power is adopted. The equation of the dummy winding, whose parameters are referred to with subscript , is added to the the usual equations of a symmetrical machine referred to stator frame (1). A short-circuit resistance can be introduced, which is included in the operational impedance . The shorted circuit variables and parameters, like the rotor ones, are all referred to the stator, and to identify them the superscript is added. The following symbols

1320

IEEE TRANSACTIONS ON INDUSTRY APPLICATIONS, VOL. 36, NO. 5, SEPTEMBER/OCTOBER 2000

are defined: , where is the derivative operator, the mutual inductance coeffithe self-inductance and the resistance cient, , , , and is the rotor electrical of stator and rotor respectively, speed

(1) The EMF induced in the shorted turns is proportional to , while for the current transformation the ampere turns equiva, the lence must be considered. Therefore, defining relationships among the actual variables and those referred to , , and the stator with turns are . Starting from the dynamic model represented by (1) and introducing suitable simplifications, an approximated relationship between a few machine parameters and the fault percentage can be obtained. At first, steady-state conditions must be considered, i.e., constant speed and sinusoidal symmetrical voltcomponents can be replaced by symmetages supply. The rical components and the stator asymmetry produced by the dummy windings causes a backward-sequence current at frerotating with respect to rotor at frequency quency . The dummy winding being a single-phase one, it features forward and backward components of equal amplitude , and . Naming , and , (2) is obtained

Fig. 3. Steady-state equivalent circuit of the machine with a stator short circuited.

Fig. 4. Simplified equivalent circuit of the machine with a stator short circuited.

(2) , Introducing the leakage reactance , and , expressing the magnetizing terms and dividing the third and the we obtain the fourth equations respectively by and standstill equivalent model

As , the above equations correspond to the equivalent and, circuit of Fig. 3, where no current flows in the branch thus, it can be omitted. Moreover, if the magnetizing currents are neglected, the circuit becomes as in Fig. 4. With further assumptions from the circuit of Fig. 4, it stems that the stator backward-sequence current is linked to the shortcircuit characteristics. Due to the dissimilar resistance values, can be neglected with respect to . the voltage drop This voltage, neglecting stator voltage drop, is equal to the ap. This current is plied voltage , therefore, split into the stator and rotor backward components at node . Owing to common parameters values, it can be assumed that the amplitude of these backward currents are quite equal, so that the following relationship holds in terms of amplitude: (4) is composed by the impedance of The equivalent impedance the shorted turns and by the fault impedance. For zero fault and can be found. In impedance, a relationship between

(3)

BELLINI et al.: DIAGNOSIS OF INDUCTION MOTORS FAULTS

1321

Fig. 5. Steady-state equivalent circuit of the machine with a rotor additional resistance.

Fig. 6. Simplified equivalent circuit of the machine with a rotor additional resistance.

fact, for a low value of the turns reactance can be neglected and, in comparison with the turns resistance, which is . The stator backward-sequence curtherefore, rent becomes (5) In spite of the heavy assumptions adopted, this relationship can be used as a first-level diagnosis of short circuit turns in the case of no short-circuit resistance [23]. B. Rotor Bar Breakage Model contiguous bars in a rotor with bars A breakage of can be simulated by the dummy resistance in one of the windings of the three-phase equivalent slip-ring becomes in the machine [34]. The additional resistance (Fig. 2). In the following, the two-phase model equal to is introduced. The dynamical model of symbol the faulted machine referred to rotor frame, which will be used for simulation purposes, is, therefore, reported in (6)

can be replaced by symmetrical components. Therefore, doing the asymmetry of the rotor, provoked by the dummy resistance, causes in the rotor windings an additional backward-sequence current rotating at frequency , when referred to the stator. Moreover, the symmetrical component impedance transand the backformation introduces the forward impedance ward impedance , therefore, relationship (7) holds

(7)

and . where Expressing the magnetizing terms, dividing the second equation and the third and the fourth by , the model equations by become

(8) (6) With further simplifications, it is possible to derive an approximated relationship among a few machine parameters and the from the dynamic model represented by fault percentage (6). Again, steady-state conditions must be considered, i.e., constant speed and sinusoidal symmetrical voltages. components The above equations can be represented by the equivalent circuit of Fig. 5. If the magnetizing branches are neglected, it becomes as in Fig. 6. With further assumptions, an approximated relationship between the component at frequency and of the stator current can be obtained. Neglecting with rein the branch with the additional resistance spect to

1322

IEEE TRANSACTIONS ON INDUSTRY APPLICATIONS, VOL. 36, NO. 5, SEPTEMBER/OCTOBER 2000

Fig. 7.

Block diagram of a field-oriented motor drive.

and replacing the total impedance of the branch BD by the following relationship holds in terms of amplitude:

(9) The adopted model, based on the generalized rotating field theory, refers to forward- and backward-sequence currents. On the asymmetrical rotor there are effectively forward. and backward-sequence currents at the same frequency Referring the rotor backward-sequence current to the stator, . If the corresponding sequence appears at frequency this frequency is positive, it is effectively a backward sequence. However, this is not the case for usual slip values, therefore, it appears as an equivalent forward-sequence current at frequency . The constant-speed assumption required for the above-reported modelization leads to incorrect results. The speed ripple is at low frequency and it is only partially damped by the inertia, and its effect must be introduced. The rotating magnetic field contains a periodical mechanical angular variation which . Therefore, induces EMF and currents at frequencies is a right sideband current component at frequency added, beyond the modification of the left sideband current. The authors have noted that the sum of the two sideband comdepends mainly on the fault ponents at frequencies degree and not on the inertia, therefore, relationship (9) can still be adopted to determine the faulted bars number, if the sum of the amplitude of the two current components is used [33]. III. CONTROL IMPACT ON DIFFERENT VARIABLES When closed-loop control systems are used, all the manipulated variables are available beyond those at the motor terminals. These variables, i.e., the outputs of the regulators, can be sensed for diagnostic purposes, and the results are very useful, since they show directly the action of control. Namely, a fault in the machine can be sensed as a plant parameter variation. Since the regulators in presence of faults will still force the controlled variables to the reference value, the controller output changes, reflecting fault entity and type. Therefore, new diagnostic indexes can be retrieved from manipulated variables, too.

Drive topology must be considered and, according to the adopted control scheme, machine faults affect specific variables, easing the diagnostic procedure. Obviously, the control gains affect the behavior of the system, and their effects should not mask the diagnostic procedure. In this paper, a field-oriented induction motor drive is investigated adopting the common scheme depicted in Fig. 7, limited to constant flux speed range. Current loops are closed on a reference frame synchronous with the rotor flux. The angle and the amplitude of rotor flux are obtained by means of a flux observer that allows for the control of the amplitude of rotor flux as well. With this configuration, the output of the flux controller, i.e., the requested magnetizing current , and the output of the speed controller, i.e., the requested torque current , can be monitored for diagnostic purposes. is not sufIt should be expected that the torque current ficiently robust as a diagnostic index, since it is strongly affected by operating conditions. In particular, is affected by load torque value and ripple. Moreover, the amplitude of the spectrum components introduced by anomalies is, in general, dependent on the speed loop bandwidth usually set by the user and on the frequency of electrical quantities. On the other hand, the field current seems to be very attractive. Flux loop bandwidth is set by the manufacturer and, therefore, it usually features fixed and well-known control parameters. It does not depend on operating conditions if decoupling is correct. Therefore, it should be expected that anomalous spectrum lines appearing in the spectrum depend only on machine troubles, and are independent of operating conditions. Several simulations have been performed in order to prove these assertions, relying on the dynamic models reported with different fault degrees. Then, experiments have been performed on real-time-controlled machines by means of a digital-signal-processor (DSP)-based drive. Two identical cage machines have been used: one for stator faults with winding taps to realize short circuits, and the other one for rotor faults, introduced by rotor bar breakage. The data of the machines are , as follows: 1.5 kW, 50 Hz, 400 V, 4 A, 1410 rpm, , mH, mH, and . The machines are a three slots/pole/phase with a total of 36 stator slots. In every slot, two coils of 44 turns are located,

BELLINI et al.: DIAGNOSIS OF INDUCTION MOTORS FAULTS

1323

i Fig. 8. Amplitude of f spectrum lines (%) of ~ with different n=N percentage, simulation (solid line) and experimental (dotted line) results.

Fig. 10. Amplitude of 2f spectrum lines (%) of i with different n=N percentage, simulation (solid line) and experimental (dotted line) results.

Fig. 9. Amplitude of f spectrum lines (%) of ~ with different n=N v percentage, simulation (solid line) and experimental (dotted line) results.

Fig. 11. Amplitude of 2f spectrum lines (%) of i with different n=N percentage, simulation (solid line) and experimental (dotted line) results.

resulting in 264 turns/phase. The terminal connections of two coils are available and two shorted circuit tests, respec%, and 44 turns, tively, involving 22 turns, %, have been performed. The amplitude of the current flowing in the short-circuited turns is very high and the Joule losses could damage insulation [22]. In fact, the current can be expressed as the ratio between the EMF induced in the short-circuited turns, and the turns resistance, neglecting the reactance. A suitable choice of the speed set point, that leads to a frequency lower than the rated one, limits the EMF, which is dependent on frequency and, therefore, the current. However, in order to further reduce the current amplitude, an external additional resistance of 1.5 is added, which can be considered as a fault resistance. As for rotor faults, a breakage of one bar and then of another contiguous one are operated in sequence. The speed reference values forced for stator and rotor breakage are, respectively, 41 and 141 rad/s; this choice and a load equal to 60% of the rated one result in frequencies respectively close to 15 and 45 Hz. Therefore, the anomalous

spectrum components, respectively, at frequencies and , are close to 30 Hz and 6 Hz. The speed loop proportional gain is used as a parameter, in order to analyze the different impact of regulator parameters on the anomalous harmonics contents, while the integrative gain is set to zero. Specifically, the speed loop bandwidth is varied from 10 up to 40 Hz. The fixed speed reference value leads obviously to a shift of supply frequency as speed control gain or fault degree changes. The bandwidths of the current loops are set to 600 Hz. The field-oriented induction motor drive was simulated with Matlab and the different variables are examined. Obviously, in rotor faults simulation the transformation and inverse transformation blocks from fixed frame to rotor frame are needed, since the model is in that reference frame. Considering stator faults simulation, the space vector of currents must be computed in order to span the whole frequency range in order to account for both negative and positive sequences. The space phasors are built from currents and voltages

1324

IEEE TRANSACTIONS ON INDUSTRY APPLICATIONS, VOL. 36, NO. 5, SEPTEMBER/OCTOBER 2000

Fig. 12. Amplitude of s f spectrum lines (%) of i with one and two broken bars, simulation (solid line) and experimental (dotted line) results.

(1 6 2 )

i Fig. 15. FFT spectrum of ~ for a faulted machine (stator) with different values, n=N : .

= 8 3%

Fig. 13. Amplitude of s f spectrum lines (%) of v with one and two broken bars, simulation (solid line) and experimental (dotted line) results.

(1 6 2 )

Fig. 16. FFT spectrum of ~ for a faulted machine (stator) with different v values, n=N : .

= 8 3%

Fig. 14. Amplitude of sf spectrum lines (%) of i and i with one and two broken bars, simulation (solid line) and experimental (dotted line) results.

in the stator reference frame and are indicated in the following and . The amplitude of spectrum as of the current and voltage space phasors lines at frequency as a function of the proportional gain are reported in Figs. 8 of and 9, while the correspondent amplitude at frequency and current spectrum are in Figs. 10 and 11. The two above-noted fault percentages are considered. Notice that the theoretical steady-state rms values of backward current components according to the relationship (4) are, respectively, and . normalized by the test current, component of the current These results address that the space phasor is no more effective as a diagnostic index for stator faults as far as closed-loop systems are concerned: the current component is not constant with proportional gain variation and is not consistent with (4). Relying also on the corresponding voltage component, a mixed benchmark should be investigated to retrieve a new diagnostic index from the combination of current and voltage components.

BELLINI et al.: DIAGNOSIS OF INDUCTION MOTORS FAULTS

1325

Fig. 17.

FFT spectrum of i current varying the k parameter, n=N

= 8 3%.
: i Fig. 19. FFT spectrum of ~ for a faulted machine (stator) with different n=N percentage, k .

= 10

Fig. 18.

FFT spectrum of i current varying the k parameter, n=N

= 8 3%.
:

Figs. 10 and 11 report simulation results for the spectrum components of the controlled variables and , and address compotheir peculiarities. As is known, the currents in the nents are referred to a reference frame rotating with the rotor flux, therefore, the spectrum is shifted in frequency. It results harmonic amplitude of the current can lead to a that the new diagnostic index, quite robust with the control parameters harand dependent on the fault extent. On the contrary, the monic amplitude of the is strongly dependent on the proportional gain . As far as rotor faults are concerned, the usual slip values lead to positive values of the critical spectrum components at fre, therefore, the negative range of frequency quency does not contribute to an effective fault analysis. This is the reason why only the spectrum of one phase voltage and current is analyzed. Results are reported in Figs. 1214 for one and two broken bars. The same consideration reported above for

Fig. 20. FFT spectrum of ~ for a faulted machine (stator) with different n=N v percentages, k .

= 10

stator faults on input current and voltage space vectors can be adopted in this case, simply referring current and voltage spec. Notice that the theotrum components at frequencies retical steady-state values of backward current components according to the relationship (9) are, respectively, and . spectrum component at freAgain, the features of the seem very attractive for diagnostic purposes. In quency fact, this component is independent of control parameters and is dependent quite linearly on the fault degree. Notice that, in Fig. 14, the same chart reports both and shapes, that have the same order of magnitude, since the speed ripple is within the control bandwidth for the considered value of frequency.

1326

IEEE TRANSACTIONS ON INDUSTRY APPLICATIONS, VOL. 36, NO. 5, SEPTEMBER/OCTOBER 2000

Fig. 21. FFT spectrum of i for a faulted machine (stator) with different n=N percentages, k 10.

Fig. 23. FFT spectrum of i values and two broken bars.

for a faulted machine (rotor) with different k

Fig. 24. FFT spectrum of v for a faulted machine (rotor) with different k values and two broken bars. Fig. 22. FFT spectrum of i for a faulted machine (stator) with different n=N percentage, k = 10.

IV. EXPERIMENTAL RESULTS Using a fully digitally controlled drive, the experiments corresponding to the simulated conditions have been performed and the different variables have been sampled for fast Fourier transform (FFT) computation. As for stator faults, different values of the proportional gain have been used and the electrical quantities , , , and (in order to build the phasors and ), , and have been processed. Some FFT spectra of the different variables are reported in Figs. 1518 for different proportional gain values and a fixed fault percentage and in Figs. 1922 for the two fault percentages and a fixed value. The healthy machine has also been analyzed for comparison, showing the intrinsic anomalies of the

manufacturing process. Notice that the amplitudes of the space phasors are normalized. component at freExperimental results confirm that the changes in function of control gain while the quency harmonics of the current is constant with variations and should be a better diagnostic index. So far, it is not possible to provide a quantitative evaluation of the fault degree, since the number of parameters is too high. Moreover, the additional resistance avoids stating a proportional relationship with the fault . Notice that, in the experimental results, specpercentage trum lines appears in correspondence of mechanical speed and of converters bias. As for rotor faults, Figs. 2326 report the FFT spectra of input current, voltage, and of controlled variables for different values of proportional speed gain . Figs. 2730 report the results obtained with one or two broken bars and with the healthy machine

BELLINI et al.: DIAGNOSIS OF INDUCTION MOTORS FAULTS

1327

Fig. 25. bars.

FFT spectrum of i current varying the k parameter and two broken

Fig. 27. FFT spectrum of i for a faulted machine (rotor) with zero, one, and two broken bars, k 10.

Fig. 28. FFT spectrum of v for a faulted machine (rotor) with zero, one, and two broken bars, k = 10.

Fig. 26. bars.

FFT spectrum of i current varying the k parameter and two broken

for a constant value of proportional gain . The experimental results, indicated by the dots in Figs. 814, confirm that the only effective quantity to be considered for diagnostic purposes in a controlled machine is still the current spectrum. Notice that for rotor faults the speed ripple is slightly damped by mechanical inertia and, thus, the amplitude of the compofrequency is comparable with the corresponding nents at components for the considered fault percentages and frequency. On the contrary, for stator faults mechanical inertia strongly smoothes the amplitude of the component at frequency . It should also be considered that the information contained in the sideband components introduced by the speed ripple is

lost when the frequencies of the input variables are shifted by and the manipulated variables are analyzed. This consideration holds both for rotor faults, for which the components at frebecome a single component at , and for quencies stator faults, for which the sideband components at frequencies , corresponding to the speed ripple, become a single component at . V. CONCLUSION As far as voltage-supplied machines are concerned, suitable components of the current space-vector spectrum can be used to detect electrical faults and to evaluate their entity. Specifically, can diagnose the negative-sequence component at frequency stator faults, while the sideband components at frequencies can diagnose rotor faults.

1328

IEEE TRANSACTIONS ON INDUSTRY APPLICATIONS, VOL. 36, NO. 5, SEPTEMBER/OCTOBER 2000

REFERENCES
[1] W. T. Thomson, On-line current monitoring to diagnose shaft misalignment in three phase induction motor drives, in Proc. ICEM94, Paris, France, 1994, pp. 238243. [2] G. B. Kliman and J. Stain, Methods of motor current signature analysis, Elect. Mach. Power Syst., no. 20, pp. 463474, 1992. [3] R. R. Schoen, T. G. Habetler, B. K. Lin, J. H. Schlag, and S. Farag, An unsupervised on-line system for IM fault detection using stator current monitoring, in Conf. Rec. IEEE-IAS Annu. Meeting, Denver, CO, Oct. 1994, pp. 103108. [4] F. Filippetti, G. Franceschini, C. Tassoni, and P. Vas, Recent developments of induction motor drives fault diagnosis using AI techniques, in Proc. IEEE IECON98, Aachen, Germany, Sept. 1998, pp. 19661973. [5] J. Penman, J. G. Hadwick, and B. Barker, Detection of faults in electrical machines by examination of the axially directed fluxes, in Proc. ICEM78, Brussels, Belgium, 1978, pp. R/S1R/S10. [6] J. S. Hsu, Monitoring of defects in induction motors through air-gap torque observation, IEEE Trans. Ind. Applicat., vol. 31, pp. 10161021, Sept./Oct. 1995. [7] D. Kastha and B. K. Bose, Investigation of faults modes of voltage-fed inverter system for induction motor drives, in Conf. Rec. IEEE-IAS Annu. Meeting, Houston, TX, Oct. 1992, pp. 858866. [8] F. Filippetti, G. Franceschini, and C. Tassoni, Integrated diagnostics systems for failure identification in power converter, in Proc. EPE95, vol. 3, Seville, Spain, 1995, pp. 270274. [9] J. R. Cameron, W. T. Thomson, and A. B. Dow, On-line current monitoring of induction motors, in Proc. IEE-EMD Conf., 1987, pp. 282286. [10] G. Gentile, N. Rotondale, F. Filippetti, G. Franceschini, M. Martelli, and C. Tassoni, Analysis approach in IM stator faults to on-line diagnostics, in Proc. ICEM90, Cambridge, MA, Aug. 1990, pp. 491497. [11] C. Hargis, B. G. Gaydon, and K. Karnish, The detection of rotor defects in induction motors, in Proc. IEE EMDA Conf., London, U.K., 1982, pp. 216220. [12] F. Filippetti, G. Franceschini, and C. Tassoni, Neural network aided on-line diagnostics of IM rotor faults, IEEE Trans. Ind. Applicat., vol. 31, pp. 892899, July/Aug. 1995. [13] G. B. Kliman, W. J. Premerlani, and R. A. Koegl, A new approach to on-line turn fault detection in AC motors, in Conf. Rec. IEEE-IAS Annu. Meeting, San Diego, CA, Oct. 1996, pp. 687693. [14] G. Joksimovic and J. Penman, The detection of interturn short circuits in the stator windings of operating motors, in Proc. IEEE IECON98, Aachen, Germany, 1998, pp. 19741979. [15] R. S. Wieser, C. Kral, and F. Pirker, The Vienna induction machine monitoring method; on the impact of FOC structure on real operational behavior of a faulty machine, in Proc. IEEE IECON98, 1998, pp. 19011907. , High-sensitive rotor cage monitoring during dynamic load op[16] eration. The Vienna monitoring method, in Proc. ICEM98, Instanbul, Turkey, Sept. 1998, pp. 432437. [17] R. S. Wieser, M. Schagginger, C. Kral, and F. Pirker, The integration of machine fault detection into an indirect field-oriented induction machine drive control schema: The Vienna monitoring method, in Conf. Rec. IEEE-IAS Annu. Meeting, St. Louis, MO, Oct. 1998, pp. 278285. [18] A. Bellini, F. Filippetti, G. Franceschini, and C. Tassoni, Closed loop influence on induction machine asymmetries effects, in Proc. SDEMPED99, Gijn, Spain, Sept. 1999, pp. 4350. [19] F. Filippetti, G. Franceschini, C. Tassoni, and P. Vas, Transient modeling oriented to diagnostic of induction motor machines with rotor asymmetries, in Proc. ICEM1994, vol. 3, Paris, France, Sept. 1994, pp. 6267. [20] P. Vas, Simulation and monitoring of induction motors with rotor asymmetries, in Proc. ICEM92, Manchester, U.K., Sept. 1992, pp. 435439. [21] F. Filippetti, G. Franceschini, and C. Tassoni, Neural network aided on-line diagnosis of induction motor rotor faults, IEEE Trans. Ind. Applicat., vol. 31, pp. 892899, July/Aug. 1995. [22] F. Filippetti, G. Franceschini, C. Tassoni, S. Meo, and A. Ometto, A simplified model of induction motor with stator shorted turns oriented to diagnostics, in Proc. ICEM96, vol. III, Vigo, Spain, Sept. 1996, pp. 410413. [23] G. Gentile, S. Meo, A. Ometto, N. Rotondale, F. Filippetti, G. Franceschini, and C. Tassoni, 19891998: State of art of model based diagnostic procedure for IM stator inter-turn short circuits, in Proc. SDEMPED99, Gijn, Spain, Sept. 1999, pp. 1931.

Fig. 29. 10.

FFT spectrum of i current with zero, one, and two broken bars, k

Fig. 30. 10.

FFT spectrum of i current with zero, one, and two broken bars, k =

As for controlled machines, these indexes are no more effective, as their information is masked by control action. Therefore, the behavior of other variables have been investigated with the aim of finding new indexes suitable for diagnostic purposes. Simulation and experimental results have proved that the spectrum of the field current can be an effective diagnostic index when the field-oriented control scheme is adopted. In fact, the and amplitudes of the spectrum components at frequencies are almost constant with proportional gain variations and they seem quite linearly dependent on stator and rotor faults degrees, respectively. On the contrary, the corresponding component of torque current is dependent on control gain, load conditions and, overall, on the frequency.

BELLINI et al.: DIAGNOSIS OF INDUCTION MOTORS FAULTS

1329

[24] M. Jufer and M. Abdelaziz, Influence dune ropture de barre on dun anneau sur les caracteristiques extremes dun moteur asynchrone a cage, Bull. Assoc. Suisse Electriciens, pp. 921925, 1978. [25] S. Williamson and A. C. Smith, Steady-state analysis of three-phase cage motors with rotor bar and end ring faults, Proc. Inst. Elect. Eng., pt. B, vol. 129, no. 3, pp. 93100, 1982. [26] S. Williamson and K. Mirzoian, Analysis of cage induction motors with stator winding faults, IEEE Trans. Power App. Syst., vol. PAS-104, pp. 155163, July 1985. [27] G. Gentile, N. Rotondale, F. Filippetti, G. Franceschini, M. Martelli, and C. Tassoni, Analysis approach of induction motor stator faults to on-line diagnostics, in Proc. ICEM90, Cambridge, MA, Aug. 1990, pp. 491497. [28] G. Gentile, N. Rotondale, M. Martelli, and C. Tassoni, Harmonic analysis of induction motors with stator faults, Elect. Mach. Power Syst., no. 22, pp. 215231, 1994. [29] G. J. Retter, Matrix and Space-Phasor Theory of Electrical Machines. Budapest, Hungary: Akademiaikiado, 1987. [30] P. Vas, Electrical Machines and Drives, A Space-Vector Approach. Oxford, U.K.: Oxford Science, 1992. [31] C. Tassoni, What are the links between terminal behavior and machine faults, presented at the IEEE-IAS Annu. Meeting, Lake Buena Vista, FL, Oct. 1995. [32] F. Filippetti, G. Franceschini, C. Tassoni, and P. Vas, Integrated condition monitoring and diagnosis of electrical machines using minimum configuration artificial intelligence, in Proc. EPE97, vol. 2, Trondheim, Norway, Sept. 1997, pp. 983988. [33] F. Filippetti, G. Franceschini, C. Tassoni, and P. Vas, AI techniques in induction machine diagnosis including the speed ripple effect, IEEE Trans. Ind. Applicat., vol. 34, pp. 98108, Jan./Feb/ 1998. [34] F. Filippetti, G. Franceschini, M. Martelli, and C. Tassoni, Development of the knowledge base of an expert system to diagnose rotor electric faults of induction motors, in Conf. Rec. IEEE-IAS Annu. Meeting, Houston, TX, Oct. 1992, pp. 9299.

Fiorenzo Filippetti was born in Fano, Italy, in 1945. He received the Masters degree in electrical engineering from the University of Bologna, Bologna, Italy, in 1970. He joined the University of Bologna in 1976 as an Assistant Professor and is currently an Associate Professor of Electrotechnics in the Department of Electric Engineering. He has authored or coauthored more than 90 published scientific papers and is the holder of an industrial patent. His main research interests include the simulation and modeling of electric circuits and systems and the application of diagnostic techniques to circuits and systems. Prof. Filippetti is a member of the Italian Association of Electric and Electronics Engineers (AEI) and of the European Consortium for Research on Condition Monitoring of Electric Systems and Drives (CRCM).

Giovanni Franceschini (A97) was born in Castelnovo ne Monti, Italy, in 1960. He received the Masters degree in electronic engineering from the University of Bologna, Bologna, Italy. He is currently an Associate Professor at the University of Parma, Parma, Italy. His current research interests include high-performance electric drives and diagnostic techniques for industrial electric systems. He has authored or coauthored more than 60 published technical papers on these topics.

Alberto Bellini (S96A98) was born in Forl, Italy, in 1969. He received the M.S. and Ph.D. degrees in electronics engineering from the University of Bologna, Bologna, Italy, in 1994 and 1998, respectively. Since 1997, he has been with the University of Parma, Parma, Italy, where he currently is an Assistant Professor of Electrical Engineering. His research interests include applications of artificial intelligence to nonlinear signal processing, industrial drive control, and diagnostics. He has coauthored more than 30 published scientific conference and journal papers.

Carla Tassoni (A90SM92) was born in Bologna, Italy, in 1942. She received the Masters degree in electrical engineering from the University of Bologna, Bologna, Italy, in 1966. She joined the University of Bologna as an Assistant Professor and then became an Associate Professor of Electrical Machines in the Department of Electric Engineering. She is currently a Full Professor of Electrical Engineering at the University of Parma, Parma, Italy. She has authored or coauthored more than 100 published scientific papers and is the holder of an industrial patent. Her main research interests include the simulation and modeling of electric systems and applications of diagnostic techniques. Prof. Tassoni is a senior member of the European Consortium for Research on Condition Monitoring of Electric Systems and Drives (CRCM).

Das könnte Ihnen auch gefallen