See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/228792581

Indirect vector control of induction motor

Article · September 2006

CITATIONS READS

2 554

3 authors, including:

Vasile Horga Marcel Ratoi

Gheorghe Asachi Technical University of Iasi Gheorghe Asachi Technical University of Iasi
46 PUBLICATIONS   98 CITATIONS    26 PUBLICATIONS   73 CITATIONS   

SEE PROFILE SEE PROFILE

All content following this page was uploaded by Marcel Ratoi on 26 May 2015.

The user has requested enhancement of the downloaded file.

Proceedings of the 6th WSEAS International Conference on Simulation, Modelling and Optimization, Lisbon, Portugal, September 22-24, 2006 98

Indirect Vector Control of Induction Motor

ALEXANDRU ONEA1 VASILE HORGA2 MARCEL RĂŢOI2
1
Department of Automatic Control and Technical Informatics
2
Department of Electric Drives
"Gh. Asachi" Technical University of Iaşi
Bd. Mangeron. 53A, 6600 Iaşi
ROMANIA
e-mail: aonea@ac.tuiasi.ro, horga@tuiasi.ro, ratoi@tuiasi.ro

Abstract: - The paper emphasis a solution for induction motor speed control. The paper presents a vector
control model of induction machine with four parameters only. The main advantage of this minimum number
of parameters model consists in the measurements possibilities of them. This parameters result univocally
from the identification procedures. This minimum number of parameters model preserves the dynamic
behavior of the motor. This model minimizes the uncertainty of the model parameters. The indirect vector
control structure was reformulated according to this model, simulated with MATLAB, and implemented with
a floating-point DSP processor. The experimental results show good transient and steady-state performances.
The accuracy of the results is given by the precision of the motor model used.

Key-Words: - induction motor, vector control, parameter estimation, symmetrical T model, asymmetrical Γ.

1 Introduction sensors and the ability to operate at low speeds has

Squirrel-cage induction motor is a simple low cost
and robust electrical machine. For the sake of its
constructive simplicity this machine is maintenance
free, and it is used in hostile environment without
modifications.
The principle of torque generation in the
induction machine is similar with the one of the d.c.
machine. Unfortunately this similarity was not
highlighted before ’70. This is one of the reasons of
the vector control techniques development after this
period. The theory of the space vector allowed to
simplify and to compress the expression of the
electromagnetic torque. This is delivered as a vector
product between the an active current and a reactive
current in a specific coordinates system.
The absence of the autonomous inductor device
and the lack of rotor currents measurements increase
the complexity of the control structure. This control
structure must allow creating in the machine an
equivalent permanent magnetic through the control
of rotor flux, and the decoupling between this flux
and the electromagnetic torque. This is why the
mathematical models used have to describe with the
highest possible accuracy the operating of the
machine in both steady-state and transient regime.
In the direct vector control schemes the flux position
is measured or estimated on the basis of the flux
components in the stator coordinates. In the indirect
control schemes the flux position is determined
indirectly through the rotor speed and the slip
estimation. The absence of the rotor flux position
increased the popularity of the indirect control
strategy.
Robustness of the solution adopted is very
important, since some parameters can change as a
function of time or may be unknown. Several
modern non-linear control methods used in the
control are reported in the literature as feedback
linearization techniques or robust techniques based
on variable structure systems.
An analysis of several control algorithms in the
field showed a variety of computational and
interfacing problems. Some authors reveal that the
sample rates as low as hundred of μs are often
required to ensure dynamic performances. Therefore
a powerful processing unit is required.
Two interface classes can be considered: process
interface and user interface. Process interface is the
most requesting in terms of hardware resources,
while the user interface is strongly software
dependent. Most of the industrial solution for digital
drives control are closed systems, and do not allow
to the user to modify the control algorithm or to
embed new supplementary functions. These systems
are not useful for research. Papers reporting
university research reveal a poor experimental
activity or only simulation results are presented.
This paper focuses on an indirect vector control
solution implemented on an experimental set-up
build in the Technical University “Gh. Asachi”,
Jassy. The experimental results were obtained with
a software package called DMCDS (Digital Motion
Proceedings of the 6th WSEAS International Conference on Simulation, Modelling and Optimization, Lisbon, Portugal, September 22-24, 2006 99

Control Development Shell) developed also in the ⎧ e e

disd ⎛ L2 ⎞ e L
Technical University “Gh. Asachi”. ⎪usd = σLs + ⎜⎜ Rs + Rr m2 ⎟⎟isd e
− ωeσLsisq − m2 Rrψer
⎪ dt ⎝ Lr ⎠ Lr
⎨ e
(4)
2 Indirect vector control rotor field ⎪ e disq ⎛ L2m ⎞ e Lm e
⎜ ⎟ e
oriented of induction machine ⎪usq = σLs dt + ⎜ Rs + Rr L2 ⎟isq + ωeσLsisd + ωr L ψr
⎩ ⎝ r ⎠ r
The analysis of vector control structures highlights On the basis of this equivalent model of vector
several possibilities to control the instantaneous controlled induction machine, the equations of the
values of the electromagnetic torque of the induction decoupling and performance regulators can be
machine. Because the equations that describes the computed. Eq. (4) is used to design current
machine becomes significantly less complicated regulators and the decoupling regulator of the
when the reference orthogonal frame is synchronous command voltages. Taking into account only the
and synphasic with the rotor flux, almost all
practical schemes use this type of orientation [1]. ⎛ dΨ r ⎞
permanent magnetic state ⎜⎜ r ≅ 0 ⎟⎟ , eq. (4)
The position of rotor flux can be estimated directly ⎝ dt ⎠
or indirectly. The first type means the use of becomes:
transducers or flux estimators. The indirect
⎧ e e
di sd e e
estimation schemes of the rotor flux position ensure ⎪u sd = σL s + R s i sd − ωe σL s i sq
a good behaviour in all speed range, and they are the ⎪ dt
⎨ e
(5)
common solution in practice. ⎪ e di sq e e
The induction machine equations in the flux ⎪⎩u sq = σL s dt + R s i sq + ωe L s i sd
coordinate frame can be written as follows:
The control structure based on the inverse
⎧ d ψ se dynamic model of the induction machine assumes
⎪u se = R s i se + + jωe ψ se knowledge about five parameters of the machine: Rs,
⎪ dt
⎪ Rr, Lm, Ls şi Lr
d ψ er
⎪0 = R i e + + j(ωe − ωr )ψ r The parameters of the machine are usualy
e
r r
⎪ dt established with the short-circuit and no load tests of
⎪ e e e the induction machine . The value of the stator and
⎨ψ s = L s i s + L m i r (1)
⎪ e rotor leakege inductance is difficult to obtain
e e
⎪ψ r = L m i s + L r i r because from this tests data is possible to determine
⎪ only the sum of these paramaters L Σσ , but not each
⎪ dt
( m r )
⎪J dΩ = pL i e x i e − m − DΩ
s r one of them. The test standards recommend several
⎪ empirical distributions of this parameter:
⎩ The equivalent T circuit is incompletely specified
Simbols u, i, ψ denote voltage, current, and flux because the values of the stator and rotor leakege
linkage, respectively, rotor speed is ωr, leakage inductance have not unique values.
inductances are identified with index σ, index m The modern parameter estimation methods allow
denotes parameters and variables associated with to compute with high accuracy four parameters
magnetizing flux, index s denotes parameter and associated to the induction motor model: the
variables associated with the stator,and index r is for parameters of the stator circuit (Rs, Ls), the time
parameters and variables associated with the rotor. constant of the rotor Tr, and the global leakage
Indices e upperscript denotes the frame coefficient σ [2],[3]. If the designer knows the value
synchronous with the reference flux. of the stator leakage inductance L σs , the other
If the position of the rotor flux is known and the parameters can be computed as follows.
coordinate frame is symphasic with this one, then L2m L
e Lm = Ls − Lσs ; Lr = ; Lσr = Lr − Lm ; R r = r (6)
ψ r = ψ erd = ψ er ; ψ erq = 0 (2) (1− σ)Ls Tr
and (1) can be rewritten as follows. The sum of the leakage inductances of the machine
⎧dψer 1 e Lm e e is constant [4]. A good approximation of the stator
RrLm isq L
⎪ + ψr = isd;ωsl = e
; me = p m Ψreisq
e
leakage inductance L σs is [5]:
⎪ dt Tr Tr Lr ψr Lr
L σs ≈ 12 σL s (7)
⎨ (3)
⎪θ = ω dt = ⎛⎜ω + Lm isq ⎞⎟dt = ⎛⎜pΩ+ Lm isq ⎞⎟dt
e e
This approximation allows computing the rotor
⎪s ∫
e
⎝
∫
⎜ r Tr ψer ⎟
⎠
⎜
⎝
∫ Tr ψer ⎟⎠ leakage inductance as:
⎩
Proceedings of the 6th WSEAS International Conference on Simulation, Modelling and Optimization, Lisbon, Portugal, September 22-24, 2006 100

2−σ It is defined a matrix T

L σr ≈ σL s (8)
4(1 − σ) ⎡1 0 ⎤
T=⎢ 1⎥
(15)
The value of the ratio of stator leakage inductance ⎣0 α ⎦
and the rotor leakage inductance is defined as: This allows writing the equivalent model of the
L 2(1 − σ) motor (16).
k σ = σs = <1 (9)
L σr 2−σ ⎡R +L (p+ jωe) αLm(p+ jωe) ⎤ e
There are cases when this ratio can not ⎡use⎤ ⎢ s s
⎡ ( + ω ) ⎤ ⎥⎡is ⎤(16)
⎢ ⎥=⎢ αL (p+ jω ) α2 R p j e
+Lr (p+ jωe)⎥⎥⎢i'e⎥
characterize well enough the leakage in the stator ⎣ 0⎦ ⎢ m e ⎢ r( ( )) ⎣r⎦
⎣ ⎣ p+ j ω −ω ⎦⎦⎥
and the rotor of the induction machine. A better e r

approach seems to be the use of a minimum number The structure of the equivalent model is
of parameters which can define the transient presented in Fig. 1.
electromagnetic state and can be detected without ( L σs + L m (1 − α))(p + jωe ) (α2 Lσr + Lm (α2 − α))(p + jωe )
approximation.
i se Rs 'e
ir

3 Squirrel-cage induction motor

compact model us
e
αLm (p + jωe ) p + jωe
α2 R r
p + j(ωe − ωr )
The model of the squirrel-cage induction motor has
a degree of freedom because rotor voltage is null
(ur=0). This makes possible the obtain several
models which are energetically equivalent with the Fig. 1. The generalized phasor circuit of the squirrel-
classical symmetrical T model of the machine. cage induction motor
The induction machine can be assimilated with a
transformer. Then a turns ratio can be explained in The turns ratio matrix (15) does not modify the
the same manner as the turns ratio used for a input impedance at the motor terminals. The
transformer. The classical equivalent circuit of the equivalent circuit generates the same torque and
machine is obtained with the new state variables: speed for the same input signals. Some remarkable
N properties can be obtained with “asymmetric”
' ' N ' N
ur = s ur ; ir = r ir ; ψr = r ψr (10) equivalent circuits by choosing appropriate values
Nr Ns Ns for α.
and parameters A “resistive” circuit is obtained when α is chosen
⎛N ⎞
2
⎛N ⎞
2
N to vanish rotor leakage inductance in the phasor
R 'r = ⎜⎜ s ⎟⎟ R r ; L'σr = ⎜⎜ s ⎟⎟ Lσr ; L'm = s Lm (11) generalized circuit.
⎝ Nr ⎠ ⎝ Nr ⎠ Nr
α 2 L r − αL m = 0 (17)
This model is completely characterized by five
For this particular case, the factor α yields:
parameters.
Because the induction motor is squirrel-cage, the Lm
α= . (18)
N
turns ratio α = s can be chosen arbitrarily and any Lr
Nr The new asymmetrical Γ model is completely
other value that ensures the power invariance is specified by four parameters (19) and is presented in
allowed except 0 or ∞. A more general model of the Fig.2.
2
induction motor results from eq. (10)-(11), with the L ⎛L ⎞
new variables defined in (12) and the parameters Rs4 =Rs5; Ls4 =Ls5; Lσs4 =Lσs5 + m5 Lσr5; Rr4 =⎜⎜ m5 ⎟⎟ Rr5(19)
Lr5 ⎝ Lr5 ⎠
defined in (13).
' i rg ' Ψ Lm
i rg = ; Ψr = r (12) (Lσs + Lσr )(p + jωe )
α α e
Lr
'e
is ir
R 'r = α 2 R r ; L' r = α 2 L r ; L' m = αL m (13)
Rs

The phasor equations (1) can be rewritten in a

2
matriceal form: e
us L2m ⎛ Lm ⎞
⎜⎜ ⎟⎟ R r
p + jωe
R +L (p+ jωe) Lm(p+jωe)
(p + jωe ) p + j(ωe − ωr )
⎝ Lr ⎠
⎡use⎤ ⎡ s s ⎤⎡ e⎤ Lr
⎢
⎢ ⎥= L (p+jω ) R (p + jω ) ⎥⎢is ⎥(14)
⎣ 0⎦ ⎢⎣ m
e
+L (p+ jω )
(p+j(ωe −ωr )) r e ⎥⎦⎣ir ⎦
e
e r
Fig.2. The equivalent asymmetric equivalent circuit.
Proceedings of the 6th WSEAS International Conference on Simulation, Modelling and Optimization, Lisbon, Portugal, September 22-24, 2006 101

The leakage inductance has the value ⎧ e e

di sd
e e
L m5 ⎪u sd = R s i sd + L σs 4 − ωe L σs 4 i sq
L m 4 = L s 4 − L σs 4 = L m5 (20) ⎪ dt
L r5 ⎨ e
(26)
⎪ e e
di sq e
and the global leakage coefficient is ⎪⎩u sq = R s i sq + L σs 4 dt
+ ωe L s i sd
def L2m 5 L This equivalent model is a nonlinear process
σ = 1− = 1 − m4 (21)
L s5 L r 5 L s4 with two inputs (isd, isq) and two coupled outputs
The rotor time constant of this new model is (Ψr, me). The decoupling regulator is obtained with
L m4 L r5 the inverse model of the machine. This leads to a
Tr 4 = = = Tr 5 (22) feedforward structure presented in Fig.4.
R r4 R r5
This denotes a dynamic behavior equivalent with
the one of the model with five parameters. e
isd Lm ψr
e
mr
Finally the parameters of this model are univocal Tr s + 1
me Ω
determined on the basis of the estimated parameters p
Lm
+
-
1
Lr Js
of the motor ( R̂ s , L̂ s , T̂r , σ̂ ). e
isq

Rs4 = R̂s; Ls4 = L̂s; Lm4 =(1−σ

ˆ )L̂s; Lσs4 =σ
ˆL̂s; Rr4 =
(1−σˆ)L̂s (23) Lm
T̂r Tr
1

θs +
2
ωsl
1 p
+
s ωr
4 Indirect vector control of induction Fig.3. The induction motor model
motor in the rotor flux frame
The minimum number of parameters model of
induction motor is used for a vector control
structure. On this purpose the decoupling regulators * (Tr s + 1) i sd
e*

Ψr
must be designed taking into account the new state L m (Tr 0 s + 1)
variables and the new defined parameters.
With (12), (13) and (18), the phasor model from me
* 2
Lr i sq
e*

(1) becomes: 1 pL m 1
Lm
⎧ e e
dψ s
e
e Tr
ω̂sl
⎪u s = R s i s + + jω e ψ s
2

⎪ dt
Fig. 4. Structure of the decoupling regulator
⎪ dψ r 4
e
⎪0 = R r 4 i er 4 + e
+ j(ω e − ω r ) Ψ r 4
⎪ dt The transfer function of the reactive component is
⎪ e e e (24) not a proper one, and, for practical implementations
⎨ψ s = L s i s + L m 4 i r 4 is replaced with
⎪ e
( e
⎪ψ r 4 = L m 4 i s + i r 4
e
) e*
i sd Tr s + 1
⎪ = (27)
e
⎪m e = pL m 4 (i r 4 x i s )
e
ψ r L m (Tr 0 s + 1)
*

⎪ The active protection of the group inverter-motor is

⎪ realised through the limits of the current component,
⎩
±iisq max, which generates the torque. The decoupling
In the reference frame synchronous with the
rotor flux phasor, the phasor equations in of the stator currents allows a cascade control
structure like the one used for DC motor speed
cartezian coordinates are:
control. The optimal speed regulator parameters can
⎧ dψ r 4 1 L m4 be obtained with the pole-placement method.
⎪ dt + T ψ r 4 = T i sd
⎪ r r The inverse model used for current reference
⎪ L m4 values leads theoretically at an over-all system
⎨ωsl = i sq (25) without dynamic. In steady-state (27) becomes:
⎪ Tr ψ r 4
1 *
⎪m e = pψ r 4 i sq e*
i sd = Ψr (28)
⎪ Lm
⎩
Proceedings of the 6th WSEAS International Conference on Simulation, Modelling and Optimization, Lisbon, Portugal, September 22-24, 2006 102

The feedforward control structure induces a 1 ⎛ 1 ⎞ ⎛ 1 ⎛ 1 ⎞⎞

2

k pv = J ⎜1 + ⎟; k iv = J ⎜⎜ ⎜1 + 2 ⎟ ⎟⎟ (31)
limitation of the torque response. The transfer Tr ⎝ σ 2 ⎠ ⎝ 2 Tr ⎝ σ ⎠⎠
function is given by
The torque generation mechanism is
2
me 1ω2m ⎛ i*sq ⎞ underdamped. In this case it is desirable to impose a
= 2 , ωm = ⎜ ⎟ +1 , cut-off frequency for the speed loop much smaller
* 2
me s + 2ς mωms + ωm Tr ⎜ i*sd ⎟
⎝ ⎠ than the one of the torque generation mechanism:
−1 ωBv = KωBm ; 0 < K < 1 (32)
⎛ 2 ⎞
⎜ ⎛ i *sq ⎞ ⎟ Eq. (31) can be rewrittten with this coefficient:
ςm = ⎜ ⎜ ⎟ +1⎟ (29)
⎜ ⎟* 2
⎜ ⎝ i sd ⎠ ⎟ k pv = J
K⎛ 1 ⎞ ⎛ K ⎛
⎜1 + 2 ⎟; k iv = J ⎜⎜
1 ⎞⎞
⎜1 + 2 ⎟ ⎟⎟ (32)
⎝ ⎠ Tr ⎝ σ ⎠ σ ⎠⎠
⎝ 2Tr ⎝
The speed loop is analyzed with (29). A PI regulator The value of this coefficient can be
is sufficient in order to ensure a null steady-state experimentally determined.
error. The structure of the speed loop is given in This control structure can be used for current-fed
Fig.5a. The electromagnetic time constants are small inverters. If it is used a voltage-fed inverter the
enough to be neglected in this model. The simplified orthogonal components of the stator current will be
model of the speed loop is given in Fig.5b. indirectly controlled through the stator voltage
Neglecting the viscosity coefficient, D, the components. Two suplimentary loops must be
transfer function of the control system is: added.
2
ωv
Ω (s + z ) In the steady-state magnetic regime, the system
*
= 2 z becomes the system described by (26).
Ω s + 2ς v ωv s + ω2v
The tuning of the PI current regulators can be
k iv k pv k done on the basis of the open loop transfer function
ωv = , ςv = , z = iv (30)
J 2 Jk iv k pv Tii s + 1 K 1 σL s
H (s) = K Ri ; K= ; T= (33)
Tii s Ts + 1 Rs Rs
mr
Ω
The time constant of the electric circuit, T, is
Ω* + * 2
me -
ki − v me ωm 1 compensated with the integral time constant of the
kp− v + 2 2
s + 2ζmωms +ωm + Js + D
s
-
PI regulator.
1 T
H 0 (s ) = ; τ= (34)
(a) τs + 1 K Ri K
mr For a specific time constant of the current loop,
Ω * + m*e me - 1 Ω the tuning parameters of the current loop can be
kp− v + k
i− v

- s + Js + D computed:
σL s
K Ri = ; Tii = T = σTs (35)
τ adm
(b)
Finally the complete control structure is
Fig.5. Speed control loop structure
presented in Fig.6.
Neglecting the viscosity coefficient, D, the
transfer function of the control system is:
ωv 2 5 Experimental results
Ω (s + z ) This control structure was digitally implemented and
*
= 2 z
Ω s + 2ς v ωv s + ω2v tested. The voltage-feed induction motor with a
k iv k pv k PWM inverter has the nominal parameters:
ωv = , ςv = , z = iv (30) PN=1.5kW, UN=380V, IN=3.8 A, cos ϕ=0.79,
J 2 Jk iv k pv
ΩN=1410 rev/min, fN=50 Hz [8]. The results were
The design objective for the speed regulator is to obtained by numerical simulation with Matlab, and
obtain a minimum transient time for step input. Due with aDSP board C31MCDB [6].
to the limited capacity of the torque response of the The experimental results are presented in Fig. 7-14.
machine, the cut-off frequency of the loop can not Fig. 7-10 show the motor speed, the motor estimated
be bigger than the one of the transfer function (29). torque and the values of the orthogonal components
If the cut-off frequencies are equal, the parameters of the motor current when the electric drive is not
of the speed regulator are: loaded.
Proceedings of the 6th WSEAS International Conference on Simulation, Modelling and Optimization, Lisbon, Portugal, September 22-24, 2006 103

ψr4 (Tr s + 1) PI id + usd* uA*(t)

* +
L (T s + 1 isd* - (kr-i ,Ti-i) - – uB*(t)
P
1
usq* (θ ) uC*(t)
Ls
θs
L σs 4
isq * isd iA(t)
me*
pΩ + *
PI Ω 2 + PI iq +
1 + iB(t)
- (kp-v , ki-v) (kr-i ,Ti-i) P (θs)
p - isq iC(t)
1
1 θs
ω̂sl
L m4 + ω̂e 1 θs
2 T + s
pΩ

Fig.6. The analog control structure

6 Conclusion
1 5 0 0

1 0 0 0

5 0 0 The induction motor is not a system completely

observable. As a consequence the symmetrical
r e v /m i n
0

-5 0 0
phasor circuit can not be completely specified on the
basis of the measurements at the motor terminals
-1 0 0 0

-1 5 0 0

configured in a single topology. The phasor model

0 0 .5 1 1 .5 2 2 .5 3 3 .5 4 4 .5 5 5 .5 6 6 .5
tim e [s ]

Fig.7. Speed of the vector control induction motor must be reformulated in order to be completely
drive specified with a minimum set of parameters. The
6
new vector control scheme based on an
4
asymmetrical equivalent model uses only the
information obtained from the parameter
2
A
0

-2
identification experiments without approximation.
-4

-6 0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 5.5 6 6.5

time[s]
References:
Fig.8. The reference/measured value of the reactive [1] Vas P., Vector Control of A.C. Machines. Clarendon
component of the stator current Press, Oxford, 1990.
25 [2] Horga V., Graur I., Răţoi M., Indirect Parameter
20
15
Estimation of Induction Motor Based on Time Domain
10 Analysis. Proc. of 11th Int. Symp. on Power Electronics,
5
Nm
0 Novi-Sad, Yugoslavia, 246-250, 2002.
-5
-10
[3] Onea A., Horga V., Diaconescu M., Boţan C., Off-
-15 Line Parameter Estimation of Induction Motor Based on
-20
-25 0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 5.5 6 6.5
a Multitime Scale Approac: Mastorakis N., Antoniou G.
time [s]

Fig.9. The estimated value of motor electromagnetic (Eds.). Recent Advances in Circuits, Systems and Signal
Processing, WSEAS Press, 175–180, 2002.
torque
[4] Bassi E., Benzi F., Bolognani S., Buja G., A Field
25
20 Orientation Scheme for Current-Fed Inductiom Motor
15
10
Drives Based on the Torque Angle Closed-Loop Control.
A
5
0
IEEE Trans. on Ind. Applications, 28, 5, 1038-1044
-5 (1992).
[5] Ben-Brahim L., Gastli A., Al-Hamadi M., Auto-
-10
-15
-20
-25 0
Tuning for Sensorless AC Motor Drive Systems. IEEE Int.
0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 5.5 6 6.5
time [s]
Symp. on Ind. Electronics, Bled, Slovenia, 367-372, 1999.
Fig. 10 The reference/measured value of the active [6] Onea A, Horga V., Educational Software Environment
component of the stator current for Motion Control. Proc. of the IASTED 5th Int. Conf. on
Comp. and Adv. Techn. in Education, 334-339, 2002.

View publication stats