05508416

7/28/2019 05508416

1/9

IEEE TRANSACTIONS ON INDUSTRY APPLICATIONS, VOL. 46, NO. 5, SEPTEMBER/OCTOBER 2010 1873

An Active Stator Temperature Estimation Techniquefor Thermal Protection of Inverter-FedInduction Motors With Considerations

of Impaired Cooling DetectionPinjia Zhang, Member, IEEE, Bin Lu, Senior Member, IEEE, and Thomas G. Habetler, Fellow, IEEE

AbstractThermal protection is one of the most importantaspects of any motor control system. This paper proposes astator winding temperature estimation method for the thermalprotection of inverter-fed electric motors. By modifying the spacevector pulsewidth modulation in an open-loop motor drive, a dcvoltage can be intermittently injected into the motor. The statortemperature can be estimated by measuring only the dc compo-

nent of the phase current under both constant- and variable-loadconditions. The evaluation of the resultant torque pulsation andthe compensation for serial resistances are also discussed. Theproposed stator temperature estimation method is validated fromexperimental results under variable-load conditions and bothhealthy and impaired cooling conditions. The error in the statortemperature estimation is within 8 C under different operatingconditions. The significance of this method lies in its non-intrusivenature: only current sensors are required for implementation thenormal operation of the motor is not interrupted.

Index TermsCooling, induction motor protection, motordrives, signal injection, stator resistance estimation, temperatureestimation, thermal protection.

I. INTRODUCTION

THERMAL protection is one of the most important aspects

of any condition monitoring system for electric motors.

Thermal overload can lead to failures of the stator winding

insulation, bearings, motor conductors, cores, etc. [1], [2].

An often-quoted rule of thumb says that the motors life is

reduced by 50% for every 10 C above the temperature limit.Therefore, since a motor must be tripped off immediately when

the temperature limit is reached, accurate stator temperature

monitoring is critical for prolonging a motors lifetime. In

Manuscript received June 9, 2009; revised September 15, 2009 and

December 13, 2009; accepted December 24, 2009. Date of publication July 12,2010; date of current version September 17, 2010. Paper 2009-EMC-173.R2,presented at the 2009 IEEE International Electric Machines and Drives Con-ference, Miami, FL, May 36, and approved for publication in the IEEETRANSACTIONS ON INDUSTRY APPLICATIONS by the Electric MachinesCommittee of the IEEE Industry Applications Society.

P. Zhang is with the Electrical Machines Laboratory, General Electric (GE)Global Research, Niskayuna, NY 12309, USA (e-mail: [email protected]).

T. G. Habetler is with the School of Electrical and Computer Engi-neering, Georgia Institute of Technology, Atlanta, GA 30332 USA (e-mail:[email protected]).

B. Lu is with the Innovation Center, Eaton Corporation, Milwaukee,WI 53216 USA (e-mail: [email protected]).

Color versions of one or more of the figures in this paper are available onlineat http://ieeexplore.ieee.org.

Digital Object Identifier 10.1109/TIA.2010.2057391

addition, a thermal overload can be caused not only by a

motor overload but also by cooling capability deterioration.

In the case of cooling capability deterioration, it is crucial to

inspect and repair the motor as early as possible to avoid the

temperature rise caused by the impaired cooling capability

and the resultant deterioration of motor components [3]. With

the popularization of motor drives, online thermal protection

of inverter-connected electric motors is becoming highly

desirable. Because of the high cost of embedded thermal

sensors and their installation, sensorless monitoring of the

stator winding temperature is crucial for the thermal protection

of, in particular, small- to medium-size motors. The estimation

of stator temperature should not only be accurate but also

robust to the variations in the motors cooling capability in

order to allow for detection of a possible cooling problem.

Many thermal-model-based stator temperature estimation

techniques have been proposed over the years [4][10]. An

empirical thermal model for inverter-fed induction machines

is proposed in [10]. The effects of operating conditions onthe stator temperature estimation, such as rotor speed, input

frequency, etc., are considered. However, these methods cannot

adapt to the changes in the motors cooling capability, and an

accurate identification of thermal parameters is difficult without

embedded thermal sensors [11]. Aside from thermal-model-

based approaches, motor-parameter-based methods, which es-

timate the stator winding temperature (Ts) by estimating thestator winding resistance (Rs), are preferred because of theirrobustness to the variations in the motors cooling capability

[12]. Many electrical-model-based stator resistance estimation

methods have been proposed for improving rotor flux esti-

mation or sensorless speed estimation accuracy in the low-speed region for inverter-fed machines [13][17]. However,

these approaches are shown in [18] to be highly sensitive to

motor parameter variations. As a result, dc-signal-injection-

based stator resistance estimation methods, which use a motors

dc model, are proposed for thermal protection of line- and

soft-starter-connected electric machines [3], [19][22]. These

approaches are shown to be accurate and robust to the variations

in the motors cooling capability.

With the wide application of motor drives, the sensorless

estimation of stator winding resistance draws more attention

not only for thermal protection purposes but also for improving

closed-loop speed control performances. It is proposed in [23]

0093-9994/$26.00 2010 IEEE

7/28/2019 05508416

2/9

1874 IEEE TRANSACTIONS ON INDUSTRY APPLICATIONS, VOL. 46, NO. 5, SEPTEMBER/OCTOBER 2010

Fig. 1. Typical scalar control scheme for induction motors.

Fig. 2. Modified SVPWM for dc-signal injection.

and [24] to estimate stator winding resistance via dc-signal

injection. However, the application of the proposed technique is

limited due to its requirements of voltage measurements, which

are typically not available in motor drives. Plotkin et al. [25]

proposed a dc-signal-injection-based voltage-sensorless stator

resistance/temperature estimation technique, but a lookup table

is needed for the compensation of system errors.This paper proposes a nonintrusive dc-signal-injection-based

stator winding temperature estimation method for thermal

protection of open-loop drive-fed electric motors, using only

current sensors. Section II shows a dc-signal injection method

using open-loop motor drives by modifying the space vector

pulsewidth modulation (SVPWM). The resultant torque pulsa-

tion caused by the injected dc signal can also be estimated based

on the monitoring of phase current. Section III introduces the

stator temperature estimation approach based on the dc-signal

injection by monitoring only the dc component of the phase

current under both constant- and variable-load conditions. The

practical implementation considerations of series resistances

and their compensation are also discussed. The experimentalvalidation of the proposed method is shown in Section IV

with experimental results of two induction motors under both

constant- and variable-load conditions, when the motors cool-

ing capability is healthy or impaired. The errors in the stator

temperature estimation from experimental testing are within

8 C under different operating conditions.

II. NONINTRUSIVE DC-SIGNAL INJECTION METHOD

USING OPE N-L OO P MOTOR DRIVES

A. Open-Loop Drive and SVPWM

The structure of a typical open-loop motor drive using scalarcontrol, which is also known as V/f control, is shown in

Fig. 1. The boost voltage is V0 for the motor start-up, is

the input frequency command, and V1 is the calculated inputvoltage magnitude by the field-weakening function K(). Thethree-phase input voltage control signal (vabc) is calculatedbased on . Using the dq transform, the three-phase voltagecontrol signal (vabc) can be transformed to the dq voltage

control vector (vdqs). Then, given the voltage control vector,the SVPWM can achieve accurate input voltage control bycontrolling the power switches in the converter.

B. Modified SVPWM for DC-Signal Injection

To the injected dc signals, a dc-voltage control vector

(vdc,dqs) is added to the original voltage control vector (vdqs)in the stationary dqreference frame, as shown in Fig. 2. Whena dc voltage (vdc) is injected between phase a and phases band c, the dc-voltage control vector can be calculated as

vdc,dqs =2

3 1

1

2 1

20 3

2

32

vavbvc

=2

3 1

2

vab +1

2

vac

32vbc

=

2

3

12vdc +

12vdc

32 0

=

2

3

vdc

0

. (1)

By adding the dc-voltage control vector, a controllable dc

voltage can be intermittently injected into the motor for thermal

protection purposes. Meanwhile, the original voltage control

vector, which is given by scalar control, is still applied to

maintain the normal operation of the motor. The modified

SVPWM is shown in Fig. 2. Since the magnitude of the dc-

voltage control vector is typically much smaller than that of the

original voltage vector, the operating region of the SVPWM isnot largely affected.

7/28/2019 05508416

3/9

ZHANG et al.: TECHNIQUE FOR THERMAL PROTECTION OF INDUCTION MOTORS 1875

C. Evaluation of Torque Pulsation

While the fundamental-frequency current induces a constant

output torque, the injected dc signal induces an output torque

oscillating at the fundamental frequency. Let

i dqs and

dqsbe the stator current and total flux linkage space vectors in

thed

q

stationary reference frame, respectively. The air-gap

torque (Tag) can be calculated as the cross product of

dqs and

i dqs, i.e.,

Tag =3P

4

dqs i dqs (2)

where P is the number of poles. It is shown in [21] thatthe torque pulsation caused by the injected dc signal and the

constant torque induced by the fundamental-frequency current

can be evaluated, respectively, as

Tdcag

3P4

dqs

i

dqs

Tag 3P4

dqs

idc

dqs

(3)where is the fundamental frequency. Therefore, the relativetorque pulsation can be shown as

TagTdcag

dqs

idc

dqs

dqs i dqs =

dqsi

dc

dqs

dqsi dqs

cos()=

Ia,dcIa,peak cos()

(4)

where cos() is the power factor and Ia,dc and Ia,peak representthe magnitude of the dc component and the peak value of the ac

component in phase-a current, respectively, assuming that thedc signal is injected between phases a, b, and c.

Therefore, by using (3) and (4), the torque pulsation can be

evaluated by monitoring the dc current and the power factor.

Since the injected dc current can be controlled by adjusting

the dc-voltage command (vdc), given the allowable relativetorque pulsation, the torque pulsation can be controlled within

an acceptable range.

III. REMOTE AND SENSORLESS STATOR

TEMPERATURE ESTIMATION SCHEME

A. Current-Based Stator Temperature Estimation

With the dc signal injected using the modified SVPWM, re-

sistance and stator temperature can be calculated, respectively,

as [21]

Rs =2 vdcab3 idca

(5)

Ts =Ts0 +(Rs Rs0)

Rs0(6)

where idca and vdcab represent the dc components in the phase

current (ia) and line-to-line voltage (vab), respectively; Ts0 andRs0 represent Ts and Rs at room temperature, respectively;

Ts and Rs are the estimated Ts and Rs from the dc-signalinjection, respectively; and is the temperature coefficient ofresistivity with respect to Ts0.

As only current sensors are present in most of the motor

drives, it is preferred to avoid using voltage measurements.

Because of the nonideal switching of insulated-gate bipolar

transistors (IGBTs), such as the dead time and dwell time,the actually injected dc voltage does not accurately follow the

dc-voltage command. The compensation of the nonideality of

the switching is difficult, which makes it difficult to accurately

estimate the injected dc voltage from the dc-voltage command

[25]. However, under constant-load conditions, the injected dc

voltage is found to be nearly constant from the experimental

results. Therefore, it can be assumed that the injected dc voltage

remains constant under constant-load conditions. Based on this

assumption, voltage measurements can be avoided during the

estimation of stator resistance, as

Ts =Ts0 1

+

RsRs0

=Ts0 1

+

1

v

dcab/i

dca

vdcab/idca0

= Ts0 1

+

1

i

dca0

idca(7)

where idca0 represents the dc component in the phase cur-rent when the stator temperature is Ts0. Therefore, the statorwinding temperature can be estimated using only the magni-

tude of the injected dc current under constant-load conditions.

Typically, the stator temperature can be considered as the room

temperature right after cold start, and therefore, the dc current

right after starting can be estimated as idca0.In the case of load change, the injected dc voltage changes,

as the variations of the magnitude of the current may lead tothe variations of the injected dc voltage, due to the nonideality

of the switching of IGBTs. However, it can be assumed that,

before and after the load change, the stator winding tempera-

ture variations can be neglected. Therefore, the change of the

magnitude of the injected dc current caused by the load change

can be compensated using a rescaling process, as

idca = idca,load2

idca,load1(t0)idca,load2(t0+)

(8)

where idca,load1 and idca,load2 represent the measured dc currents

under load-1 and load-2 conditions, respectively; t0 representsthe time when the load condition is changed from loads 1 to 2;

idca,load1(t0) and idca,load2(t0+) are the dc currents measuredright before and after the load change, respectively; and idcais the rescaled dc current after compensation for the load

change, which is periodically updated for the estimation of

stator winding resistance and temperature. Based on this, the

stator winding temperature can be monitored under the new

load condition using (7).

For continuous-load-variation conditions, however, it is dif-

ficult to estimate the stator winding temperature only based

on the stator current measurement. This is due to the inac-

curate injection of the dc-voltage signal, which is because

of the nonideal switching of the power switches. In the caseof continuous-load-variation applications, additional voltage

7/28/2019 05508416

4/9


Fig. 3. DC model of the motor drive system.

sensors are required for measuring the injected dc voltage.

With the measurement of both dc voltage and current, the

stator winding resistance and temperature can be continuously

monitored online.

B. Compensation for Series Resistances

Since motor drives are normally installed in the motor control

center, long cables may be present between the motor and motor

drives. The cable resistance and the internal resistance of motor

drives may be comparable to the stator resistance, which largely

decreases the accuracy of stator temperature estimation when

(7) is used. Therefore, the compensation of series resistances,

including motor drive internal resistance and cable resistance, is

crucial for the accurate estimation of stator temperature. The dc

model of the motor drive system is shown in Fig. 3, neglecting

contact resistances. Rcable represents the cable resistance, andRdrive denotes the internal resistance of motor drives.

As suggested in [3] and [21], the cable resistance can be

estimated based on the length and the size of the cable. It

is shown in [21] that the stator temperature estimation error

can be neglected after the compensation of cable resistancefor small- to medium-size induction motors. For more accurate

estimation of the cable resistance, the cable resistance can be

experimentally measured during the installation of motor drives

with the motor terminals shorted. Such testing can be one of

the initial tests required during the installation of motor drives

for obtaining the parameters of the motor system. Therefore,

the stator temperature can be remotely monitored in the motor

control center.

The internal resistance of motor drives consists of the equiva-

lent resistance of IGBT, cable contact resistances, etc. The cable

and contact resistances can be assumed constant under normal

operating conditions. The equivalent resistance of IGBT mayvary slightly with different temperature, different magnitude of

the gate signal, and different magnitude of the phase current.

For some applications, these variations may be negligible com-

pared to the stator resistance. Therefore, the internal resistance

of motor drives can be assumed constant and predetermined

before installation. However, for some other applications, these

variations might not be negligible. Therefore, the internal resis-

tance of motor drives needs to be predetermined with different

magnitudes of input current and different temperatures to form

a lookup table for online compensation. Since temperature

sensors are typically attached to the IGBT modules to monitor

their temperature in motor drives, both the temperature of

IGBT and the magnitude of the phase current can be monitoredonline. Therefore, the internal resistance of motor drives can

be estimated online using the predetermined lookup table. The

pretesting of the internal resistance of motor drives is only

required once for each model of motor drives and stored in the

signal processing chip for online compensation.

Under constant-load conditions, it can be assumed that the

injected dc voltage is constant, as

idca (Rs + Rcable + Rdrive) = idca0(Rs0 + Rcable + Rdrive0)

(9)

where Rdrive and Rdrive0 represent the internal resistances ofmotor drives when the stator resistance is Rs and Rs0, respec-tively. Based on the estimation of cable and internal resistances

of motor drives, the stator temperature can be estimated as

Ts =Ts0 +RsRs0

1

=Ts0 1

+ i

dc

a0(Rs0 + Rcable + Rdrive0) idc

a (Rcable + Rdrive) idca Rs0.

(10)

Therefore, the effects of series resistances can be compen-

sated for improving the accuracy of the stator temperature

estimation.

C. Overall Thermal Protection Scheme

The overall thermal protection scheme for open-loop drive-

fed induction motors is shown in Fig. 4. The dc signals

are injected by using the modified SVPWM, as shown in

Section II-B. The relative torque pulsation caused by the in-jected dc signal can be estimated using (4) by monitoring

the phase current. Therefore, the relative torque pulsation can

be controlled within an acceptable range by adjusting the dc-

voltage command. On the other hand, it is understandable that

the accuracy of the stator temperature estimation is highly de-

pendent on the magnitude of the injected dc voltage. Therefore,

the determination of the dc-voltage command is a tradeoff

between acceptable torque pulsation and the accuracy of the

stator temperature estimation.

Since dc-signal injection causes undesirable torque pulsa-

tion, it is not necessary to inject the dc signal and estimate the

stator temperature and resistance continuously. DC signals canbe periodically injected for a minimal time interval that is suf-

ficient to obtain an accurate estimate of the stator temperature

while small enough not to cause unacceptable torque pulsation.

From the experimental results of this paper, it is suggested to

inject dc signals for 1 s each time to obtain an accurate estimate

of the stator temperature. Given a typical motor thermal time

constant, a period of 510 min for the stator temperature update

is sufficient for thermal protection purposes, depending on

the requirements of practical application. Therefore, the motor

performance is only affected by dc-signal injection for 1 s every

510 min. In this paper, for validation purposes, the dc signals

are injected for 1 s every 1 min.

The importance of the proposed thermal protection schemelies in its nonintrusive nature: Only current sensors are required

7/28/2019 05508416

5/9


Fig. 4. Overall thermal protection scheme.

TABLE INAMEPLATE INFORMATION OF THE EXPERIMENTAL SETUP

Fig. 5. Experimental setup.

for implementation; normal operation of the motor is not

interrupted.

IV. EXPERIMENTAL VALIDATION

A. Experimental Setup

The proposed thermal monitoring scheme is tested on two

induction motors, whose ratings and parameters are shown in

Table I. An Eaton SPX9000 motor drive is programmed toinject the dc signal using the modified SVPWM, as stated

in Section II. The switching frequency of the inverter is set

as 5 kHz. A 10-hp dc generator supplying a resistor bank is

connected to the tested motor to vary the load conditions by

adjusting the resistance of the resistor bank. The motor phase

current is measured using Hall-effect sensors. The data are

then acquired and stored using a NI LabView system with 16-b

A/D conversion at 100-kHz sampling frequency. The motors

are each equipped with nine K-type thermocouples at different

locations (three in each phase) in the stator windings to record

the average stator winding temperature for validation purposes,as shown in Fig. 5.

7/28/2019 05508416

6/9


Fig. 6. Stator current with dc-signal injection.

B. DC-Signal Injection

The stator phase current of motor 1 during dc-signal injection

is shown in Fig. 6. A low-pass filter with a cutoff frequency

of 500 Hz is used to remove the current harmonics caused by

the switching of IGBTs. It can be observed that, by using the

modified SVPWM, a dc signal is successfully injected into the

motor. While the dc-voltage command is set as 5 V, the injected

dc current is around 7 A. Therefore, based on the dc-signal

injection, the stator winding resistance and temperature can be

estimated using the motors dc model.

C. Stator Temperature Estimation UnderConstant-Load Condition

Based on the monitoring of the stator current, the stator

winding temperature can be determined. The internal resistance

of motor drives is assumed constant and must be predetermined

or estimated a priori, together with the cable resistance. The

effects of series resistances can be compensated using (10). The

estimated stator winding temperature based on dc injection for

motor 1 under constant-load conditions is shown in Fig. 7. The

tested motor is operated under no load, 30%, 60%, and 90% of

the rated load with a rated input frequency command of 60 Hz.

The estimated stator winding temperature for motor 1 with an

input frequency of 30 Hz is shown in Fig. 8, where the motor isoperated under 30% and 45% of the rated load, respectively.

The measured temperatures are calculated using the average

temperature measured from the preinstalled thermocouples for

validation purposes. It can be seen from Figs. 7 and 8 that the

stator winding temperature can be accurately monitored using

only the current measurements under constant-load conditions

with different input frequencies. The maximum error of Tsestimation is within 8 C.

D. Stator Temperature Estimation Under

Variable-Load Conditions

To test the feasibility of the proposed stator temperatureestimation scheme under variable load conditions, motor 2 is

operated under variable-load conditions (no load 100% 50% 75% of the rated load). The effects of the load changeon the dc-signal injection are compensated using (8) for each

load change. To remove the measurement noise, the measured

currents before and after the load change for rescaling are

obtained by nonlinear curve fitting of the dc current measured

under each load condition. The stator temperature estimationresults are shown in Fig. 9. The maximum error in the stator

temperature estimation is within 7 C. It can be observedin Fig. 9 that, by using (8), the proposed thermal protection

scheme is capable of providing accurate estimation of the stator

temperature under variable-load conditions.

E. Stator Temperature Estimation With Impaired Cooling

As stated in Section I, it is crucial that the stator temperature

can be accurately estimated when the motors cooling capability

is deteriorated so that the user can be warned for inspection or

repair of the motor with impaired cooling. To test the feasibility

of the proposed stator temperature estimation scheme in the

case of impaired cooling, a paper foil is attached to the end

of motor 2 to partly block the ventilation, as shown in Fig. 10.

For comparison purposes, the motor is operated again under

variable-load conditions (no load 100% 50% 75% ofthe rated load). The stator temperature estimation results are

shown in Fig. 11. It can be observed that the proposed stator

temperature estimation scheme can provide accurate estimation

of the stator temperature for determining whether the cooling

capability of the motor is healthy or impaired. It can be ob-

served from the comparisons of Figs. 9 and 11 that impaired

cooling induces an increased stator temperature rise under the

same load condition. Therefore, the stator temperature estima-tion, in addition to improved traditional protection, can also be

used to detect the abnormal cooling capability of the motor

so that the user can be warned for inspection or repair of the

induction motor. The development of such an impaired cooling

detection technique is out of the scope of this paper, but from

Fig. 11, it can be observed that the proposed technique can

provide accurate stator temperature estimation under impaired

cooling conditions, which is essential for the online detection

of impaired cooling.

V. CONCLUSION

An active stator temperature estimation scheme has been

proposed in this paper for the thermal protection of inverter-fed

induction motors. DC signals are intermittently injected into the

motor using a modified SVPWM pattern. The stator winding

temperature can then be estimated based on the monitoring of

only the stator phase current under both constant- and variable-

load conditions.

The torque pulsation caused by the injected dc signals has

been evaluated so that the torque pulsation can be controlled

within an acceptable range by adjusting the dc-voltage com-

mand in the modified SVPWM. In addition, a compensation

technique for series resistances, including the internal resis-

tance of motor drives and cable resistance, has been suggestedto improve the accuracy of the stator temperature estimation.

7/28/2019 05508416

7/9


Fig. 7. Stator temperature estimation with an input frequency of 60 Hz (motor 1). (a) No load. (b) 30% load. (c) 60% load. (d) 90% load.

Fig. 8. Stator temperature estimation with an input frequency of 30 Hz (motor 1). (a) 30% load. (b) 45% load.

The proposed stator temperature estimation scheme has been

validated from experimental results on two induction motors

with different ratings. It has been shown that the proposed

stator temperature estimation scheme is capable of providing

accurate stator temperature estimation under both constant-and variable-load conditions and both healthy and impaired

cooling conditions. The errors in the stator winding temperature

estimation from experimental testing are within 8 C underdifferent operating conditions.

The proposed stator temperature estimation scheme can pro-

vide reliable protection of inverter-fed induction motors underboth healthy and impaired cooling conditions, which makes

7/28/2019 05508416

8/9


Fig. 9. Stator temperature estimation (motor 2).

Fig. 10. Impaired cooling by blocking ventilation (motor 2).

Fig. 11. Stator temperature estimation with impaired cooling (motor 2).

it feasible for impaired cooling detection. The importance of

the proposed stator temperature estimation scheme lies in its

nonintrusive nature.

1) Only current sensors are required for implementation.2) The normal operation of the motor is not interrupted.

REFERENCES

[1] Motor Reliability Working Group, Report of large motor reliability sur-vey of industrial and commercial installationsPart I, IEEE Trans. Ind.

Appl., vol. IA-21, no. 4, pp. 853864, Jul. 1985.[2] S. Grubic, J. M. Aller, B. Lu, and T. G. Habetler, A survey

on testing and monitoring methods for stator insulation systems oflow-voltage induction machines focusing on turn insulation prob-lems, IEEE Trans. Ind. Electron., vol. 55, no. 12, pp. 41274136,Dec. 2008.

[3] S.-B. Lee and T. G. Habetler, A remote and sensorless thermal protectionscheme for small line-connected ac machines, IEEE Trans. Ind. Appl.,vol. 39, no. 5, pp. 13231332, Sep./Oct. 2003.

[4] K. D. Hurst and T. G. Habetler, A thermal monitoring and parametertuning scheme for induction machines, in Conf. Rec. 32nd IEEE IAS

Annu. Meeting , 1997, vol. 1, pp. 136142.[5] M. S. Abou-El-Ela, A. I. Megahed, and O. P. Malik, Thermal model

based digital relaying algorithm for induction motor protection, in Proc.Can. Conf. Elect. Comput. Eng., 1996, vol. 2, pp. 10161019.

[6] A. Bousbaine, M. McCormick, and W. F. Low, In-situ determinationof thermal coefficients for electrical machines, IEEE Trans. EnergyConvers., vol. 10, no. 3, pp. 385391, Sep. 1995.

[7] P. H. Mellor, D. Roberts, and D. R. Turner, Lumped parameterthermal model for electrical machines of TEFC design, Proc. Inst.

Elect. Eng.Electric Power Appl., vol. 138, no. 5, pp. 205218,

Sep. 1991.[8] J. F. Moreno, F. P. Hidalgo, and M. D. Martinez, Realisation of tests to

determine the parameters of the thermal model of an induction machine,Proc. Inst. Elect. Eng.Electr. Power Appl., vol. 148, no. 5, pp. 393397,Sep. 2001.

[9] H. Nestler and P. K. Sattler, On-line estimation of temperatures in elec-trical machines by an observer, Electr. M ach. Power Syst., vol. 21, no. 1,pp. 3950, Jan. 1993.

[10] J. T. Boys and M. J. Miles, Empirical thermal model for inverter-drivencage induction machines, Proc. Inst. Elect. Eng.Electr. Power Appl.,vol. 141, no. 6, pp. 360372, Nov. 1994.

[11] D. Staton, A. Boglietti, and A. Cavagnino, Solving the more difficultaspects of electric motor thermal analysis in small and medium sizeindustrial induction motors, IEEE Trans. Energy Convers., vol. 20, no. 3,pp. 620628, Sep. 2005.

[12] B. Lu, T. G. Habetler, and R. G. Harley, A nonintrusive and in-servicemotor-efficiency estimation method using air-gap torque with consider-ations of condition monitoring, IEEE Trans. Ind. Appl., vol. 44, no. 6,pp. 16661674, Nov./Dec. 2008.

[13] L. Zhen and L. Xu, Sensorless field orientation control of inductionmachines based on a mutual MRAS scheme, IEEE Trans. Ind. Electron.,vol. 45, no. 5, pp. 824831, Oct. 1998.

[14] E. D. Mitronikas, A. N. Safacas, and E. C. Tatakis, A new stator resis-tance tuning method for stator-flux-oriented vector-controlled inductionmotor drive, IEEE Trans. Ind. Electron., vol. 48, no. 6, pp. 11481157,Dec. 2001.

[15] M. Tsuji, C. Shuo, K. Izumi, and E. Yamada, A sensorless vector con-trol system for induction motors using q-axis flux with stator resistanceidentification, IEEE Trans. Ind. Electron., vol. 48, no. 1, pp. 185194,Feb. 2001.

[16] F. Zidani, D. Diallo, M. E. H. Benbouzid, and R. Nait-Said, Direct torquecontrol of induction motor with fuzzy stator resistance adaptation, IEEETrans. Energy Convers., vol. 21, no. 2, pp. 619621, Jun. 2006.

[17] R. Beguenane, M. El Hachemi Benbouzid, M. Tadjine, and A. Tayebi,Speed and rotor time constant estimation via MRAS strategy for induc-tion motor drives, in Conf. Rec. IEEE Int. Electr. Mach. Drives Conf.,1997, pp. TB3/5.1TB3/5.3.

[18] S.-B. Lee, T. G. Habetler, R. G. Harley, and D. J. Gritter, An evaluationof model-based stator resistance estimation for induction motor statorwinding temperature monitoring, IEEE Trans. Energy Convers., vol. 17,no. 1, pp. 715, Mar. 2002.

[19] S.-B. Lee and T. G. Habetler, An online stator winding resistance es-timation technique for temperature monitoring of line-connected induc-tion machines, IEEE Trans. Ind. Appl., vol. 39, no. 3, pp. 685694,May/Jun. 2003.

[20] P. Zhang, Y. Du, B. Lu, and T. G. Habetler, A remote and sensorlessthermal protection scheme for soft-starter-connected induction motors,in Conf. Rec. IEEE IAS Annu. Meeting, 2008, pp. 17.

[21] P. Zhang, B. Lu, and T. G. Habetler, A remote and sensorless stator wind-

ing resistance estimation method for thermal protection of soft-starter-connected induction machines, IEEE Trans. Ind. Electron., vol. 55,no. 10, pp. 36113618, Oct. 2008.

7/28/2019 05508416

9/9


[22] D. A. Paice, Motor thermal protection by continuous monitoring of wind-ing resistance, IEEE Trans. Ind. Electron. Control Instrum., vol. IECI-27,no. 3, pp. 137141, Aug. 1980.

[23] L. A. de Souza Ribeiro, C. B. Jacobina, and A. M. N. Lima, Linearparameter estimation for induction machines considering the operatingconditions, IEEE Trans. Power Electron., vol. 14, no. 1, pp. 6273,Jan. 1999.

[24] S. D. Wilson, G. W. Jewell, and P. G. Stewart, Resistance estimation for

temperature determination in PMSMs through signal injection, in Proc.IEEE Int. Conf. Electr. Mach. Drives, 2005, pp. 735740.

[25] J. Plotkin, M. Stiebler, and D. Schuster, A novel method for online sta-tor resistance estimation of inverter-fed ac-machines without temperaturesensors, in Proc. 11th Int. Conf. OPTIM, 2008, pp. 155161.

Pinjia Zhang (S06M10) received the B.Eng. de-gree in electrical engineering from Tsinghua Uni-versity, Beijing, China, in 2006, and the Masterand Ph.D. degrees in electrical engineering from theGeorgia Institute of Technology, Atlanta, in 2009and2010, respectively.

Since May 2010, he has been with the ElectricalMachines Laboratory, General Electric (GE) GlobalResearch Center, Schenectady, NY. His research in-terests include electric machine design, protectionand diagnostics, motor drives, power electronics, and

artificial intelligence and its applications in power systems. He has publishedover 20 papers in refereed journals and international conference proceedingsand has four patent applications in these areas.

Mr. Zhang was the recipient of the second prize in the student paper and

poster contest of the IEEE Power Energy Society General Meeting, Pittsburgh,PA, in July 2008. He was also the recipient of the GE Student Intern/Co-opContribution Award (SICCA) in 2009.

Bin Lu (S00M06SM09) received the B.Eng.degree in automation from Tsinghua University,Beijing, China, in 2001, the M.S. degree in electricalengineering from the University of South Carolina,Columbia, in 2003, and the Ph.D. degree in electricalengineering from the Georgia Institute of Technol-ogy, Atlanta, in 2006.

In summer 2006, he was with the Manufacturing

Research Laboratory, General Motors R&D Center,Warren, MI, as a Research Engineer Intern. SinceOctober 2006, he has been with the Innovation

Center, Eaton Corporation, Milwaukee, WI, where he is currently a SeniorEngineering Specialist. Since February 2010, he has been with the Eaton ChinaInnovation Center, Shanghai, China, responsible for its setup and operation.His research interests include controls and diagnostics of electric machinesand power electronics, computational intelligence applied to energy systems,integration and protection of renewable energy sources, modeling and simula-tion, and electric load identification and monitoring in building applications. Hehas published over 50 papers in refereed journals and international conferenceproceedings and has 13 U.S. and international patent applications in these areas.

Dr. Lu was the recipient of the Second Prize Transactions Paper Award fromthe IEEE TRANSACTIONS ON INDUSTRY APPLICATIONS in 2008.

Thomas G. Habetler (S82M83SM92F02)received the B.S.E.E. and M.S. degrees in electricalengineering from Marquette University, Milwaukee,WI, in 1981 and 1984, respectively, and the Ph.D.degree from the University of Wisconsin, Madison,in 1989.

From 1983 to 1985, he was a Project Engineerwith the Electro-Motive Division, General Motors.Since 1989, he has been with the Georgia Insti-tute of Technology, Atlanta, where he is currentlya Professor of electrical engineering in the School

of Electrical and Computer Engineering. His research interests are in electricmachine protection, condition monitoring, and drives, and he has publishedover 150 papers in these fields. He is a regular Consultant to industry in thefield of condition-based diagnostics for electrical systems.

Dr. Habetler has served on the IEEE Board of Directors as Division IIDirector, is the Past-President of the IEEE Power Electronics Society, and the

Past-Chair of the Industrial Power Converter Committee of the IEEE IndustryApplications Society. He has received four Conference Prize Paper Awardsfrom the IEEE Industry Applications Society.

05508416

Documents

Transcript of 05508416