SYMMETRY BASED CONTROL OF INDUCTION MOTOR

M. Monika1 Singh N. M.2 Bhil S.K.3

1- Faculty, Dept. of Electronics Engineering, Pillais Institute of information Technology, Mumbai, India.
2- Faculty, Department of Electrical Engineering, V.J.T.I. Mumbai, India.
3- Faculty, Department of Electrical Engineering, V.J.T.I. Mumbai, India.
Abstract- In this paper symmetry based control of induction motor is proposed. The fifth order model of
Induction motor is reduced to the base coordinates which is decoupled from the fiber dynamics by using a
regular static feedback. This makes the control of Induction motor similar to the control of separately
excited D.C. motor. This paper shows that the selection of a particular frame of reference for the two phase
equivalent model depends on the control objectives which are to be taken as the base coordinates.
1. Introduction
The field oriented control method [10,11] given by
German researchers Hasse and Blaschke is most widely
used method for the control of induction motor. The
advantage of this method is that it gives independent
control of flux and torque similar to the control of
separately excited D.C. machines.
This method consists of nonlinear change of
coordinates together with a nonlinear asymptotically
decoupling state feedback.
The whole idea of vector control is based on the
symmetry. This paper shows that Direct control
method of vector control consists in reducing the
dynamics to the base dynamics which is decoupled
from the fiber dynamics. This paper shows that in the
direct control method the symmetry group action on
the manifold reduces the control to the base
coordinates.
Section 2 recalls the classical model of induction motor
Section 3 of this paper gives the brief idea of
symmetries and fiber bundles. Section 4 gives
generalization of the direct field control method to a
control problem invariant under a transformation group
of finite dimension. This section also explains the
reason for choosing a particular frame of reference for
the two phase equivalent model of induction motor
methods using symmetry concepts.
2. Model of Induction Motor
(A)The two phase equivalent representation of a
three phase Induction Motor
It is given by the following set of equations [1]. The
equations for the stator and rotor are
d
s
R s is +
= us
--(1)
dt
d r
R r ir +
=0
--(2)
dt

Where

s = Ls i s + M i r e j

--(3)

r = L r i r + M i s e-j

--(4)

Rs - Stator resistance,
is - Stator current ,
s - stator flux,
Rr - rotor resistance,
ir - rotor current ,
r - rotor flux,
us - applied voltage to the stator ,
Ls - stator inductance,
Lr - rotor inductance and
M - mutual inductance between stator and rotor.
The rotor flux and currents are expressed in stator
reference frame, is the angle of rotation of the rotor
d
is the instantaneous angular velocity of
and =
dt
the rotor.
The stator current is expressed in the complex notation
as
i s = isa + jisb
where isa and isb are the currents in two stator phases
and similarly us can be written as
u s = usa + jusb
where usa and usb are the voltages applied to two stator
phases.
The torque produced by the motor is computed with
the help of Lorentz force law [1] and it is given by
2
= M Img[ i s (i r e j )*]
3
As the rotor currents can not be measured with the cage
rotors, the rotor current vector is substituted by the
rotor flux or the equivalent magnetizing current vector
imr which is defined in stator coordinates as
r e j = M i mr = Mimr e j = M [i s + (1 + r )i r e j
Eliminating the rotor currents in the torque we get

2 M
Img[i s (i mr i s )*]
3 (1 + r )

This can be written as

2
= (1- ) Ls Img[ i s i mr *]
3

Where is the total leakage factor of the motor and it

is given by

= 1

1
(1 + s )(1 + r )

Where s and r are the stator and rotor leakage factor

Given by

Ls = (1 + s ) M
Lr = (1 + r ) M
The motion of the motor is given by

Lr
= rotor time constant
Rr
The equation (6) and (7) represent the model for the
current fed induction motor.
(C) Model with rotor flux reference frame
In this method the electrical variables are expressed in
d-q axis frame or rotor flux reference frame. The
variables are rotated through an angle i.e. rotor flux
angle .The stator current and the magnetizing current
then are
Where Tr =

i mr = imr e j

d
J
= L
dt
2
= (1- ) Ls Img[ i s i mr *]- L
3

-- (5)

Where J is the moment of inertia of rotor and L is the

load torque.
The equations (1) to (5) represent the two phase
equivalent model of three phase induction motor.
(B)The current fed explicit model
We can get the reduced model for the induction motor
by neglecting (leakage factor) as it is very small for
most of the motors and substituting the rotor current
vector by rotor flux or equivalent magnetizing current
vector imrin the equation for rotor current.

isd + jisq = i s e j
So the equations (6) and (7) becomes

d
2
= Ls isq imr - L
dt
3
di
Tr mr + imr = isd
dt
d
= Tr imr + isq
Tr imr
dt
J

r = M i mr e

and

1
(i mr i s )e j
(1 + r )
(This equation is obtained from the equation for the
magnetizing current given by

i mr = i s + (1 + r )i r e

in the equation for the rotor current given by

d r
dt

=0

Substituting the equation of stator current and

magnetizing current in the equations (6) we get

By solving the above equation we get

di mr
+ (1 jTr )i mr = i s
dt

d 2
= Ls Img[ i s i mr *]- L
dt
3

di
Tr mr + (1 jTr )i mr = is
dt

Tr

--(6)

d 2
= Ls imr isqr sin L
dt
3

Consider equation (7) and multiplying both sides of

the equation by e-j ,we get
di
Tr mr e j + (1 jTR )i mr e j = i s e j
dt
Substitution of the equation of stator current and
magnetizing current in the above equation gives

This yields the classical reduced model as

d 2
= Ls Im g[i s imr e j ( ) ] L
3
dt

Simplifying this we get

d 2
J
= Ls Im g[(i s e j )imr e j ] L
dt
3
This can be written as

we get

dMimr e j
Rr
(imr is )e j +
=0
M(1+r )
dt
Tr

- -(10)

i s e j =isdr + jisqr and i mr = (imr e j )e j

ir =

Rr ir +

- -(9)

(D)Model with rotor reference frame

Now instead of rotor flux reference frame if we choose
rotor reference frame then the variables are rotated
through an angle i. e. the angle of rotation so the
stator current and magnetizing current now becomes

Substituting
j

- -(8)

dimr e j ( ) j
e
+ (1 jTr )imr e j ( ) e j = i s e j
dt

By solving the above equation we get

--(7)

isdr + jisqr =
dimr j
+ jTr imr ( mr )e j + (1 jTr )imr e j
e
dt
d
d
Where mr =
and =
dt
dt
On simplification this reduces to
d
Tr imr e j + (1 jmr Tr )imr e j = isdr + jisqr
dt
d
Tr imr + (1 jmr Tr )imr = (isdr + jisqr )e j
dt
d
Tr imr + (1 jmr TR )imr = (isdr + jisqr )(cos + j sin )
dt
Comparing the real and imaginary parts, we get
d
Tr imr + imr = isdr cos isqr sin and
dt
mr Tr imr = isdr sin + isqr cos
= Tr

So the equation (6) and (7) becomes

d
2
Ls imr [ isqr sin ] - L
=
dt
3
di
Tr mr + imr = isdr cos isqr sin
dt
d
Tr imr
= isdr sin + isqr cos
dt
J

- -(11)
- -(12)
- -(13)

The currents isqd and isqr are given by

isdr = isa cos + isb sin and

isqr = isb cos isa sin

manifold M , then the G is a symmetry group[4] for the

system if for each g G there exists a regular static
feedback U=k g ( x, u ) such that for all x and u :

Dg( x). f ( x, u ) = f ( g ( x), kg ( x, u ))

This shows that the vector field f is invariant up to a
regular
static
feedback
i.e.
the
equation
dx
= f ( x, u ) remains unchanged after the change of
dt
variables ( x, u ) ( X ,U ) = ( g ( x), kg ( x, u )) that is
dX
= f ( X ,U )
dt
Locally around any point of M, it is always possible to
choose rectifying coordinates (y,z) R s R n s close
to zero such that the action of g G near the identity
reads:
g.(y,z) = ( (y,z),z)
With a smooth map.
The coordinates, which are left unchanged by the
action of G, are called base coordinates and the ycoordinates are called fiber or orbit coordinates [4].
Here z-coordinates are the base and y-coordinates are
the fiber coordinates.
Let us assume that f y and f z are the components of f in
(y, z) coordinates, then:

f=f y

+fz
y
z

Invariance of the vector field implies that for each

g G close to the identity, there exists a feedback
v = k g (y,z,u) such that

3. Symmetries and fiber bundles

Fiber Bundle
A fiber bundle with fiber F is a map f : E B where
E is the total space and B is the base space of the fiber
bundle ,with the condition that every point in the base
space b B has a neighborhood U such that f 1 (U) is
homeomorphic to U F in a special way. Locally E
looks like the product B F. A fiber bundle also
comes with a group action on the fiber. This group
action represents the various ways the fiber can be
viewed as equivalent. The group on a vector bundle is
the group of invertible linear maps, which reflects the
equivalent description of a vector space using different
vector space bases.
Symmetry
A control system
dx
= f ( x, u )
xM
dt
Where M is the smooth manifold and let G is a Lie
transformation group of dimension s acting on the

f z (y,z,u) = f z ((y,z), z, kg(y,z,u))

As (y,z) and (0,z) belong to same orbit there exists
g G close to the identity which sends (y,z) to (0,z)
so we now have to control only base coordinates i.e.
= f z (0,z,v)
which is independent of the fiber variables y.
The feedback v= kg (y, z, u) can be obtained by solving
the over determined system
f z (y,z,u) = f z (0,z,v)
4. Generalization of Direct Field Oriented Control
Of Current Fed Induction Motor
The control of D.C. motor is simple ,as in this the flux
and the torque can be controlled separately .The
similarity of field oriented control of induction motor
[10,11] with the control of D.C. motor is achieved by
selecting a particular frame of reference i.e. a particular
transformation group .
Direct Field Oriented control
For this control we consider the model of the current
fed induction motor with rotor flux reference frame.

In this reference frame all the variables are rotated

through an angle i.e. rotor flux angle or we can say
that they undergo a transformation g-.
Consider the equation given in section 2 ( C ) we can
see that the speed and the rotor flux remain unchanged
under the transformation and so these form the base
coordinates and the rotor flux angle forms the ycoordinates i.e. fiber or orbit coordinates.
The equations of the induction motor can be written as
12
1

=
Ls isq imr - L
J 3
J
1
1
i
mr = isd imr
Tr
Tr
1
isq
Tr imr
Where the first two equations represents
= fz (0, z, v) where z corrosponds to and imr and
v= kg (y, z, u)
So the controller now is to be designed for the above
two equations with isd and isqas controls. This method
needs a flux observer[9].
Section 2 (A) gives the two phase equivalent model of
three phase induction motor in stator reference frame.
In this model all the electrical variables are rotated
through an angle [4] , we can say that they undergo a
transformation g , R and can be expressed as

= +

u s U s
i
s Is
= e j
s s
r r

I
i r r

us
i
s

s
r

i r

Here remains unchanged and the expressions (1) to

(5) remain unchanged by the transformation and so we
can say that the transformation g is the symmetry of
the system
The reduced model given in section 2 (B) is also
invariant under the transformation group G i. e. g
The section 2 (D) gives the model of the induction
motor in rotor reference frame.Here the variables are
rotated through an angle i. e. the angle of rotation or
we can say that they undergo a transformation g-
The equations are

2
d
=
Ls imr [ isqr sin ] - L
dt
3
di
Tr mr + imr = isdr cos isqr sin
dt
d
Tr imr
= isdr sin + isqr cos
dt

- -(11)
- -(12)
- -(13)

. The currents isqd and isqr are given by

isdr = isa cos + isb sin and

isqr = isb cos isa sin
From these we can see that imr is independent of but
is dependent on (i. e. rotor flux angle), with this the
imr (rotor flux ) is controlled by isdr and isqr both ,
whereas depends rotor position as well as on the
rotor flux angle and so it can not be taken as the base
coordinate, with this the equations are coupled and so
the control is difficult . This transformation is not
suitable for the control of induction motor.
5. Conclusion
In this paper it is shown that the field oriented control
of Induction Motor mainly rely on the rotational
invariance of the system, and by controlling the base
coordinates the control of the motor is achieved .we
have also seen that in the rotor reference frame and
imr are unchanged but the equations are coupled and so
the speed and rotor flux (i.e. imr) can not be controlled
separately. This transformation based on rotor position
(i.e. ) is generally used in synchronous machines.
---------------------------------------References
[1] W. Leonhard. Control of Electrical Drives.
Springer 3rd Edition.
[2] Olver P. Equivalence ,Invariants and Symmetry
[3] R.Ortega, P.J.Nicklasson A.Loria, H.Sira-Ramirez.
Passivitybased Control of Euler Lagrange
Systems. Springer 1998.
[4] Ph. Martin, P. Rouchon. Symmetry and field
oriented control of induction motors, CAS
Technical report no 493 ,September 4 ,1998.
[5] F. Fagnani and J.Willems. Representations of
symmetric linear dynamical systems . SIAM J.
Control and Optim., 31:1267-1293, 1993.
[6] V.I. Arnold. Ordinary Differential Equations
Springer Verlag.
[7] P.Martin, P. Rouchon. Two remarks on Induction
Mtotor.CAS 1996.
[8] Ph. Martin ,R. M. Murray ,P. Rouchon. Flat
systems, Equivalence and trajectory generation
Technical Report , April 2003.
[9] G Espinosa and R Ortega. State observer are
unnecessary for induction motor control Systems
and Control letters,23:315-323,1994.
[10] F Blaschke. The Principal of field orientatio as
applied to the new Transvector closed loop control
system for rotating field machines. Siemens
review, page 217,1972.
[11] K. Hasse. Drehzahlregelverfahern fr schnelle
umrichterantriebe
mit
stromrichtergespeisten
Asynchron-Kurzschublufermotoren.
Regelungstechnik, page 60,1972.