Rotor Flux Estimation Using Voltage Model of

Induction Motor

Pavel Brandstetter, Martin Kuchar

Department of Electronics
VSB – Technical University of Ostrava
Ostrava, Czech Republic
[email protected]

Abstract—The paper deals with mathematical models of an resistance. Modern estimation methods include soft computing
induction motor which are used for control algorithms of AC techniques, for example genetic algorithm [6].
drives with induction motors. The first part of the paper
describes a current model of the induction motor which is used in In the paper, there are described a rotor flux estimator
the model reference adaptive system as a reference model for a which uses a frequency adaptive integrator and a stator
stator resistance estimator. The next part deals with a voltage resistance estimator using the model reference adaptive
model, which uses for the determination of the individual rotor system.
flux vector components integration without feedback, whose
basic disadvantage is time dependent offset of output quantity. In II. MATHEMATICAL MODELS OF THE INDUCTION MOTOR
the paper, there is described a solution of the integrator with the
offset elimination. The voltage model of the induction motor The mathematical model of the induction motor is used
works with the voltage drop on the stator resistance which is with the consideration of an electromagnetic process in the
depending on temperature changes. Therefore there is described simulations. This mathematic model represents a multi-
the stator resistance estimator which uses the model reference parameter, non-linear system with the following conclusions,
adaptive system. The correct activity of the described rotor flux for simplification of the mathematical description:
estimator and stator resistance estimator is confirmed by
simulations in Matlab-Simulink. • The machine is symmetrical.
• Air gap is constant.
Keywords— AC drive; induction motor, current model, voltage • Iron losses are neglected,
model; model reference adaptive system; vector control • Resistances and inductances are the same and constant
at all phases (e.g. changes resistances depending on
I. INTRODUCTION temperature, magnetic circuit saturation or hysteresis,
etc. are neglected).
At present, there are two control methods of AC drives
with induction motors: a vector control and a direct torque The mathematical model of the induction motor is
control method. The vector control method is based on the designed like a flux model in the fixed machine coordinates
transformation of a three phase speed and time dependent (α, β). Therefore, we use an equivalent two phase system
system into a two coordinate time invariant system using so- instead of a three phase coordinate system [7].
called orienting quantities. These quantities may be obtained
by using different flux estimators. In the technical literature, A. Current Model of the Induction Motor
there are many methods for evaluating the orienting angle The exact value and position of the magnetizing current
which is necessary for vector rotation from the stator vector or rotor flux vector allows the creation of
coordinate system to the rotating coordinate system. One of determinations by the use of so-called current model:
the methods is the use of mathematical models of the
induction motor [1]-[5].
⎡⎛ 1 ⎞ 1 ⎤
A rotor flux estimator for calculation rotor flux vector Ψ SR (CM ) = ∫ ⎢⎜ jωm − ⎟ Ψ RS (CM ) + Lh i SS ⎥ dt (1)
components and orienting angle which uses a voltage model of ⎢⎣⎝ TR ⎠ TR ⎥⎦
the induction motor is very simple. The greatest disadvantage
is the necessity to use an integrator and a temperature change ⎡ 1 1 ⎤
of the stator resistance. Adaptation of this temperature change Ψ Rα ( CM ) = ∫ ⎢ − Ψ Rα ( CM ) − ωeΨ Rβ (CM ) + Lm iSα ⎥ dt (2)
is thus required, and it is necessary to estimate this parameter ⎣ TR TR ⎦
in order to maintain it equal to its actual value. There are many
⎡ 1 1 ⎤
general algorithms to estimate the stator resistance. The best Ψ Rβ ( CM ) = ∫ ⎢ − Ψ R β (CM ) + ωeΨ Rα (CM ) + Lm iS β ⎥ dt (3)
known methods use different types of estimators which often ⎣ TR TR ⎦
use an adaptive mechanism to update the value of stator
The paper was supported by the projects: IT4Innovations Centre of
Excellence project, reg. no. CZ.1.05/1.1.00/02.0070 funded by the European
Regional Development Fund and the national budget of the Czech Republic,
Center for Intelligent Drives and Advanced Machine Control (CIDAM)
project, reg. no. TE02000103 funded by the Technology Agency of the Czech
Republic, project reg.no. SP2015/81 funded by the Student Grant Competition
of VSB-Technical University of Ostrava.

978-1-4673-6788-2/15/$31.00 ©2015 IEEE

 where ΨSR(CM) – rotor flux vector in [α, β] coordinate A. Frequency Adaptive Integrator
system; iSS – stator current vector in [α, β] coordinate system; One of the simplest modifications is the usage of a
ΨRα(CM), ΨRβ(CM) – rotor flux vector components in [α, β] frequency adaptive (FA) integrator which consists of the
coordinate system; iSα, iSβ – stator current vector components integrator and PI controller for DC component elimination of
in [α, β] coordinate system; Lm – magnetizing inductance; ωe – the output signal. In Fig. 2, the block scheme of the frequency
electrical rotor angular speed; TR – rotor time constant. adaptive integrator is shown [8].
B. Voltage Model of the Induction Motor For the depiction of the amplitude and phase response of
adaptive integrator (see Fig.3, Fig.4), we can derive according
From the system of basic equations, a voltage flux model to Fig.2 the transfer function of the integrator:
can also be derived:
1
L
Ψ SR (VM ) = R ⎡⎣ ∫ ( u SS − RS i SS ) dt − σ LS i SS ⎤⎦ (4) FS ( s ) =
T1 s
(7)
Lm 1 (1 + K RT2 s )
1+ ⋅
T1 s T2 s
LR
⎡ ( uSα − RS iSα ) dt − σ LS iSα ⎤
Lm ⎣ ∫
Ψ Rα (VM ) = ⎦ (5)
We can modify (7) to following form:
LR
⎡ ( uS β − RS iS β ) dt − σ LS iS β ⎤ T2 s
Lm ⎣ ∫
Ψ R β (VM ) = ⎦ (6) FS ( s ) = (8)
2
T1T2 s + K RT2 s + 1

where ΨSR(VM) – rotor flux vector in [α, β] coordinate where T1 – time constant of the integrator; KR –
system; uSS – stator voltage vector in [α, β] coordinate system; proportional gain of the PI controller; T2 – time constant of the
iSS – stator current vector in [α, β] coordinate system; ΨRα(VM), PI controller.
ΨRβ(VM) – rotor flux vector components in [α, β] coordinate
system; uSα, uSβ – stator voltage vector components in [α, β] 1
coordinate system; iSα, iSβ – stator current vector components sT1
a
in [α, β] coordinate system; Lm – magnetizing inductance; RS –
stator resistance; LS – stator inductance; LR – rotor inductance; KR
σ = (1-L2m /LSLR) – leakage coefficient. a
1
III. ROTOR FLUX ESTIMATION sT2

For rotor flux vector components estimation, we can use Fig. 2. Block scheme of the frequency adaptive integrator.
voltage model of the induction whose block scheme is shown
in Fig. 1. We can derive for calculation of frequency adaptive
integrator parameters following equations:
LR
Lm
1 1
T1 = , T2 = , K R = 2cT1 (9)
L
RS ⋅ R ωcr 2c 2 T1
Lm ∫ dt
LR LS The crossover frequency ωcr = 2πfS can be chosen in
σ⋅
Lm consideration of stator voltage frequency fS = 50Hz that the
LR
gain of integrator would equal 1.
Lm

LR
Integrator phase deviation is:
RS ⋅
Lm ∫ dt
ωT2 K R
L L Δϕ = ϕ − ϕ0 = π − arctg (10)
σ⋅ R S 1 − ω 2T1T2
Lm

Fig. 1. Block scheme of the rotor flux estimator using IM voltage model. where φ = – π/2 is a phase shift of the ideal integrator.
Main problem of the estimator using IM voltage model is We can modify (10) to following form:
the computation of the rotor flux components ΨRα, ΨRβ. The
problem arises during integration process, because the voltage ω T2 K R
and current components include a small DC component part. tg (Δϕ ) = − (11)
This DC component causes the integrator drift, which is 1 − ω 2T1T2
unacceptable. The drift problems of the pure integrator can be
solved by modifying the integration procedure. We will obtain using (9) following equation:
 2 (ω c ) The DC component PI controller effect is continuously set-
tg (Δϕ ) = − (12) up so that phase and amplitude deviation was the same in the
2 − (ω c )
2
wide range of the working frequencies. The following
equation illustrates the weighted factor a, by which the
For the small phase deviation Δφ < 5° there can be written: integrator adaptation is gained:

2 (ω c ) ω T
Δϕ = tg (Δϕ ) = − (13) a= , K R′ = K R ⋅ a, T2′ = 2 (16)
2 − (ω c )
2 ωcr a

Modification (13) there is achieved following equation: where parameter a takes the interval value <0, 1>.
The described frequency adaptive integrator can be used
2Δϕ c 2 + 2ω c − ω 2 Δϕ = 0 (14) for signal processing in the frequency band from 1 Hz where
integrator phase deviation Δφ is less than 5°.
There is only acceptable the positive equation root,
because with the negative equation root the whole system B. Simulation Results
would be unstable: The described rotor flux estimation method was simulated
by Matlab-Simulink on the vector controlled AC drive with
2 the induction motor. The simulated induction motor has the
ω ⎛ ω ⎞ 1 2 following parameters:
c=− + ⎜ ⎟ + ω (15)
2 Δϕ ⎝ 2 Δϕ ⎠ 2 • Pn = 2.7 kW, Tn = 19 Nm
• ωn = 1360 rpm, pp = 2
Maximal phase deviation should not be higher than 3° • ISn = 7.51 A, IRn = 17 A
(0.05 rad), then c = 7.984 s-1. We can calculate for the chosen • ISxn = 6.8 A, ISyn = 8.16 A
crossover frequency ωcr = 2πfS parameters of the PI controller
• USn = 400/230 V, ΨSn = 0.877 Wb
T2, KR using (7).
• RS = 1.83 Ω, RR = 2.19 Ω
• LS = 0.137 H, LR = 0.137 H
• Lm = 0.129H, TR = 0.0546 s
Following figures show simulation results that confirm
right activity of the described rotor flux estimator using
integrator with the offset elimination.

Fig. 3. Amplitude frequency characteristic of the FA integrator.

Fig. 5. Reference speed (blue) and real speed (green) of the IM drive.

Fig. 4. Phase frequency characteristic of the FA integrator. Fig. 6. Stator current vector components iSα (blue), iSβ (green) of the IM.
 B. Simulation Results
Simulations have been realized using software product
Matlab-Simulink. Following figures show simulation results
that confirm right activity of the described stator resistance
estimator using MRAS. There was simulated AC drive start at
constant speed 100 rpm and following load jump 10 Nm (see
Fig.9 and Fig.10).

Fig. 7. Rotor flux vector components ΨRα(VM) (blue), ΨRβ(VM) (green) from the
voltage model with additional time dependent offset.

Fig. 9. Reference speed (blue) and real speed (green) of the IM drive.

Fig. 8. Rotor flux vector components ΨRα(VM) (blue), ΨRβ(VM) (green) from the
voltage model with offset elimination.

Fig. 7 shows rotor flux vector components ΨRα(VM), ΨRβ(VM)

from the voltage model of the induction motor with additional
time dependent offset. Fig.8 shows these quantities after
elimination by using the frequency adaptive integrator.
Fig. 10. Motor torque (green) and load torque (blue).
IV. STATOR RESISTANCE ESTIMATOR
One class of algorithms for estimation of the stator
resistance uses the stator voltages, stator currents and the
motor speed which are measured or estimated and the stator
resistance is calculated using the measurement results.

A. Model Reference Adaptive System

The Model Reference Adaptive System (MRAS) for the
stator resistance estimation employs two models – reference
current model described by (1) and adaptive voltage model
described by (4). The difference between output quantities of
the models is evaluated by adaptation mechanism. The Fig. 11. Rotor flux vector components ΨRα(CM) (blue), ΨRβ(CM) (green) from the
adaptation mechanism has two parts: evaluation of the refrence current model.
adaptation signal Φ(e) and its following minimization using
classical PI controller which produces the estimated value of
the stator resistance for the adaptive voltage model.
The adaptation algorithm is described by the following
equations [9]:

Φ ( e ) = iSα ⎡⎣Ψ Rα (VM ) − Ψ Rα (C M ) ⎤⎦ + iS β ⎡⎣Ψ R β (VM ) − Ψ R β (C M ) ⎤⎦ (17)

t
RSest = K1Φ ( e ) + K 2 ∫ Φ ( e ) dt (18)
0
Fig. 12. Rotor flux vector components ΨRα(VM) (blue), ΨRβ(VM) (green) from the
where K1 > 0, K2 > 0. adaptive voltage model.
Fig. 13. Stator current vector components iSα (blue), iSβ (green) of the IM. Fig. 17. Adaptation signal Φ(e).

The difference between the real and the estimated stator

resistance is negligible (see Fig.16). The adaptation signal
based on the adaptation mechanism is shown in Fig.17.

V. CONCLUSION
The appropriate and functional solution of the rotor flux
estimator is necessary not only for the vector control but also
for modern control methods including so-called sensorless
control. The presented simulation results show that the
designed rotor flux estimator with elimination temperature
Fig. 14. Time dependent change of the stator resistance. changes of the stator resistance is suitable for control methods
of AC drives with induction motors.

REFERENCES
[1] P. Vas, Sensorless Vector and Direct Torque Control. Oxford University
Press, 1998.
[2] I. Neborak, “Analysis of Setting Methods of Magnetic Field in Induction
Motor Vector Control,” Proceedings of the 13th International Scientific
Conference on Electric Power Engineering 2012, pp. 981-984, Brno,
Czech Republic, 2012.
[3] P. Chlebis, P. Vaculik, P. Moravcik, and Z. Pfof, “Direct Torque Control
Methods for Three-level Voltage Inverter,” Proceedings of the 10th
International Scientific Conference on Electric Power Engineering 2009,
pp. 352-356, Kouty nad Desnou, Czech Republic, 2009.
Fig. 15. Estimated stator resistance.
[4] V. Sladecek, P. Palacky, D. Slivka, and M. Sobek, “Influence of Power
Semiconductor Converters Setup on the Quality of Electrical Energy
from Renewable Sources,” Proceedings of the 11th International
Scientific Conference on Electric Power Engineering 2010, pp. 527-531,
Brno, Czech Republic, 2010.
[5] I. Neborak, P. Simonik, and L. Odlevak, “Electric Vehicle Modelling
and Simulation,” Proceedings of the 14th International Scientific
Conference on Electric Power Engineering 2013, pp. 693-696, Kouty
nad Desnou, Czech Republic, 2013.
[6] P. Palacky, P. Hudecek, A. Havel, “Real-Time Estimation of Induction
Motor Parameters Based on the Genetic Algorithm,” International Joint
Conference CISIS'12-ICEUTE'12-SOCO'12 Special Sessions, Advances
in Intelligent Systems and Computing, 2013, vol. 189, pp. 401-409.
[7] B. Dobrucky, M. Benova, and P. Spanik, “Using Complex Conjugated
Fig. 16. Difference between real and estimated stator resistance. Magnitudes- and Orthogonal Park/Clarke Transformation Methods of
DC/AC/AC Frequency Converter,” Elektronika Ir Elektrotechnika, Issue
5, pp. 29-34, 2009.
Figures 11 and 12 show time courses of the rotor flux
[8] O. Skuta, Modified concepts of the artificial neural network architecture
vector components from the reference current model and in the modern control of electrical drives, PhD. Thesis, VSB-Technical
adaptive voltage model of the induction motor. The stator University of Ostrava, 2008.
current vector components are other variables for the [9] V. Vasic, S. N. Vukosavic, and E. Levi, "A stator resistance estimation
adaptation process of the model reference adaptive system scheme for speed sensorless rotor flux oriented induction motor drives,"
(see Fig.13). Figures 14 and 15 confirm the correct function of IEEE Trans. on Energy Conversion, Vol. 18, No. 4, pp. 476-483, 2003.
the stator resistance estimator.