You are on page 1of 6

A Tutorial on PI Velocity Control of Induction

Motors
Liuping Wang and Lu Gan
School of Electrical and Computer Engineering
RMIT University, Australia
Abstract-This paper presents vector-based PI control of
induction motors in a tutor ial style. Based on a general nonlinear
model for induction motors, a cascade control structure is
proposed for velocity control, where the inner-loop systems are
designed using proportional controllers with dynamic decoupling
whilst the outer velocity loop is controlled by PI controller.
Desired closed-loop performance parameters, such as desired
bandwidth and closed-loop gain, are used for selection of the
controller parameters. Robustness of the control system against
parameter variations is characterized using gain and phase
margins of the closed-loop system. Exper imental results are
obtained to illustrate the design.

I.

INTRODUCTION

Almost all electric drives in the current industrial applica


tions are controlled by the PI controllers. The controller struc
tures are already defined in the commercial drives, however
the controller parameters may be adjustable for performance
improvement. In a typical drive, there are a number of PI
controllers used, depending on the actual applications. To our
knowledge from working with a number of industries, the
PI control systems for the drives are in cascade feedback
structure in which both inner-loop and outer-loop systems are
controlled using PI controllers. This structure at least could
be traced back for the past two decades, (See [1],[2],[3],[4]).
It is necessary to use PI controllers for the outer-loop system
because of the closed-loop performance requirement for the
disturbance rejection of load torque disturbance. However,
question seldom raised whether it is necessary to use PI
controllers for the inner-loop current control.
In this paper, a cascade vector control is proposed for
velocity control of induction motor, where the inner-loop
current isq is controlled by a proportional controller and
the outer-loop velocity is controlled by a PI controller. We
believe that this structure has been proposed before for the
application of induction motor control, but it was not widely
received in the current industrial practice. In order to bring
this simpler control architecture to the attention of industrial
practitioners, this paper will discuss how to design the inner
loop proportional controller, how to design the outer-loop PI
controller as well as how to integrate the inner-loop system
with the outer-loop system. All the design approaches will
be based on the knowledge of the physical induction motor
model with desired closed-loop performance specifications.
Model uncertainty and parameter variations will be addressed
using frequency response analysis tools in terms of the desired
closed-loop performance specifications.
978-1-4673-2421-2/12/$31.00 2012 IEEE

This paper is organized as follows. In Section 2, the cascade


PI control system is discussed in detail to guide the users on
the design and performance specification. In Section 3, the de
sired closed-loop performance parameters, such as the steady
state gain for the inner-loop controller and the bandwidth for
the outer-loop control system, are discussed and Nyquist plot
is used as a graphic guide for the selection of these parameters.
Experimental results are used in Section 4 to demonstrate the
closed-loop control performance on a test bed.
II.

CASCADE

PI

CONTROLLER DESIGN

In this section, a cascade PI velocity control system is


proposed where the inner-loop is controlled by proportional
controller and outer-loop is controlled by PI controller. There
is also an independent PI control loop for the d-axis current
because of the requirement of zero steady-state error for
the d-axis current. The starting point is on the assumption
of vector control with correct Park-Clark transforms so that
the dynamics of the induction motors are described by the
nonlinear equations (1- 4) .
A.

Cascade Control System Structure

From the literature, several standard models of induc


tion motor are available to use for control system design
([1],[5],[6]). Among them is a mathematical model with four
differential equations in direct-quadrature (d - q) coordination,
which is originated from the field oriented control theory
([ 1],[7]). This model includes both electrical and mechanical
components. When the parasitic effect such as eddy currents,
magnetic field saturation are neglected, the dynamic model of
an induction motor is governed by the following differential
equations [7]:
.

disd
dT
.
/ disq
tsq + Ta
dT
d'I/Jr d
'l/J rd + T rb
tsd + Ta

f dW+

dw
J dt

1
.
kr
/
TaWstsq + --'l/J rd + -Usd
raT r
ra
1
.
kr
/
-TaWstsd - -w'l/J rd + -usq
ra
ra
Lh isd
3ZpLh
.
sq
-----:u:;-'l/J r d t - TL

(1)
(2)
(3)
(4)

where isd and isq are the stator current in the d - q co


ordination, 'l/J r d is the rotor flux in d-axis and the input
variables Usd and usq represent the stator voltages in the dq
coordination. Ws and ware the synchronous and rotor velocity

1648

respectively. For notational simplicity, the constants inside the


model are derived from the motor parameters, T;
"rLs,
=

L . , Tr
L
1 - LLi'
R s + Rrkr2 ' kr
I!::
5t and a
sLr'
where Rs and L s are the stator resistance and inductance, Rr
and Lr are the rotor resistance and inductance respectively,
Lh is the mutual machine inductance; TL is the load torque,
id is the friction coefficient, J is the inertia constant and Zp
r"

is the number of pole pairs.

The manipulated variables are Usd (t) and usq(t) , and the
outputs of the system are the velocity w()t and rotor flux
1/;r d (t) . Thus, the entire system is a fourth order nonlinear
system with two inputs and two outputs. As we know, a PI
controller is designed based on a model of a single-input
and single-output system. If the system dynamics are more
than first order, there will be approximation errors in the
design when using a PI controller structure. It appears that the
straightforward application of PI controllers for the induction
motor would not be fruitful, because of the system order, the
number of inputs and outputs and the existence of nonlinearity.

outer-loop velocity control to produce desirable closed-loop


performance, the variation of 1/;r d (t) needs to be minimized
so that it can be closely approximated by a linear time
invariant parameter. In other words, its value needs to converge
to an expected steady-state value in a closed-loop feedback
control. From the model equation (3), the dynamic response
of 1/;r d from isd is a first-order system with time constant Tr.
Therefore, if the d-axis current isd is regulated to a constant
value, then convergence of 1/;r d to a constant is ensured. In
the steady-state, 1/;; d (t)
Lh i;d (t) where 1/;; d (t) and i;d (t)
are the set-point signals for 1/;r d (t) and isd (t) respectively. To
maintain the correct set-point value 1/;;d (t) , a PI controller is
required for the d-axis current control.
Figure 1 shows the cascade feedback PI control system of
an induction motor with compensation for the nonlinearity.
=

A cascade control system architecture is best suited in


this situation because with the cascade control system the
higher order system can be separated into several lower
order subsystems and hence PI controllers can be applied
effectively. In addition, the nonlinearity can be compensated
using a technique called feedback linearization. The potential
disadvantage of using cascade feedback control structure is
the deployment of additional sensors for the subsystems. For
the induction motor control, all the currents and voltages are
measured, therefore, there is no additional sensor required.
Upon the selection of cascade control system architecture,
the structures of inner-loop and outer-loop are determined via
the analysis of the time constants and the interactions of the
subsystems. There are four subsystems in the induction motor
model described by (1), (2), (3) and (4). The time constant
T; is much smaller than the time constant . Therefore,
there will be an inner-loop controller for controlling the q
axis (isq) current by manipulating the voltage usq. The inner
loop control objective is secondary in terms of delivering
the ultimate control objectives. As a result, the accuracy
of the inner-closed-loop system response in steady-state is
not critical. Instead, the inner-loop dynamic response speed
reflected in terms of closed-loop pole location, easiness in
control system implementation and tuning, and robustness of
the closed-loop cascade control system are of the important
considerations when selecting the controller structure for the
inner-loop control system. For these reasons, it is proposed
in this paper to use a proportional controller for the inner
loop control systems. Furthermore, because of the existence of
nonlinearity and the dynamic coupling, input-output feedback
linearization and feedforward compensation is deployed to
form a linear closed-loop system for the inner-loop. The outer
loop system is for the velocity control by manipulating the set
point of the current isq, which achieves the control objective.
The electrical-mechanical equation (4) reveals the dependence
of velocity w()t on both isq(t) and 1/;r d (t) . In order for the

----'-----1
Fig. l.

B.

Eo<

Nonlinear cascade control of induction motor

Cascade Control of Velocity

The cascade control of velocity involves the dynamics of


the current isq and the dynamics of the velocity w, which
are described by models (2) and (4). The isq current model
will be used for the inner-loop design of proportional feedback
linearization control system and the motion model for velocity
will be used for the design of a PI controller for the outer-loop
system.
In the first stage of the design of a cascade control system,
the inner-loop proportional controller is considered. From (2),
an auxiliary variable usq(t) is defined by the relationship:

1 A
-, usq(t)
r"T
"

I
.
kr
-, usq(t) - ws(t) Zsd (t) - -, W()t 1/;r d (t)
r"T
r"T
"

"

(5)

By substituting (5) into (2), we obtain the first order differen


tial equation:
(6)

which will be used as the basis for the design of proportional


controller for the inner-loop system. The Laplace transfer

1649

i8q(t)1/Jrd(t), the product is approximated using Taylor series


expansion:

function of (6) is
(7)

Since the proportional control will lead to a steady-state error


in the closed-loop system, its closed-loop steady-state gain
is not unity, and this steady-state gain should be taken into
consideration in the design of the outer-loop control system.
Along this line, the desired inner-loop control performance is
specified via the steady-state gain of the the proportionally
controlled system. Let Kg denote the proportional feedback
controller gain. The closed-loop transfer function between the
18q(s) and set-point signal I;q(s) is
(8)

The closed-loop pole is found by letting the denominator S +


Kq
1
,.
= 0 leading to
TO' + T(TTa
, _c

Kg
s=- - - -1
T

Tra

For Kg > 0, the closed-loop pole is located at the left


hand side of the open-loop pole --1;in the complex plane,
T"
which means that the closed-loop dynamic response is always
faster than the open-loop response. The steady-state gain of
the closed-loop transfer function (8) is calculated by letting
S = 0, leading to
Kq
T'-..
,, r ,,
--,
-=-;;-;;;--:K2
1
+
T
Tr(T
__C_
,

= cx

(9)

In the design of the proportional controller, the desired closed


loop performance is specified by choosing a desired value
for cx, which from (9) gives the solution of the proportional
controller gain Kg,

cx
q
Kc = __ra
1 - CX

(10)

For instance, if the steady-state gain cx = 0. 9, then the


proportional controller gain is Kg = 9ra, and the closed-loop
.
pole is at 10
T"
Assuming the set-point signal to the current control loop is
i;q, the auxiliary control signal u8q(t) is calculated using the
feedback error signal i;q(t) - i8q(t)

i8 q(t) 1/Jrd (t)

-i 1/Jd

+ 1/Jd isq(t) + i1/Jrd (t)

Substituting this approximation into (4), we obtain the lin


earized model for the design of PI control of the velocity,
3Z L
dw(t)
= iJd w(t) + 2Lp Jh X
dt
r
ss 88 + S8 q(t) + S8q rd ()t ) TL
( -t8q1/Jrd 1/Jrdt8
-J
t8 1/J
3Z
L
d
p h ss
i
- Jw()t +
1/Jrdt8q ()t +
2Lr J
3ZpLh
3ZpLh
TL
88.i.
88.i.88
t8q'f/rd
t8q'f/rd (t)
_

2Lr J

2Lr J

where /'i,t =
The last three terms on the right-hand side
of (12) are considered as disturbances in the feedback control
system. In particular, the constant term /'i,t i 1/Jd is an input
constant disturbance and if the load torque f changes in a
step signal manner, then both disturbances will be completely
rejected without steady-state error by the PI controller. The
term /'i,tig1/Jrd(t) is a linear function of the rotor flux 1/Jrd(t).
Because i8d(t) is under PI control, in the steady-state, 1/Jrd(t)
will converge to its set-point signal 1/J;d' where 1/J;d = Lhi ;d'
hence becoming a constant. This disturbance term will also be
completely rejected without steady-state error. In short, with
PI control of velocity, all the constant terms on the right-hand
side of (12) are regarded as constant disturbances that will be
completely rejected by the action of integrator in the controller.
To design the PI velocity controller, based on (12), consider
the Laplace transfer function between the velocity n ( s) and
the current I sq( s
),

3;LLj'.

/'i,t 1/Jd

S+

li
J

By substituting the current feedback signal 18q(s) with its set


point signal I;q(s) using the closed-loop transfer function (8),
the relationship between the velocity n( s) and the set-point
signal for the inner-loop current control is established:
(13)

u8 q(t)

= Kg(i;q(t)

- i8q()t )

Thus, from the relationship between u8q(t) and usq(t) given


by (5), the feedback control signal u8q(t) with linearization is
obtained as

u8q(t)

= Kg(i;q(t)

where Kg = Io:ra and cx is the desired steady-state gain


which is specified in the design of inner-loop proportional
controller. Generally, cx is chosen close to 1 in range of
o < cx < 1. Then, the assumption can be made that the time
constant is far larger than the time constant (1 - cx) T
from the inner-loop current control. The transfer function (13)
is approximated as following

- i8q()t ) + raTwsisd (t) + k r w1/Jrd (t)

(11)

In the second stage of the design of a cascade control


system, the outer-loop PI control system is considered. The
mathematical model for the outer-loop control of w(t) is based
on the motion model (4). In order to linearize the bilinear term

(12)

n (S)
-I; q( s
)

Let a = if and b =
represented by

1650

/'i,t 1/Jd
S + Jy

--xcx

/'i,t 1/Jd x cx

C S + Co
C ( s) = --S
I

and the PI controller be

1
= K c ( 1 + -)
TIS

, where K c is the proportional gain and T1 is the integral


time constant. A standard procedure to find the PI controller
parameters is to specify the desired closed-loop bandwidth
Wn and a damping coefficient
0. 707, and solving the
polynomial equation
=

(14)

where the left-hand side of the equation (14) is the polynomial


that determines the actual closed-loop poles and the right-hand
side is the polynomial that determines the desired closed-loop
poles. By equating these two polynomials, the actual closed
loop poles are assigned to the desired closed-loop poles, which
leads to the controller coefficients:

2wn - a

Usd

Kc

D.

2wn - a

(16)

The control signal i;q(t) is computed using the position form


from the outer-loop PI controller, where

i;q(t)

r (W*(T )
Jo

- w( T )) d T

In order to introduce saturation on the q-axis current isq(t) ,


the feedback error from inner-loop current control, eisq
i;q(t) - isq(t) , needs to be considered, because of the steady
state error between the set-point signal i;q(t) and the feedback
signal isq(t) . Assuming that the maximum current allowed
is iax, the implementation of saturation and anti-windup
scheme should be based on the following inequalities:
=

C. PI

_iax + eisq(t)

i;q(t)

Slip Estimation

(15)

Kc
K c (w *(t) - w(t)) +
T1

I Jor (i;d (T ) - i Sd (T )) dT

K
T[

kr
'l/J r d (t) - Traws(t) isq(t)
Tr

Ws (t)

and the integral time constant is calculated as

T1

K (i;d (t) - iSd ()t ) +


-

or in terms of PI controller parameters:

2wn -

The position of the field is required in the vector control of


induction motor, the angle Bs is estimated for the position of
the d - q coordinates. In this paper, the simple open-loop slip
estimation is used, where the model equation is defined as

Co

Let a
-l;and b
_
,__1 . The desired closed-loop control
T(T
TaTadesigned is identical to the one used in the previous section,
by choosing the bandwidth parameter Wn and a damping
coefficient , leading to the proportional control gain and the
integral time constant via the equations (15) and (16).
The manipulated variable Usd is computing using the posi
tion form of the PI controller where

iax + eisq(t)

W(t) +

isq(t)
.
T r Zsd ()t

(18)

By assuming the motor is started from stillness and the initial


values Ws and W are equal to zero, then integrating both sides
of the equation (18) leads to

Bs(t)

B(t) +

{t Sq(T )
T r Jo Zsd (T )

dT

where B(t) is the measured rotor position from the encoder.


Because the current feedback signals isd and isq typically
contain a significant amount of noise, direct application of
(II-D) leads to inaccurate estimation of Bs(t) . The traditional
approach is to use the reference signals i;d and i;q to replace
the actual isd and isq in order to reduce the effect of noise.
In addition, because the inner-loop proportional control causes
a steady-state error between the reference current i;q and the
feedback current isq, the steady-state gain for the inner-loop
current control system ( cx)is taken into consideration, which
yields the estimation of Bs,
=

Control of d-axis Current

The PI control of isd current with linearization by using the


model equation (1), firstly, let the auxiliary control variable
Usd (t) be defined by

1
-, Usd (t)
Tara
A

.
1
kr
-, Usd (t) + -, -'l/J r d (t) + Ws Zsq(t)
Tara
TaraT r

By substituting this auxiliary variable into model equation (1),


a linear time-invariant (LTI) model is obtained with respect to
the auxiliary control variable Usd (t):

disd (t)
dt

--

1 .
1
-,Zsd (t) + -, Usd (t)
Ta
Tara
A

The Laplace transfer function of (17) is

(l7)

III. TUNING INNER-LOOP AND OUTER-LOOP


CONTROLLER
The tuning process for the inner-loop controller is to reach
a compromise between the closed-loop response speed and
the amplification of input noise from PWM, measurement
noise and measurement disturbance, both from current sensors.
These undesired factors will affect the choice of bandlimit of
the inner-loop closed-loop system.
The inner-loop proportional controllers for q-axis current
control is calculated using
cx
q

Kc

ra

__

1 - CX

(19)

where cx is the steady-state gain of the inner-loop control


system. With this calculation of the proportional gain, the
1651

closed-loop transfer function for the inner-loop system is

T1(s)
,

T1(s)
,

(1

- a To-S

+1

(20)

The closed-loop time constant is (1 - a ) To- and the closed


loop bandwidth is Wb
la ;a' It is clear that as a increases,
the controller gain increases and the closed-loop bandwidth
increases. In theory, 0 < a < 1. But in practice, when a is
too small, the closed-loop cascade control system performance
deteriorates because there is a large mismatch at the steady
state between the desired q-axis current i;q and the actual
q-axis current isq(t). On the other hand, when a is too large
(close to 1 ), the controller gain Kg increases dramatically (to
00 in the extreme case). The extremely large controller gain
will amplify the noise and cause the closed-loop instability
because of modeling errors. Through the analysis, simulation
studies and experimental trials, a is found to be in the
range between 0. 7 and 0. 9 in order for the cascade closed
loop control systems to work well. For simplicity, we choose
a
0. 7 for a slow response speed, a
0. 8 for a medium
response speed and a
0. 9 for a fast response speed in the
inner-loop system.
The tuning process for the outer-loop controller is to reach a
compromise between the closed-loop bandwidth and the effect
of measurement noise from the encoder, load disturbance,
possible sinusoidal disturbance from inner-loop current control
system due to current sensor bias, and the neglected dynamics
in the design. A faster closed-loop response in the outer
loop control system will require a larger desired closed-loop
bandwidth, which is necessary for high performance to set
point response, load disturbance rejection and the current
sensor bias compensation. On the other hand, the measurement
noise and the neglected dynamics in the design restrict the
width of the closed-loop bandwidth.
The parameter is the damping coefficient, which is chosen
to be 0. 707, and the parameter Wn then corresponds to the
bandwidth of the outer closed-loop system. The choice of Wn
in theory is between 0 and 00, which is extremely wide. Thus,
in practice, it is useful to narrow down the range of parameter
selection so that this parameter can be selected easily.
Instead of selecting Wn directly, a parameter "( is defined as
the ratio between the open-loop bandwidth and the closed-loop
bandwidth:
=

"(

Wn

= -

system as Wn
a. We can also have a rough estimate of the
closed-loop settling time for the choice of"( parameter. For the
approximate second order from the outer closed-loop system,
the settling time is estimated as E!n' and the open outer- loop
system is a first order system with settling time estimated as
i. Therefore, the ratio between the closed outer -loop system
nd the open outer -loop system is approximately E'
Determining the maximum of the closed-loop bandwidth
is more complicated. The closed-loop control system will
become unstable if the parameter Wn is chosen to be too large.
At a quick look, one could not easily think a first order system
controlled by a PI controller to become unstable. The cause of
instability for the induction motor drive here was the neglected
inner-loop dynamics in the design that was first order model
with a small time constant, sensor dynamics and possible time
delay due to the PWM. So, the actual outer-loop system is at
best a second order system controlled by a PI controller, and
this closed-loop system could become unstable. Thus, we need
to carefully select the factor "( for the outer-loop PI control
system.
Nyquist plot ([8]) of the outer-loop control system provides
an effective means to determine the bandwidth Wn to be used
in the design. A good design should have reasonable gain and
phase margins to ensure that the closed-loop system is robustly
stable in the presence of factors known and unknown. Figure
2 shows the Nyquist plot of the cascade control loop of w.
The Nyquist loci show that when the inner-loop proportional
controller gain is fixed with the a
0. 9, increase of the outer
loop bandwidth Wn
"(a via the value of "( will result in the
changes of Nyquist curves. In particular, larger "( leads to a
smaller gain margin as shown in Figure 2.
=

. ,
-2
3

'.
5

- -

0.
5

('
-0
, )

(21)

This parameter is to normalize the closed-loop bandwidth Wn


against the parameter a (the dominant pole of the open outer
loop system). The parameter a also approximately represents
the bandwidth of the open outer-loop system when the inner
loop dynamics are neglected. By substituting the parameter "(
into the equations for computing the PI controller parameters
(see (15) and (16)), it is evident that in order to ensure a
positive T[, 2"( - 1 > O. A lower bound on "( is reached,
0. 707. Giving it some safety margin, the
that is "( > 21E
minimum of "( is set to 1 in the tuning process. This, in
turn, sets the minimum of the bandwidth of the outer-loop

5=---_--'-, -----:0
- :"::
.
5--'------=
0
"
.
=
5----'----'
: =
'.
'
5-----'
-2L---"'.

Fig. 2. Nyquist plots with respect to changes in "f. Key: line(l) I


(2) I = 7.5; line (3) I = 10; line (4) I = 12.5;.

IV.

5; line

EXPERIMENTAL RESULTS

The cascaded control system proposed in this paper is


implemented on an industrial size induction motor using

1652

_U 2

MATLAB xPC Target software. The characteristics of the


motor have 2 pole pairs, star connection, 0. 75 kW power with
supply frequency of 50 Hz, the rated velocity is 1435 RPM,
supply voltage is 415 V and rated current is 1. 75 A. The
parameters used in the experiments correspond to the nonlinear
0. 2;
model (see 1-4), which are listed as follows. 1fJ; d
Rs 11. 2; Rr 8. 3; L s 0. 0455; Lr 0. 068; Lh 0. 57;

- 0

--- 1

10

15

20

r :i. '.Ilr'falll* "1

Lh + L s; Lr Lh + Lr; kr Lh/Lr; T r Lr/Rr;


(J
1 - LV(L sx Lr
) ; fd 0.0023; J 0. 0052; Zp 2;
R" Rs + Rr * k ;; T " ( (J * L)s /R,,) . The desired closed
loop specifications for inner-loop current control for isq is
a:
0. 9 and for is d is "(is d
10, the outer-loop specification
Ls

5
- 0

is "(W
6. 667, The sampling interval for the implementation is
/:).t
100e-6. With these specifications, the inner-loop P con
160.4251, the PI controller for
troller for isq has gain: Kg
isd has K sd 234. 2206 and T j sd 7. 8326*10-4. The outer
0. 0716 and
loop PI controller has the proportional gain K':;
0. 0990.
the integral time constant T J
In the experiment, the set-point of the velocity is chosen to
be low-speed range 300 RPM, half of the rated speed 700 and
rated speed 1400 RPM. At ti
2sec, a step disturbance is
introduced to the feedback of current isq to represent the load
change.
Figure 3 shows the velocity closed-loop responses for the
sequence of set-point changes whilst Figure 4 shows the
closed-loop responses for currents control, and the manipu
lated voltage variables (Usd and usq) are displayed in Figure
5. All figures show that the PI cascade closed-loop control
of the induction motor has produced satisfactory results in all
speed ranges and disturbance rejection.

10
Time (sec)

15

20

Fig. 4. Current closed-loop responses. Key: line(l) Actual feedback; line(2)


Set-point signal.

200 ------------.---------,

200
00
----------- -------- -----0
-5
-10
15
20
Time (sec)

-4

5 00 --------------------------
o
5
10
15
20
Time (sec)
Fig. 5.

Manipulated variables

F-=1l

-2
proposed structure has advantages over the conunonly used
PI controllers for the inner-loop current control in terms of
model uncertainty and temperature variations for induction
motor control, which will be published in our future work.
REFERENCES

O---0---5--10--15---=2
Time (sec)

Fig. 3. Velocity closed-loop response. Key: line(l) Set-point signal; line(2)


Actual feedback.

V.

CONCLUSIONS

This paper has discussed the design of vector-based in


duction motor control using a cascade control structure. The
difference between what we discussed here and the common
practice in industries is the structure of the inner-loop current
controller. With the proportional controller used in the inner
loop current control, the design is simplified and the exper
imental results show satisfactory results. Both experimental
and analytical results have been obtained to show that the

[l] Nguyen Phung Quang and Jorg-Andreas Dittrich. Vector Control of Three
Phase AC Machines. Springer, 1st edition, 2008.
[2] Robert D. Lorenz, Thomas A. Lipo, and Donald W. Novotny. Motion
control with induction motors. Proceedings of the IEEE, VoI.82:121540, 1994.
[3] L. Harnefors and H.-P Nee. Model-based current control of ac machines
using the internal model control method. IEEE Transactions on Industry
Applications, VoI.34:133-41, 1998.
[4] Fernando Briz del Blanco, Michael W. Degner, and Robert D. Lorenz.
Dynamic analysis of current regulators for ac motors using complex
vectors. IEEE Transactions on Industry Applications, Vo1.35:1424-32,
1999.
[5] E.R Filho and R.M de Souza. Three-phase induction motor dynamic
mathematical model. In IEEE International Electric Machines and
Derives Conference Record, page MB1I2.1, May 1997.
[6] J Holtz. The induction motor-a dynamic system. In 20th International
Conference on Industrial Electronics, Control and Instrumentation, vol
ume I, page PI, Sep 1994.
[7] Arne Linder, Rahul Kanchan, Ralph Kennel, and Peter Stolze. Model
Based Predictive Control of Electric Drives. Cuvillier Verlag Goettinggen,
2010.
[8] G. C. Goodwin, S. Graebe, and M. Salgado. Control System Design.
Prentice-Hall, Englewood Cliffs, NJ, 2000.

1653

You might also like