International Journal of Power Electronics and Drive System (IJPEDS)
Vol. 9, No. 1, March 2018, pp. 433~442
ISSN: 2088-8694, DOI: 10.11591/ijpeds.v9n1.pp433-442
433
Real Time Validation of EKF Estimator for Low Speed
Estimation in DTC IMD
Mini R1, Binoj P.V.2, B. Hariram Satheesh 3, Dinesh M.N.4
1,2
Electrical and Electronics Engineering Department, Amrita University, India
3
ABB GISL, Bengaluru, India
4
R V College of Engineering, Bengaluru, India
Article Info
ABSTRACT
Article history:
Low speed estimation in DTC IMD is not accurate due to the presence of
transient offset, drift and domination of ohmic voltage drop in the measured
stator voltages and currents used for estimating the stator flux required for
accurate estimation of speed. EKF is a nonlinear, recursive adaptive
algorithm capable of estimating speed ranging from very low speed to rated
speed using equation of motion from noisy measured currents and voltages
based on state space technique. In the previous work a new state space model
of IM was developed for estimation in EKF by feeding load torque profile as
an input variable instead of estimating it by considering load torque as
constant, validated using MATLAB-Simulink software. In this paper real
time validation of the EKF controller with load profile fed as input for speed
estimation in DTC IMD is carried out using OPAL-RT simulator and real
time results validates the simulation results and proves the effectives of the
new EKF for low speed estimation in DTC IMD.
Received Jul 28, 2017
Revised Jan 9, 2018
Accepted Jan 21, 2018
Keyword:
DTC
EKF
IMD
PIL
Real time testing
Sensorless
Copyright © 2018 Institute of Advanced Engineering and Science.
All rights reserved.
Corresponding Author:
Faculty Electrical and Electronics Department,
Amrita School of Engineering, Bengaluru,
Amrita VishwaVidyapeetham, Amrita University, India
Email: minir_sujithr@yahoo.com, mini_sujith@blr.amrita.edu.
1.
INTRODUCTION
The advancement in power electronics, vector control and direct torque control resulted in
unprecedented growth in induction motor (IM) variable speed drives. The control of IM is more complex due
to rotating stator magnetic field and inaccessibility of rotor current for measurement. [1,2] Takahashi and
Noguchi [3] introduced DTC for low and medium voltage induction motor drives in 1985 and DTC
developed by M. Depenbrock [4] for high voltage induction motor drives became very popular in industry.
DTC offers excellent independent control of electromagnetic torque and stator flux linkage by the optimum
selection of voltage space vector to obtain excellent dynamic torque response and greater efficiency.
Conventional DTC has variable switching frequency and posses torque ripples due to the hysteresis band
controllers and the voltage space vector selection was based on a switching table with limited optimum
voltage switching states. [5,7]. Space vector modulation (SVM) technique used for optimum voltage space
vector selection offers reduction of ripples in torque, flux and speed and gives constant switching frequency.
[6,8,9].
Sensorless control is the norm of industry due to reduction in complexity and cost, ruggedness,
reliability, noise immunity and operation in a hostile environment. Sensorless control needs different types of
speed estimation techniques to estimate the rotor speed from measured stator voltages and currents. Open
loop estimators and closed loop observers are used for speed estimation. Open loop estimators are used
where no closed loop speed control is required and it lacks the speed estimation correction term which most
of the closed loop observers are having. [10] Open loop estimator performance degrades during low speed
Journal homepage: http://iaescore.com/journals/index.php/IJPEDS
434
ISSN: 2088-8694
operation due to parameter variation, pure integrators used, drift and dc offset in measurements, noise in the
system etc. Hence low speed estimation of DTC IMD is a research challenge.
Closed loop observers like MRAS, EKF, Luenberger, Sliding mode control were investigated by
pervious researchers [11-13]. Among that Extended Kalman filter(EKF) is a is a promising solution for low
speed estimation because it is a recursive, stochastic algorithm suitable for nonlinear systems. EKF is suitable
for joint parameter and state estimation and hence EKF is investigated as the most suitable observer for low
speed estimation in DTC IMD. In sensorless DTC IMD, EKF estimates the rotor flux linkages, stator currents
and rotor speed. [14-18]. Previous researches shows that rotor speed is estimated in EKF considering it as a
constant [1]. From literature survey it is observed that speed information can be extracted from the equation
of motion, relating load torque and electromagnetic torque. Electromagnetic torque can be obtained using the
rotor flux linkages and stator currents obtained from the EKF state variable estimate.
The load torque required for the speed estimation is taken as an additional state variable in EKF and
considered it as a constant [19]. But this method offers sluggish performance during low speed operation. A
new method was proposed by the authors [ 20] which also uses the equation of motion for rotor speed
estimation by EKF but the load torque required for speed estimation is fed as an input to the EKF for
estimating the speed. This method offers fast convergence and reduced estimation even at very low speed
including zero speed. The software validation of new approach of estimating speed using EKF observer is
carried out using MATLAB-Simulink software and observed the effectiveness of the technique for low speed
estimation in sensorless DTC IMD. In this paper real time validation of the new EKF approach using the load
profile input to EKF estimator for rotor speed estimation is carried out using Processor- In-The -Loop (PIL)
real time validation technique using OPAL-RT real time simulator OP4500. The results and analysis are
presented.
2.
SENSORLESS SVM DTC INDUCTION MOTOR DRIVE
DTC is a powerful control technique which can directly and independently control the
electromagnetic torque and stator flux linkages of an induction motor. Space vector modulation(SVM) is
used for the optimal selection of inverter voltage switching space vector. In conventional DTC optimal
switching-voltage vectors are to be selected from the optimum switching-voltage vector look-up table. But in
conventional DTC hysteresis band controllers the optimum selection of voltage space vector from the look up
table and cause torque ripples and gives variable switching frequency. In SVM PI controllers replace the
hysteresis band controllers and offers constant switching frequency and reduction in torque ripple.
Sensorless drives reduce the complexity and cost of system but sensorless drive performance greatly
depends on the speed estimator. Basic open loop estimators have low speed estimation issues due to stator
resistance parameter variation during low speed operation of the drive. Along with that system and
measurement noise, drift, dc offset etc present in the system will affect the low speed estimation. EKF is a
closed loop observer which can be used for any nonlinear adaptive system to estimate the parameter from any
noisy environment using state space technique and recursive algorithm and hence EKF estimator is an ideal
choice for mitigating low speed estimation issues in sensorless DTC IMD.
3.
INDUCTION MOTOR MODEL DESCRIPTION
Speed estimation using EKF needs state space model of induction motor for the prediction of states
at any instant from previously estimated values. The two axis state space model of induction motor in
stationary reference frame can be used where the stator current and rotor flux linkages are the state variables
and rotor speed is augmented as the fifth state variable and is estimated using equation of motion. The model
developed in this paper is feeding the load torque profile as an input to EKF estimator instead of considering
load torque as constant as given by previous researchers. The mathematical model of the induction motor
with five state variable.
Int J Power Electron & Dri Syst, Vol. 9, No. 1, March 2018 : 433 – 442
Int J Power Electron & Dri Syst
-(
𝑖𝑑𝑠
𝑑 𝑖𝑞𝑠
ψdr =
𝑑𝑡
ψqr
[ ωr ]
𝑅𝑠 𝑅𝑟′ Lm2
)
+
𝐿′𝑠 𝐿′𝑠 (𝐿′𝑟 )2
0
Lm
Tr
0
𝑃 3 𝑃
∗ ∗ ∗ ψqr
[ 2𝐽 2 2
−
ISSN: 2088-8694
ωrLm
Lm
𝐿′𝑠 𝐿′𝑟 Tr 𝐿′𝑠 𝐿′𝑟
𝑅𝑠 𝑅𝑟′ Lm2
Lm
ωrLm
-( ′ + ′ ′ 2) − ′ ′
𝐿𝑠 𝐿𝑠 (𝐿𝑟 )
𝐿𝑠 𝐿𝑟 𝐿′𝑠 𝐿′𝑟 Tr
1
0
−
−ωr
𝑇𝑟
Lm
1
ωr
−
𝑇𝑟
Tr
𝑃 3 𝑃
∗ ∗ ∗ ψdr
0
0
2𝐽 2 2
0
𝑖𝑑𝑠
𝑖
𝑖𝑑𝑠
1 0 0 0 0 𝑞𝑠
[𝑖 ] = [
] ψdr
0 1 0 0 0
𝑞𝑠
ψqr
[ ωr ]
435
(1)
0
1
Ls'
0 𝑖𝑑𝑠
𝑖𝑞𝑠
0
0 ψdr +
0
ψqr
0
[
]
0 ωr
0
[
0
]
0
1
Ls'
0
0
0
0
0
0
0
𝑃
−
2𝐽]
𝑉𝑑𝑠
[ 𝑉𝑞𝑠 ]
𝑇𝑙
(2)
The electromagnetic torque required for estimating the rotor speed is derived from stator currents and rotor flux
linkages in Equation (3).
𝑇𝑒 =
3 𝑝 𝐿𝑚
(𝜓𝑑𝑟𝑖𝑞𝑠 − 𝜓𝑞𝑟 𝑖𝑑𝑠 )
2 2 𝐿′𝑟
(3)
The load torque required to be fed to EKF estimator for rotor speed estimation is incorporated in the
mathematical model of machine by feeding the load profile as the third element to the input matrix along
with the stator voltages.
4.
EXTENDED KALMAN FILTER ALGORITHM
Extended kalman filter algorithm (EKF) is a stochastic and recursive adaptive observer which can be
used for joint state and parameter estimation of a nonlinear dynamic system. EKF will take care of the noise
in the system during estimation by using covariance matrices of the state variable (P), system noise (Q) and
voltage measurement noise (Ru) and the current measurement noise (Re). These noise sources take care of
measurement and modeling inaccuracies. EKF algorithm consists of two stages of calculation, first stage is
prediction of states using mathematical model which contains previous estimates and in second stage the
predicted states are continuously corrected by a feedback correction scheme. The prediction stage needs the
discretized model of the IM. The weighted difference of the measured and estimated output is added to the
predicted values.[20].’
𝑖𝑑𝑠 (𝑘 + 1)
𝑖𝑞𝑠 (𝑘 + 1)
ψdr(𝑘 + 1) =
ψqr(𝑘 + 1)
[ ωr(𝑘 + 1) ]
1-(
𝑅𝑠 𝑅𝑟′ Lm2
)𝑇
+
𝐿′𝑠 𝐿′𝑠 (𝐿′𝑟 )2
0
Lm𝑇
Tr
0
𝑃 3 𝑃
− ∗ ∗ ∗ ψqr ∗ 𝑇
[ 2𝐽 2 2
ids(𝑘)
iqs(𝑘)
ids(𝑘)
1 0 0 0 0 ψ (𝑘)
[
]= [
] dr
iqs(𝑘
0 1 0 0 0
ψqr(𝑘)
[ ωr(𝑘) ]
0
𝑅𝑠 𝑅𝑟′ Lm2
1-( ′ + ′ ′ 2)𝑇
𝐿 𝐿 (𝐿 )
𝑠
𝑠
0
𝑟
Lm𝑇
Tr
𝑃 3 𝑃
∗ ∗ ∗ ψdr ∗ 𝑇
2𝐽 2 2
Lm𝑇
𝐿′𝑠 𝐿′𝑟 Tr
ωrLm𝑇
− ′ ′
𝐿𝑠 𝐿𝑟
𝑇
−
𝑇𝑟
𝑇ωr
0
ωrLm𝑇
0
𝐿′𝑠 𝐿′𝑟
Lm𝑇
0
𝐿′𝑠 𝐿′𝑟 Tr
−𝑇ωr
𝑇
−
𝑇𝑟
0
ids(𝑘)
iqs(𝑘)
0 ψdr(𝑘) +
ψqr(𝑘)
0 [ ωr(𝑘) ]
0
]
(4)
T
Ls'
0
[
0
0
0
0
T
Ls'
0
0
0
0
0
0
0
𝑃
− 𝑇
2𝐽 ]
𝑉𝑑𝑠 (𝑘)
[ 𝑉𝑞𝑠 (𝑘) ]
𝑇𝑙 (𝑘)
(5)
The EKF algorithm was developed to estimate the states and the estimated states of previous instant
and the stator voltages are used for predicting the values of the state variables at the present instant. The
deviation of predicted values from actual values are obtained by comparing the predicted stator currents and
the measured stator currents. This difference in value is the error which is tuned using a correction factor to
get an accurate estimate of the states. The corrected error is then summed up with the predicted values for
estimating the values of the state variables. EKF is used to estimate the rotor speed of the sensorless DTC
IMD.
Real Time Validation of EKF Estimator for Low Speed Estimation… (Mini R)
436
ISSN: 2088-8694
A 20hp DTC IMD using EKF estimator is developed in MATLAB-Simulink software and validated
the performance of the drive for speed estimation from rated speed to very low speed including zero speed.
Simulation results proves the effectiveness of this new load profile input feed EKF estimator for low speed
estimation in DTC IMD and presented the results in [ 20]. The motor parameters are given in Table 1. In this
work a real time validation of the controller using Processor- In-The -Loop (PIL) real time validation
technique is carried out and real time performance of the controller is validated and compared with
simulation results.
Table 1. Motor Parameters
Power(kw)
Frequency(Hz)
J(kg/m2)
B(Nm/rad/s)
No of poles P
Voltage(V)
Current(A)
15
50
0.102
.009541
2
400
36
Rs (ohm)
Rr'(ohm)
Lls(H)
𝐿′𝑙𝑟 (H)
Lm(H)
Nm(rpm)
Tl(Nm)
0.2147
0.2205
0.000991
0.000991
0.06419
1460
98
5.
REAL TIME SIMULATION
Model driven development approach has gained tremendous demand in testing and validating the
controller in real time simulator environment. This helps to reduce the time taken for development of
embedded system and to produce rapid and reliable product in a short interval of time. RT-PIL is a
verification and validation technique used towards the development of a hardware prototype. The
performance analysis obtained from this real time PIL will provide the details required for hardware and
computational resources for the implementation of future prototype. The advantages of this approach are the
simulation time scale is the same to clock time scale, the real time controller code generation from the
simulated algorithm can be done automatically, the time taken for developing, cost to design and prototype
the controller and testing of algorithm at extreme conditions with high accuracy can be achieved. In PIL the
specific driver functions installed in a simulation integrated environment of the host PC will communicate
with the non real time environment target processor.
5.1. Structure of Opal-RT real time simulation system
Opal-RT's real time simulator enables to link software simulations made in MATLAB/Simulink
SimPowerSystems models on a dedicated real-time computing platform. Real time simulation helps to link
the controller board to the simulation model. The structure of RT lab real time simulator is shown in the
block diagram Figure 1.
Figure 1. Processor in the loop block diagram
Target machine is equipped with programmable DSP/FPGA and I/O channels to communicate with
the simulation host computer and the real controller. The pulse resolution of RT-LAB simulation platform is
10ns, hence many events can be generated in one simulation step and offer high precision timing to the IGBT
Int J Power Electron & Dri Syst, Vol. 9, No. 1, March 2018 : 433 – 442
Int J Power Electron & Dri Syst
ISSN: 2088-8694
437
switches in the inverter. The minimum simulation time step in RT Lab simulator is 20μs. For real time
simulation we need to convert the simulation model made in MATLAB-Simulink to be transformed to real
time model with a time step of 20μs, then segmentation, editing, compilation and finally code will be
generated from the mathematical model, then the model will be loaded to operate in real time simulation
platform [ 23-24].
5.2. Modification required for the existing model
The existing model was running with a fixed step size of 2 μs and to work the model in real time
simulator the model step size needs to be increased to 20μs. But when we increase the step size,
reconstruction of output signal is not correct due to reduced number of samples and error in the pulse width
of PWM signal due to increase in time step. RT EVENTS module tool equipped with timestamp offered by
RT LAB give solutions for this issue. The MATLAB simulation modules affected by events can be replaced
by RTEVENTS equipped with timestamp to reduce the impact on simulation output due to increase in time
step. The PWM pulse generated by RTEVENTS with timestamp, control the pulse changing accurately
between two simulation steps.
5.3. Procedure to convert MATLAB model to RT LAB real time model
MATLAB-Simulink model needs to be grouped to three subsystems to use in RT-LAB and the
naming of top level subsystem should use SM, SC and SS to ensure functions of different parts by RT LAB.
SM (master subsystem) will take care of all real time calculations, SS( slave subsystem) is required only if
computational elements are distributed across multiple nodes if complexity of the SM subsystem is too large
to handle and it will take care of real time calculations and SC (console subsystem) consists of real time
monitoring and data communication of key parameters, scopes, displays etc and it is the only subsystem
available to us while simulation is running [21-22] Next step is to add OpComm blocks to enable and save
communication setup details between console and computation nodes. All signals must go through the
Opcomm module before going to subsystem on the top model. Figure 2 shows the RT lab compatible model
of the DTC SVM IMD with Load profile input feed EKF estimator.
Figure 2. RT lab compatible model of the overall
system
Figure 3. RT Lab real time simulation diagram of
SM (master) subsystem
The RT Lab compatible model of the overall system consists of three subsystems namely the SM_
Master subsystem, SS_ Slave subsystem and SC_Console subsystem. The real time simulation models of
each subsystem is shown in Figure 3,4 and 5 respectively.
The Master subsystem consists of dc input source, inverter, 20hp induction motor. Here the
universal inverter bridge used in the MATLAB Simulink model is replaced by 2 level times tamped bridge.
From this subsystem 3 stator voltages and two stator currents are taken out through analog output block
provided by RT Lab I/O. The actual values of voltages and currents in from the model has to be scaled down
in between ±16V, which is the maximum allowable limit of analog out card of the simulator. The 2 level
TSB will receive gating pulses from Time Stamped Digital in (TSDI) port through event detector block. An
Real Time Validation of EKF Estimator for Low Speed Estimation… (Mini R)
438
ISSN: 2088-8694
Opctrl block needs to be added in this subsystem which provides an interface to the Opal-RT evaluation
board to control all the I/O lines.
Figure 4. RT lab real time simulation diagram of SS(slave) subsystem
The SS (Slave) subsystem consists of the main controller of sensorless DTC SVM IMD with Load
profile input feed EKF as the estimator. The controller will receive three stator voltages and two stator
currents required for the estimation through the analog in block. The signals received from analog in block
has to be rescaled before giving to the controller. The blocks used to generate PWM signals needs to be
replaced by RTEVENT - blocks as shown in Figure 5 The PWM pulses generated by the controller taken out
via Time Stamped Digital Out(TSDO) through Event generator block.
Figure 5. RT lab real time simulation diagram of SC (console)subsystem
The SC subsystem consists of all the user interface blocks like scope, slider gain etc. Using slider
gain we can control the system even under running condition.
Int J Power Electron & Dri Syst, Vol. 9, No. 1, March 2018 : 433 – 442
Int J Power Electron & Dri Syst
ISSN: 2088-8694
439
Figure 6. Pulse generation using RT-events
6.
REAL TIME PROCESSOR-IN-THE LOOP SIMULATION RESULTS & ANALYSIS OF EKF
CONTROLLER
The real time performance validation of the new load profile input feed EKF estimator for sensorless
DTC IMD for low speed estimation was performed in Real-Time Processor- In- The-Loop (RT-PIL) using
Opal-RT digital simulator. In this approach both plant and controller are running in the same simulator and
they exchange real time signals via loop back cables (hardwired) through the I/Os of the simulator. Real
Time - PIL results are presented here to prove the effectiveness of new EKF estimator for low speed
estimation in DTC IMD. To validate the effectiveness of the new EKF algorithm during steady state and
transient state, the speed and torque reversal are tested for all ranges of speeds from rated speed to very low
speed including zero speed.
Table 2. Profile of Reference Speed and Applied Load
Time(s)
Speed(rpm)
(a)
0
5
1
-5
Time(s)
Torque(Nm)
0
0
0.5
98
1.5
-98
(b)
Figure 7. Plots taken through opwrite file of real time simulator at 5 rpm subjected to full load torque of 98
Nm with speed and torque reversal a) Speed plots b) Torque plots
Real Time Validation of EKF Estimator for Low Speed Estimation… (Mini R)
440
ISSN: 2088-8694
(a)
(b)
Figure 8. MATLAB-simulink simulation plots of speeds at 5 rpm subjected to full load torque of 98 Nm with
speed and torque reversal a) Speed plots b) Torque plots
From Figure 7 and Figure 8 clearly proves that the load profile input feed EKF estimator provides
quick response, accurate speed and torque estimation at very low speeds at rated torque with speed and
torque reversal. The speed and torque estimated by EKF is closely following the reference speed and torque.
This proves the effectiveness of EKF estimator for speed estimation from rated speed to very low speed. To
validate the performance of the drive under varying load conditions, the drive is subjected to run at full load
torque, 3/4th load, 1/2th load and 1/4th load and speed and torque estimation is observed to be satisfactory for
the entire ranges of speeds. Figure 8 and Figure 9 shows the operation of the drive at 1 rpm and subjected to
variable torques. The test set speed and torque values are shown in Table. 3.
(a)
(b)
Figure 9. Plots taken through opwrite file of real time simulator (a) Speeds and (b) Torques at 1 rpm
subjected to varying load conditions
(a)
(b)
Figure 10. MATLAB-simulink simulation plots (a) Speeds and (b) Torques at 1 rpm subjected to varying
load conditions
Int J Power Electron & Dri Syst, Vol. 9, No. 1, March 2018 : 433 – 442
Int J Power Electron & Dri Syst
ISSN: 2088-8694
441
Figure 11. Real time Simulator output displayed in DSO with an applied constant torque of 98
Nm(scale:1division=100Nm) and speed varies from 2 rpm to 5 rpm (scale: 1 division= 10 rpm)
(a)
(b)
Figure 12. Real time Simulator output displayed in DSO with a 2 rpm speed subjected to a torque change a)
from 49Nm to 98Nm (half load to full load torque) b) from 98 Nm to 49Nm (speed scale:1 div=10 rpm
and torque scale: 1div =100Nm)
Figure 13. Real time Simulator output displayed in DSO at 0rpm subjected to full load torque of 98Nm
(speed scale: 1 div = 10 rpm, Torque scale 1 div=98Nm)
The real time output seen in the DSO validates the excellent performance of the drive with EKF
estimator with Load profile fed to EKF as an input for estimation.
7.
CONCLUSION
To improvise the performance of the sensorless DTC IM drive during low speed operation an EKF
estimator with load torque required for speed estimation is fed to the EKF model is simulated in OPAL-RT
real time simulator using the Processor-In-The Loop technique to evaluate the real time performance of the
EKF estimator for low speed estimation. The output results obtained from real time simulator shows the
effectives of the EKF for low speed estimation at various torque and speed conditions. The steady state and
transient state performance at rated full load torque and varying load conditions are verified and observed the
EKF estimator gives excellent performance from rated speed to very low speed including zero speed. The
Processor-In-The -Loop (PIL) validation scheme helps to analyze the performance of EKF estimator to
realistic conditions using a real time simulator. The real time simulator relates the analysis to realistic
problems associated communication and exchange of data between embedded processor and sensorless DTC
Real Time Validation of EKF Estimator for Low Speed Estimation… (Mini R)
442
ISSN: 2088-8694
controller. The future work is to validate the controller in Hardware-In-The -Loop real time validation
technique using OPAL-RT real time simulator.
ACKNOWLEDGEMENTS
We express our gratitude to Dr. Ilam Parithi and Mr. Amit Kumar K.S of OPAL-RT, Bangalore for
their support, valuable guidance and providing access to real time simulator facility for validating our EKF
controller.
REFERENCES
[1]
[2]
[3]
[4]
[5]
[6]
[7]
[8]
[9]
[10]
[11]
[12]
[13]
[14]
[15]
[16]
[17]
[18]
[19]
[20]
[21]
[22]
[23]
P. Vas, Sensorless Vector and Direct Torque Control. New York: OxfordUniv. Press, 1998.
Bimal.K.Bose,Modern Power Electronics and AC Drives.Upper Saddle River,NJ: Prentice Hall PTR,2002.
Takahashi and T. Noguchi, “A New Quick-Response and High-Efficiency Control Strategy of an Induction Motor”,
IEEE Tran. On Industry Applications, vol. IA-22, No. 5, pp. 820-827. Sept.-Oct.1986.
M. Depenbrock, “Direct self control for high dynamics performance ofinverter feedAC machines,” ETZ Arch.., vol.
7, no. 7, pp. 211–218, 1985.
CristianLascu, Ion Boldea and FredeBlaabjerg, “A Modified Direct Torque Control for Induction Motor Sensorless
Drive”, IEEE Tran. On Industry Applications, vol. 36, No. 1, pp. 122-130.Jan/Feb.2000.
J. Holtz, “Sensorless control of induction motors—Performance and limitations,”in Proc. IEEE-ISIE Annu.
Meeting, Puebla, Mexico, 2000, vol. 1, pp. PL12–PL20.
K. Rajashekara, Sensorless Control of ACMotors. Piscataway, NJ: IEEE Press, 1996.
R. Ortega, N. Barabanov, G. Escobar, and E. Valderrama, “Direct torque control of induction motors: Stability
analysis and performance improvement,” IEEE Trans. Autom. Control, vol. 46, no. 8, pp. 1209–1222,Aug. 2001.
J. Holtz, “Sensorless control of induction machines—With or without signal injection?” IEEE Trans. Ind.Electron.,
vol. 53, no. 1, pp. 7–30, Feb. 2006.
M.Cirrincione and M. Pucci, “An MRAS-based sensorlesshighperformance induction motor drive with a predictive
adaptive model,” IEEE Trans. Ind. Electron., vol. 52, no. 2, pp. 532–551, Apr. 2005.
M. Cirrincione, M. Pucci, G. Cirrincione, and G.-A. Capolino, “An adaptive speed observer based on a new total
least-squares neuron for induction machine drives,” IEEE Trans. Ind. Appl., vol. 42, no. 1, pp. 89–104,
Jan./Feb.2006.
G. Edelbaher, K. Jezernik, and E. Urlep, “Low-speed sensorless control of induction machine,” IEEE Trans. Ind.
Electron., vol. 53, no. 1, pp. 120–129, Feb. 2006.
Wade, M.W. Dunnigan, and B.W. Williams, “Comparison of stochastic and deterministic parameter identification
algorithms for indirect vector control,” in IEE Colloq.—Vector Control and Direct Torque Control Induction
Motors, London, U.K., 1995, vol. 2, pp. 1–5.
F. Chen and M. W. Dunnigan, “Comparative study of a sliding-mode observer and Kalman filters for full state
estimation in an induction machine,” Proc. Inst. Electr. Eng.—Electr. Power Appl., vol. 149, no. 1, pp. 53–64,
Jan.2002.
Alsofyani, I.M., Idris, N.R.N., Alamri, Y.A., Sutikno, T., Wangsupphaphol, A. and Nordin, N.M., 2015.
Comparison of Estimated Torques Using Low Pass Filter and Extended Kalman Filter for IM Drives at Low
Speed. International Journal of Power Electronics and Drive Systems, 6(1), p.92.
F. Cupertino, A. Lattanzi, L. Salvatore, and S. Stasi, “Induction motor control in the low-speed range using EKFand LKF-based algorithms,” in Proc. IEEE-ISIE, Bled, Slovenia, 1999, vol. 3, pp. 1244–1249.
K. L. Shi, T. F. Chan, Y. K. Wong, and S. L. Ho, “Speed estimation of an induction motor drive using an optimized
extended Kalman filter,” IEEE Trans. Ind. Electron., vol. 49, no. 1, pp. 124–133, Feb. 2002.
Godpromesse Kenn´e, Tarek Ahmed-Ali, Franc¸oise Lamnabhi-Lagarrigue, and Amir Arzand´e, “Real-Time Speed
and Flux Adaptive Control of Induction Motors Using Unknown Time-Varying Rotor Resistance and Load
Torque,” IEEE Trans.Energy Conversion, vol. 24, no. 2, pp. 375-389, June.2009.
Murat Barut, Seta Bogosyan and MetinGokasan, “Speed-Sensorless Estimation for Induction Motors Using
Extended Kalman Filters,” IEEE Trans. Ind. Electron., vol. 54, no. 1, pp. 272-280, Feb.2007.
Mini, R., C. Saranya, B. Hariram Satheesh, and M. N. Dinesh. "Low Speed Estimation in Sensorless Direct Torque
Controlled Induction Motor Drive using Extended Kalman Filter." International Journal of Power Electronics and
Drive Systems 6, no. 4, 2015.
Mikkili, Suresh, Anup Kumar Panda, and Jayanthi Prattipati. "Review of real-time simulator and the steps involved
for implementation of a model from MATLAB/SIMULINK to real-time." Journal of The Institution of Engineers
(India): Series B 96, no. 2 (2015): 179-196.
Jonsdottir, G. M., Muhammad Shoaib Almas, Maxime Baudette, M. P. Palsson, and Luigi Vanfretti. "RT-SIL
performance analysis of synchrophasor-and-active load-based power system damping controllers." In2015 IEEE
Power & Energy Society General Meeting, pp. 1-5. IEEE, 2015.
Martins-Filho, Luiz S., Adrielle C. Santana, Ricardo O. Duarte, and Gilberto Arantes Junior. "Processor-in-theLoop Simulations Applied to the Design and Evaluation of a Satellite Attitude Control." (2014).
Int J Power Electron & Dri Syst, Vol. 9, No. 1, March 2018 : 433 – 442