Professional Documents
Culture Documents
Induction Motor Modelling PDF
Induction Motor Modelling PDF
Espoo 2000
Mircea Popescu
Teknillinen korkeakoulu
Shk- ja tietoliikennetekniikan osasto
Shkmekaniikan laboratorio
Raportti 63
Popescu M., Induction Motor Modelling for Vector Control Purposes, Helsinki University of
Technology, Laboratory of Electromechanics, Report, Espoo 2000, 144 p.
Keywords: Induction motor, vector control, d-q models, continuos time domain, discrete time
domain, linearization
Abstract
Widely used in many industrial applications, the induction motors represent the starting point when
an electrical drive system has to be designed. In modern control theory, the induction motor is
described by different mathematical models, according to the employed control method. In the
symmetrical three-phase version or in the unsymmetrical two-phase version, this electrical motor
type can be associated with vector control strategy. Through this control method, the induction
motor operation can be analysed in a similar way to a DC motor. The goal of this research is to
summarize the existing models and to develop new models, in order to obtain a unified approach on
modelling of the induction machines for vector control purposes. Starting from vector control
principles, the work suggests the d-q axes unified approach for all types of the induction motors.
However, the space vector analysis is presented as a strong tool in modelling of the symmetrical
induction machines. When an electrical motor is viewed as a mathematical system, with inputs and
outputs, it can be analysed and described in multiple ways, considering different reference frames
and state-space variables. All the mathematical possible models are illustrated in this report. The
suggestions for what model is suitable for what application, are defined as well. As the practical
implementation of the vector control strategies require digital signal processors (DSP), from the
continuos time domain models are derived the discrete time domain models. The discrete models
permit the implementation of the mathematical model of the induction motors, in order to obtain
high efficiency sensorless drives. The stability of these various models is analysed.
Distribution:
Helsinki University of Technology
Laboratory of Electromechanics
P.O. Box 3000
FIN-02015 HUT
Tel: +358-9-451-2384
Fax: +358-9-451-2991
E-mail: electromechanics@hut.fi
Mircea Popescu
ISBN 951-22-5219-8
ISSN 1456-6001
Picaset Oy
Helsinki 2000
Contens
Abstract ..
Preface ....
9
9
11
13
14
17
17
17
21
24
26
27
29
29
31
32
34
41
43
45
47
51
52
54
56
58
60
3.
65
65
65
68
70
71
72
73
89
94
95
97
99
101
104
4.
108
108
110
113
115
117
121
5.
123
123
125
127
129
131
6.
135
135
136
138
References ..
141
3.7.
3.8.
74
76
78
79
81
83
84
87
aij
Bm
H
H
I (a,b,c)s
I (a,b,c)r
I (d,q)s
I (a,b)s
A, B, C , D , E , F
I (d,q)r
I (a,b)r
Is
Ir
i(a,b,c)s
i(a,b,c)r
i(a,b)s
i(a,b)r
i(A,B,C)s
i(X,Y)s
i(d,q)s
i '(d,q)r
i '(d,q)R
is( R, I )
i 'r ( R, I )
is
i 'r
J
j
K p,i
s,r
s,r
K
k
L(a,b,c)(s,r)
Ll(s,r)
Ll(m,a)
lM
ls
l 'r
M (a,b,c),(s,r)
auxiliary
magnetisation inductance in per unit system
total stator inductance in per unit system
total rotor inductance in per unit system
mutual inductance for stator phase, respectively rotor
N s,r
P
p
Rs
rs
rm
ra
rr
r 'r
s
sm
T
Te
TL
t
U (a,b,c)s
per unit stator phase resistance for the three-phase induction machine
stator phase resistance for the symmetrical induction machine
main stator phase resistance for the unsymmetrical single-phase machine
auxiliary stator phase resistance for the unsymmetrical single-phase machine
rotor phase resistance for the symmetrical induction machine
referred rotor phase resistance for the symmetrical induction machine
Laplace operator
critical slip for the induction machine
sampling period
instantaneous electromagnetic torque
load torque
time
instantaneous stator phase voltage for the three-phase induction machine
U (a,b,c)r
U s ( R, I )
(d,q)s
U
(a,b)s
U
Us
Ur
u(d,q)s
u '(d,q)r
us
u 'r
Wc
Xb
x
xm
xls
xl(m,a)
x 'lr
xL
xL(d,q)
xM
xM(d,q)
z
(d,q)
R, I
(d,q)s
(d,q)m
(d,q)r
s(R , I )
'r(R, I )
(d,q)s
stator flux linkage in d-q co-ordinates and flux linkage units per second.
(a,b,c)s
'(d,q)r
referred rotor linkage fluxes in d-q co-ordinates and flux linkage units per
second
models referred rotor linkage fluxes in d-q co-ordinates and flux linkage
units per second
rotor flux linkage in abs or ab co-ordinates
referred rotor flux linkage space vector for three-phase induction machine
stator flux linkage space vector for three-phase induction machine
'(d,q)R
(a,b,c)r
'r
s
m
s,r
r
k
b
n
r
10
machine design requires very fine tolerance levels as stated by Nabae et al. (1980) and Murata et al
(1990). However, in the development of suitable models for control purposes, it is possible to make
certain assumptions that considerably simplify the resulting machine model. Nonetheless, these
models must incorporate the essential elements of both the electromagnetic and the mechanical
systems for both steady state and transient operating conditions (Nowotny and Lipo - 1996).
Additionally, since modern electric machines are invariably fed from switching power conversion
stages, the developed motor models should be valid for arbitrary applied voltage and current
waveforms. This work presents suitable models for use in digital current control of the induction
motors. In addition, the limits of the validity of these models are summarised and, in some cases,
the models are extended to account for some non-idealities of the machine.
Usually, the following assumptions are made (Lorenz et al. 1994):
No magnetic saturation, i.e. machine inductance is not affected by current level.
No saliency effects i.e. machine inductance are not functions of position.
Negligible spatial mmf harmonics i.e. stator windings are arranged to produce sinusoidal mmf
distributions.
The effects of the stator slots may be neglected.
There is no fringing of the magnetic circuit.
The magnetic field intensity is constant and radially directed across the air-gap.
Eddy current and hysteresis effects are negligible.
The modern control theory for an electrical drive system requires the existence of a real-time,
stabile, and precise mathematical model for each component of the system. The analysis and the
design of the numerical command for such systems depend on the hardware and software resources.
If in communication techniques the real-time response of the system is not always compulsory, in
industrial processes the real-time response of the drive systems is essential.
The soft numerical command for the electrical drive systems is far more flexible to implement
than the hardware version. For the latter, lately there is an intense research effort for implementing
ASIC (application specific integrated circuit). The numerical command of the electrical drive
systems is a challenging task mainly due to the DSP (digital signal processing) technology. Now it
is possible to realise linear and non-linear techniques for implementing continuos and discrete
mathematical models of the entire element of an electrical drive system, including the electrical
machine (Xu and Nowotny - 1990, 1992).
For the AC drives there are several solutions for implementing the command and the control of
the system. A quick summary of the existing technologies already out there in the field is given
below:
DC Drives
Initially the DC drives were used for variable speed control because they could easily achieve a
good torque and speed response with high accuracy. Field orientation of the motor is achieved using
a mechanical commutator with brushes. In DC, torque is controlled using the armature current and
field current. The main drawback of this technique is the reduced reliability of the DC motor - the
fact that brushes and commutators wear down and need regular servicing; that DC motors can be
costly to purchase; and that they require encoders for positional feedback.
AC Drives
The evolution of AC variable speed drive technology has been partly driven by the desire to
emulate the performance of the DC drive, such as fast torque response and speed accuracy, while
using robust, cheap to purchase and relatively maintenance-free AC motors (Kelemen and Imecs 1987).
AC Drives, frequency controlled using PWM
With this technique, sometimes known as scalar control, the field orientation of the motor is not
used. Instead, the frequency and the voltage are the main control variables and are applied to the
stator windings. The status of the rotor is ignored, meaning that no speed or position signal is fed
11
back. The drive is therefore regarded as an open-loop drive. This type of drive is suitable for
applications such as pumps and fans, which do not require high levels of accuracy or precision.
AC Drives, flux vector control using PWM
Here, field orientation is achieved by mathematical modelling using microprocessors and
feedback of rotor speed and angular position relative to the stator field by means of an encoder (Vas
- 1990). This results in a drive with greater stability and capable of fast torque response and
accurate speed control. But the drawback is the need for the encoder, which reduces drive system
reliability and adds cost. The controlling variables in a DC drive for torque are armature current and
field current, and armature voltage for torque. AC drives using the PWM principle; however, use
voltage and frequency as the controlling variables and these are controlled by a device called a
modulator. A modulator adds considerable delay in the responsiveness of a motor to changes in
torque and speed. Furthermore, with flux vector AC drives, a tacho-generator or position encoder is
invariably needed to obtain any real degree of accuracy. Such devices are costly and compromise
the simplicity of the AC induction motor.
AC Drives, sensorless flux vector
The flux vector controlled drive with encoder feedback does offer very high levels of
performance across a wide power range and should not be confused with sensorless vector - or open
loop vector - drives, which offer performance only slightly superior to that of a standard inverter
using scalar control (Rajashekara et al. - 1996).
1.2.Algorithm of vector control
The induction motors are very common because they are inexpensive and robust, finding use in
everything from industrial applications such as pumps, fans, and blowers to home appliances.
Traditionally, induction motors have been run at a single speed, which was determined by the
frequency of the main voltage and the number of poles in the motor. Controlling the speed of an
induction motor is far more difficult than controlling the speed of a DC motor since there is no
linear relationship between the motor current and the resulting torque as there is for a DC motor.
The technique called vector control can be used to vary the speed of an induction motor over a
wide range. It was initially developed by Blaschke (1971-1973). In the vector control scheme, a
complex current is synthesised from two quadrature components, one of which is responsible for
the flux level in the motor, and another which controls the torque production in the motor.
Essentially, the control problem is reformulated to resemble the control of a DC motor. Vector
control offers a number of benefits including speed control over a wide range, precise speed
regulation, fast dynamic response, and operation above base speed.
The vector control algorithm is based on two fundamental ideas. The first is the flux and torque
producing currents. An induction motor can be modelled most simply (and controlled most simply)
using two quadrature currents rather than the familiar three phase currents actually applied to the
motor. These two currents called direct (I d) and quadrature (Iq) are responsible for producing flux
and torque respectively in the motor. By definition, the Iq current is in phase with the stator flux,
and Id is at right angles. Of course, the actual voltages applied to the motor and the resulting
currents are in the familiar three-phase system. The move between a stationary reference frame and
a reference frame, which is rotating synchronous with the stator flux, becomes then the problem.
This leads to the second fundamental idea behind vector control.
The second fundamental idea is that of reference frames. The idea of a reference frame is to
transform a quantity that is sinusoidal in one reference frame, to a constant value in a reference
frame, which is rotating at the same frequency. Once a sinusoidal quantity is transformed to a
constant value by careful choice of reference frame, it becomes possible to control that quantity
with traditional proportional integral (PI) controllers.
12
Vector transforms
The Park and Clarke vector transforms are one of the keys to vector control of induction motors.
I) Clarke transform
The forward Clarke (1943) transform does a magnitude invariant translation from a three phase
system into two orthogonal components. If the neutral - ground connection is neglected, the
variables in a three-phase system (A, B, and C) sum is equal to zero, and there is a redundant
information. Therefore, the system can be reduced to two variables, called X and Y. The Clarke
transform is given by:
iAs (t )
iXs (t ) 2 1 cos( ) cos(2 )
(1)
i (t ) = 3 0 sin( ) sin(2 ) iBs (t )
Ys
i (t )
Cs
where:
2
3
Using the relation:
iAs ( t) + i Bs (t ) +i Cs (t ) = 0
and the fact that:
(2)
1
2
4
cos
= cos
=
2
3
3
(3)
iX s ( t )
= 1 i ( t) i ( t )
( As
)
Cs
iB s ( t )
(4)
The Clarke transform can also be understood using a vector diagram as shown in Fig. 1.1. In the
figure, A, B, and C are the axes of a three phase system, each offset 120 from the other. X and Y
are the axes of a two variable system where X is chosen to be coincident with A. To perform the
Clarke transform of a three variable system (iA, iB, iC), iX is equal to iA and iY is the scaled
projection of iB and iC onto the Y axis. The scaling is necessary to preserve the signal magnitudes
through the transform.
Y
is
iB
i Y
A, X
0
iX = i A
iC
13
sin( )
sin( ) In x (t )
cos( ) In y (t )
(5)
where is the angle to rotate the vector through. A reverse vector rotation can be accomplished
simply by changing the sign on the sin ( ) input value. The vector rotation is illustrated by Fig. 1.2.
Some references (Vas -1990, Nowotny and Lipo - 1996) describe the Park transform as a
combination of the Clark and Park transforms presented here. Breaking into a three-variable-to-two
transform (i.e. the Clarke transform) and a vector rotation is done for efficiency of calculation: with
separate Park and Clarke transforms, only two trigonometric calculations are required as opposed to
6 in the traditional Park transform.
Y
is
iY
iQ
iD
X
iX
Flux
s
Stator
14
Flux
command
Torque
command
FOC
Inverter
IM
Voltage
Current
s
1
s
sr
Slip frequency
calculation
r
+
Speed
sensor
Voltage
FOC
Inverter
Current
IM
command
Flux vector
measurement
or estimation
Speed
sensor
15
quantities. In practice either one of them is controlled and we speak of "torque control" or "speed
control". When a variable speed drive operates in torque control mode the speed is determined by
the load. Torque is a function of the actual current and actual flux in the machine. Likewise when
operated in speed control the torque is determined by the load.
Variable speed drives are used in all industries to control precisely the speed of electric motors
driving loads ranging from pumps and fans to complex drives on paper machines rolling mills
cranes and similar drives.
The idea is that motor flux and torque are used as primary control variables which is contrary to
the way in which traditional AC drives control input frequency and voltage, but is in principle
similar to what is done with a DC drive, where it is much more straightforward to achieve. In
contrast, traditional PWM and flux vector drives use output voltage and output frequency as the
primary control variables but these need to be pulse width modulated before being applied to the
motor. This modulator stage adds to the signal processing time and therefore limits the level of
torque and speed response time possible from the PWM drive.
In contrast, by controlling motor torque directly, DTC provides dynamic speed accuracy
equivalent to closed loop AC and DC systems and torque response times that are 10 times faster. It
is also claimed that the DTC does not generate noise like that produced by conventional PWM AC
drives. And the wider spectrum of noise means that amplitudes are lower which helps to control
EMI and RFI emissions. The basic structure of direct torque and stator flux vector control is
presented in Fig. 1.6.
Flux
command
Flux
controller
Inverter
Torque
command
Voltage
Voltage
IM
vector
Torque
controller
selection
Current
Flux vector
measurement
Speed
sensor
or estimation
Fig. 1.6. Basic structure of direct torque and flux vector control
In DTC field orientation is achieved without feedback using advanced motor theory to calculate
the motor torque directly and stator flux without using a modulator or a requirement for a
tachogenerator or position encoder to feed back the speed or position of the motor shaft. Both
parameters are obtained instead from the motor itself. DTC's configuration also relies on two key
developments - the latest high-speed signal processing technology and a highly advanced motor
model precisely simulating the actual motor within the controller. A DSP (digital signal processor)
is used together with ASIC hardware to determine the switching logic of the inverter.
The motor model is programmed with information about the motor, which enables it to
determine parameters including stator resistance, mutual inductance saturation coefficients and
motor inertia. The model also encompasses temperature compensation, which is essential for good
static speed accuracy without encoder.
16
In normal operation, measurements of the two motor phase currents and the drive DC link
voltage, together with information about the switching state of the inverter are fed into the motor
model . The motor model then outputs control signals, which are accurate estimates of the actual
motor torque and actual stator flux. All control signals are transmitted via optical links for high
speed. In this way, the semiconductor switching devices of the inverter are supplied with an
optimum switching pattern for reaching or maintaining an accurate motor torque.
Also, both shaft speed and electrical frequency are calculated within the motor model. There is
no need to feedback any shaft speed or position with tachometers or encoders to meet the demands
of 95% of industrial applications. However, there will always be some special applications where
even greater speed accuracy will be needed and when the use of an encoder improves the accuracy
of speed control in DTC. But even then, the encoder does not need to be as costly or as accurate as
the one used in traditional flux vector drives, as DTC only has to know the error in speed, not the
rotor position.
The drive will have a torque response time typically better than 5ms. This compares with
response for both flux vector PWM drives and DC drives fitted with encoders. The newer
sensorless flux vector drives now being launched by other drives manufacturers have a torque
response measured in hundreds of milliseconds.
DTC also provides exceptional torque control linearity. For the first time with an open loop AC
drive, torque control can be obtained at low frequencies, including zero speed, where the nominal
torque step can be increased in less than 1ms. The dynamic speed accuracy of DTC drives is better
than any open loop AC drives and comparable to DC drives, which use feedback.
DTC brings other special functions, not previously available with AC drives, including
automatic starting in all motor electromagnetic and mechanical states. There is no need for
additional parameter adjustments, such as torque boost or starting mode selection, such as flying
start. DTC control automatically adapts itself to the required condition. In addition, based on exact
and rapid control of the drive intermediate DC link voltage, DTC can withstand sudden load
transients caused by the process, without any overvoltage or overcurrent trip.
17
2.1. Introduction
Until the last decades the three-phase induction machine was mainly used in constant speed
drives due to the control system performance, not to the operating principle of the machine.
Nowadays, this situation is completely changed. With the technical progress in power electronics
and microelectronics, the three-phase induction machine control becomes very flexible and highly
efficient. Since 1983, the year when the digital signal processor (DSP) appeared, the control theory
for this type of machine was permanently improved.
New mathematical models have to be implemented for the three-phase induction machine in
order to analyse its operation both dynamically and in steady-state.
2.2. Voltage and flux linkage equations
The first mathematical model for the dynamic analysis of the induction machine was based on
the two real axis reference frame, developed initially by Park (1929) for the synchronous machine.
Using the symmetric configuration of the induction machine, Kovacs and Racz (1959) have
elaborated the space complex vector theory, and obtained a model for the steady-state analysis of
the machine. Both theories are used for modelling the three-phase induction machine. The
following assumptions are made when a complete equations system is written to describe the
continuous-time linear model of the induction machine (Krause et al. 1995):
Geometrical and electrical machine configuration is symmetrical;
Space harmonics of the stator and rotor magnetic flux are negligible;
Infinitely permeable iron;
Stator and rotor windings are sinusoidally distributed in space and replaced by an equivalent
concentrated winding;
Saliency effects, the slotting effects are neglected;
Magnetic saturation, anisotropy effect, core loss and skin effect are negligible;
Windings resistance and reactance do not vary with the temperature;
Currents and voltages are sinusoidal terms.
End and fringing effects are neglected
All these assumptions do not alter in a serious way the final result for a wide range of induction
machines.
18
ar
as
U I
as as
U
cr
Uar
I
ar
U
I
I cr
cr
bs
bs
U I
br br
bs
U I
cs cs
cs
br
U a s = rs I as +
U bs
Ucs
(6-8)
where U a s , U bs , U cs are the instantaneous stator voltages, I as , I bs , I c s are the instantaneous stator
currents, rs = ras = rbs = rcs is the stator winding resistance and as , bs , cs are the total magnetic
fluxes for the three stator windings.
The flux-current relations are determined after detailing the total flux of a stator winding. For the
other two windings, there are valid similar relations:
as = asas + bsas + csas + aras + bras + cras
(9)
where the flux components are:
asas the magnetic flux produced by stator phase current as in the stator phase winding as
bsas the magnetic flux produced by stator phase current bs in the stator phase winding as
csas the magnetic flux produced by stator phase current cs in the stator phase winding as
aras the magnetic flux produced by rotor phase current ar in the stator phase winding as
bras the magnetic flux produced by rotor phase current br in the stator phase winding as
cras the magnetic flux produced by rotor phase current cr in the stator phase winding as
These components are computed with the expressions:
asas = Las I as
aras = M aras I ar
bsas = M bsas I bs
bras = M bras I br
csas = M csas I cs
cras = M cras I cr
19
Self-inductance Las has two components, one created by the linkage magnetic flux Lmas and the
second created by the leakage magnetic flux Llas :
Las = Lmas + Llas
(10)
The mutual inductance, which is considered to be equal due to the machine symmetry, can also be
split in two components. However, the leakage flux created component in the mutual inductance
can be neglected. It results that:
1
M asbs = M bsas = M ascs = M csas = Lmas
(11)
2
The mutual inductance within the stator and rotor windings varies with the relative space position
between them. The stator flux created by current from rotor phase ar in stator phase as depend on
the angle value :
wk
1
M aras = M asar = 2 w 2 Lmas cos = Lmas cos
(12)
ws k ws
kt
where k t represents the turn ratio multiplied by the winding factor ratio. In a similar way the
relations for the others mutual inductances can be written:
1
2
(131
4
kt
3
14)
Due to the symmetrical configuration of the induction machine, we can deduce the total magnetic
flux for the stator phase winding as expressed as follows:
1
1
1
as = ( Lmas + Llas ) I as Lmas I bs Lmas I cs + Lmas I ar cos +
2
2
kt
(15)
1
2 1
4
Lmas I br cos +
+ Lmas I cr cos +
kt
3 kt
3
For the rotor windings, by using a rotor reference frame, it can be developed a similar equation
system to the stator case:
d
Uar = rr I ar + ar
dt
d
U br = rr I br + bar
(16-18)
dt
d
Ucr = rr I cr + cr
dt
where: U ar , U br ,U cr are the instantaneous rotor voltages, I ar , Ibr , Icr are the instantaneous rotor
currents, rr = rar = rbr = rcr is the rotor winding resistance and ar , br , cr are the total magnetic
fluxes for the three rotor windings.
The total rotor magnetic flux for the winding ar is described by:
ar = arar + brar + crar + asar + bsar + csar =
(19)
= Lar I ar + M brar I br + M crar I cr + M asar I as + M bsar I bs + M csar I cs
In this case the mutual inductance is:
20
1
M brar = M arbr = Lmar
2
1
M crar = M arcr = Lmar
2
M asar = M aras = kt Lmar cos
(20-21)
Due to the symmetrical windings and motor configuration, one can write the following relation:
1
Lmas = kt Lmar = Lm
(25)
kt
Through a similar algorithm as that one for the stator and rotor phase as, respectively ar, it is
possible to obtain another four equations: two for the stator phases bs, and cs, and two for the rotor
phases br and cr. All six final equations can be grouped in a matrix form as follows:
d [ ]
[Us ] = rs [ I s ] + s
dt
d [ ]
[U r ] = rr [ I r ] + r
(26-29)
dt
[ s ] = [ Ls ] [ I s ] + [ M s ] [ Ir ]
[ r ] = [ Lr ] [ I r ] + [ M r ] [ I s ]
where: [Us ] , [ Ur ] , [ I s ] , [ I r ] , [ s ] , [ r ]
Lmas
1 + s
Lmas + Llas 2 Lmas
2
2
2
1
1
1
1
1+ s
(32)
[ Ls ] = Lmas Lmas + Llas Lmas = Lmas
2
2
2
2
1
1
1 Lmas
1
Lmas
Lmas + Llas
1 + s
2
2
2
Lmar + Llar
1
[ Lr ] = Lmar
2
1 Lmar
2
1
Lmar
2
Lmar + Llar
1
Lmar
2
= L
mar
+ Llar
1
Lmar
2
1
Lmar
2
Lmar
1 + r
1
2
1
2
1
2
1+ r
1
2
1+r
1
2
1
(33)
21
cos
1
4
[ M s ] = Lmas cos +
kt
3
cos +
3
[ M r ] = kt Lmar
cos
cos +
3
cos +
3
2
4
cos +
cos +
3
3
cos
cos +
cos +
cos
(34)
cos +
cos +
cos
(35)
cos +
cos
cos +
3
Note:
1) [ M s ] = [ M r ] the mutual stator inductance matrix equals the transpose matrix of the mutual rotor
inductance;
L
L
L
L
2) s = las = as 1; r = lar = ar 1 are the stator, respectively rotor leakage factors.
Lmas Lmas
Lmar Lmar
3) The matrix system determined above represents the flux-current equations set for the three-phase
induction machine in a reference frame attached separately to each armature.
T
2
2
1 xa + xb + xc
3
1
x0 (t ) = ( xa + xb + xc )
3
where:
x (t ) =
=e
2
=e
4
3
(36)
1
3
+ j
2
2
1
3
j
2
2
[ xa (t )
xb (t )
T
2
xc (t ) ] = Re 1 x( t )
22
The phase voltage for the induction machine can be expressed with the help of the space vector
transformation:
2
1 U as + U bs + 2 U cs =
3
2
2 d
2
2
= 1 I as + I bs + I cs rs +
1 as + bs + cs
3
3 dt
Us =
(37)
or in a condensed form:
d s
U s = rs I s +
(38)
dt
where U s , I s , s are space vectors for stator voltage, current and flux. Similarly, we get the rotor
equation:
d r
U r = rr I r +
(39)
dt
where U r , I r , r are space vectors for rotor voltage, current and flux, respectively.
Since the machine is considered magnetically linear, the stator and flux linkage will be determined
as follows, making the notation:
2
2
[ ] = 1
3
3
we get:
2
2
2
[ ] [ s ] = [ ] [ Ls ] [ I s ] + [ ] [ M s ] [ I r ]
3
3
3
where:
1
1
1 + s
2
2
1
1
3
1+ s
= Lmas + s [ ]
[ ] [ Ls ] = 1 Lmas
2
2
2
1
1
1+ s
2
2
2
4
cos
cos +
cos +
3
3
4
2 3
j
cos
cos +
[ ] [ M s ] = 1 2 Lm cos + 3
= 2 Lm [ ] e
3
2
4
cos
cos +
cos +
3
3
The condensed stator flux-current equation results from the last three equations:
3
3
s = Lmas + s I s + Lm I r e j
2
2
where: s , I s , I r are space vectors notations for stator flux linkage, current and rotor current.
Similarly, the rotor flux linkage-current equation is deductible:
2
2
2
[ ] [ r ] = [ ] [ L r ] [ I r ] + [ ] [ M r ] [ I s ]
3
3
3
where:
(40)
(41)
23
[ ] [ Lr ] = 1
[ ] [ M r ] = 1
1 + r
1
2
Lm
2
1
2
= L 3 + []
mar
r
1+ r
1
2
1
2
1
1+ r
1
2
cos
2 Lm cos +
3
cos +
cos +
3
cos
cos +
cos +
3
4 3
j
cos +
= L [ ] e
3 2 m
cos
Finally, we get the condensed form of the rotor flux linkage-current equation:
3
3
(42)
r = Lmar + r I r + Lm I s e j
2
2
where r , I r , I s are the space vector notations for rotor flux linkage, current and stator current
respectively.
For an easier manipulation of the equations we make the notations:
3
Ls = Lmas + s
2
Lr = Lmar + r
2
3 1
3
M = Lmas = kt Lmar
2 kt
2
The general set of voltage and flux linkage equations, written in space vector notations, is:
d s
U s = rs I s +
dt
d r
U r = rr I r +
(43-46)
dt
s = Ls I s + M e j I r
r = Lr I r + M e j I s
Some important conclusions have to be drawn regarding this mode of describing the machine
equations, through space vector notations:
The 12 scalar equations system written in natural reference frame is transformed in a 4 vector
equations system. This form is equivalent to substituting the real induction machine equipped
with three-phase windings on stator and rotor with a fictive machine equipped with single phase
winding on stator and rotor;
An inconvenience of the developed system is that the stator equations system is written in stator
reference frame, and the rotor equations system is written in rotor reference frame, making the
analysis of the machine difficult;
The mutual inductance depends on the relative rotor position.
In Fig. 2.2 is illustrated the new fictive model of the induction machine from the space vector point
of view theory.
24
ar
as
U
s
U
r
I
r
cr
cs
bs
br
x (t) = x(t) e jk ( t)
(47)
where: k(t) is the time variable relative angle between the new reference frame and the stationary
k
reference frame initially considered.; x k(t) represents the space vector for the new reference frame.
The reverse transformation relation is:
k
x (t ) = x (t ) e j k ( t )
(48)
The homopolar component being a scalar variable, is independent from the chosen reference frame.
k
Re
0
Rek
k
x
25
U s = rs I s +
d L s I s + M e j I r
k
s
= U e
dt
k
s
j k
= rs I e
d
dI
= rs I sk e jk + Ls e j k
+ j e j k I sk k
dt
dt
k
s
=e
j k
j k
+ M
U r = rr I r +
) =U
k
r
j (k )
k
r
= rr I e
j ( k ) +
j (k )
k
r
)
+
I kr
dt
d
dI
e jk
+ j e j k I kr k
dt
dt
k
k
d Ls I s + M I r
k
k
k
rs I s +
+ j k Ls I s + M I r
dt
d L r I r + M e j I s
d L s I sk e jk + M e
= e j k
(49)
k d ks
k
rs I s +
+ j k s
dt
d L r I r e j (k ) + M e j (k ) I s
k
)=
)=
dt
dt
k
j (k ) dI r
j (k ) dI ks
k
k d (k )
k d (k )
j ( k )
j ( k )
= rr I r e
+ Lr e
+ j e
Ir
+ j e j (k ) I s
+ M e
=
dt
dt
dt
dt
=e
j ( k )
k
k
d L r I r + M Is
k
k
k
rr I r +
+ j (k ) Lr I r + MI s
dt
= e j (k ) r I kr + d r + j ( ) k
r
k
s
dt
(50)
where k , are the angular velocity of the arbitrary reference frame, respectively the angular
velocity of the induction machine. Finally, if we consider P the number of poles for the induction
machine, the electrical angular velocity of the rotor is (P / 2 ) and the vectorial equations of the
induction machine written in a common arbitrary reference frame which rotates with angular
velocity k are:
k
d s
k
+ j k s
dt
k
d r
U kr = rr I rk +
+ j ( k ) kr
dt
k
k
k
s = Ls I s + M I r
k
U s = rs I s +
(51-54)
r = Lr I r + M I s
The above equations system represent the mathematical model of the induction machine with stator
and rotor equipped with one fictive windings each in a common arbitrary reference frame. It has to
be highlighted that the mutual inductance does not depend on the relative rotor position (Kovacs 1984).
k
r
k k
U I
r r
k
k
k
k
Us
k
k
Is
s
26
1
Ir
kt
U 'r = kt U r
(55-58)
'r = k t r
r 'r = k t2 rr
Now the stator and rotor flux linkage equations can be written as:
3
k 3 1
k
k
k
k
s = L mas + s I s + Lmas I r = s L mas I s + k t M I s + I 'rk
2
2
k
(59)
3 1
1
3
2 kt
kt
2
(60)
3
3
Lmas = k t2 Lmar the main magnetisation inductance
2
2
k
k
= I s + I r the magnetisation current space vector
LM = k t M =
k
IM
k
kM = LM I M
the magnetisation flux
(61)
(62)
(63)
k
s
(64)
k
k
k
U 'r = r 'r I 'r +
+ j k 'lrk + M
(65)
dt
2
Finally we can write a complete equations system, in a common arbitrary reference frame, with
rotor variables referred to the stator, which define the induction machine. The result is a
mathematical model of the three-phase induction machine equipped with two fictive windings,
rotating with angular velocity k. The simplified representation of this system is given in Fig. 2.5.
27
d sk
k
U = rs I +
+ j k s
dt
d 'rk
P
U 'kr = r 'r I 'rk +
+ j ( k ) 'kr
dt
2
k
k
k
s = Ls I s + LM I 'r
k
s
k
s
(66-69)
S
R
Us
k
k
Ur
k
I k k
r
r
Ik
k
s s
(70)
*
where m is the phase number (3), U s ,U 'r are the stator and rotor voltage space vectors, I sk , I 'kr
are the conjugate stator and rotor current space vectors.
Using a detailed relation for the space vectors with rotor variables referred to the stator, we get:
Pi (t ) = Uas I as + U bs I bs + Ucs I cs + U 'ar I 'ar + U 'br I 'br + U 'cr I 'cr
(71)
From the complete equations system of the induction machine, the following expressions can be
obtained for the input power of the machine:
k
3
d 'rk k *
P k k *
k k * d s k *
k k*
k
k*
Pi = Re rs I s I s +
I s + j k s I s + r 'r I 'r I 'r +
I 'r + j k 'r I 'r (72)
2
dt
dt
2
k
3
d s k * d 'rk k *
P k k *
2
k k*
2
(73)
Pi = Re rs I s + r 'r I 'r +
Is +
I 'r + j k s I s + k 'r I 'r
2
dt
dt
2
The first two terms from the power equation represent the Joule effect loss, the following two terms
represent the electromagnetic power due to the time variation of the magnetic energy, and the last
term stands for the mechanical power available at the machine shaft, if the hysterezis loss, eddy
current loss and the stray losses are neglected. The mechanical power will be:
28
Pm =
*
*
3
P
k
k*
k
Re j k Ls I s + LM I 'kr I s + k Lr I 'rk + LM I s I 'kr =
2
2
(74)
3
P
*
3
P
*
= Re j LM I sk I 'rk =
LM Im I sk I 'kr
2
2
4
Using the flux linkage relations, three expressions are deductible for the induction machine
mechanical power:
3P
*
3P
3P
*
k*
k
k
Pm =
Im 'kr I 'rk =
Im s I s =
LM Im I s I 'kr =
4
4
4
(75)
*
3P
LM
*
3
P
L
k
k
k
k
M
=
Im 'r I s =
Im s 'r
4
L 'r
4
where:
= Ls L 'r L2M
which gives four expressions for the instantaneous electromagnetic torque if the mechanical power
is divided by the rotor angular velocity:
}
}
3P
*
3P
3P
*
k*
k
k
Im 'kr I 'rk =
Im s I s =
LM Im I s I 'rk =
4
4
4
*
3 P LM
*
3P L M
k
k
=
Im 'kr I s =
Im s 'kr
4 L 'r
4
Te =
(76)
The torque and the rotor speed a related by the mechanical equation:
2 d
Te = J
+ TL
(77)
P dt
where: J is the inertia of the rotor and in some cases the connected load. The first term on the righthand side of the equation is the inertial torque. The load torque TL is positive for a torque load on
the shaft of the induction machine.
Considering the voltage, flux linkage and mechanical equations, the complete relations system of
the induction machine can be written:
k
ds
k
k
k
U s = rs I s +
+ j k s
dt
d 'rk
P
k
k
+ j ( k ) 'kr
U 'r = r 'r I 'r +
dt
2
k
s = Ls I s + LM I 'rk
k
r
(78-82)
k
s
Im 'kr I s = J
+ TL
4 L 'r
P dt
The above set of equations is valid for the following conditions:
All the equations are written in a common reference frame, which rotates with the arbitrary
angular velocity k .
The rotor variables are referred to the stator;
The stator and rotor variables are described by the space vector notations;
The machine parameters are constant in a common reference frame for both armatures;
The mechanical equation is written in real domain, and is independent of the chosen reference
frame.
k
r
29
k
s
LM i s
dss
LM i'r
s
dk
qs
m s
m
r
s
r
s
k
dm
rk
drs
r
dr
Fig. 2.6. Definitions of the electrical angles between different reference frames
30
c) If the flux linkage per second is used in the induction machine equations, it is per unitised by
dividing by base voltage.
The following values of the machine parameters are used as base variables:
U b = U sn,max =
I b = I sn.max =
Zb =
2 U s(rms),n
2 I s(rms),n
Ub
Ib
-base impedance
3
U b I b = 3U s(rms),n I s(rms)n
2
= sn = 2 f n
Sb =
-base power
sb
2
2
4
sb = sn =
fn
P
P
P
U
b = b
sb
b =
Lb =
b
Ib
Tb =
Sb
3 P U sn.max I sn,max
=
b
sn
4
H =
1 J b2
2 Sb
-base torque
-inertia constant
Us
r I I
d s 1
j k s
= s s b +
+
Ub
Ub Ib
dt U b
Ub
(83-84)
k
1 d s
k
k
+ j k sk
u s = Rs i s +
sn dt
k
1 d 'r
+ j ( k ) 'kr
sn
dt
(85)
In the previous per unit expressions we should note that the multiplying coefficient of the time flux
derivative is necessary, as the time is not referred. The relative time, defined as follows, can be
used:
t
t relative = sb t = sn t = 2
(86)
Tsn
where Tsn represent the stator voltage supply period.
Finally the complete equations set in per unit system for the induction machine is:
31
1 d s
k
u =Ri +
+ j k s
sb dt
k
s
k
s s
k
1 d ' r
u ' = R 'r i ' +
+ j ( k ) 'kr
sb
dt
k
r
k
r
sk = xs i sk + xM i 'kr
(87-91)
3P
k*
k
Im s i s TL
4
sb
dt
k
1 d 'r
j 'kr
sb
dt
(92-96)
sk = xs i sk + xM i 'rk
'rk = x 'r i 'rk + xM i ks
d
1 3P
k*
k
Im s i s TL
dt
2H 4
xM2
and if we note: = xs x r x M2 the
xs xr
x
x
k
i 'rk = s 'rk M s
is =
(97-98)
32
u qs = Rs iqs +
1 d qs
sb dt
u ds = Rs ids +
1 d ds
sb dt
1 d 'qr
'dr
sb
dt
1 d 'dr
+ 'qr
sb
dt
(99-107)
qs = xs iqs + xM i 'qr
ds = xs ids + xM i'dr
'qr = x'r i'qr + xM iqs
'dr = x'r i 'dr + xM ids
d
1
=
dt
2H
3P
ds iqs q sid s ) TL
(
4
If the induction machine is equipped with cage rotor, then the rotor voltage is zero. A complete
block diagram for the induction machine in stationary reference frame using as inputs stator voltage
and currents and load torque, and as output the speed, is illustrated in Fig. 2.7.
i
as
i ds
Rs
i bs
+
sn
1/s
+
Rs
-1/2
3
sn
*
qs
iqs
i cs
2
1/3
ds
1/s
1/s
uas
u
ds
+
+
1/(2H)
+
-
u
bs
1/3
uqs
ucs
2
3-1/2
T
L
Fig. 2.7. Block diagram of the induction machine in stationary reference frame
33
systems. Transformation from the arbitrary frame to the rotor frame is made by substituting
k = , or from the stationary frame with the matrix relations:
sin r xds
xdr cos r
x =
qr sin r cos r xqs
xds cos r sin r xd r
x =
cos r xq r
qs sin r
(108-109)
where r is the electrical angle between the magnetic axis of the stator flux and the magnetic axis
of the rotor flux.
The complete equations set results as follows:
k
1 d s
k
k
k
u s = Rs is +
+ j
s
sb
dt
u ' rk = R 'r i 'rk +
k
s
k
1 d 'r
dt
sb
= x s i s + x M i 'kr
k
(110-114)
d
1 3P
*
Im 'rk i 'rk T L
=
2H 4
dt
+ ds
sb dt
1 d ds
u ds = Rs ids +
qs
sb dt
1 d 'q r
u 'qr = R 'r i 'qr +
dt
sb
1 d 'd r
u 'dr = R 'r i 'dr +
dt
sb
qs = xs iqs + xM i'q r
(115-123)
ds = xs ids + xM i'd r
'qr = x'r i 'qr + xM iqs
'dr = x 'r i 'dr + xM ids
d
1
=
dt
2H
3P
(
4
The block diagram of the induction machine model in rotor reference frame is presented in Fig.
2.8. The input variables are stator currents and load torque, and as output variable is chosen the
angular velocity of the rotor. Also, the rotor voltages are considered zero (case of squirrel-cage
rotor). It has to be observed that two algebraic loops appear in this diagram. This inconvenient leads
to a limited use of this reference frame to the wound rotor induction machine, when the block
diagram is similar to that from Fig. 2.8, where inputs will be rotor voltages and currents.
34
as
3 ---> 2
i bs
ds
+
+ e-j sr 1/3
i cs
x
M
1/x'
+
i qs
i'
dr
-1/2
3
x
M
i' qr
+
1/x'
1/s
dr
1/s
qr
1/s
+
-
-R'
r sn
1/(2H)
-
-R'
r sn
T
L
Fig.2.8. Block diagram of the induction machine in rotor reference frame (cage rotor case)
i
dr
ar
+
+
i br
+
dr
Rr
1/3
Rr
-1/2
3
sn
1/s
qr
iqr
+
sn
i cr
2
1/s
1/s
uar
u
dr
+
u
br
+
+
1/(2H)
-
1/3
ucr
uqr
2
3-1/2
T
L
Fig. 2.9. Block diagram of the induction machine in rotor reference frame (wound rotor case)
2.7.4. Synchronous reference frame equations. Block diagram
The synchronously rotating reference frame, with angular velocity equal to that one of the
power supply system, is particularly convenient when incorporating the dynamic characteristics of
an induction machine into a digital computer program used to study the transient and dynamic
stability of power systems. The synchronously rotating reference frame may also be useful in
variable frequency applications if we may assume that the stator voltages are a sinusoidal balanced
set. It was systematically developed by Kovacs (1984) and Krause et al (1995), or Lorenz et al.
(1994).
35
When a two-axis co-ordinates reference frame is employed, it has to be fixed to different variables
of the machine. The main configurations are the synchronously reference frame fixed to: stator flux,
air-gap flux and rotor flux.
I) Stator flux fixed synchronous reference frame.
In order to link the d-axis of the synchronous reference frame to the stator flux space vector, the qcomponent of this flux vector is defined equal to zero:
ss = dss
(124)
qss = 0
The following equations set is obtainable:
u s qs = Rs i s qs + s s ds
u s ds = Rs i s ds +
1 d s ds
dt
sb
1 d 's qr
+ ( s ) 's dr
dt
sb
1 d 's dr
( s ) 's qr
dt
sb
(125-133)
0 = xs i s qs + xM i 's qr
s ds = xs i s ds + xM i 's dr
's qr = x 'r i 's qr + xM i s qs
's dr = x'r i 's dr + xM i sd s
d
1 3P s s
=
ds i qs TL
dt
2H 4
The block diagram from Fig. 2.10 presents the induction machine model in synchronous reference
frame linked to the stator flux, with inputs stator voltages and currents, and load torque. The outputs
are synchronous and rotor angular velocity.
i
i
as
i ds
+
3 ----> 2
bs
+
i cs
1/3 -j ss
e
s
2
ds
Rs
sn
1/s
+
iqs
-1/2
N
R
u
as
ds
3 ----> 2
bs
+
u cs
/
+
1/(2H)
-
1/3 -j ss
s
e
2
D
+
1/s
+
u
u qs
+
3-1/2
TL
Fig. 2.10. Block diagram of the induction machine in synchronous reference frame
(linked to stator flux)
36
+ s r ds
dt
sb
1 d r ds
u r ds = Rs i r ds +
s r qs
sb
dt
u 'r qr = 0 = R 'r i 'r qr + ( s ) 'r dr
u 'r dr = 0 = R 'r i 'r dr +
1 d 'r dr
dt
sb
(135-143)
r qs = xs i r qs + xM i 'r qr
r ds = xs i r ds + xM i 'r dr
0 = x 'r i 'r qr + xM i r qs
1
3P r
' dr i 'r qr TL =
4
2H
3P L M r r
' dr i qs TL
4 L 'r
The block diagram from Fig. 2.11 presents the induction machine model in synchronous
reference frame linked to the rotor flux, with inputs stator currents, and load torque. The outputs are
synchronous and rotor angular velocity. The rotor voltages are considered equal to zero (case of the
cage rotor).
i
i
as
3 ---> 2
bs
+
i cs
i ds
+
1/3
rs
e -j s
2
dr
+
x M R'r /x'r
sn
1/s
*
-
s
+
iqs
-1/2
N
R' /x'
r r
x
R' /x'
M r r
D
/
xM
1/s
x'
r
+
1/(2H)
+
-
TL
Fig. 2.11. Block diagram of the induction machine in synchronous reference frame
(linked to rotor flux)
37
+ s m ds
sb
dt
u m ds = Rs i m ds +
1 d m ds
s
dt
sb
m
qs
1 d 'm qr
+ ( s ) 'm dr
dt
sb
1 d 'm dr
( s ) 'm qr
dt
sb
0 = xM ( i m qs + i 'm qr )
dm
(145-155)
= x M ( i m ds + i 'm dr )
ds
= xs i m ds + x M i 'm dr =
qs
= xs i m qs + x M i 'm qr = xls i m qs
+ xls i m ds
dm
3P
dm
+ x'lr i 'm dr
dm
i m qs T L
Fig. 2.12 illustrates the induction machine model in synchronous reference frame linked to the airgap flux, with inputs stator voltages and currents, synchronous angular velocity and load torque.
The output is rotor angular velocity.
i
as
bs
+
Rs
i
cs
+
x ls
sn
1/s
dm
qs
*
-1/2
1/s
ls
*
+
as
+
3 ----> 2
bs
+
+
u
ds
m
1/3-j
e
s
2
3 ----> 2
u
u
cs
+
1/(2H)
-
m
1/3
-j
e
s
2
+
ds
qs
-1/2
T
L
Fig. 2.12. Block diagram of the induction machine in synchronous reference frame
(linked to air-gap flux)
38
+ s sc ds
dt
sb
1 d scds
s sc qs
dt
sb
u sc ds = Rs i scds +
1 d 'sc qr
+ ( s ) 'scdr
dt
sb
1 d 'sc dr
( s ) 'sc qr
dt
sb
(157-165)
3 P sc
qs i sc sd TL
The block diagram from Fig. 2.13, uses as inputs the stator voltage space vector, the d-axis stator
current component, the load torque, and the synchronous angular velocity. As output is considered
the rotor angular velocity (speed).
i as
i ds
i bs
+
i cs
3 ---> 2
1/3
-j sc
e
s
2
Rs
1/s
+
iqs
3-1/2
+
ds
sn
1/s
u as
+
+
3 ----> 2
u bs
+
1/3e
-j sc
s
u qs
+
1/(2H)
-
u cs
2
u ds
3 -1/2
sn
qs
1/s
TL
Fig. 2.13. Block diagram of the induction machine in synchronous reference frame
(linked to stator current space vector)
39
+ s rc ds
dt
sb
u rc ds = Rs i rc ds +
u 'rc qr = 0 =
1 d rc ds
s rcqs
dt
sb
1 d 'rc qr
+ ( s ) 'rc dr
dt
sb
1 d 'rc dr
( s ) 'rc qr
dt
sb
(167-175)
rc ds = xs i rc ds + xM i 'rc dr
rc qs = xs i rc qs
'rcqr = xM i rc qs
'rcd r = x 'r i 'rcdr + xM i rc ds
d
1 3P rc
1 3P
xM i rcqs i 'rc dr TL
' qr i'rc dr TL =
dt
2H 4
2H 4
In Fig. 2.14, the block diagram that describes the induction machine model has as inputs the d-axis
stator voltage component and current space vector, the load torque, and the synchronous angular
velocity. As output is considered the rotor angular velocity (speed).
i
as
i bs
+
i
ds
cs
3 ---> 2
1/3
rc
-j
e
s
2
+
Rs
+
+
i
qs
x
+
-
1/s
s
as
ds
ubs
+
cs
3 ----> 2
1/3
rc
-j
e
s
2
+
+
3P/4
ds
1/s
qs
-1/2
3
sn
x
u
3
1/(2H)
-
qs
-1/2
TL
Fig. 2.14. Block diagram of the induction machine in synchronous reference frame
(linked to rotor current space vector)
40
1 d m cqs
+ s m cds
dt
sb
u mc ds = Rs i mc ds +
1 d m cds
s mc qs
dt
sb
1 d ' mc q r
+ ( s ) 'mcd r
sb
dt
1 d ' mc dr
( s ) 'mc qr
dt
sb
i mc qm = 0 = i mc qs + i ' m cq r
mc ds = xs i m cds + x M i 'mc dr
mc qs = xs i m cqs + x Mi 'mc qr = xls i mc qs
'mcqr = xM i m cqs + x 'r i 'mc qr = x'lr i'mc q r
'mcdr = x 'r i 'mc dr + x M i mc ds
d
1
=
dt
2H
3P
xM i mc dm imc qs TL
4
(177-185)
A block diagram for the mathematical model of the induction machine in this reference frame is
presented in Fig. 2.15. As inputs are chosen the stator current space vector, the d-axis component of
the stator voltage space vector, the synchronous angular velocity and the load torque. The output is
the machine (rotor) angular velocity. The similarity to the synchronous reference frame linked to
the air-gap flux space vector can be observed, as the magnetising current is directly proportional
with the air-gap flux, through the magnetising inductance.
i
i
as
bs
i cs
+
3 --> 2
j m
+ e
s 1/3
2
ds
Rs
+
+
1/s
dm
+
iqs
sn
x ls
x ls
-1/2
qs
1/s
3P/4
*
+
u as
u
+
bs
+
u cs
3 --> 2
m
1/3
j
e
s
2
ds
u
1/(2H)
ds
-
u qs
+
-1/2
Fig. 2.15. Block diagram of the induction machine in synchronous reference frame
(linked to magnetising current space vector)
41
(186-191)
s = ds + j qs
'r = 'dr + j 'qr
m = dm + j qm
[ x ] = [is
i'r im s 'r m ]
(192)
The d-q axes are orthogonal and fixed to the stator, d axis coincides to the magnetic axis of the as
winding.
As there are four voltage equations, it is necessary to consider two of the space vectors as statevariables in order to obtain a solution for the equations system. Let the selected pair of state-space
variables be denoted as x 1 , x 2 . The set of six state-space variables will be expressed in terms of the
two selected state-space variables:
[ x ] = [i s
i 'r
im
'r
m ]
a11
a
21
a31
=
a41
a51
a61
a12
a 22
a32
[ x1
a 42
a52
a62
x2 ]
(193)
Only fourteen out of the fifteen state-space possibilities represent valid mathematical models for
the three-phase induction machine (the pair of state-space variables that comprises airgap flux
linkage per second space vector and the magnetising current space vector cannot be selected as
these vectors have the same direction). There are three types of models:
1. current state-space variables models;
2. flux linkages state-space variables models;
3. mixed currents-flux linkages state-space variable models.
In conjunction with the mechanical equation:
P
(194)
p r =
(Te TL )
2J
we obtain a complete version of the three-phase induction machine model, viewed as the key for a
motion control system.
The starting point for the state-variable models is given by the voltage equations system written in
stationary reference frame:
u = L px + R x
(195)
where: x is the selected set of state-variables and represents also the output of the model, u is the
input vector (stator voltages), L is the coefficients matrix (it can be formed by reactance values, or
non-dimensional elements) for multiplying the time derivative of the state-variables, R is the
coefficients matrix (it can be formed by resistances and reactances values or non-dimensional
elements) for multiplying the state-variables and p stands for the differential operator (d/dt).
42
(196)
B = L1
All the following mathematical models permit a discretisation for the implementation of
controllers in the drive systems with three-phase induction machines. Classically, the current statespace and the flux state-space models are the chosen options for the manufacturers of drive systems.
The original vector orientation scheme was based on the alignment of the synchronous reference
frame to the rotor flux linkage space vector. Then the vector control strategy was extended by
considering as well the stator and air-gap flux space vectors, as alignment of the reference frame.
However, for a complete analysis of the vector control strategies, each of the six space vectors that
can be selected as state-space variables represents a possible basis for a reference frame. If the
magnetising current space vector selection as alignment gives similar results to the air-gap flux
space vector case, the stator and rotor current space vectors are still to be further analysed as new
options for vector control.
An easy to follow steps algorithm for implementing vector oriented control systems is obtained
as follows:
1. A complete mathematical model of the three-phase induction machine is developed in stationary
reference frame, according to the chosen set of state-space variables;
2. The rotor based variables are completely expressed in the new state-variable system;
3. The rotor angular velocity r term is substituted with ( r - k) where k is the angular
velocity of the synchronous frame;
4. The synchronous reference frame is selected linked to one of the space vectors, which means
that the q-axis component of the reference space vector is null;
5. The torque equation is computed according to the selected flux or current space vector in the
synchronous reference frame.
The block diagram of modelling an induction motor for vector control purpose is detailed in Fig.
2.16.
synchronous frame
x
u
p x=Ax+Bu
[a ij]
-j
T (closed loop)
e
Te
(open loop)
Tl
r
+
+
P
2J
stationary frame
Fig. 2.16. Block diagram form of the induction machine mathematical model
Fig. 2.17 shows the general two-axis equivalent circuit for the induction machine, in an arbitrary
reference frame. This equivalent circuit represents the starting point for determining the matrix
equations, which will be derived further.
43
k ds
-
r
s
L
ls
L'
lr
( k) 'dr
+
r'
+
i'
i qs
L
M
qs
qr
u'
qr
k qs
+
L
ls
L'
lr
( k) 'qr
+
-
r'
+
i
ds
LM
i'
dr
u'
dr
ds
-
I) As complex state variables, the currents space vectors x = iqs , ids , i 'qr , i 'dr are usually
assumed. The stator current space vector is considered generally as the right choice, because it
corresponds to directly measurable quantities. This model is readily available from voltages and
flux linkages per second equations, and it can be expressed in a matrix form as follows:
rs x'r
2
iqs xm r
p ids D b
=
b i 'qr xmrs
D
i 'dr
xx
s m r
D b
xm2 r
D b
xmr'r
D
rs x' r
D
xm x 'r r
D b
xs xm r
D b
xmrs
D
xs r'r
D
xs x 'r r
D b
xm x'r r
x 'r
D b
D
xmr'r iqs
0
D
ids +
xs x' r r i 'qr xm
D b i 'dr D
xs r'r
0
xm
D
x 'r
D
xs
D
xm
D
xm uqs
D uds
u'
0 qr
u 'dr
xs
(197)
where: D = xs x 'r x
If the stator, rotor and total leakage factor definitions are used:
2
m
xs = xls + xm = (1 + s ) xm
x 'r = x 'lr + xm = (1 + r ) xm
=1
1
(1 + s )(1 + r )
we obtain:
D = xm2 ( s r + s + r ) = xm2
44
(i i ' i i ' )
(198)
4 b qs dr ds qr
II) The stator and the magnetising space vector currents as state-space variables
T
x = iqs , ids , qm
i , idm represent another mathematical d-q axis model based on currents space
vectors. By selecting the magnetising current space vector among the state-space variables, it is
possible to include the saturation effect in modelling the induction motor. Also, the stator current
space vector is a measurable quantity, and determines a precise and accurate option for
implementing controllers. The state matrix elements are all non-zero, which implies a
computational effort similar to the previously analysed model.
xm x'lr r
xmr'r
xm x 'r r
rs x'r r'r xm
b
b
i
r
r
iqs
rs x'r r 'r xm xm x'r
xmr 'r qs
xm x'lr
i
b
b
p ds 1
ids +
=
i
b iqm D
xls x'lr r
xls r'r
xls x ' r r qm
r 'r xls rs x lr'
b
b idm
idm
(199)
x 'r
1 0
+
D x 'lr
xm
x' r
0
x 'lr
0
xls
0
0 uqs
xm uds
0 u 'qr
xls u 'dr
The instantaneous electromagnetic torque may be expressed in terms of the state-space variables:
Te =
3 P xm
( iq sid m id siq m
4 b
(200)
III) The third possible combination of current space vectors as state-space variables is obtained
T
by selecting magnetising and rotor currents x = i 'qr , i 'dr ,i qm , idm . When compared to the other
currents space vector models, the similar computational burden for obtaining the outputs of the
system is obvious. The main difference between them is the presence of the global parameters (selfreactances) in the stator and rotor current state-space variables model, while in the other two models
an accurate determination of the leakage reactances or leakage factors is necessary. A complete
description of the state-variables in matrix notation is given below:
45
rs xm r 'r xs
r
i 'qr
xs x'lr
i '
p dr 1
b
=
b iqm D
rs x 'lr r 'r xls
idm
r
xls x 'lr
b
x m
1 0
+
D x 'lr
0 xs
xm 0
0
xls
x 'lr
xs x'lr
r
b
rs xm
rs xm r 'r xs
xls x 'lr
r
b
xs xm
r
b
rs x 'lr
xls xm
r
b
i '
rs xm qr
i 'dr +
i
xls xm r qm
b idm
rs x 'lr
xs xm
r
b
(201)
u qs
u
ds
u ' qr
xls u ' dr
0
xs
0
(202)
One option is to select the stator and rotor flux linkages space vectors x = qs , ds , ' qr , 'dr for
describing the mathematical model of the machine. For this system the matrix equations are as
follows:
rs x 'r
D
qs
0
p ds
=
b 'qr r 'r xm
D
'dr
0
rs xm
D
rs x 'r
D
r 'r xs
D
rs xm qs u qs
D ds u ds
+
r 'qr u 'qr
b ' u '
dr dr
r 'r xs
D
0
(203)
r
b
The electromagnetic torque is determined with the equivalent relation:
3P
x
(204)
Te =
m ( qs 'dr ds 'qr )
4 b D
By comparison with the current models, it can be observed that the computational burden is
essentially lower. Due to this important feature, this model is the most suitable for discretisation in
motion control strategies.
II) An alternative to model the three-phase induction machine with flux space vectors as statevariable system, is the selection of air-gap flux space vector among the set of independent variables.
This choice presents the advantage of an easier saturation effect modelling, but the disadvantage of
an increased computational burden. The air-gap flux is a measurable quantity, and this advantage
imposes it in many practical solutions for vector control schemes.
r 'r xm
D
46
A first approach is given by the stator and air-gap flux space vectors selected as state-space
T
variables x = qs , ds ,qm , dm .
r
rs
s
0
0
xls
xls
rs
rs
qs
0
qs
x
x
ls
ls
p ds
ds
=
+
r
x
'
x
r
'
x
x
x
'
r
x
x
'
r
'
x
x
x
x
'
x
b qm
s lr m
r m
m lr
r
s m lr
r s
ls m
r
lr s
qm
1
+
x
D
D
D
x
D
D
D
x
x
dm
ls
b
ls
b
ls m
dm
r x ' x r' x
x x x' x
rs x'lr xm r 'r xs
xm x'lr r
s lr m + r m ls m r 1+ lr s
D
x
D
D
D
x
x
x
D
D
b
ls
b
ls
m
ls
1
0
x 'lr xm
0
xls xm
D
0
x 'lr xm
D
0
0 uqs
uds
0
u'
qr
xls xm u 'dr
D
(205)
The electromagnetic torque expression depends on stator leakage factor:
3P
1
Te =
qs dm ds qm
4 xls b
(206)
III) The third option of selecting flux space vectors as state-space variables is similar to the
previous model. It comprises the air-gap (magnetising) flux and rotor flux space vectors
T
x = qm , dm , ' qr , 'dr in the set of independent variables. The same computational effort is
required, and the electromagnetic torque is determined in a suitable form for vector control if the
rotor leakage factor is known:
rs xm r'r xm xls
xm xls r
rs x 'r r 'r xm xls
D + x ' D
D
x 'lr D
D b
lr
D
x 'lr D
D b
D
x 'lr D dm
p dm
+
=
'qr
r 'r
r 'r
b 'qr
r
0
x 'lr
x 'lr
b
'dr
'dr
(207)
r 'r
r'
0
r
r
x 'lr
x 'lr
b
xm x 'lr
D
+ 0
0
0
0
xm x'lr
D
0
0
xm xls
D
0
1
0
0
uqs
xm xls uds
D u 'qr
0 u '
dr
1
The resulting relation for computing the instantaneous electromagnetic torque is as follows:
47
Te =
3P
1
( qm 'dr dm 'qr )
4 x 'lr b
(208)
2.8.3. Models with mixed currents flux space vectors state-space variables
I) For obtaining an acceptable computational effort as well as measurable output quantities, the
general accepted solutions are the mixed currents-flux state-space variable models. If the stator
variables are chosen for modelling the three-phase induction machine system, then a mixed flux
T
qs 0
p ds r 'r
=
b iqs D
x 'r r
ids
D b
1
0
+ x 'r
D
0
1
0
x 'r
D
rs
0
0
0
x 'r r
D b
rs x 'r + r 'r xs
D
r 'r
D
r
b
qs
ds
r
+
b
iqs
r x ' + r 'r xs ids
s r
0
rs
(209)
0
0 uqs
uds
1 0
u '
qr
u '
0 1 dr
0
0
The instantaneous electromagnetic torque relation depends only on the magnitude of the output
vector components.
3P 1
Te =
( dsiqs qsids )
(210)
4 b
II)
Another
important
T
mixed
flux
linkages-currents
state-space
variables
model
x = 'qr , ' dr, i' qr, i ' dr is that expressed in rotor quantities. This model represents the optimum
solution for rotor flux oriented control strategy in a drive system with an induction machine.
However, as it is impossible to measure the rotor currents if the machine is equipped with cage
rotor, there are limitations in using this model for vector control strategies. The state-space
variables system is detailed below:
r
r 'r
0
0
b
0
1 0
0
'
'
qr
qr
r
0
0 1 uqs
0
0
r 'r
0
'
u
'dr x
p dr
b
m
=
+
ds
0
0
0
i 'qr
u '
b i 'qr rs
xs r
r x ' + r' x
D
0
s r r s
qr
D b
D
x
u '
i 'dr D
i 'dr 0
m 0 0 dr
xs r
D
rs
r x ' + r 'r xs
0
s r
D
D
D b
(211)
48
The electromagnetic torque expression depends only on the magnitude of the output vector
components:
Te =
3P 1
(212)
III) The most used alternative model for rotor flux oriented control strategies is that which
T
comprises the stator current and rotor flux linkage x = 'qr , ' dr , iqs , ids as state-space variables. It
contains the advantages of measurable output quantities (stator currents) and acceptable
computational burden. The matrix equation and the electromagnetic torque relation are presented
below:
rs x 'r2 + r 'r xm2
0
x
'
D
r
iqs
rs x'r2 + r 'r xm2
0
p ids
x 'r D
=
b 'qr
r 'r xm
0
x 'r
'dr
r 'r xm
0
x 'r
x
x 'r
0 m
0
D
uqs
D
x 'r
x
u
0
m ds
+ 0
D
D u 'qr
0
0
1
0
u 'dr
0
0
0
1
and
3P xm
Te =
r 'r xm
x 'r D
xm r
D b
r'
r
x 'r
r
b
xm r
D b
r 'r xm iqs
x 'r D ids
+
'qr
r
b 'dr
r 'r
x 'r
(213)
(214)
IV) A theoretical mathematical model is that with mixed rotor currents space vector and stator
T
flux linkages space vector as state-space variables x = qs , ds , i 'qr , i ' dr . It should be noted that
the advantage of this model is that only the stator winding parameter is necessary and as such the
influence of rotor parameters is minimised. This model can be used for an unconventional stator
flux oriented control with rotor current components producing the torque and the flux. The
expressions for implementing this model are:
49
xs D
i 'qr
r
i '
p dr
b
=
b qs
rs xm
ds
xs
r
b
rs xm
xs D
r x + r 'r x
xs D
x
m
D
r
s
xs
rs xm
xs
2
s m
2
s
xm r
D b
xm
rs xm i 'qr D
xs D i 'dr
+ 0
qs
0 1
ds
0
rs
xs
xs
D
xm
D
0
0
u qs
xs u ds (215)
D u 'qr
0 u '
dr
0
and
Te =
3P xm
( qs i 'dr ds i 'qr )
4 xs b
(216)
T
V) The mixed stator current space vector and air-gap flux space vector x = qm , dm , iqs , ids as
state-space variables belongs to one of the most complex types of models. It preserves information
regarding both stator and rotor parameters. This mathematical model is the suitable choice for the
air-gap flux orientation control strategy. The greatest advantage of this model is that by using Hall
sensors or tapped stator windings, all the output vector components are measurable. Due to its
versatility, this model is widely used in controllers implementation, especially for medium speed
applications. It permits also the simulation or modelling of the saturation effect. By comparison to
the previous mixed models, the state-space matrix contains only non-zero elements, which leads to
greater computational effort.
xm x'lr ( r / b )
r 'r
x 'r ( r / b ) iqs
iqs
rs x 'r r 'r xm
rs x 'r r 'r xm
x 'r ( r / b )
r 'r
p ids 1 xm x 'lr ( r / b )
ids +
=
b qm D xm (r 'r xm rs x 'lr ) xls x mx 'lr ( r / b )
xls r ' r
xls x 'r ( r / b ) qm
xls r 'r
dm
xls xm x 'lr ( r / b ) xm (r 'r xm rs x 'lr ) xls x 'r ( r / b )
dm
xm
0
0 uqs
x 'r
0
x 'r
xm uds
0
1
+
xls xm
0
0 u 'qr
D x 'lr xm
x 'lr xm
xls xm u 'dr
0
0
(217)
The electromagnetic torque is computed as:
3P 1
Te =
( iqs dm ids qm )
4 b
(218)
VI) Another theoretical model, similar in form to the precedent one, is the mixed rotor current
T
space vector and air-gap flux space vector as state-space variables x = qm , dm , i 'qr , i ' dr Its main
shortcoming is the presence of the unmeasurable rotor currents among the state-space variables.
The matrix equation of the system is given below:
50
i 'qr
rs xm r 'r x s
i'
p dr 1 x s x 'lr ( r / b )
=
b qm D rs x m x 'lr r 'r xm x ls
dm
x m x ls x 'lr ( r / b )
xm
1 0
+
D x m x 'lr
0
xm
0
x m x 'lr
x s x 'lr ( r / b )
rs x m r 'r x s
x m x ls x 'lr ( r / b )
rs x m x 'lr r 'r xm xls
rs
x s ( r / b )
rs x 'lr
x m x ls ( r / b )
xs ( r / b ) i 'qr
i'
rs
dr +
xm xls ( r / b ) qm
rs x 'lr
dm
0 u qs
x s u ds
0 u ' qr
x m x ls u ' dr
xs
0
x m xls
0
(219)
The electromagnetic torque relation becomes:
3P 1
Te =
( idr qm iqr dm )
4 b
(220)
VII) If the magnetising current space vector is selected as state-space variable together with one
T
of the flux linkages space vectors (i.e. stator flux linkages) x = iqm , idm , qs , ds , the computation
of the state matrix elements gives several null results. This choice for a set of state-space variables
presents only theoretical importance, as the output vector components are unmeasurable. Also the
computational demand of the model does not make it a practical option for vector control
implementation.
rs x 'lr
iqm
p idm
=
b qs
ds
x 'lr
D
+ 0
1
0
xls
D
x 'lr
D
0
xm r 'r xs xls
xls D
r
b
rs xm
xls
r
b
rs xm x'lr r'rx s xls
xls D
0
rs xm
xls
D b
rs
xls
0
x 'lr r
D b
+
qs
0
ds
r
s
xls
(221)
0
uqs
xls uds
D u 'qr
0 u '
dr
0
3 P xm
( idm qs iqm ds )
4 b x ls
(222)
VIII) One other model analysed for theoretical reasons, is the mixed current-flux linkages space
vector model which realises a the connection between the magnetising current space-vector and the
T
rotor flux linkage space vector, selected as state-space variables x = iqm , idm , 'qr , 'dr As the
output vector components cannot be measured, this model, like the previous one, is prohibitive for
implementation in vector control strategies. However, from computational effort, the state matrix
contains the same number of zero elements (four) and the instantaneous electromagnetic torque is
determined in a suitable form comparable to the classical vector control implementations.
51
x 'lr D
iqm
0
p idm
=
b 'qr
r 'r xm
x 'lr
'dr
xls
x 'lr
0
0
D
uqs
D
x 'lr
xls uds
0
+ 0
D
D u 'qr
0
1
0 u '
0
dr
0
0
0
1
0
r 'r xls xm rs x ' r x 'lr
x 'lr D
xls r
D b
r 'r
x 'lr
r 'r xm
x 'lr
r
b
xls r
D b
rs x 'lr r 'r
x 'lr D
r
b
r'
r
x 'lr
xls iqm
idm +
'qr
'dr
(223)
(224)
52
Induction
Controller
Transformation
motor
current space vectors x = qs , ds , iqs , ids is selected. The mathematical model is given in the
chapter dedicated to d-q models of the three-phase induction machine.
The rotor based variables expressed in state-space variables terms are:
1 s
i 'sqr =
qs xsiqss
xm
1 s
ds xsidss
xm
1
x 'r qss D iqss
'sqr =
xm
i 'sdr =
(225-228)
1
x 'r dss D idss
xm
To obtain the machine equations in the synchronuous stator flux reference frame, one has to
eliminate the rotor flux from the rotor voltage equations and then force the q component of the
stator flux to be zero. The stator voltages equations remain unchanged. If a cage rotor is considered,
the resultant equations are as follows:
p s s
p s
x 'r dss D idss ) x 'r
qs = 0
(
r 'r xs + D
iqs
b
b
b
(229-230)
p s s
p
s
s
D iqss = r 'r + x 'r
x 'r qss
r 'r xs + D
ids
ds
b
b
b
b
where the definition used for the slip speed is: s = e r .
If the special reference frame is fixed to the stator flux linkage vector, the q-component of this
flux vector is defined equal to zero:
qss = 0
(231)
ss = dss
From stator flux linkage equations, the q-current components are given by:
'sdr =
53
iqss = iqss
(232)
xs s
iqs
xm
The electromagnetic torque relation and slip speed can be derived in stator field orientation control
as:
3P 1 s s
Te =
ds iqs
4 b
(233-234)
( b r 'r xs + D p ) i qss
s =
x 'r dss D idss
The second dynamic equation of the machine, shows that there is a coupling between the stator
s
current components. Consequently, any change in the torque producing component iqs without
iqrs =
changing idss accordingly will cause a transient in the stator flux. A decoupler is necessary to
overcome this disadvantage. Therefore the command current of the d-axis component of the stator
current can be calculated as follows:
K
s
idss = Kp + i dss + idq
p
(235-236)
s
D
i
s
qs
s
=
idq
b r 'r xs + D p
where Kp and Ki are proportional, respectively integral coefficients of the flux controller. This
controller can be PI type or soft computing technique type (fuzzy, neural-network) as demonstrated
by Xu and Nowotny (1988).
Steady-state perfomance of a stator flux oriented system
Letting the derivative operator p = 0, one can obtain the steady-state voltage equations of the
induction machine. After several manipulations of the system equations, the d current components
are given by:
b r 'r xs s
x 'r xs x 'r xm2 s s
s
ids =
2 iqs +
iqs
s
xm
xm
xm
b r 'r
(237)
2
'
x
x
x
m
s
idrs = s r
iqss
'
x
r
m
b r
it yields the slip speed equation:
s2 iqss D 2 s b r 'r xm2 ss + ( b r 'r ) 2 xs2iqss = 0
(238)
The solutions of the above equations have to be real, for a given stator flux linkage. This means that
the determinant of the second order equation satisfy the condition:
2
2
2
(239)
= ( xm2 b r 'r s s ) 4D 2 ( b r 'r xs ) ( iqss ) 0
The maximum values for the q component of the stator current, the slip speed and the
electromagnetic torque (pull-out torque) are:
54
( )
xm2 ss
2 Dxs
( s )max
r ' r xs
D
iqss
max
(240-242)
( )
( Te )max
2
s
3 P x m s
=
4 2 b Dxs
If the angular slip velocity is larger than ( s )max static instability will arise. This maximum
value depends only on machine parameters and not on stator flux level. Neveretheless, the pull-out
torque depends on the square of the flux magnitude. The pull-out torque determines the limit for the
system stability operation zone, when the stator flux oriented control is employed. It is possible to
limit the torque command not to exceed the pull out torque for a given stator flux. Some important
conclusions cand be drawn for this vector control strategy:
The electromagnetic torque and stator flux producing components of the stator current are not
decoupled;
A parameter dependent decoupling network has to be included;
Torque and flux control does not require speed feed back;
For operation at low speed it is difficult to estimate the stator flux;
It is a good alternative for medium performance drive.
2.9.2. Rotor flux field orientation (RFO)
For this vector control strategy, the set of state-space variables formed by rotor flux linkage and
T
stator current space vectors x = 'qr , ' dr , iqs , ids is selected. If the induction motor is equipped
with wound rotor, as the rotor currents are also measurable, the rotor current space vector can be
also selected as state-space variables together with the rotor flux linkage space vector
T
x = 'qr , ' dr, i' qr, i ' dr . The mathematical model is given in the chapter dedicated to d-q models
of the three-phase induction machine. The first case will be analysed, for the second one the
algoritm is similar.
The stator flux linkage and rotor current space vectors components expressed as functions in
terms of state-space variables are:
1
qsr =
xm 'rqr + D iqsr
x 'r
1
xm 'rdr + D idsr
dsr =
x 'r
(243-246)
1
r
r
r
i 'qr =
'qr xm iqs
x 'r
1
'drr xm idsr
x 'r
The voltage equations re-written in terms of the state-space variables become:
i 'drr =
55
D p r s D r
p r
uqsr = rs +
iqs +
ids + xm
'qr + xm s 'rdr
x 'r b
b x 'r
b
b
D p r s D r
p r
udsr = rs +
ids
iqs + xm
'dr x m s 'rqr
x 'r b
b x 'r
b
b
r'
r'
p r s r
'dr
u ' rqr = 0 = xm r iqsr + r +
'qr +
b
x 'r
x 'r b
(247-250)
r 'r r r 'r
p r s r
ids +
+
'
'dr
x 'r
b qr
x 'r b
If the special reference frame is fixed to the rotor flux linkage vector, the q-component of this
flux vector is defined equal to zero:
'rqr = 0
(251)
'rr = 'rdr
From stator flux linkage equations, the q-current components are given by:
u ' rdr = 0 = xm
iqsr = iqsr
(252)
xm r
iqs
xr
The flux producing component of the stator current is determined as follows:
x 'r p r
1 +
'dr
r 'r b
r
ids =
(253)
xm
The above relation shows that there is no need of a current decoupler in rotor field orientation
scheme. Both stator current components (torque and flux producing) can be controlled
independently.
i 'rqr =
3P
x
3P
x2
m 'rdr iqsr =
m 2 iqsr
4 x 'r b
4 ( x 'r )
( )
r 'r
s
The most important features of rotor field oriented (RFO) vector control are:
56
As it is the original vector control strategy developed by Blashke (1971), the RFO represents the
most popular approach in vector control strategies;
It provides complete decoupling of the torque and flux producing components of the stator
current;
There are possible direct vector control strategy (the usage of sensors or model to provide
feedback of the flux magnitude and orientation) and indirect vector control strategy (the usage
of assumed slip frequency relationship to achieve field orientation);
The direct vector control strategy is the optimum choice for medium and high-speed
applications;
The indirect vector control strategy is the optimum choice for low-speed applications;
and stator current space vectors x = qm , dm , iqs , ids is selected. If the induction motor is equipped
with wound rotor, as the rotor currents are also measurable, the rotor current space vector can be
also selected as state-space variables together with the air-gap flux linkage space vector
T
x = qm , dm , i 'qr , i ' dr . The mathematical model is given in the chapter dedicated to d-q models of
the three-phase induction machine. As both cases are similar to analyse, only the first case will be
detailed.
The stator and rotor flux linkages and rotor current space vectors components expressed as
functions in terms of state-space variables are:
m
m
qsm = qm
+ ( xs x m )i qs
m
dsm = dm
+ ( xs x m )i dsm
'mqr =
x 'r m
qm ( x 'r xm ) iqsm
xm
(258-261)
x 'r m
qm ( x 'r xm ) idsm
xm
1 m
i 'qrr =
qm iqsm
xm
qsm =
(262-263)
1 m
m
i' =
dm ids
xm
The voltage equations of the machine expressed in state-variables terms, are readily deductible
now:
p
p m s m
uqsm = rs + ( xs xm ) iqsm +
qm +
dm + ( xs xm ) idsm
b
b
b
r
dr
p m
p m s m
m
uds = rs + ( xs xm ) ids +
dm
qm + ( xs xm )iqsm
b
b
b
p
u ' = 0 = r 'r + ( x 'r xm )
b
dm ( x 'r xm )idsm
iqs +
qm +
b xm
xm xm b
p
u 'mdr = 0 = r 'r + ( x 'r xm )
b
m r 'r x 'r p
+
ids +
xm xm b
m
qr
m s x ' r m
qm ( x 'r xm )iqsm
dm
b xm
(264-267)
57
If the special reference frame is fixed to the air-gap flux linkage vector, the q-component of this
flux vector is defined equal to zero:
m
qm
=0
(268)
m
mm = dm
From air-gap flux linkage equations, the q-current components are given by:
iqsm = iqsm
(269)
i 'mqr = iqsm
The flux producing component of the stator current is determined as follows:
p m r 'r x 'r p m
s m
+
i
(270)
r 'r + ( x 'r xm ) ids =
dm + ( x 'r xm )
b
b qs
xm xm b
One can note that the d-axis component of stator current for AFO is not only controllled by the daxis air-gap flux componennt, but also by the q-axis (torque producing) component of the stator
current. It is necessary to decouple the stator current components, in order to achieve a linear
control. For this reason, the command of current of the d-axis component is computed as follows:
K m
idsm = K p + i dm
+ idqm
p
(271-272)
m
(
x
'
x
)
s
r
m
qs
m
idq
=
b r 'r + ( x 'r xm ) p
The electromagnetic torque relation and slip speed can be derived in air-gap field orientation
control as:
Te =
s =
3P 1 m m
dm iqs
4 b
(273-274)
x 'r m
dm ( x 'r xm ) idsm
xm
r 'r
b
It yields the slip speed equation:
2
2 ( x 'r xm )
m
s
iqsm s dm
+ r 'r biqms = 0
(276)
r 'r b
The solutions of the above equations have to be real, for a given stator flux linkage. This means that
the determinant of the second order equation satisfy the condition:
m
= ( dm
) 4 ( x 'r xm ) (iqsm ) 0
2
(277)
The maximum values for the q component of the stator current, the slip speed and the
electromagnetic torque (pull-out torque) are:
58
( )
iqsm
max
( s )max
( Te )max
m
dm
2( x 'r xm )
r 'r
=
x ' r xm
(278-280)
m
3P dm
=
8 x ' r xm
When the angular slip velocity is larger than (s ) max instability will occur. One should note that
the maximum (pull-out) angular slip velocity depends only on the rotor parameters and does not
depends on the air-gap flux. However, the maximum (pull-out) torque is proportional to the square
of the air-gap flux magnitude and thus a small increment of air-gap flux will detetmine a significant
increment of the electromagnetic torque.
Some important conclusions cand be drawn for the air-gap field orientation (AFO) control
strategy:
The electromagnetic torque and air-gap flux producing components of the stator current are not
decoupled;
A parameter dependent decoupling network has to be included;
Torque and flux control does not require speed feed back;
No sophisticated parameter estimation methods or model based observers is required;
It is a good alternative for low and medium performance drives as the air-gap flux can be
measured directly;
2.9.4. Stator current orientation (SCO)
If the field orientation control is well established with multiple practical solutions for different
industrial applications, an unconventional method of controlling the speed and torque for the
induction machine, is given by the selection of currents space vectors as linking basis for the
synchronuous reference frame. The set of state-space variables is identical with the stator flux field
T
orientation control (SFO): the stator flux and current space vectors x = qs , ds , iqs , ids . The
mathematical model is identical to that used in SFO case.
The rotor based variables expressed in state-space variables terms are:
1 sc
i 'sc
qs xs iqssc
qr =
xm
1 sc
ds xs idssc
xm
1
x 'r qssc D iqssc
'scqr =
xm
i 'sc
dr =
(281-284)
1
x 'r dssc D idssc
xm
To obtain the machine equations in the synchronuous stator current reference frame, one has to
eliminate the rotor flux from the rotor voltage equations and then force the q component of the
stator current to be zero. The stator voltages equations remain unchanged. If a cage rotor is
considered, the resultant equations are as follows:
'scdr =
59
p sc s
p sc
x 'r dssc D idssc r 'r + x 'r
r 'r xs + D
iqs
= 0
b
b
b qs
(285-286)
p sc s
p sc s
D iqssc r 'r + x 'r
x 'r qssc = 0
r 'r xs + D
ids
ds +
b
b
b
b
If the special reference frame is fixed to the stator current space vector, the q-component of this
current vector is defined equal to zero:
sc
iqs = 0
(287)
iss = idssc
Different from the SFO case, we have to express the q-flux components, by considering the flux
linkages equations:
qssc = qssc
(288)
x'
'scqr = r qssc
xm
The electromagnetic torque relation and slip speed can be derived in stator current orientation
control as:
Te =
s =
3 P 1 sc sc
qs ids
4 b
( b r 'r
xs + x 'r p ) qssc
(289-290)
From the second dynamic equation of the machine, a relation between the d-axis component of the
stator current and stator flux linkage space vector components can be deduced:
p sc
p sc s
(291)
x 'r qssc = 0
r 'r xs + D
ids r 'r + x 'r
ds +
b
b
b
The previous dynamic equation of the machine, shows that there is a coupling between the stator
flux linkage components. Consequently, any change in the torque producing component dssc
without changing qssc accordingly will cause a trensient in the stator flux. A decoupler is necessary
to overcome this disadvantage. Therefore the command current of the d-axis component of the
stator current can be calculated as follows:
K
dssc = K p + i idssc + dqsc
p
s
(291-292)
x 'r qssc
b
sc
dq =
p
r 'r + x 'r
where Kp and Ki are proportional, respectively integral coefficients of the current controller. This
controller can be PI type or soft computing technique type (fuzzy, neural-network).
60
x r'
x 'r D s
2
qssc s 2 r b
r 'r xm b
xm s
r'
= r b qssc
xm s
dssc = qssc
sc
dr
(293)
s x 'r 1 sc s 1 sc
1
(294)
qs + ids + 2 qssc = 0
xm
b r 'r xm
b r 'r
The solutions of the above equations have to be real, for a given stator current. This means that the
determinant of the second order equation satisfy the condition:
2
2
sc
idssc
x 'r qs
= 4
(295)
2 0
r
'
r
'
x
r
r
m
The maximum values for the q component of the stator flux linkage, the slip speed and the
electromagnetic torque (pull-out torque) are:
x 2 i sc
( qssc ) max = 2mx 'ds
r
r'
(296-299)
( s )max = r
x 'r
( )
( Te )max
2
sc
3P xm ids
=
4 2 b x ' r
If the angular slip velocity is larger than ( s )max static instability will arise. This maximum value
depends only on machine parameters and not on stator current or flux level. Nevertheless, the pullout torque depends on the square of the stator current magnitude. The pull-out torque determines
the limit for the system stability operation zone, when the stator current space vector oriented
control is employed. It is possible to limit the torque command not to exceed the pull out torque for
a given stator current. Some important conclusions can be drawn for the stator current orientation
(SCO) vector control strategy:
The electromagnetic torque and stator current related components of the stator flux linkage are
not decoupled;
A parameter dependent decoupling network has to be included;
Torque and flux control does not require speed feed back;
An operation at low speed is easy to realise as there is no need to estimate the stator flux;
It is a good alternative for wide range speed drives.
2.9.5. Rotor current orientation (RCO)
Another unconventional vector control strategy is the one wich comprises a synchronuous
reference frame linked to the rotor current space vector. The rotor current orientation control
strategy (RCO) is analysed for a cage rotor induction motor. The selected set of state-space
variables is given by two options if we consider the criteria of direct measurable quantities:
T
I) The stator flux linkage and rotor current space vectors: x = qs , ds , i 'qr , i ' dr ;
When the stator flux and and the rotor current are selected as state-space variables, one can derive
the rotor flux and stator current functions in terms of state variables as follows:
61
1 rc
qs xm i 'qrrc
xs
1 rc
idsrc =
ds xm i 'drrc
xs
x
D
'rcqr = m qsrc + i 'qrrc
xs
xs
iqsrc =
(300-304)
xm rc D rc
ds + i 'dr
xs
xs
To obtain the machine equations in the synchronuous rotor current reference frame, one has to
eliminate the rotor flux from the rotor voltage equations and then force the q component of the rotor
current to be zero. The stator voltages equations are re-written also. The resultant equations have
the following form:
rx
r
p rc s rc
uqsrc = s m i 'qrrc + s +
qs +
xs
b ds
xs b
'rcdr =
uds =
rc
rs xm rc rs
p rc s rc
i 'dr + +
ds
xs
b qs
xs b
D p rc D s rc xm p rc xm s rc
0 = r 'r +
i ' dr +
qs +
ds
i ' qr +
xs b
xs b
xs b
xs b
(305-308)
D p rc D s rc xm p rc xm s rc
0 = r 'r +
i 'qr +
ds
qs
i 'dr
xs b
xs b
xs b
xs b
If the special reference frame is fixed to the rotor current space vector, the q-component of this
current vector is defined equal to zero:
rc
i 'qr = 0
(309)
irrc = i 'drrc
Different from the RFO case, we have to express the q-flux components, by considering the flux
linkages equations:
qsrc = qsrc
(310)
x
'rcqr = s qssc
xm
The electromagnetic torque relation and slip speed can be derived in stator current orientation
control as:
Te =
s =
3 P xm rc rc
i'
4 xs b qs dr
xm p qsrc
(311-312)
D i 'rcdr + x m dsrc
From the second dynamic equation of the machine, a relation between the d-axis component of the
rotor current and stator flux linkage space vector components can be deduced:
D p rc xm p rc xm s rc
0 = r 'r +
(313)
i ' +
xs b dr x s b ds x s b qs
A coupling between the stator flux linkage components appears, as can be deduced from the
previous dynamic equation of the machine. Consequently, any change in the torque producing
62
rc
component dsrc without changing qs accordingly, will cause a transient in the stator flux. The
command current of the d-axis component of the stator current can be calculated in order to
overcome this disadvantage:
K
dsrc = K p + i i 'drrc + dqrc
p
(314-315)
1
rc
sc
dq = s qs
p
where Kp and Ki are proportional, respectively integral coefficients of the current controller. This
controller can be PI type or soft computing technique type (fuzzy, neural-network).
xs r 'r b qs
(316)
'rcdr = 0
It yields the slip angular velocity equation:
D x D
idrrc m2 s qsrc = 0
(317)
xs xs r 'r b
which for a given rotor flux has always real solution. Thus the slip speed and the electromagnetic
torque are:
xs r ' r bi 'drrc
s =
xm qsrc
(318-319)
3P xm rc rc
Te =
i'
4 xsb qs dr
The most important features of rotor current oriented (RCO) vector control with mixed flux and
current state-space variables are:
The electromagnetic torque and rotor current related components of the stator flux linkage are
not decoupled;
The decoupling network that has to be included is parameter independent and easy to
implement;
Torque and flux control require speed feed back;
There is no stability limit as there is no pull-out slip speed or electromagnetic torque,
An operation at low speed is easy to realise as there is no need to estimate the stator flux;
It is a good alternative for wide range speed drives.
T
II) The stator and rotor currents x = iqs , ids , i 'qr , i 'dr . When the stator and the rotor current are
selected as state-space variables, one can derive the stator and rotor flux as functions in terms of
state variables from the classical flux linkage equations.
To obtain the machine equations in the synchronuous rotor current reference frame, one has to
eliminate the rotor flux from the rotor voltage equations and then force the q component of the rotor
current to be zero. The stator voltages equations are re-written also. The resultant equations have
the following form:
63
p rc
p rc
uqsrc = rs + xs
i 'qr + xs s i dsrc + xm s i 'drrc
iqs + xm
b
b
b
b
uds = xs
rc
s rc
rc
p
i qs xm s i 'qr + rs + xs
b
b
b
p rc
p
0 = xm
iqs + r 'r + x 'r
b
b
rc
p rc
i'
ids + xm
b dr
rc
s rc
(320-323)
s rc
p rc
p rc
iqs x 'r s i 'qrrc + xm
ids + r 'r + x 'r
i 'dr
b
b
b
b
If the special reference frame is fixed to the rotor current space vector, the q-component of this
current vector is defined equal to zero:
rc
i 'qr = 0
(324)
irrc = i 'drrc
Different from the (I) case of RCO control strategy, for the (II) case we have to express the stator
flux linkage q-components, by considering the flux linkages equations:
qsrc = xs iqsrc
(325)
x
'rcqr = xm iqssc = m qsrc
0 = xm
xs
From the second dynamic equation of the machine, a relation between the d-axis component of the
rotor current and stator current space vector components can be deduced:
p rc
p rc
xm
ids = x m s iqsrc r 'r + x 'r
(326)
i 'dr
b
b
b
The electromagnetic torque relation and slip speed can be derived in stator current orientation
control as:
3P xm rc rc
Te =
i i'
4 b qs dr
(327-328)
xm piqsrc
s = rc
xm ids + x 'r i 'drrc
A coupling between the d-axis and q-axis stator current components appears, as can be deduced
from the previous dynamic equation of the machine. Consequently, any change in the torque
rc
producing component idsrc without changing iqs accordingly, will cause a transient in the stator flux.
The command current of the d-axis component of the stator current can be calculated in order to
overcome this disadvantage:
K
idsrc = K p + i
p
1
rc
idq
= s iqscs
p
rc r c
i 'dr + idq
(329-330)
where Kp and Ki are proportional, respectively integral coefficients of the current controller. This
controller can be PI type or soft computing technique type (fuzzy, neural-network).
Steady-state perfomance of a rotor current oriented system
After several manipulations of the system equations if we let the derivative operator p = 0, the d
flux components are obtained from the steady-state voltage equations of the induction machine:
64
dsrc =
D s rc
iqs
r 'r b
(331)
' =0
It yields the slip angular velocity equation:
D x D
idrrc m s iqsrc = 0
(332)
xs xs r 'r b
which for a given rotor flux has always real solution. Thus the slip speed and the electromagnetic
torque are:
r ' i 'rc
s = r brc dr
xm iqs
(333-334)
3P xm rc rc
Te =
i i'
4 b qs dr
The rotor current oriented (RCO) vector control with currents state-space variables presents the
same features as the mixed flux-current state variables model. However the current model is easier
to be implemented as it provides directly the estimated values for the two-axis co-ordinate stator
current. Thus the command for a current PWM inverter is readily obtainable. Also, as feedback
measurements, the stator currents sensors are more reliable and cheaper than the flux sensors.
rc
dr
65
a)
b)
c)
d)
e)
f)
g)
In Fig. 3.1 is illustrated a single-phase induction machine. The following assumptions have been
made:
electrically orthogonal stator windings with sinusoidal distribution;
only the fundamental-space-harmonic-component of the air-gap flux distribution will be
considered;
magnetic-saturation effects, core loss and stray load losses are negligible;
magnetic-diffusion (i.e. deep-bar) effects in the rotor will be ignored. This assumption is
typically valid in small induction machine. It is further justified by the fact that under most
operating conditions, the single-phase induction motor will be operating at low slips and hence
the rotor currents will be at frequencies sufficiently low that magnetic-diffusion effects are
insignificant.
temperature effect on windings resistance and reactance value is negligible;
the lamination magnetic permeability is considered infinite;
in steady-state operation, the voltages and currents are sinusoidal.
66
bs axis
br axis
ar axis
as
bs
Te
br
ar
TL
r
s
br
bs
ar
as axis
as
-
u as
rr
rs
rotor
stator
Ls
i as
Lr
iar
Lr
Ls
+
rs
rr
i bs
-
ubs
ibr
Fig.3.1. The real model of the symmetrical single-phase induction machine with
squirrel-cage rotor represented by two identical windings
The machine rotor is described by two identical magnetically orthogonal windings. Considering
an arbitrarily reference frame, the spatial position of the stator winding is characterised by the
electrical angle s and the spatial position of the rotor winding is characterised by the electrical
angle r. The angular speed of the rotor is r and the displacement between stator and rotor
windings is r. These angles are linked through the relation:
s = r + r
(335)
The voltage equations related to the machine from Fig.3.1, can be expressed as follows:
uas = Vs 2cos t
(336-337)
ubs = Vs 2cos( t + )
d
uas = rs ias + as
dt
d bs
ubs = rs i bs +
dt
(338-342)
d ar
0 = rr iar +
dt
d
0 = rr ibr + br
dt
As the machine is magnetically linear, the fluxes are easily determined from the currents and
inductance values. Particularly, it can be written:
67
(343-346)
(352)
68
In the above relation, the superscript () stands for referred rotor windings to the stator according to
the expressions:
N
i 'abr = r iabr
Ns
'abr =
Ns
abr
Nr
(356-357)
N
L 'rr = s Lrr
Nr
The electromagnetic torque relation can be simplified and presented in the following form:
P
Te = Lms ( ias i 'ar + ibsi 'br ) sin r + ( iasi 'br ibsi 'ar ) cos r
(358)
2
This relation shows that there are two components of the instantaneous electromagnetic torque:
an average component with constant value for a given value of the rotor speed and a pulsating
component with a frequency double of the currents frequency ( e). The pulsating torque
component determines an important magnetic noise for the single-phase induction machine
compared to the three-phase induction machine. The mechanical equation that links the torque and
the rotor speed is:
2 d
2
Te = J r + Bm r + TL
(359)
P dt
P
where J is the rotor inertia, B is the viscous friction coefficient associated to the rotational system of
the machine and with the mechanical load, and P represents the number of poles for the analysed
machine.
3.3. Analysis of the single-phase induction machine in stationary reference frame
In order to eliminate the time dependence of the voltage and flux equations terms, a variables
transformation into a new reference frame is necessary. This transformation is given by the
following relations (Krause et al - 1995):
f qds = K ss fabs
(360)
f 'qdr = K sr f 'abr
cos sin
K ss =
sin cos
cos( r ) sin( r )
K sr =
sin( r ) cos( r )
(361)
(362)
(363)
where f can be fluxes, voltages, currents in new co-ordinates (d-q) or in classical co-ordinates (a-b).
Index (s) stands for statoric terms and index (r) stands for the rotoric terms. One should note that the
two transformation matrixes depend on the generalised co-ordinate which expresses the
periferical displacement of the chosen reference frame.
Reference frames linked to stator, rotor or arbitrarily can be chosen for the polyphase induction
machine. As for the single-phase induction machine, the stator windings are not identical, the only
transformation that maintains the windings parameters (resistance, inductance) unchanged is the
stationary reference frame transformation ( =0).
The voltage and linkage flux equations are expressed for the previous assumptions made at the
beginning of this chapter.
69
Note: For compactness, the flux linkages will be further described by flux linkage units per
second or volts, and the inductance equivalent circuit elements will be transformed in reactance
elements.
u qs = rs iqs +
p
qs
b
u dq = rs ids +
p
ds
b
(364-367)
0 = r 'r i 'qr +
p
'qr r 'dr
b
b
0 = r 'r i 'dr +
p
'dr + r 'q r
b
b
(368-371)
xs
(r/ b)dr
xr
+
iqs
xm
iqr
rr
uqs
_
rs
xs
xr
(r/b)QR
ids
xm
idr
+
rr
uds
_
70
3.4. Analysis of the steady-state operation for the symmetrical single-phase induction machine
In a wide range of applications the single-phase induction machine is equipped with a cage rotor.
During the steady-state operation, the stator parameters are variable with the stator voltage
frequency , and the rotor parameters are variable with the slip frequency - r. From the
stationary reference frame equations, by letting the differential operator p be replaced by the
complex operator j , the following matrix relation is established (Krause and Thomas - 1965):
0
j
xm
0
rs + j xss
b
b
qs
U
I
0
rs + j
xss
0
j
xm qs
ds
b
b
U
(373)
I ds
=
0
Iqr
r
xm
xm
r 'r + j
x 'rr
x 'rr
b
b
b
b
0
I dr
r
xm
j r xm
x 'rr
r 'r + j
x 'rr
b
b
b
b
where
xss = xs + xm
(374-375)
x 'rr = x 'r + xm
A symmetrical two-phase system is defined by the identities:
FQS = jFDS
(376-377)
FQR = jFDR
~
where F represents a complex variable with current or voltage significance. The four equations
from the above matrix relation are interdependent. If it is used, the variable slip s is defined as:
r
(378)
s=
and also through the inverse transformation from the stationary reference frame (d-q) to the
physical one (a-b), the relations for the steady-state operation analysis of the symmetrical singlephase induction machine can be written:
as = ( r + jx ) I + jx ( I + I )
U
s
s
as
m
as
ar
(379-380)
r'
The above equations suggest the equivalent circuit from Fig. 3.3.
rs
jxs
jxr
r/s
+
U as
I as
jxm
I ar
71
With slight modifications (x m becomes 3/2x m) it can be observed that the equivalent circuit from
Fig. 3.3 can be used also for the three-phase induction machine performance in steady-state
operation.
The electromagnetic torque equation is:
P x
Te = 2 m Re jIas* Iar =
2 b
Te =
(381)
(382)
( rs r 'r + s ( xm2 xss x 'rr )) 2 + ( rr xss + srs x 'rr )2
It is important to highlight that the positive values of the torque are obtainable when the slip s is
positive (motor operation) and the negative values when the slip s is negative (generator operation).
By setting the torque/slip derivative equal to zero, the relation for the critical slip can be obtained:
sm = r ' r G
G=
rs2 + xss2
( xm2 xss x 'rr ) 2 + rs2 x 'rr2
(383-384)
The positive value corresponds to motor operation, and the negative one to generator operation.
If at start-up (s = 1) the torque is directly dependent to the variation of rotor resistance, as the
magnetisation reactance value is considered to be much higher than the stator or rotor resistance
value, the maximum torque value is not dependent to the rotor resistance value:
2
as
2(P /2)( x2m / )G U
Te,max =
(385)
2
rs + G(xm2 xss x 'rr ) + (xss + Grs x 'rr )2
3.5. Analysis of the unsymmetrical single-phase induction machine
A new equations set and a new equivalent circuit for the unsymmetrical single-phase induction
machine can be obtained by eliminating from the initial assumptions the one referring to identical
stator windings (Krause - 1965):
Voltage equations:
u qs = rm iqs +
p
qs
b
u ds = ra ids +
p
ds
b
(386-389)
1
p
'qr r 'qr
0 = r 'r i 'qr +
b
k b
0 = k 2 r 'r i 'dr +
p
'dr + k r 'dr
b
b
Flux equations:
qs = xlm iqs + xm (iqs + i 'qr )
(390-393)
72
P xm
k (iqsi 'dr idsi 'qr )
(394)
2 b
The mathematical model presented in Fig. 3.4 permits the analysis of steady state as well as
dynamic operation of the unsymmetrical single-phase induction machine. This model is also
suitable for implementing non-linearities effects such as: core loss or saturation of the main or
leakage inductance.
Te =
rm
xlm
(1/k) ( r/ b)dr
xsr
+
iqs
iqr
xm
uqs
rr
ra
xla
k (r/b)qr
k2 xsr
ids
uds
idr
+
k 2 rr
k2 x m
73
The original field orientation scheme, developed for the three-phase induction machine, was
based on the alignment of the rotor flux linkage because the torque and the rotor flux are related to
each other in a straightforward manner, without any de-coupling circuit. To calculate the rotor flux,
the stator and rotor leakage inductance and the main inductance are necessary. If some other
equivalent circuits are implemented, it can be obtained a model like that one illustrated by
Slemon (1989), where the stator and rotor leakage inductance are viewed as a total measurable
inductance.
3.6.1. Linear model of the symmetrical single-phase induction machine
The classical T-form circuit model can be transformed into simpler models with no loss of
information or accuracy. Since rotor variables can be seen from the stator reference frame as
referred variables depending on the transformation turn ratio, we can choose a value such that the
magnetisation inductance is equal to the total stator inductance. This would give the following set
of rotor variables related to those from the T-form equivalent circuit:
'dR
'
qR = 1 '
i 'dR dr
i 'qR 1
'qr
i 'dr
(395)
i 'qr
where = xm /( x s + x m ) . The above relation combined with the voltage equations gives the
equivalent d-q modified model for the symmetrical single-phase induction machine presented in
Fig. 3.5. This configuration has been denoted as the form model due to its inductance structure.
The parameters of this equivalent circuit are related to those of the T-form d-q mathematical model
through the relations:
xM = xm = xs + xm
xL = ( xr + xm ) 2 ( xs + xm ) = xs + x r 2
(396)
r 'R = 2 r 'r
rs
( r/b) dR
xL
+
iqs
xM
iqR
r R
uqs
_
rs
( r/b ) qR
xL
uds
ids
xM
idR
+
rR
74
These parameters can be derived directly from the usual no load and standstill measurements on the
machine.
This model represents an appropriate solution for the analysis of scalar control and vector
control (stator flux oriented) single-phase induction machine (Slemon - 1994). The following
equations describe the complete operation of the system:
Stator and rotor voltage:
u qs = rs iqs +
p
qs
b
u ds = rs ids +
p
ds
b
0 = r 'R i'qR
0 = r 'R i 'dR
(397-400)
p
+
'qR r 'dR
b
b
p
+
' dR + r ' qR
b
b
2
2
P qs + ds r 'R sb
Te = 2
2
2
r '2R + ( sxL )
where the slip s is given by: s = (b r ) / b
(401-404)
(405)
(406)
i ''qR 1
'
'qr
i 'dr
i 'q r
(407)
where: '' = xm /( xm + xr ) . Using the voltage equations given for T-form equivalent circuit, it will
result a new configuration denoted as inverse form model shown in Fig.3.6 in which:
75
( xs + xm )( xr + xm ) xm2
x'L = xs + ' xr =
( xr + xm )
x 'M = ' xm = '2 ( xr + xm )
(408)
rs
( r/ b) dR
xL
+
iqs
xM
i qR
rR
uqs
_
rs
( r/ b) qR
xL
ids
xM
idR
+
r R
uds
_
Fig. 3.6. Equivalent linear d-q inverse form circuit for the symmetrical
single-phase induction machine
The following equations can be written to describe the operation of the machine:
Stator and rotor voltage:
u qs = rs iqs +
p
qs
b
u ds = rs ids +
p
ds
b
(409-412)
p
0 = r ''R i ''qR +
''qR r ''dR
b
b
p
0 = r''R i ''dR +
''dR + r ''qR
b
b
ds = x'L ids + x 'M (ids + i ''dR ) = ( x'M + x'L )ids + x 'M i 'dR
''qR = x 'M i 'qM = x 'M ( iqs + i ''qR )
(413-416)
(417)
76
2
2
P ''dR + ''qR sb
Te = 2
r ''R
2
where s denotes the slip for the analysed machine.
(418)
(421-423)
i'
i 'dr
; i 'qR = qr
(423)
77
+ (1- )x
x+
s
( x+(1)
r
i'
ds
(1) x
ds
+ i'
dR
( x+(1)
r
x)
i'
qs
qs
i'
dR
qR
qR
Fig. 3.7. Flux linkage equivalent circuit with arbitrary turn ratio
The rotor quantities are now calculated with the equations:
p
0 = r 'R i 'dR +
' dR + r 'qR
b
b
p
'qR r 'dR
0 = r'R i'qR +
b
b
(428-430)
r 'R = 2 rr
Through variation of the arbitrary turn ratio the appropriate reference vector flux for controlling
the machine can be selected. Table 3.I summarise the choices that have to be made:
TABLE 3.I . Specific choices of the turn ratio
Turn ratio
=
xm
xm + xr
=1
x + xs
= m
xm
Mathematical model
Inverse -form
d(q)
d(q)r
Flux selected
Rotor flux
T-form
d(q)m
d(q)s
Air-gap flux
Stator flux
-form
The universal mathematical model given in Fig. 3.7 permits a choice between different flux vectors
by selecting only one parameter , denoted the arbitrary turn ratio. All other basic machine
parameters remain unaffected.
3.6.4. Linear model of the unsymmetrical single-phase induction machine
Usually the single-phase induction machine is equipped with non-identical sinusoidally
distributed windings arranged in space quadrature. From the T-form equivalent circuit developed
78
for this type of machine in previous paragraphs, it can be developed a linear -form equivalent
circuit as shown in Fig. 3.8.
Since rotor variables can be seen from the stator reference frame as referred variables depending
on the transformation turn ratio, we can choose a value such that the magnetisation inductance is
equal to the total stator inductance on each of the two-axis co-ordinates. This would give the
following set of rotor variables related to those from the T-form equivalent circuit:
d
'dR q
'
qR = 1
i 'dR d
i 'qR 1
q
'dr
'qr
i 'dr
(431)
i 'q r
where
q = xm /( xlm + xm )
(432-433)
d = k 2 xm /( xla + k 2 xm )
For an identical distribution of the windings, we have with a good approximation xla = k 2 xlm , and
this gives equal turn ratio for each axis ( d = q ).
The above relations combined with the voltage equations give the equivalent d-q modified model
for the unsymmetrical single-phase induction machine. The parameters of the -form model are
related to those of the T-form d-q mathematical model through the relations:
xMq = q xm = xlm + xm
2
(434-438)
r ' R = q r'r
xMd = d k 2 xm = x sa + k 2 x m k 2 xMq
2
( r /k b ) dR
xLq
+
iq s
xMq
iqR
rR
u qs
_
r sa
(k r / b) q R
xLd
i ds
xM d
idR
+
k 2 rR
u ds
_
79
b qs
b ds
p
1
'qR r 'dR
0 = r 'R i 'qR +
b
k b
p
0 = k 2 r 'R i 'dR +
'dR + k r 'qR
b
b
Flux linkage:
qs = xMq iqs + i 'qR
(439-442)
(443-447)
Te = k
iqs i 'dR idsi 'qR ) = iqs ds k ids qs
(448)
(
2
b
2 k
This -form model represents an appropriate solution for the analysis of scalar control and vector
control (stator flux oriented) of the unsymmetrical single-phase induction machine drives.
3.6.5. Linear inverse model of the unsymmetrical single-phase induction machine
Another possible approach on analysing the unsymmetrical single-phase induction machine is
the one when an arbitrary turn ratio is used such that the magnetisation inductance in each axis is
equal to the correspondent total rotor inductance. The following set of variables is readily
obtainable:
d'
''dR q '
''
(449)
qR = 1 '
'qr i 'dr i 'qr
i ''dR d ' dr
i ''qR
'
where:
q ' =
xm
xr + xm
d ' =
k 2 xm
=q '
k 2 ( xr + xm )
(450)
The linear inverse -form model of the unsymmetrical single-phase induction machine is
presented in Fig. 3.9. This configuration is deductible from the T-form model using the above
expressions and new equivalent machine parameters defined as follows:
80
x 'Mq = q ' xm
x 'Md = d ' k 2 xm = k 2 x 'Mq
x'Lq = xlm + q ' xr
(451-455)
r 'R = q(d)
'
r'r
r sm
( r/kb )dR
xLq
+
iqs
x Mq
i qR
rR
uqs
---
r sa
(kr/b) qR
xLd
uds
ids
xMd
idR
+
k2 r R
---
0 = k 2 r'R i 'dR +
(456-459)
'dR + k r 'qR
b
b
Flux linkage:
qs = x 'Lq iqs + x 'Mq (i qs + i 'qR ) = ( x 'Mq + x 'Lq )iqs + x 'Mq i 'qR
(460-463)
81
x 'Mq
P
P 1
Te = k
( iqs i 'dR ids i'qR ) = iqs 'dR k ids 'qR =
2
2
k
b
1
P
(464)
The circuit model from Fig. 3.9, is particularly suitable for analysis and understanding the vector
control systems with single-phase induction machine when the rotor flux vector is chosen as
reference.
3.6.6. Universal model of the unsymmetrical single-phase induction machine
All the developed mathematical models for the unsymmetrical induction machine can be
generically presented with a universal model that preserves the characteristics for each of the
equivalent form circuits.
The basic idea is to calculate in the stationary reference frame (the only valid option for singlephase induction machine that maintains constant parameters) an arbitrary flux vector. The flux
vector is determined from the flux linkage equations given in various forms of equivalent circuits
(T-form, -form, inverse -form) by using an arbitrary turn ratio transformation.
This turn ratio, that characterises the two-axis reference frame, has two values: q and d. By
multiplying the real rotor vector flux r = dr + j qr with the turn ratio we can define the flux
linkage equations as:
qs = ( xlm + x m ) iqs + xm i 'qr q xm iqs + q xm iqs
(465-466)
ds = xla + k 2 xm ids + xm i 'dr d k 2 xm ids + d k 2 xm ids
(467-468)
(
= (x
la
+ k 2 (1 d ) xm
)
)i
ds
(
)
= ( x + (1 ) x ) i
(469)
+ 'd d
qs
+ 'q q
(470-473)
= 'd d + d k 2 ( d xr + ( d 1) xm ) i 'dR
= 'q q + q q xr + ( q 1) xm i 'qR
This flux linkage can be represented by the equivalent circuit given in Fig. 3.10. The rotor
quantities are now calculated with the equations:
82
'dR + k r 'qR
b
b
p
1
0 = k 2 r 'R i 'qR +
'qR r 'dR
b
k b
0 = r 'R i 'dR +
(474-476)
r 'R = q(d) rr
Through variation of the arbitrary turn ratios q and d there are several options to select the
appropriate reference vector flux for controlling the single-phase induction machine. Table 3.II
summarises the available choices:
x
+ (1- )x
sm
q m
x +
sa
(
x +(1)
q
q
r
q
x )
m
i'
ds
(1 ) x
k
m
d
(
d
+ i'
ds
dR
x+ ( 1)
r
d
x)
m
i'
qs
qs
i'
dR
qR
qR
Fig. 3.10. Flux linkage equivalent circuit with arbitrary turn ratio
for unsymmetrical single-phase induction machine
TABLE 3.II
Specific choices of the turn ratio q and d
Turn ratio
q =
xm
xm + xr
d =
k 2 xm
= q
k 2 ( xm + xr )
q = d =1
q =
xm + xlm
xm
d =
k 2 xm + xla
q
k 2 xm
Mathematical
model
Inverse -form
d(q)
Flux selected
d(q)r
Rotor flux
T-form
d(q)m
Air-gap flux
-form
d(q)s
Stator flux
83
The universal mathematical model given in Fig. 3.10 permits a choice between different flux
vectors by selecting only two parameters q and d that have practically equal values, and
denote the arbitrary turn ratio for each axis of the system. As in the symmetrical single-phase
induction machine case, all other basic machine parameters remain unaffected.
3.7. D-Q axes models of the single-phase induction machine
The single-phase induction machine can be modelled by using different state-space variables,
keeping as inputs the stator voltages and the load torque, and as outputs the electromagnetic torque
and rotor angular velocity. Let us define the possible set of currents and flux linkages per second
space vectors as follows:
i s = ids + j iqs
i r = i 'dr + j i 'qr
i m = idm + j iqm
(477-482)
s = ds + j qs
'r = 'dr + j 'qr
m = dm + j qm
[ x ] = [i s
i'r im s 'r m ]
(483)
Two-space vectors have to be selected from the six available vectors. Let the selected pair of
state-space variables be denoted as x 1 , x 2 . It is possible to express the set of six state-space
variables in terms of the two selected state-space variables as follows:
a11 a12
a
21 a22
a
a
T
T
(484)
[ x ] = [ is i'r i m s 'r m ] = 31 32 [ x1 x 2 ]
a41 a42
a51 a52
a61 a62
where the coefficients aij, depend on the chosen set of state-space variables.
Out of fifteen possible state-space versions obtained from combining two space vectors from the
total of six, only fourteen represent valid mathematical models for the single-phase induction
machine. Please note that the pair of state-space variables that comprise the airgap flux linkage
space vector and the magnetising current space vector cannot be selected as these vectors have the
same direction. These models can be classified in three types: currents space vectors, flux linkages
space vectors and mixed currents-flux linkages space vectors models.
In conjunction with the mechanical equation:
pr =
P
(Te TL )
2J
(485)
we obtain a complete version of the single-phase induction machine model, viewed as the key for a
motion control system.
For an unitary approach on all the possible models, and considering that the magnetising
reactance has generally much greater value than the leakage reactance, the following approximation
is made without altering the final results:
xla + k 2 xm k 2 ( xlm + xm ) = k 2 xs
(486)
84
The starting point for the state-variable models is given by voltage equations system written in
stationary reference frame:
u = L px + R x
(487)
where: x is the selected set of state-variables and represent also the output of the model, u is the
input vector (stator voltages), L is the coefficients matrix (it can be formed by reactance values, or
non-dimensional elements) for multiplying the time derivative of the state-variables, R is the
coefficients matrix (it can be formed by resistances and reactances values or non-dimensional
elements) for multiplying the state-variables and p stand for the differential operator (d/dt). It results
the general form of the state-variables system:
px = A x + B u
(488)
where:
A = L 1 R
(489-490)
B = L1
All the following mathematical models can theoretically be discretised and used for
implementation of controllers in the drive systems with single-phase induction machines. However,
practically only few of them are suitable for this purpose as the computational effort, and implicitly
the cost of the hardware, is a decisive factor to be considered.
When the vector control strategy has to be chosen, each of the six space vectors that can be
selected as state-space variables represents a possible basis for a reference frame. The algorithm for
implementing field oriented control systems is obtained as follows:
1. The rotor flux space vector is expressed as function of state-variables;
2. The rotor angular velocity r term is substituted with ( r - k) where k is the angular
velocity of the synchronous frame;
3. The dynamic equations that relates the rotor angular velocity to rotor flux linkages and voltages
are deduced in the new state-variable system;
4. The synchronous reference frame is selected linked to one of the space vectors, which means
that the q-axis component of the reference space vector is null;
5. The torque equation is computed according to the selected flux or current space vector in the
synchronous reference frame.
3.7.1. Models with currents space vectors as state-space variables
I) The most used model of the single-phase induction machine is the currents space vectors stateT
space variables x = iqs , ids , i 'qr , i 'dr . An important feature of this model is that the stator and
leakage reactance, arbitrarily considered equal in many implementations, are now included in the
stator and self-reactance values. As the stator currents are easily measurable quantities, a controller
based on these state variables gives good performance accuracy. The model is readily available
from the equations of voltages and flux linkages per second, and it can be expressed in a matrix
form as follows:
85
p
b
x 'r rm
D
iqs x m2 r
i
D k b
ds
=
i 'q r
x m rm
D
i 'd r
x x
s m r
D k b
x 'r
xm
0
D
D
1
0
0
k 2 x 'r D
xm
D
xs
D
1
k 2 xm D
x m2 k r
D b
x m r'r
D
ra x 'r 1
D k2
x m x 'r r
D
k b
xs xm k r
b
D
x m ra 1
D k2
x s r 'r
D
x s x 'r r
D
k b
1 u qs
2
k xm D u ds
u
'
qr
0
u '
dr
1
k 2x s D
x m x 'r k r
b
D
iqs
x m r 'r
D
ids +
xs x ' r k r i 'q r
b i 'd r
D
x r'
s r
D
(491)
where:
D = xs x 'r xm2
The electromagnetic torque can be computed as:
x
P
Te = k m ( iqs i 'dr ids i 'qr )
(492)
2
b
As it can be observed, a shortcoming for this model is the full system matrix, with all 16 elements
non-zero values.
II) Another mathematical d-q axis model based on currents space vectors is the one which
comprises the stator and the magnetising space vector currents as state-space variables
T
x = iqs , ids , qm
i , idm . The state matrix elements are all non-zero, which implies a computational
effort similar to the previously analysed model.
86
iqs
p ids
b iqm
idm
x 'r
A
1
0
+
x 'lr
A1
rm x 'r + r 'r x m
A1
x m x 'lr k r
A2
b
=
r 'r x lm rm x 'lr
A1
x x'
r
la
lr
A2
k b
xm
A1
x 'r
A2
x ls
A1
x 'lr
A2
xm
A2
x la
k 2 A2
x m x 'lr k r
b
A1
x m r 'r
A1
ra x 'r + k 2 r 'r x m
A2
x x'
k r
lm lr
A1
b
r 'r x la ra x 'lr
A2
x m x 'r k r
A2
b
r' x
r ls
A1
x x'
r
la r
A2
k b
x m x 'r k r
b
A1
iqs
k 2 r 'r xm
A2
ids
x lm x 'r k r iqm
A1
b idm
r' x
r la
A2
uqs
u
ds
u 'q r
u 'd r
(493)
where:
A1 = xlm x 'r + xm x'lr = D
A2 = k 2 xm x 'lr + xla x'r k 2 D
The instantaneous electromagnetic torque may be expressed in terms of the state-space variables:
x
P
Te = k m ( iqs idm ids iqm )
(494)
2
b
III) The third possible combination of current space vectors as state-space variables is obtained
T
by selecting magnetising and rotor currents x = i 'qr , i 'dr ,i qm , idm . When compared to the other
current space vector models, the similar computing burden for obtaining the outputs of the system is
obvious. The main difference between them is the presence of the global parameters (selfreactances) in the first model, while in the other two models an accurate determination of the
leakage reactance is necessary. A complete description of the state-variables in matrix notation is
given below:
87
rm x m + r ' r x s
A1
x s x ' lr k r
i ' qr
b
A2
p i ' dr
=
r x ' r' x
b iq m
m
lr
r lm
i
A
1
d m
x x'
r
la lr
A2
k b
xm
A
1
0
+
x ' lr
A1
xs
A1
xm
A2
x lm
A1
x ' lr
A2
x s x ' lr k r
b
A1
ra x m + k 2 r ' r x s
A2
rm x m
A1
x lm x ' lr k r
b
A1
ra x ' lr k 2 r ' r x la
A2
xs
A2
x la
k 2 A 2
xs x m k r
b
A2
rm x ' lr
A1
x la x m
r
A2
k b
xs x m k r
b
A1
ra x m
A2
x lm x m k r
b
A1
r x'
a lr
A2
i 'q r
i 'd r
i qm
i d m
u qs
u
ds
u 'q r
u ' d r
(495)
where:
A1 = xlm x m + x' lr x s = D
A2 = xla xm + k 2 x 'lr x s k 2 D
The instantaneous electromagnetic torque is expressed as follows:
Te =
P
x
k m ( i 'dr iqm i 'qr idm )
2
b
(496)
I) In this case the flux linkages space vectors, x = qs , ds , ' qr , 'dr are used for describing the
mathematical model of the machine. The matrix equations for this system are as follows:
rm xm
rm x 'r
0
0
D
D
ra x m qs u qs
ra x 'r
qs
0
0
k2D
k 2 D ds u ds
p ds
(497)
=
+
r 'r x s
r 'qr u 'qr
b 'qr r 'r x m
0
D
D
k b ' u '
'dr
dr dr
r
'
x
k
r
'
x
0
r m
r
r
s
D
D
b
where D has the same significance as in the previous cases.
The electromagnetic torque is determined with the equivalent relation:
P
xm
(498)
Te =
( qs ' dr ds 'qr )
2 k b D
As opposed to the current models, it can be observed that the computational burden is substantially
lower. Due to this important feature, this model is the most suitable for discretisation in motion
control strategies.
88
II) An alternative to model the single-phase induction machine with flux space vectors as statevariable system, is the selection of air-gap flux space vector among the set of independent variables.
A first approach is given by the stator and air-gap flux space vectors selected as state-space
T
variables x = qs , ds ,qm , dm .
r
rm
m
0
0
xlm
xlm
r
r
a
qs
0
a
0
qs
x
x
la
la
p ds
ds
=
+
2
rm x'lr xm r 'r xm
xlmx mx 'lr kr
rm xmx 'lr r 'r xs
xlmxm r
k x 'lr xs qm
b qm
1 +
x A + A
xla A1 b
xlm A1
A1
A1 kb
xla xm dm
dm lm 1
1
ra k2 x'lr xm k 2 r'r xm
xla xm kr x 'lr xs
ra k2 x'lr xm k 2r 'r xs
xla xmx'lr kr
1 +
xlm A2 b
xla A2
A2
A2 b xlm xm
xla A2
A2
1
0
x' x
+ lr m
A1
0
1
0
k 2 x 'lr xm
A2
0
0
xlm xm
A1
0
0
0 uqs
uds
0
u 'qr
xla xm u 'dr
A2
(499)
where:
A1 = xlm xm + x 'lr xs = D
A2 = xla xm + k 2 x 'lr xs k 2 D
The electromagnetic torque expression shows that this model is prohibitive for implementation for
modelling the single-phase induction machine in a vector control system and becomes:
1
P 1
k k
1
(500)
Te = qs ds
qs dm
dm qm
+
kxlm
2 b
kxlm xla xla
which can be simplified only if the approximation xla = k 2 xlm is reasonable to be made:
P
1
Te =
qsdm dm qm
2 kxlm b
(501)
III) The third option of selecting flux space vectors as state-space variables is the one which
T
comprises the air-gap (magnetising) flux and rotor flux space vectors x = qm , dm , ' qr , ' dr . By
comparison to the previously analysed model, this one presumes the same computational effort, but
the electromagnetic torque is determined in a suitable form for vector control, without any
supplementary approximations.
89
rm x'r
A
1
qm
p dm
=
b 'qr
'dr
xm x'lr
0
A
1
xm x'lr
+ 0
A2
0
0
0
0
r 'r x m xlm
x 'lr A1
rm xm r 'r x m xlm
A1
x 'lr A1
xm xla k r
A2
b
r 'r
x 'lr
r 'r
x 'lr
r 'r
x 'lr
k r
b
ra xm r 'r x m x la qm
A2
x 'lr A2 dm
+
'qr
r
k b
'dr
r'
r
x 'lr
xm xlm r
A1
kb
u
qs
xm xla uds
A2 u 'qr
0 u 'dr
xm x'lr
A1
0
1
0
(502)
where:
A1 = xm x 'lr + xlm x 'r = D
A2 = k 2 xm x 'lr + xlm x 'r k 2 D
The resulting relation for computing instantaneous electromagnetic torque is as follows:
P
1
Te =
( qm 'dr dm 'qr )
2 kx 'lr b
(503)
3.7.3. Models with mixed currents flux space vectors state-space variables
I) If the stator variables are chosen for modelling the single-phase induction machine system,
T
then a mixed flux linkages-currents state-space variables model x = qs , ds , iqs , ids is developed.
The matrix equations and the electromagnetic torque relation are:
0
0
qs
r'
p ds
r
=
D
b iqs
ids x 'r r
D k b
rm
0
r x ' + r'r xs
m r
D
x 'r r
D k b
r 'r
k 2D
r
k b
1
0
ra
qs
ds
k r
x'
+ r
b
D
iqs
2
0
k D
0
1
0
x 'r
k2D
0 0
0 0 u qs
u ds
0 0
u'
qr
u '
0 0 dr
(504)
Te =
II)
P 1
2 b
k ds iqs k qsids
Another
important
T
(505)
mixed
flux
linkages-currents
state-space
variables
model
x = 'qr , ' dr , i'qr , i'dr is that expressed in rotor quantities. This model can be easily used for the
rotor flux oriented control of the machine. We can describe the system as follows:
90
k r
'qr
'
p dr
b
=
rm
b i 'qr
D
i 'dr
x
s r
D k b
r
k b
rr
xs r
D b
rm x 'r + r 'r xs
D
ra
k4D
'
qr
k 2 r 'r
0
'd r + xm
i 'qr D
0
i 'dr 0
r x ' + k 2 r 'r xs
a r 2
k D
0
0
0
xm
k2D
0 0
0 0 uqs
uds
0 0
u'
qr
u '
0 0 dr
(506)
The electromagnetic torque relation is given below:
Te =
P 1
2 b
(507)
III) A similar model to the previous one is that with stator current and rotor flux linkage
T
x = 'qr , ' dr , iqs , ids as state-space variables. It represents an alternative model for rotor flux
oriented control strategies. The matrix equation and electromagnetic torque relation are presented
below:
rm x '2r + r 'r xm2
x 'r D
iqs
0
p ids
=
b 'qr
r 'r xm
x 'r
'dr
r 'r x m
x 'r D
x m k r
D b
r 'r
x 'r
k 2 r 'r xm
x 'r
kr
b
x 'r
D
0
0
0
x 'r
k2D
0
0
xm r
D k b
r 'r x m iqs
k 2 x 'r D ids
' +
r
qr
k b
'dr
r'
x 'r
x
0
m
u qs
D
u
x
2m ds
0
k D u 'qr
1
0 u '
dr
0
1
(508)
(509)
IV) A rarely used mathematical model is that with mixed rotor currents space vector and stator
T
flux linkages space vector as state-space variables x = qs , ds , i 'qr , i ' dr . However, the advantage
of this model is that merely only stator winding parameters are necessary, so the influence of rotor
parameters is minimised. It can be used for an unconventional stator flux oriented control with rotor
current components producing the torque and the flux. The expressions for implementing this
model are:
91
xs D
i 'qr
r
i '
p dr
k b
=
b qs
rm xm
xs
ds
k r
b
rm xm
xs D
ra x + k r 'r x
k 2x s D
2
m
2
s
xm
D
r
m
xs
ra xm
xs
xm r
D k b
xm
D
i
'
ra xm
qr
4
k Dxs i 'dr
0
+
qs
0
ds 1
0
r
2a
k xs
0
xm
k 2D
0
1
xs
D
0
0
0
0
uqs
xs uds
k 2 D u 'qr
0 u 'dr
0
(510)
where the significance of D was already stated.
The main system output, the electromagnetic torque, can be computed as:
Te =
P xm
2 xsb
k qs i 'dr ds i 'qr
k
(511)
V) The mixed stator current space vector and air-gap flux space vector as state-space variables
T
x = qm , dm , iqs , ids belongs to one of the more complex model types. It preserves information
regarding both stator and rotor parameters. Different from the previous mixed models, the statespace matrix contains only non-zero elements, which leads to greater computational effort.
rm x 'r + r 'r xm
xm x 'lr k r
r 'r
x'
r r
A1
A1
A1
A1 k b
b
iqs
i
x
x
k
r
x
+
k
r
x
x
k
r
'
'
'
'
'
qs
m
lr
r
r m
r
r
r
a r
A2
A2
A2 b
A2
b
ids
p ids
+
xlm x m x 'lr k r
xlm r 'r
xlm x 'r r qm
b qm xm (r 'r x m rm x 'lr )
A1
A1
A1
A1
k b dm
b
dm
x x x ' k
x x ' kr
x r'
x m k 2 ( x 'lr ra + xla r 'r )
r
la m lr
la r
la r
A2
A2
A2
b
A2
b
x 'r
A
1
0
+
x 'lr xm
A1
xm
A1
x 'r
A2
xlm x m
A1
k 2 x 'lr x m
A2
x m uqs
A2 uds
u 'qr
0
u 'dr
xla xm
A2
0
(512)
where:
92
Te =
P 1
2 b
iqsdm k idsqm
k
(513)
VI) A similar model to the precedent one, is the mixed rotor current space vector and air-gap flux
T
space vector as state-space variables x = qm , dm , i 'qr , i ' dr . The matrix equation of the system is
given below:
rm x m + r 'r x s
A1
x x ' k r
i 'qr
s lr
i'
b
A2
p dr
=
b qm
x m ( r 'r x lm + rm x 'lr )
A1
dm
x x x ' k r
la m lr
b
A2
xm
A
1
0
+
x 'lr x m
A1
xs
A1
xm
A2
x lm x m
A1
k 2 x 'lr x m
A2
x s x 'lr k r
b
A1
ra x m + k 2 r 'r x s
A2
rm
A1
x lm x m x 'lr k r
b
A1
x m ( k 2 x 'lr ra k 2 x la r 'r )
A2
xs k r
A2 b
x lm rm
A1
x la x m k r
A2
b
iqs
ra
2
k A2 ids
x lm x m r qm
A1
k b dm
x la ra
A2
xm r
A1 k b
x s u qs
A2 u ds
u 'q r
0
u 'd r
x la x m
A2
0
(514)
where:
A1 = xlm xm + x 'lr xs = D
A2 = xla xm + k 2 xlr xs k 2 D
The electromagnetic torque relation becomes:
P 1
1
Te = k i 'dr qm i 'qr dm
2 b
k
(515)
VII) If the magnetising current space vector is selected as state-space variable together with one
T
of flux linkages space vectors x = iqm , idm , qs , ds , the state matrix computation gives several
null elements. Nevertheless, the torque expression is more complicated and the final relation can be
written in the same form as in previous cases only through some new assumptions,.
93
rm x'lr xm r'r x s x ls
xls A1
xla A1 r
iqm
i x A k
p dm
lm 2
b
=
rm xm
b qs
xlm
ds
x 'lr
A
1
+ 0
xlm
A1
x 'lr
A2
0
1
0
0
xlm A2 k r
xla A1 b
xla x 'lr r
xlm A2 k b
r
m
xlm
0
k 2 ra x m
xla
xlm x' lr k r
xla A1 b
a
xla
0
u
qs
xla u ds
k 2 A2 u 'qr
0 u 'dr
(516)
where:
A1 = x m xlm + xs x 'lr = D
A2 = xm xla + k 2 x s x 'lr k 2 D
The electromagnetic torque relation, with the approximation: xla k 2 xlm can be written as:
k3
1
P 1
1
k
xm
idm qs
iqm ds + qs ds
2 b
kxlm
xla
xla
kxlm
(517)
P
xm
1
k idm qs iqm ds
2 b xlm
k
The presence of oscillating terms given by the product between flux linkages, makes this model
prohibitive for implementing in vector control strategies.
Te =
VIII) The last mixed current-flux linkages space vector model realises the connection between
the magnetising current space-vector and the rotor flux linkage space vector
T
x = iqm , idm , 'qr , 'dr , selected as state-space variables. The state matrix contains the same
number of zero elements (four) and the instantaneous electromagnetic torque is determined in a
suitable form for vector control implementation.
94
iqm
p idm
b 'qr
'dr
x 'lr
A
1
+ 0
x 'lr A1
x 'lr A2
=
r 'r xm
0
x 'lr
k 2 r 'r xm
0
x 'lr
xlm
0
0
A1
u
qs
x 'lr
xla u ds
0
A2
k 2 A2 u 'qr
0
1
0 u 'dr
0
0
1
xla r
A2 k b
r 'r
x 'lr
r
b
xlm r
A1 k b
'qr
r
kb
'dr
r'
r
x 'lr
(518)
where:
(519)
95
5. The synchronous reference frame is selected linked to one of the space vectors, which means
that the q-axis component of the reference space vector is null;
6. The torque equation is computed according to the selected flux or current space vector in the
synchronous reference frame.
The transformation of reference frames for the induction motor vector control can be summarised as
shown in Fig. 3.11.
Synchronuous reference frame
Induction
Controller
Transformation
motor
current space vectors x = qs , ds , iqs , ids is selected. The mathematical model is given in the
chapter dedicated to d-q models of the single-phase induction machine.
The rotor based variables expressed in state-space variables terms are:
1
i 'sqr =
( qss xs iqss )
xm
s
xsids
dss
k 2 xm
xm
(520-523)
1
s
s
s
'qr =
( x'r qs D iqs )
xm
1
'sdr =
( x'r dss k2 D idss )
xm
To obtain the machine equations in the synchronuous stator flux reference frame, one has to
eliminate the rotor flux from the rotor voltage equations and then force the q component of the
stator flux to be zero. The stator voltages equations remain unchanged. If a cage rotor is considered,
the resultant equations are as follows:
i 'sdr =
p s s 1
p s
x 'r dss k D idss r'r + x 'r
r 'r xs + D
iqs
= 0
b
b k
b qs
s
s
p s
p s
k 2 r 'r xs + D
D iqss = r 'r + x 'r
x 'r qss
ids k
ds k
b
b
b
b
(524-525)
96
where the definition used for the slip speed is: s = e r . The electromagnetic torque expression
has to be re-written:
P 1 1
s
2 s
If we make the notation: iqs = k iqs1 the above torque relation may be expressed as for the
symmetrical induction machine:
P k
s
Te =
dss iqs1
qss idss )
(527)
(
2 b
If the special reference frame is fixed to the stator flux linkage vector, the q-component of this flux
vector is defined equal to zero:
qss = 0
(528)
dss = ss
From stator flux linkage equations, the q-current components are given by:
s
s
iqs1
= iqs1
(529)
k 2 xs s
s
i 'qr =
iqs1
xm
The electromagnetic torque relation and slip speed can be derived in stator field orientation control
as:
Te =
P k s s
d siq s 1
2 b
s
k 2 ( b r 'r xs + D p ) i qs1
s =
1
x 'r dss k D idss
k
(530-531)
The second dynamic equation of the machine, shows that there is a coupling between the stator
s
current components. Consequently, any change in the torque producing component iqs1
without
changing idss accordingly will cause a transient in the stator flux. A decoupler is necessary to
overcome this disadvantage. Therefore the command current of the d-axis component of the stator
current can be calculated as follows:
K
s
idss = Kp + i dss + idq
p
(532-533)
s
k s D iqs1
s
idq =
b r'r xs + D p
where Kp and Ki are proportional, respectively integral coefficients of the flux controller. This
controller can be PI type or soft computing technique type (fuzzy, neural-network).
Steady-state perfomance of a stator flux oriented system
Letting the derivative operator p = 0, one can obtain the steady-state voltage equations of the
induction machine. After several manipulation of the system equations, the d current components
are given by:
97
idss =
kD s s
s
iqs1 + 2ds
r'r xs b
k xs
(534-535)
x
s
i 'sdr = 2 ds s idss
k xm xm
it yields the slip speed equation:
2
s s k 2 D 2 s xm2 s
(536)
s + k 2r 'rx siqss 1 = 0
iqs1
r
'
x
kx
r s
b
s
b
The solutions of the above equations have to be real, for a given stator flux linkage. This means that
the determinant of the second order equation satisfy the condition:
2
x2 s
2
s
0
= m s 4 k 4 D 2 iqs1
kx s
The maximum values for the q component of the stator current, the slip speed and the
electromagnetic torque (pull-out torque) are:
(i )
s
qs1
max
( s ) max
(Te ) max
x m2 ss
2 k 2 D xs
r 'r xs
kD
( )
(537-539)
( )
2
s
P x m s
=
2 k 2 2 b D xs
The problems related to static stability of the single-phase induction machine when this vector
control strategy is applied, are similar to the polyphase machine case. It has to pointed out that the
critical values for current q-component, slip angular velocity and electromagnetic torque depend
also on the turns ratio k of the unsymmetrical stator windings. If we detail this consideration, it can
be observed that having a motor equipped with a main stator winding with fixed parameters
(resistance, reactance) the pull out torque and critical rotor speed are such that (Popescu -2000):
A maximum torque value and minimum rotor speed are obtained if the turns ratio k < 1, which
is the case of the usual split-phase motors configuration;
The medium values are obtained if the turns ratio k = 1, which is the case of the symmetrical
motors configuration;
A minimum torque value and maximum rotor speed are obtained if the turns ratio k > 1, which
is the case of the capacitor run motors configuration;
Some important conclusions can be drawn for this vector control strategy, besides those ones
valid for the polyphase motor:
The estimated equivalent value for the torque component of the stator current has to be
corrected by dividing to the square of the turns ratio value;
There is no need for changing the number of co-ordinates from the real machine system to the
control system.
3.8.2. Rotor flux orientation (RFO)
For this vector control strategy, the set of state-space variables formed by rotor flux linkage and
T
stator current space vectors x = 'qr , ' dr , iqs , ids is selected. The mathematical model is given in
the chapter dedicated to d-q models of the single-phase induction machine.
The stator flux linkage and rotor current space vectors components expressed as functions in
terms of state-space variables are:
98
qsr =
1
xm 'rqr + D iqsr
x 'r
1
r
xm 'drr + k 2 D ids
x'r
1
r
i 'qr =
'rqr xm iqsr
x 'r
dsr =
)
(540-543)
r
1 'dr
r
2 xm ids
x 'r k
The rotor voltage equations re-written in terms of the state-space variables become:
r'
r'
p r 1 s r
u 'rqr = 0 = xm r iqsr + r +
'dr
'qr +
x 'r
k b
x'r b
i 'drr =
r'
r'
p r
s r
u ' = 0 = k xm r idsr + r +
'qr
'dr k
x 'r
b
x 'r b
The electromagnetic torque expression has to be re-written:
P x 1
r
dr
(544-545)
(546)
r
If we make the notation: iqsr = k 2iqs1
the above torque relation may be expressed as for the
symmetrical induction machine:
P x k
r
Te = m
'rdr iqs1
'qrr idss
(547)
2 x 'r b
If the special reference frame is fixed to the rotor flux linkage vector, the q-component of this
flux vector is defined equal to zero:
'rqr = 0
(548)
'rdr = 'rr
From rotor flux linkage equations, the q-current components are given by:
r
r
iqs1
= iqs1
(549)
k 2 xm r
i'rqr =
iqs1
x 'r
The flux producing component of the stator current is determined as follows:
x'r p r
1 +
'dr
r
'
r
b
idsr =
(550)
k 2 xm
The above relation shows that there is no need of a current decoupler in rotor field orientation
scheme. Both stator current components (torque and flux producing) can be controlled
independently.
99
r
'rdr
b k r 'ri qs1
i = 2 =
k xm s
x 'r
r
ds
(551)
i 'drr = 0
It yields the slip speed equation:
s 'drr
x
r
k m b r'r iqs1
=0
(552)
2
k
x 'r
which for a given rotor flux has always real solution. Thus the resulting current controlled slip
speed and the electromagnetic torque are:
k 3 xm r ' r b iq rs 1
s =
r
x'r 'dr
k x m r' r b iqsr
x 'r 'drr
x
P
P x
r
Te = k m 'rdr iqs1
= m
2
x 'r b
2 x 'r
(i )
r
qs1
k 4 r' r
P x
= m
s
2 x 'r
(553-554)
(i )
r
qs
k 2 r 'r
By comparison with the three-phase induction motor, for the rotor field orientation control
strategy, three differences have been highlighted by Popescu and Navrapescu (2000):
The stady-state values for the slip angular velocity and torque are proportional with the turns
ratio value, respectively the square of turns ratio k which determines the option for k > 1, as the
torque response is the determinant factor in an electrical drive system;
The estimated equivalent value for the torque component of the stator current has to be
corrected by dividing to the square of the turns ratio value;
There is no need for changing the number of co-ordinates from the real machine system to the
control system;
The problem of the stator windings asymmetrical configuration is overcome by using the value
of turns ratio k in computing the estimated value for the torque and flux producing components
of the stator current.
3.8.3. Air-gap flux field orientation (AFO)
For this vector control strategy, the set of state-space variables formed by air-gap flux linkage
T
and stator current space vectors x = qm , dm , iqs , ids is selected. The mathematical model is given
in the chapter dedicated to d-q models of the single-phase induction machine.
The stator and rotor flux linkages and rotor current space vectors components expressed as
functions in terms of state-space variables are:
m
m
qsm = qm
+ ( xs x m )i qs
m
dsm = dm
+ k 2 ( xs x m )i dsm
'mqr =
x'r m
qm ( x'r xm ) iqsm
xm
'mdr =
x 'r m
qm k 2 ( x'r xm ) idsm
xm
i'rqr =
1 m
qm iqsm
xm
1
m
i ' = 2 dm
idsm
k xm
r
dr
(555-558)
(559-560)
100
The voltage equations of the machine expressed in state-variables terms, are readily deductible
now:
r'
x ' p m s 1 x'r m
p
m
u 'mqr = 0 = r 'r + ( x'r xm ) iqsm + r + r
qm +
dm k ( x'r xm ) ids
b
b k xm
xm xm b
p
u 'mdr = 0 = k 2 r 'r + ( x'r xm )
b
m r 'r x 'r p
+
ids +
xm xm b
m
s x'r m
m
dm k
qm ( x'r xm )iqs
b x m
(561-562)
The electromagnetic torque expression has to be re-written:
P 1 1 m m
m m
Te = dm
iqs k qm
i ds
(563)
2 b k
If we make the notation: iqs = k iqs1 the above torque relation may be expressed as for the
symmetrical induction machine:
P k
Te =
(564)
( dmm iqs1m qmm idsm )
2 b
If the special reference frame is fixed to the air-gap flux linkage vector, the q-component of this
flux vector is defined equal to zero:
m
qm
=0
(565)
m
dm
= mm
From the air-gap flux linkage equations, the q-current components are given by:
m
m
iqs1
= iqs1
(566)
m
i'mqr = k 2 iqs1
The flux producing component of the stator current is determined as follows:
r'
p
x' p m
m
k 2 r 'r + ( x 'r xm ) idsm = r + r dm
+ k 3 ( x'r xm ) s iqs1
(567)
b
b
xm xm b
When an air-gap field orientation control is employed, it is necessary to decouple the stator
current components, in order to achieve a linear control. For this reason, the command of current of
the d-axis component is computed as follows:
K m
idsm = K p + i dm
+ idqm
p
(568-569)
m
k s ( x'r xm ) iqs1
m
idq =
b r'r + ( x'r xm ) p
The electromagnetic torque relation and slip speed can be derived in air-gap field orientation
control as:
m
Te =
s =
2 m
P k m m
dm iqs
2 b
(570-571)
101
xm
r 'r
s
b
(572)
m
k
(
x
x
'
)
i
m
r qs1 s
i 'drm =
r 'r
b
It yields the slip speed equation:
2
m
s k ( x 'r xm ) 2 m s dm
m
+ k r'r iqs1
=0
(573)
qs1
2
b k
r 'r
b
The solutions of the above equations have to be real, for a given stator flux linkage. This means that
the determinant of the second order equation satisfy the condition:
m
dm
2
m 2
0
(574)
= 2 4k 2 ( x 'r xm ) iqs1
k
The maximum values for the q component of the stator current, the slip speed and the
electromagnetic torque (pull-out torque) are:
m
dm
m
( iqs1 ) max = 2k 3 (x ' x )
r
m
r'r
(575-577)
( s )max =
x 'r xm
( )
( )
m
dm
P 1
( Te )max =
2 2b k 2 ( x 'r xm )
For the air-gap field orientation control strategy, there have to be pointed out several important
features:
The turns ratio k determines the effects on the critical value of the electromagnetic torque of the
single-phase machine: lower torque for k > 1, and higher torque for k < 1, when the main stator
winding parameters are kept constant;
The pull-out value of the slip speed does not depend on the unsymetrical configuration of the
stator windings, and it is expressed with an identical relation to that deduced for the three-phase
induction machine;
The estimated equivalent value for the torque component of the stator current has to be
corrected by dividing to the square of the turns ratio value;
There is no need for changing the number of co-ordinates from the real machine system to the
control system;
The problem of the stator windings asymmetrical configuration is overcome by using the value
of turns ratio k in computing the estimated value for the torque and flux producing components
of the stator current.
102
and current space vectors x = qs , ds , iqs , ids . The mathematical model is identical to that used in
SFO case.
The rotor based variables expressed in state-space variables terms are:
1
i'sc
sc xs iqssc
qr =
xm qs
1
dssc k 2 xs idssc
k xm
(578-581)
1
sc
sc
sc
x'r qs D iqs
'qr =
xm
1
x'r dssc k 2 D idssc
'scdr =
xm
To obtain the machine equations in the synchronuous stator current reference frame, one has to
eliminate the rotor flux from the rotor voltage equations and then force the q component of the
stator current to be zero. The stator voltages equations remain unchanged. If a cage rotor is
considered, the resultant equations are as follows:
i 'sc
dr =
p sc s 1
p sc
x 'r dssc k D idssc r'r + x 'r
r 'r xs + D
iqs
= 0
b
b k
b qs
p
k 2 r 'r xs + D
b
sc
s
p
sc
D iqs
= r 'r + x 'r
ids k
b
b
(582-583)
sc
s
x 'r qssc
ds k
b
1 sc
If we make the notation: qssc = 2 qs1
the above torque relation may be expressed as for the
k
symmetrical induction machine:
P 1
Te =
dssc iqssc qscs 1idscs )
(585)
(
2 k b
If the special reference frame is fixed to the stator current space vector, the q-component of this
current vector is defined equal to zero:
sc
iqs = 0
(586)
iss = idssc
Different from the SFO case, we have to express the q-flux components, by considering the flux
linkages equations:
sc
sc
qs1
= qs1
(587)
x'
sc
'scqr = 2 r qs1
k xm
The electromagnetic torque relation and slip speed can be derived in stator current orientation
control as:
Te =
s =
P 1
sc sc
qs1
ids
2 k b
sc
( b r'r xs + x 'r p ) qs1
(588-589)
103
The relation between the d-axis component of the stator current and stator flux linkage space vector
components can be deduced as follows:
p sc
p sc 1 s
sc
k 2 r 'r xs + D
x 'r qs1
=0
(590)
ids r'r + x 'r
ds +
k b
b
b
Similarly to the SFO case, there is a coupling, but between the stator flux linkage components
used as control variables. Consequently, any change in the torque producing component dssc
sc
without changing qs accordingly, will cause a transient in the stator flux. A decoupler is necessary
to overcome this disadvantage:
K
dssc = K p + i idssc + dqsc
p
1 s
sc
(591-592)
x'r qs1
k b
sc
dq =
p
r 'r + x 'r
where Kp and Ki are proportional, respectively integral coefficients of the current controller. This
controller can be PI type or soft computing technique type (fuzzy, neural-network).
dr =
qs1
kxm s
it yields the slip speed equation:
2
r 'r
s x'r
1
sc
sc
+ s k 2idssc +
=0
(594)
qs1
qs1
'
x
k
r
k xm
b m
b
r
The solutions of the above equations have to be real, for a given stator current. This means that the
determinant of the second order equation satisfy the condition:
2
2
sc
x 'r qs
= k i
4
(595)
2 2 0
x
k
x
m
m
The maximum values for the q component of the stator flux linkage, the slip speed and the
electromagnetic torque (pull-out torque) are:
k 2 x2m idssc
sc
=
( qs1 ) max 2 x '
r
k r'r
(596-598)
( s ) max =
x 'r
2 sc
ds
( )
2
sc
P k xm ids
(T e )m a x =
2
2 b x' r
The analysis of this vector control strategy leads to the same conclusions as for the three-phase
induction machine case, but with the differences imposed by the asymmetrical motor configuration
and the number of phases for the supply voltage:
104
The turns ratio k determines different effects on the critical values of the rotor speed and
electromagnetic torque response of the single-phase machine: the electromagnetic torque critical
values reaches a maximum for k > 1, while the rotor speed reaches a maximum for k < 1, when
the main stator winding parameters are kept constant;
The estimated equivalent value for the torque component of the stator flux has to be corrected
by multypling with the square of the turns ratio value k;
There is no need for changing the number of co-ordinates from the real machine system to the
control system;
x = qs , ds , i 'qr , i ' dr . When the stator flux and and the rotor current are selected as state-space
variables, one can derive the rotor flux and stator current functions in terms of state variables as
follows:
1 rc
iqsrc =
qs xm i'qrrc
xs
idsrc =
1 dsrc
rc
2 xm i'dr
xs k
(599-602)
x
D
' = m qsrc + i'qrrc
xs
xs
rc
qr
xm rc k 2 D rc
i'
+
xs ds
xs dr
To obtain the machine equations in the synchronuous rotor current reference frame, one has to
eliminate the rotor flux from the rotor voltage equations and then force the q component of the rotor
current to be zero:
D p rc kD s rc xm p rc 1 xm s rc
0 = r'r +
i'dr +
qs +
ds
i'qr +
xs b
xs b
xs b
k xs b
(603-604)
D p rc kD s rc xm p rc kxm s rc
2
0 = k r 'r +
i'qr +
ds
qs
i'dr
xs b
xs b
xs b
xs b
1 rc
If we make the notation: qsrc = 2 qs1
the above torque relation may be expressed as for the
k
symmetrical induction machine:
P x
rc rc
Te = m qs1
i'dr dsrc i'qrrc
(606)
2 k b xs
If the special reference frame is fixed to the rotor current space vector, the q-component of this
current vector is defined equal to zero:
'rcdr =
105
i 'qr = 0
rc
(607)
i 'drrc = i 'rrc
Different from the RFO case, we have to express the q-flux components, by considering the flux
linkages equations:
rc
rc
qs1
= qs1
(608)
x
sc
'rcqr = 2 s qs1
k xm
The electromagnetic torque relation and slip speed can be derived in stator current orientation
control as:
Te =
P xm
rc rc
qs1
i 'dr
2 kxs b
xm
rc
p qs1
2
k
s =
1
k D i 'drrc + x m dsrc
k
(609-610)
A relation between the d-axis component of the rotor current and stator flux linkage space vector
components can be deduced:
D p
x
p
x rc
0 = k 2 r 'r + i 'drrc + m dsrc m s qs1
(611)
xs b
xs b
kxs b
Due to the coupling between the stator flux linkage components, any change in the torque
producing component dsrc without changing qsrc accordingly, will cause a trensient in the stator
flux. To overcome this disadvantage, the command current of the d-axis component of the stator
current is expressed as follows:
K
dsrc = K p + i i 'drrc + dqrc
p
(612-613)
sc
1
qs1
dqrc = s
p
k
where Kp and Ki are proportional, respectively integral coefficients of the current controller. This
controller can be PI type or soft computing technique type (fuzzy, neural-network).
Steady-state perfomance
After several manipulation of the system equations if we let the derivative operator p = 0, the d
current components are obtained from the steady-state voltage equations of the induction machine:
D s rc
dsrc =
'rcdr = 0
It yields the slip angular velocity equation:
k 2 D xm D s rc
2
(615)
idrrc
qs1 = 0
xs
kxs r'r b
which for a given rotor flux has always real solution. Thus the slip speed and the electromagnetic
torque are:
106
(616-617)
P xm
P kxm rc rc
rc rc
Te =
i' =
i'
2 kxsb qs1 dr 2 xs b qs dr
The rotor current oriented (RCO) vector control with mixed flux and current state-space
variables differs from the variant applied to the three-phase induction machine by the following:
The transformation between the reference frames (stationary to synchronous and vice-versa) is
easier as there is no need of changing the number of variables;
The asymmetry of the stator windings is modeled only by using a supplementary parameter, the
turns ratio k;
In steady-state operation, the torque and slip speed response are influenced by the unsymmetrical
configuration of the motor: higher torque and slip speed for k < 1.
T
II. The state-space variables are stator and rotor currents x = iqs , ids , i'qr , i'dr . When the stator
and the rotor current are selected as state-space variables, one can derive the stator and rotor flux as
functions in terms of state variables from the classical flux linkage equations.
To obtain the machine equations in the synchronuous rotor current reference frame, one has to
eliminate the rotor flux from the rotor voltage equations and then force the q component of the rotor
current to be zero. The stator voltages equations are re-written also. The resultant equations have
the following form:
s rc
p rc
p rc
0 = xm
iqs + r 'r + x'r
ids + kx'r s i 'drrc
i 'qr + kxm
b
b
b
b
(618-619)
s rc
s rc
p
p
rc
0 = kxm
iqs kx'r
i 'qr + k 2 xm
idsrc + k 2 r 'r + x 'r
i 'dr
b
b
b
b
107
Te =
P kxm rc rc
i i'
2 b qs dr
xm piqs
(623-624)
rc
s =
From the second dynamic equation of the machine, a relation between the d-axis component of the
rotor current and stator current space vector components can be expressed as:
p rc
p rc
k 2 xm
ids = kxm s iqsrc k 2 r'r + x 'r
(625)
i'dr
b
b
b
A coupling between the d-axis and q-axis stator current components appears, and consequently,
rc
any change in the torque producing component idsrc without changing iqs accordingly, will cause a
transient in the stator flux. To overcome this disadvantage the command current of the d-axis
component of the stator current has to be expressed in the following form:
K
idsrc = K p + i i 'drrc + idqrc
p
iqssc
1
rc
idq
= s
p
k
(626-627)
where Kp and Ki are proportional, respectively integral coefficients of the current controller. This
controller can be PI type or soft computing technique type (fuzzy, neural-network).
Steady-state perfomance
By letting the derivative operator p = 0, the d flux components are obtained from the steady-state
voltage equations of the induction machine:
kD s rc
dsrc =
i
r 'r b qs
(628)
'rcdr = 0
It yields the slip angular velocity equation:
k 2 D xm kD s rc
i 'drrc
iqs = 0
(629)
xs
xs r 'r b
which for a given rotor flux has always real solution. Thus the slip speed and the electromagnetic
torque are:
kr'r bi 'drrc
s =
xm iqsrc
(630-631)
P kxm rc rc
Te =
i i'
2 b qs dr
A comparison with the rotor current orientation control detailed for the three-phase induction
machine permits to highlight the same conclusions. However, there has to be pointed out the
necessary corrections:
The transformation between synchronous reference frame and the stationary reference frame is
made without changing the number of variables;
The influence of the unsymmetrical stator windings configuration can be evidenced by using the
turns ratio k, and it determines different effect over the rotor speed and torque response of the
motor: higher electromagnetic torque and lower rotor speed for k > 1, respectively lower torque
and higher rotor speed for k < 1.
108
Discretization method
Forward-difference method (Euler)
Backward-difference method
1 z 1
T z 1
s=
1 z 1
T
s =
2 1 z 1
T 1 + z 1
s=
2 1 z 1
,
T 1 + z 1
T
2
tan D
T
2
G D ( z ) = T Z L 1 [G ( s )]
G ( s)
G D ( z ) = 1 z 1 Z L 1
All of the above discretization methods are compared for the induction machine seen as a discrete
system. The assumptions valid for the continuous model are also valid for the discrete systems. The
results of the discretization are analysed from the point of view of computation complexity and the
response stability for a step input signal. The analysis of the different models is essential for the
design of electrical drive system with numerical command. A special attention is given to torque
estimation from the data acquisition process (voltage or current quantities).
109
The synchronous reference frame is used for determining the discrete model of the induction
machine. The following parameters can be defined:
k = s s = r
For an easy to follow analysis, the index k will be omitted in the machine equations. The general
form equations used to determine the discrete mathematical induction machine in arbitrary
synchronuos reference frames and per unit system are:
1 d s (t )
U s (t ) = rs i s (t ) +
+ j s s (t )
(632)
n dt
1 d 'r (t )
0 = r 'r i 'r (t ) +
+ j r 'r (t )
(633)
n
dt
(t ) = ls i s ( t) + lM i 'r (t )
(634)
s
r (t ) = 2 l 'r i 'r (t ) + l M is (t )
ir
2
; 'r = r ; r 'r = rr
d
1
=
( te tl )
dt TM
Depending on the synchronuous reference frame the following relations are valid:
a) Stator reference frame:
i 'r =
ls
lM
3
*
te = P Im s is
2
b) Rotor reference frame:
l
= M
l 'r
(635)
(636)
(637)
3 l
t e = P M Im r* is
2 lr
c) Airgap reference frame:
=1
3
t e = P Im m * is
2
m = l M (i s + i 'r )
(638)
(639)
(640)
If the determinant of the flux system equations is noted with and with d its inverse, the voltage
equations are re-written as follows:
1 d s ( t )
U s (t ) = rs d ( l 'r s l M 'r ) +
+ j s s (t )
(641)
n dt
1 d 'r (t )
0 = r 'r d (ls 'r l M s ) +
+ j r 'r (t )
(642)
n
dt
For zero initial conditions, if the Laplace transformation is applied, it results the matrix system:
rs lM d
s U s
s s rs l 'r d j s
+
(643)
=
r 'r ls d j r 'r 0
n 'r r 'r lM d
110
or in synthetisized form:
s Y (s) = A Y (s) + B U (s)
One can note that by comparing with the state variable system described by the equations:
s X ( s ) = A X (s ) + B U (s )
Y ( s ) = C X (s ) + D U (s )
(644)
(645)
the state variable vector is identical to the output vector (fluxes vector in this case). The input vector
is the voltage vector. Through identification we obtain:
rs lM d
rs l 'r d j s
A = n
r 'r ls d j r
(646)
r 'r lM d
B = n
4.2. Bilinear transformation method (Tustin)
The relation (3) from the Table I is known as bilinear transformation or Tustin and it makes the
connection between continuous to discrete domain. This method gives better results than the
trapezoidal approximation, or other discretization method illustrated in Table 4.II, by considering
first four order of integration operators. The describing system relation becomes:
2 z 1
Y ( z) = A Y (z ) + B U ( z)
T z +1
(647)
where T is the signal sampling period for continuous domain, and f is the sampling frequency.
TABLE 4.II
Method
1
s
T z +1
2 z 1
1
s2
T 2 ( z + 1) 2
4 ( z 1) 2
1
s3
T 3 ( z + 1) 3
8 ( z 1) 3
1
s4
T 4 ( z + 1) 4
16 ( z 1) 4
BoxerThaler
T z +1
2 z 1
T 2 z 2 + 10 z + 1
12
( z 1) 2
T 3 z2 + z
2 ( z 1) 3
T 4 z 3 + 4z 2 + z
6
( z 1) 4
Madwed
T z +1
2 z 1
T 2 z 2 + 4z + 1
6
( z 1) 2
T 3 z3 +11z2 + 11z +1
24
(z 1)3
T 4 z 4 + 26z3 + 66z 2 +1
120
(z 1) 4
Tustin
2
( z 1) Y ( z ) = ( z + 1) A Y ( z ) + ( z + 1) B U ( z )
T
2
2
zY (z ) Y ( z) =
T
T
2
I A z Y ( z) =
T
z Y ( z) = I A
T
z A Y ( z ) + A Y ( z ) + ( z + 1) B U ( z )
2
I + A Y ( z ) + ( z + 1) B U ( z )
T
I + A Y ( z) + I A
T
Y ( z ) = I A I + A z 1 Y ( z ) + I A
T
1
1
Y ( z ) = C z 1 Y ( z ) + D (1 + z 1 ) U ( z )
B ( z + 1) U ( z )
1
B (1 + z 1 ) U ( z )
(648)
111
(649)
(650)
c22 = (2 f + n rs lr d ) (2 f n r 'r ls d ) + n 2 ( rs rr l M 2 d 2 + r s ) +
+ j n [ r (2 f + n rs lr d ) + s (2 f n r 'r ls d ) ]
d11 = n (2 f + n r 'r ls d ) + j r n 2
d 21 = r 'r lM d n
Observation: d12 , and d22 are not calculated as the induction machine is considered to have cage
rotor.
1
1
s ( n) = c11 s (n 1) + c12 'r (n 1) + d11 [U s ( n 1) + U s (n ) ]
(651)
' (n ) = 1 c (n 1) + c ' ( n 1) + 1 d [U ( n 1) + U ( n) ]
21
22
r
s
s
s
21
r
where n and n 1 are two consecutive sampling periods in discrete domain.
The induction machine discrete model is presented in Fig. 4.1. For this model the stator voltage
represents the input vector and the fluxes are the expressed as output vector.
c
11
-1
1/
c
U
11
-1
12
s
+
21
12
+
1/
-1
'
r
22
U ( n 1) + U (n )
1
1
sI
s I ( n) =
sI ( n 1) +
sI
(652)
'rR ( n) R 2 + I 2
'rR ( n 1) R 2 + I 2
0
'r I ( n)
'rI (n 1)
where:
112
e 11 = e 22 = c11 R R + c 11 I I
e 12 = e 21 = c 11 R I c 11 I R
e 13 = e 24 = c 12 R R
e 14 = e 23 = c12 R I
e 31 = e 42 = c 21 R R
e 32 = e 41 = c 21 R I
e 33 = e 44 = c 22 R R + c 22 I I
e 34
f 11
f 12
f 31
=
=
=
=
(653)
e 43 = c 22 R I c 22 I R
f 22 = d 11 R R + d 11 I R
f 21 = d 11 R I d 11 I R
f 42 = d 21 R R
f 32 = f 41 = d 21 R I
The complete discrete model of the induction machine, based on the bilinear method is given in Fig.
4.2. The electromagnetic torque expressed in relative units is considering different reference frames
for the induction machine:
a) Stator reference frame
3
3
3
*
te = P Im s is = P Im{(sR jsI ) (isR + j isI )} = P (sR isI sI isR ) =
2
2
2
3
3
= P (sR ( l 'r d sI lMd ' rI ) sI ( l 'r d sR lMd 'rR )) = P lM d (sI 'rR sR 'rI )
2
2
(654)
b) Rotor reference frame:
3
3
3
*
te = P Im r is = P Im{( 'rR j 'rI ) (isR + j isI )} = P ( 'rR isI 'rI isR ) =
2
2
2
3
3
= P ( 'rR ( l 'r d sI lM d 'rI ) 'rI ( l 'r d sR lMd 'rR )) = P 2 l 'r d (sI ' rR sR ' rI ) =
2
2
3
= P lM d ( sI 'rR sR 'rI )
2
(655)
c) Airgap reference frame:
3
3
3
*
te = P Im m i s = P Im {( mR j mI ) (isR + j isI )} = P ( mR is I mI isR ) =
2
2
2
3
= P ( mR ( l 'r d sI l M d ' rI ) mI ( l 'r d sR lM d 'rR ) ) =
2
3 lM d ( sR (l 'r lM ) + 'r R ( ls lM ) ) ( l 'r d sI l Md 'rI )
= P
=
2 lM d ( s I (l 'r lM ) + 'rI ( ls l M ) ) ( l 'r d sR l Md 'rR )
( s R (l 'r lM) + 'rR (l s lM ) ) (l 'r d s I lM d 'rI ) 3
3
= P lM d
= P lM d ( sI ' rR sR 'rI )
2
( sI (l 'r lM) + 'rI (ls lM ) ) ( l 'r d s R lM d 'rR ) 2
(656)
From the torque equations, written in stator, rotor and air-gap reference frames, one can note that
the same general relation can be implemented, though the equivalent rotor flux stands for different
significance, i.e. only in air-gap reference frame it has the real physical rotor flux.
113
As it can be observed by studying the block diagram of the discrete model of the induction
machine, for a complete implementation there are necessary 31 multiplier blocks, 23 summing
blocks and 4 delay blocks. However, this model can be simplified further. The total number of
multiplier blocks can vary from one implementation to another, due to the place of this operation in
the block diagram. Also the frequency response is variable according to the implementation version
(Vainio et al - 1992).
e11
f11
e12
+
f12
-1
z
1/
e13
e14
z-1
e21
e22
f21
+
UsR(n)
1/
e23
e24
z-1
*
'
rI(n)
sI(n)
f22
z-1
f31
+
+
UsI(n)
f32
f41
+
e31
e32
1/
e33
e34
sR(n)
z-1
+
+
Torque
t
'
rR(n)
e41
e42
1/
e43
e44
-1
f42
+ j s s ( n)
(658)
n
T
1 'r (n + 1) r ( n + 1)
+ j r r ( n)
0 = r 'r l s d 'r (n) lM d s ( n) +
(659)
n
T
In matrix form it results:
114
rs l 'r d
j s
rs lM d
(n)
s (n + 1)
n T
U (n)
s + nT s
= nT
1
' (n)
'r (n + 1)
0
r 'r ls d
j r r
r 'r lM d
n T
(660)
After separating the real and the imaginary components, it will result:
1
sR (n + 1) + j sI ( n + 1) = nT ( rsl 'r d +
j s ) [ sR ( n ) + j sI ( n) ] +
nT
(661)
+ n T rs lM d [ 'rR ( n ) + j 'r I ( n) ] + n T [U sR ( n ) + j U sI ( n) ]
1
(662)
j r ) [ 'rR ( n) + j 'rI ( n) ]
n T
'd
nTs
nT rl
d
0
sR (n +1) 1 nT rl
sR (n)
UsR (n)
s r
s M
(n +1) T
U (n)
1 nT rl
0
nT rl
n
s
s 'r d
s Md sI (n)
sI
=
+ nT sI
'rR (n +1) nT r'r lMd
0
0
1nT r'r lsd
nT r rR (n)
0
nT
1 nT r'r lsd rI (n)
nT r 'r lMd
'rI (n +1)
0
+n T (r 'r ls d +
(663)
or:
( n + 1) = A (n ) +
T U (n)
(664)
i ( n) = B (n )
(665)
0
0 sR (n)
lM d
isR (n) l 'r d
i (n) 0
0
lM d sI (n)
l 'r d
sI
=
i 'rR( n) lM d
0
ls d
0 'rR( n)
lM d
0
ls d 'rI ( n)
i 'rI (n) 0
For this case the complete discrete model of the induction machine is described in Fig. 4.3.
(666)
115
a11
UsR(n)
T
n
z
-1
-1
a12
a13
a21
U
sI(n)
T
n
sR(n)
a22
z
-1
z-1
a24
rI(n)
+
Torque
+
a31
t
sI(n)
a33
+
z -1
a34
a41
rR(n)
a43
+
-1
a44
(667)
nTrl
1 + nT rs l 'r d + jsn T
s Md
I AT =
n T r 'r l M d
1 + n Tr 'r ls d + jr nT
(668)
nT rs lM d
1 + n Tr 'r ls d + j r nT
(670)
1 + nT rs l 'r d + j s nT
I AT
n T r 'r lM d
By substituting vectors Y and U with the flux vector, respectively stator voltage, it follows that:
( I A T ) 1 =
116
s (n)
s (n 1)
U s (n)
1
1
=
(
I
T
)
+ ( I A T ) B T
=
0
'
(
n
)
'
(
n
1)
r
r
s (n 1)
1
U s (n )
= 2
0
2
2
R + I 2
'r (n 1) R + I
When the real and the imaginary parts are separated, it results that:
sR ( n)
sR ( n 1)
UsR (n)
( n)
( n 1)
U ( n)
1
1
sI
sI
=
L
+ 2
M sI
2
2
2
'rR ( n) R + I
'rR ( n 1) R + I
0
'rI ( n)
'rI (n 1)
0
(671)
(672)
(673)
117
TABLE 4.III
Method \ Real operation
Forward-difference method (Euler)
Backward-difference method
Bilinear transformation method (Tustin)
Additions
11
21
23
Multiplications
17
27
31
Delays
6
4
4
l12
l11
U
sR(n)
m11
l13
m21
sR(n)
z -1
l14
m31
l21
m41
l22
+
l23
rI(n)
z -1
l24
+
l31
m12
U
sI(n)
m22
l33
l34
+
m32
l32
z
-1
Torque
t
sI(n)
+
l41
m42
l42
z -1
+
l43
rR(n)
l44
ys (n ) = C xs (n ) + D us (n)
(674)
xs (n ) = [ x1 (n), ..., x m ( n )]
where xs (n) is the vector for the state variables, us (n) and y s(n) are the input, respectively the output
vectors, and A is the state matrix. The impulse response sequence in terms of the state-variable
description is given by the relation:
T
D , for k = 0
h( k ) =
k 1
C A , for k > 0
(675)
118
H ( z ) =
h( k ) z
= D + C ( zI A ) 1 B
(676)
k =0
For the induction machine model case, there are four real outputs (the stator and rotor fluxes)
and two real inputs (stator voltage). Also, the state variables are the outputs directly. We can
establish the notation:
sR ( n)
( n)
xs ( n) = ys ( n) = sI
rR ( n)
rI ( n)
The state matrix A, the input vector us and the coefficients matrix B, C, and D have different
element according to the transformation method used for implementing the discrete mathematical
machine model.
I) Bilinear transformation method (Tustin):
The state matrix A is calculated as follows:
c11R I
c11R R + c11I I
c
c11R R + c11I I
1
11 R I
A= 2
2
c21R I
R + I c21R R
c21R R
c21R I
where:
c12R I
c12R I
c12R I
c12R I
(677)
(678)
+ j n [ r (2 f n rs l 'r d ) s (2 f + n r 'r ls d ) ]
c12 = 4 f n rs l M d
(679)
c21 = 4 f n r 'r lM d
c22 = (2 f + n rsl 'r d ) (2 f n r 'r ls d ) + n ( rs r 'r lM d 2 + r s ) +
2
+ j n [ r (2 f + n rs l 'r d ) + s (2 f n r 'r ls d ) ]
The input vector is determined with the relation:
U ( n + 1) UsR (n)
us = sR
UsI ( n + 1) UsI (n )
The coefficients matrix are:
d11 R R + d 11 I I d 11 R I d 11 I R
1
d11 R I + d11I R d11 R R + d11 I I
B= 2
2
d 21R I
d 21 R I
R + I
d 21 R I
d 21 R I
where:
2
d11 = n (2 f + n r 'r ls d ) + j r n
d 21 = r 'r l M d n
(680)
(681)
119
(682)
C =
0 0 1 0
0 0
0 0 0 1
0 0
From the transfer function matrix relation, we obtain a 4 X 2 matrix giving the transfer function
from both inputs to the four outputs:
c11 R R + c11 I I
z 2 + 2
R
I
c11R I
2
2
R + I
1
H ( z) = 2
c R R
R + I 2
21
2
2
R + I
c21R I
2
2
R + I
c11R I
2
R + I
2
I
c11R R + c11 I I
2
R + I
c21 R I
2
R + I
c21 R R
R + I
c12 R I
c12 R I
R + I
c12 R I
R + I
c22 R R + c22 I I
2
R + I
c22 R I c22 I R
2
R + I
2
2
R + I
c12 R I
2
2
R + I
c22 R I + c22 I R
2
2
R + I
c22 R R + c22 I I
z
2
2
R + I
d11R R + d 11 I I d 11R I d 11 I R
d + d d + d
11I R
11 R R
11I I
11R I
d 21R I
d 21 R I
d 21 R I
d21R I
(683)
II) Forward-difference method (Euler)
The state matrix A is described by the relation:
nTs
nT rs l M d
0
1 n T rs l 'r d
n T rs lM d
nT s
1 n T rs l 'r d
0
A=
(684)
n T r 'r l M d
n T r
0
1 nT r 'r ls d
n T r 'r l M d
0
n T
1 n T r 'r l s d
U sR (n )
nT
0
us ( n ) =
(685-686)
B=
0
0
U sI ( n )
0
0
The other matrix C and D have the same value as for the bilinear transformation case. It results the
transfer function matrix as follows:
1
nTs
nT rl
0
z 1+nT rl
s 'r d
s Md
z 1+nT rl
0
nT rl
nT s
s r' d
s Md
H (z) =
nT r 'r lMd
0
z 1+nT r'r lsd
nT r
0
nT r 'r l Md
z 1+nT r'r ls d
nT
nT 0
0 T
n
0
0
0
0
(687)
120
g + g g + g
g12 I
g12 R
1
11 R I
11 I R
11 R R
11I I
A= 2
g 21 R
g 21 I
g 22 R R + g 22 I I g 22 R I g 22 I R
R + I 2
g 21 I
g 21R
g 22 R I + g22 I R g22 R R + g22 I I
where:
g12
nTrls Md
g11R + jg11I
1+nTr 'r lsd + j
r nT
G=
=(R jI )
=
g22R +g22I
1 +nT rls 'r d+ j
g21
nT r 'rl Md
s nT
(688)
(1+nTr'r lsd) +
I r nT + j(
R r nT I (1+nTr 'r ls d))
R nTrl
s Md j
I nTrl
s Md
= R
R(1+nT rld
R nT r'r lMd j
I nT r'r lMd
s r ) +
I s nT + j(
R s nT I (1+nT rl
s 'r d))
(689)
and:
h11R R + h11I I h11R I h11I R
B= 2
(690)
h21R
h21I
R + I 2
h21I
h21 R
where:
h12
nTrls Md
h11R + jh11I
1+nTr 'r ls d + j
r nT
H =
=
T
h22R + jh22I
1+nT rls ' r d + j
h21
nT r 'r lMd
s nT
(691)
UsR (n)
As the input vector is us =
and the coefficients matrix C and D are calculated in a similar
U s I ( n)
way to the previous cases, it results the following transfer function matrix for two inputs (voltage)
and four outputs (fluxes):
g11 R R + g11I I
g g11 I R
11R I2
z
2
2
2
R + I
R + I
g11 R I + g11I R
g + g11I I
z 11 R 2R
2
2
2
R + I
R + I
H ( z ) =
g
g
221 R 2
2 21 I 2
R + I
R + I
g 21 I
g
221 R 2
2
2
R + I
R + I
h11R R + h11 I I h 11 R I h11 I R
h21 R
h21 I
R + I
h21 I
h21 R
g12 R
2
R + I
g12 I
2
R + I
g + g 22 I I
z 22 R 2R
2
R + I
g + g 22 I R
22 R 2I
2
R + I
2
2
R + I
g12 R
2
2
R + I
g 22 R I g 22 I R
2
2
R + I
g + g22 I I
z 22 R R2
2
R + I
g12 I
(692)
121
z 1 backward-difference method
s=
z T
2 z 1 biliniar transformation method
s=
T z +1
(Tustin)
Thus the interdependence between an s-domain pole and the corresponding z-domain pole is:
z pole = 1 + s poleT forward-difference method (Euler)
1
z pole =
backward-difference method
1 s poleT
z pole =
2 + s poleT
2
2 2
2 2
1 + 2 T + T + T < 1 T < 2
+ 2
<1
z pole < 1
(1 T ) 2 + ( T ) 2
backward-difference method
(694)
1 2 T + 2 T 2 + 2 T 2 > 1 T > 2
2 + 2
z pole < 1
T > 0
(2 + T ) 2 + ( T ) 2
<1
biliniar transformation method (Tustin)
(2 T ) 2 + ( T ) 2
(695)
Expressed in terms of the sampling frequency (f sample =1/T), the above conditions are:
f sample >
2 + 2
forward-difference method (Euler)
2
2 + 2
backward-difference method
2
> 0 biliniar transformation method (Tustin)
(696)
f sample <
(697)
f sample
(698)
122
These expressions prove that the sampling rate can always be selected such that the discrete-time
system is stable if the original continuos-time system is stable as well.
The poles position may be variable due to three important factors:
The simplified assumptions for linearity of the continuous-time model;
The parameters variation due to environmental effects: temperature, humidity;
The implementation of the system using fixed-point digital signal processors. The fixed point
determines round errors and scaling or quantification errors.
So, even for poles placed inside the unit circle, there is a possibility of unstable operation of the
system. A pole location very near to the unit location can be problematic, mainly in environments
with short word length of the DSP. Therefore it may be desirable to maximise the distance between
the critical pole and the unit circle and set the sampling rate accordingly.
The derivative of the modulus of the pole gives the relations:
d z pole
dT
d z pole
dT
d z pole
dT
2 + 2 2 T + 2 2 T
1 + 2 T + 2 T
+ 2T 2
2 + 2 2 T + 2 2 T
(1 2T + T + T )
2
backward-difference method
(699)
(700)
4 T 2 ( 2 + 2 ) 4T ( 2 + 2 ) + 4
(701)
( 2 + T ) + ( T )
( 2 T ) 2 + ( T ) 2
By setting the derivative equal to zero it is obtained an expression for the optimum sampling
frequency:
2 + 2
f sample, opt =
forward-difference method (Euler)
(702)
2
+2
backward-difference method
f sample, opt =
(703)
f sample, opt
(704)
One should notice that the above sampling rate values do not optimise the resemblance between
time-domain or frequency-domain responses of the continuos-time and the discrete-time system.
The stable methods (forward-differences, backward-differences and biliniar transformation) map
an s-domain point + j into the z domain as follows:
a) Forward-difference method (Euler)
(1 m ) j
(705)
z =
j
where m=1 for the optimum sampling rate; m=2 for the minimum sampling rate
b) Backward-difference method
+ j
(706)
z =
j (1 m )
where m=1 for the optimum sampling rate; m=2 for the minimum sampling rate
c) Biliniar transformation method (Tustin)
z = 1 the unit circle,
(707)
i.e. the system is stable for any stable s-domain poles ( < 0)
123
(
= (x
la
+ k 2 (1 d ) xm
)
)i
ds
(
)
= ( x + (1 ) x ) i
+ 'd d
qs
+ 'q q
(710-713)
= 'q q + q q xr + ( q 1) xm i 'qR
Rotor voltage equations (cage rotor case):
p
'dR + k r 'qR
0 = r 'R i 'dR +
b
b
p
1
0 = k 2 r 'R i 'qR +
'qR r 'dR
b
k b
r 'R = q(d) r 'r
2
(714-716)
124
Te =
=
P xm
P 1
k ( d iqs i 'dR q idsi 'qR ) =
2 b
2 b
(717)
P 1 1
2 b k
P
xm
1
'
'
qs
dR
ds
qR
2 kb xs x 'r xm2 d
q
(718)
If the flux linkages are chosen as independent variables, the currents can be deduced from the
expressions:
qs x 'r
iqs =
1
'qR x m
q
D
1
ds x 'r
'dR x m
d
ids =
k 2D
1
'qR xs qs xm
q
i 'qR =
qD
i 'dR
(719-722)
1
'dR x s ds xm
d
=
d k 2D
where:
xs = xlm + xm
x 'r = x 'lr + x m
D = xs x 'r xm2
Recapping from the continuous linear mathematical models for the unsymmetrical single-phase
induction machine, there are three specific choices of the turn ratio q and d :
a) The rotor flux is selected as reference (inverse -form model):
q =
xm
xm + x 'r
k 2 xm
d = 2
= q
k ( x m + x 'r )
x m + x lm
xm
k 2 x m + x la
=
q
k 2 xm
If the inverse of D is noted with d, the voltage equations will be re-written in a new form:
(723)
(724)
(725)
125
1
uqs = rm x 'r d qs xm d 'qR
1 d
qs
+
b dt
1
x 'r d ds xm d 'dR
1 d
ds
+
b dt
uds =
ra
k2
(726-729))
1
1 d
1
1
0 = r 'R 2 xs d 'qR xm d qs +
'qR r 'dR
k b
b dt
1
1
1 d
0 = r 'R 2 xs d 'dR xm d ds +
'dR + k r 'qR
b dt
For null initial conditions, if the Laplace transformation is applied, it results a matrix system:
qs rm x 'r d
0
s ds
=
b 'qR r 'R xm d /
0
'dR
0
rm xm d /
0
qs uqs
ra x 'r d / k 2
0
ra xm d /( k 2 ) ds uds
+
0
r 'R xs d / 2
r /(kb ) 'qR 0
r 'R xm d /
k(r /b )
r 'R xs d / 2 'dR 0
(730)
or synthetically:
sY ( s) = A Y ( s ) + B U ( s )
(731)
Y ( s ) = C X (s ) + D U (s )
It can be observed that the state variable vector is identical to the output vector (in this case the flux
linkages per second vector). The inputs vector is the voltages vector (in this case only the stator
voltages, as the rotor is short-circuited). By identification, we obtain:
rm x m d /
0
0
rm x 'r d
2
2
0
ra x 'r d / k
0
ra xm d /( k )
A = b
2
r 'R xm d /
r /( k b )
0
r 'R xs d /
(733-734)
2
0
r 'R xs d /
r 'R xm d /
k ( r / b )
B = b
2 1 z 1
T 1 + z 1
Y ( z) = A Y (z ) + B U (z )
T z +1
which gives the relation for the inputs vector Y(z):
Y ( z) = (2 f I A) (2 f I + A) z 1 Y ( z )+ ( 2 f I A) B 1 + z1 U (z )
1
where:
(735)
(736)
126
0
b r m x m d /
0
2 f b rm x 'r d
b ra x 'r d
b ra xm d
0
2f
0
2
2
k
k
(737)
b r 'R xs d
2 f I A = b r 'R xm d
r
0
2f
2
k
r' x d
r' x d
0
b R m
k r
2 f b R2 s
The expression of the flux linkages per second in the discrete time domain is obtained when using
the bilinear transformation:
qs ( n)
qs ( n 1)
uqs ( n 1) + uqs ( n)
( n)
( n 1)
ds
= C ds
+ D uds ( n 1) + uds ( n)
(738)
'qR (n )
'qR ( n 1)
0
'dR (n )
'dR ( n 1)
C = (2 f A ) (2 f + A )
1
(739)
D = b ( 2 f A)
The above relations lead to the mathematical discrete model of the single-phase induction
machine from Fig. 5.1.
The electromagnetic torque is computed using the stator and rotor flux linkages as independent
variables:
1
P xm d
xm
P
1
(740)
Te =
'
'
( qs 'dR ds 'qR )
=
qs
dR
ds
qR
2 k b ( xs x 'r x2m ) d
2
k
q
Considering the continuous time domain and using the Laplace transformation, the rotor angular
velocity value is readily available from the torque expression:
1 P 1
(741)
r = (Te TL )
s 2 J
The currents vector can be determined according to the following matrix equation:
0
xm d /
0
iqs x 'r d
qs
i
2
2
0
x 'r d / k
0
xm d /(k ) ds
ds =
(742)
i 'qR xm d /
' qR
0
xs d / 2
0
0
xm d /( k 2 )
0
xs d /( k 2 2 ) 'dR
i 'dR
or in a condensed form:
i ( n) = F ( n)
(743)
127
c11
d11
c12
+
-1
c14
d12
-1
z
c13
c21
c22
d21
+
U
ds(n)
c23
-1
c24
d22
'
qR(n)
qs(n)
c31
-1
z
d31
c32
+
+
U
qs(n)
ds(n)
d32
c33
-1
c34
_
P
x d/( k)
2 M
b
+
+
Torque
T
e
'
dR(n)
c41
c42
d41
+
c43
-1
c44
d42
128
1
1 qs (n + 1) qs ( n)
uqs ( n) = rm x 'r d qs (n ) x md 'qR ( n) +
b
ra
1
1 ds (n + 1) ds ( n)
x 'r d ds ( n) x md 'dR ( n) +
k
T
b
(745-748)
1
1
1 'qR ( n + 1) 'qR ( n) 1 r
'dR ( n)
0 = r 'R 2 xs d 'qR ( n) xm d qs ( n) +
T
k b
b
1
1 'dR ( n + 1) 'dR ( n)
1
0 = r 'R 2 xs d 'dR ( n) xm d ds ( n) +
+ k r 'qR ( n)
T
b
which gives the matrix relation for the flux linkages per second in discrete representation:
uds ( n) =
0
rm xmdbT /
0
qs (n +1) rm x'r dbT +1
qs(n)
uqs (n)
(n +1)
2
2
u (n)
(n)
0
ra x 'r dbT / k +1
0
ra xmdbT /(k )
ds
=
ds + T ds
'qR (n +1) r 'R xmdbT /
'qR (n) b 0
rT / k
0
r 'R xsdbT / 2 +1
0
r 'R xmdbT /
krT
r 'R xsdbT / 2 +1 'dR (n)
'dR (n +1)
0
(749)
or in a condensed form:
( n + 1) = E ( n) + bT u( n)
(750)
As the flux linkages per second vector is the independent variables vector, the currents are
determined according to the following matrix equation:
0
xm d /
0
iqs x 'r d
qs
i
2
2
0
x 'r d / k
0
xm d /(k ) ds
ds =
(751)
i 'qR xm d /
' qR
0
xs d / 2
0
0
xm d /( k 2 )
0
xs d /( k 2 2 ) 'dR
i 'dR
or in a condensed form:
i ( n) = F ( n)
(752)
Similar to the bilinear transformation case, the electromagnetic torque is computed with the
relation:
1
P xm d
P
xm
1
'
'
' '
(753)
qs
dR
ds
qR =
2 k b ( qs dR ds qR )
q
2 k b xs x 'r xm2 d
For the forward-differences method (Euler), the mathematical discrete model of the single-phase
induction machine is described by the block diagram in Fig. 5.2. This model is characterised by the
following number of blocks requested for implementation: 9 additions, 15 multipliers and 6 delay
blocks. By comparison with the previous discretisation method, it is obviously the simpler
implementation structure for the forward-difference method. The computational burden is 1.8 times
greater with the bilinear transformation than it is with the forward-difference method.
Te =
129
e11
U
qs(n)
b T
z -1
z -1
e13
e22
U
ds(n)
qs(n)
b T
z -1
z-1
*
dR(n)
e24
(P/2) xMd/(b k )
+
+
Torque
-
e31
Te
ds(n)
e33
+
e34
-1
e41
*
qR(n)
e43
+
-1
e44
Y ( z) = A Y (z ) + B U ( z)
(I
A T ) Y ( z ) = z 1 Y ( z) + B T U ( z )
or in a matrix form:
qs ( n)
qs ( n 1)
uqs ( n)
ds ( n) = ( I A T )1 ds ( n 1) + ( I A T ) 1 T uds ( n)
b
'qR (n )
'qR ( n 1)
0
'dR (n )
'dR ( n 1)
0
or in a condensed form:
( n) = G ( n 1) + H u( n)
where:
(754)
(755)
(756)
130
G = ( I AT )
Tr x d
0
b m m
0
1+ b Trm x 'r d
g11 g12
b Tra x 'r d
b Tra xm d
0
1+
0
2
2
k
k
= g21 g22
=
b Tr 'r xm d
g31 g32
Tr ' x d
T
0
1 + b 2r s
r
g41 g42
Tr ' x d
Tr ' x d
0
b r m
krT
1 + b 2r s
g13 g14
g23 g24
g33 g34
g43 g44
(757)
and:
H = G bT
(758)
The matrix elements gij , i,j = 1, 2, 3, 4 expressions are not detailed here, due to the space
limitations.
Like in the previous cases, the electromagnetic torque represents the output of the discrete
system:
Te =
1
P
xm
1
qs 'dR
ds 'qR
2
2 k b xs x 'r x m d
q
P xm d
( qs 'dR ds 'qR )
=
2 k b
(759)
The discrete mathematical model obtained through the backward transformation method is
illustrated in Fig. 5.3. A comparative computational burden for different methods of discretisation is
given in Table 5.I.
TABLE 5.I
Method \ Real operation
Forward-difference method (Euler)
Backward-difference method
Bilinear transformation method (Tustin)
Additions
9
21
23
Multiplications
15
27
27
Delays
6
4
4
The following conclusions can be drawn regarding different implementation options of a discrete
mathematical model for the single-phase induction machine:
the pulse invariance method and backward-difference method determine similar discrete
mathematical model for the machine;
the step invariance method and the forward-difference method determine similar mathematical
discrete model for the machine.
the frequency characteristic of the model differs from one method to another.
the forward-difference method (Euler) presents the minimum computing time.
when accuracy is the determining factor, the bilinear transformation method (Tustin) has to be
chosen for the implementation of the mathematical discrete model of the machine.
it is also possible to apply a hybrid approach where the stator and rotor equations are discretised
using different transformation methods.
131
g12
g11
U
qs(n)
h11
g13
h31
z -1
qs(n)
g14
h21
g21
h41
g22
+
g23
z -1
qR(n)
g24
(P/2) x d / ( k )
M
b
+
+
g31
g32
h12
U
ds(n)
h32
g33
g34
+
h22
-1
Torque
T
e
ds(n)
+
g41
h42
g42
z -1
+
g43
dR(n)
g44
(760)
x s ( n ) = [ x1 (n ), ..., xm ( n ) ]
where xs (n) is the vector for the state variables, us (n) and y s(n) are the input and the output vectors
respectively, and A is the state matrix. The impulse response sequence in terms of the state-variable
description is given by the relation:
D , for k = 0
h (k ) =
k 1
C A , for k > 0
(761)
h( k ) z
= D + C ( z I A ) 1 B
(762)
k=0
For the analysed system, i.e. single-phase induction machine, there are four real outputs (the
stator and rotor flux linkages per second) and two real inputs (stator voltages). Also, the state
variables are the outputs directly. We can establish the notation:
132
qs ( n)
ds ( n)
x s ( n ) = ys ( n ) =
(763)
'qR ( n )
'dR ( n )
The state matrix A, the input vector us and the coefficients matrix B, C, and D have different
elements according to the transformation method used for implementing the discrete mathematical
model for the single-phase induction machine.
u (n ) + uds ( n 1)
us (n ) = ds
(764)
0
0
b r m x m d /
2 f + b rm x 'r d
r x' d
r x d
0
2f + b a 2 r
0
b 2a m
k
k
A = b r 'R x m d
r
b r 'R xs d
0
2f +
b r 'R x m d
b r 'R x sd
0
2f +
+ k b
2
0
0
+ b r m x m d /
2 f b rm x 'r d
b ra x 'r d
b r a x m d
0
2f
0
+
a
11
k2
k 2
= a 21
+ r
b r 'R xs d
+ b r 'R xm d
0
2f
a
31
k
2
a
b r 'R xm d
b r 'R xs d 41
0
+
k r
2f
(765)
a12
a22
a32
a42
a13
a 23
a33
a 43
a14
a 24
a34
a 44
B = b b r 'R x m d
0
r x' d
2f + b a2 r
k
0
b r 'R x m d
b r m xm d /
0
2f +
b r 'R xs d
2
+ k r
r x d
b a2 m
k
b r 'R x s d
2f +
b11
b
= 21
b31
b41
b12
b22
b13
b23
b32
b42
b33
b43
b14
b24
b34
b44
(766)
The elements of the matrix A and B require too much space to be detailed here.
133
1
0
C =
0
0
1
0
0
0
0
1
0
0
0
0
0
D=
0
0
1
0
0
0
0
(767)
We obtain the transfer function described by the 4 rows, 4 columns non-zero elements matrix:
a12
a13
z a11
a
z a22
a23
H ( z) = 21
a31
a32 z a33
a42
a43
a41
a14 b11
a24 b21
a34 b31
z a44 b41
b12
b22
b32
b42
b13
b23
b33
b43
b14 h11 ( z)
b24 h21 (z)
=
b34 h31 ( z)
b44 h41 (z)
h12 ( z)
h22 (z)
h32 (z)
h42 (z)
h13 ( z)
h23 (z)
h33 (z)
h43 (z)
h14 ( z)
h24 (z)
h34 (z)
h44 (z)
(768)
where the elements hij of the transfer function matrix may be computed through linear algebraic
computation.
The transfer function for the currents vector considered as output in relation to the voltages
vector as input can be determined as well:
xm
x
x 'r
xm
x 'r
xm
h22 ( z ) 2 h42 ( z )
2 h21 ( z ) 2 h41 ( z )
2
k
k
k
k
(769)
Hi ( z) = d
xs
xm
xs
xm
h (z)
h11 ( z )
h32 ( z )
h12 ( z )
2 31
xs
xm
xs
xm
h ( z ) 2 h21 ( z )
h42 ( z ) 2 h22 ( z )
2 41
2
k
( k )
k
( k )
The corresponding input vector will be:
u1 = uqs 1 + z 1
u1
us = where:
1
u2
u2 = uds 1 + z
(
(
)
)
2
2
0
ra x 'r d bT / k + 1
0
ra xm d bT /(k )
A=
(770)
r 'R xm d bT /
0
r 'R xs dbT / 2 + 1
r T / k
0
kr T
r 'R xs dbT / 2 + 1
r 'R xm dbT /
uqs ( n )
b T
0
us ( n ) =
(771-772)
B =
0
0
uds ( n )
0
0
The other matrix C and D have the same value as for the bilinear transformation case. The transfer
function matrix results as follows:
134
0
rm xmdbT /
0
z + rmx'r dbT 1
bT 0
2
2
0
z + ra x 'r dbT / k 1
0
ra xmdbT /(k ) 0 bT
H (z) =
r'R xmdbT /
0
0
z + r 'R xsdbT / 2 1
rT / k
0
0
r 'R xmdbT /
krT
z + r 'R xsdbT / 2 1 0
0
(773)
A detailed expression for each element of the matrix H (z) with 4 rows, 2 columns of non-zero
elements, is beyond the scope of this work. The transfer function valid for considering currents
vector as outputs, when voltages vector represents the inputs, is identical in symbolic form with the
bilinear transformation method case.
III) Backward-differences method
Similar to the cases previously analysed, the state matrix A consists the main computational burden
when determining the transfer function matrix H . By comparing the established relation for the
discrete domain model through this method, we get the following notation:
1 + b Trm x 'r d
A=
b Tr 'r xm d
Tr x d
b m m
0
1+
b Tra x 'r d
k2
1+
b Tr 'r xm d
b Tr 'r xs d
2
k r T
b Tra xm d
k 2
rT
b Tr 'r xs d
1+
a11
a
= 21
a31
a41
a12
a13
a22
a 23
a32
a42
a33
a 43
a14
a 24
a34
a 44
(774)
and the input vector us and the coefficients matrix B are:
u qs (n )
u (n )
u s ( n ) = ds
0
B = A bT
(775-776)
a 41
a12
z a 22
a32
a42
a13
a23
z a33
a 43
a14
a24
a 34
a44
a11
a
21
a 31
a 41
a12
a22
a32
a42
a13
a 23
a33
a 43
a14
a 24
a 34
a 44
(777)
The complete expressions for computing the transfer function matrix elements require intense linear
algebra computation, and therefore it is recommended to determine these values according to the
concrete machine parameters. All the considerations made for the currents vector input case are also
valid.
135
136
over the positive-slope region of the torque-speed curve, and becomes negative after maximum
steady-state torque. When the eigenvalues are complex they occur as conjugate pairs and signify a
mode of oscillation of the state variables. Negative real parts correspond to oscillations, which
decrease exponentially with time, meaning a stable condition, while positive real parts correspond
to an exponential increase with time, an unstable condition.
Usually, the starting point for the machine system analysis is from the voltage equations,
combined with the mechanical equation. So, if the input vector is formed by the stator and rotor
voltages plus the load torque, the matrix equation of the system is written as follows:
u = L px + R x = [ Lp + R] x
(783)
where [ Lp + R ] is denoted as the motional impedance matrix of the machine.
It is possible to obtain the eigenvalues by substituting the differential operator p with the roots
symbol , thus the eigen-motional impedance matrix Zm can be formed, and the characteristic
equation will be:
det [ Z m ( )] = 0
(784)
The design and analysis of controls associated with machines (i.e. vector control) require the
transfer function of the actual electrical machines, viewed as a system. Using the previous statevariable set of equations, and substituting the differential operator p with the Laplace operator s, the
input-output transfer function can be expressed as follows, considering no perturbations (z = 0):
y (s)
1
H ( s) =
= C [ sI A] B + D
(785)
u( s)
If the inputs vector is zero, the output-perturbation transfer function can be computed as:
y (s)
1
H z ( s) =
= C [ sI A] E
(786)
z(s)
6.2. Three-phase induction machine linearisation
In the concrete case of the symmetrical three-phase induction machine, the per unit version is
selected for compactness, and the input vector u is chosen to be the voltages vector. Practically, it
contains two terms for the stator voltages, and two terms for the rotor voltages expressed in twoaxis coordinate system. However, a single input variables or a linear combination of several input
variables can also be selected. In this formulation, we can express u as:
u = G ui
(787)
where G is a column matrix and ui is an input variable such as a linear combination of several
input variables i.e. the amplitude of the terminal voltage.
The perturbations vector is usually given by the variation of the load torque:
z = TL
(788)
Recapping from the d-q modelling of the three-phase induction machine, there are fourteen
possible sets of state-variables: currents, flux linkages or mixed flux linkages and currents pairs. Let
the selected pair of incremental state-space variables be denoted as x1, x2 and detailed as:
x1 = x1d + j x1q
(789)
x 2 = x 2d + j x2q
The small-signal expression for electromagnetic torque can be deduced from the large-signal torque
expression:
Te = K ( x1q x2d x1d x2q )
(790)
137
xm
for flux
D
(791)
For a complete description of the linearised induction machine model, we have to include also the
mechanical equation, expressed in small-signal form:
r
1
p
=
( Te TL )
(791)
b
2H
The starting point for the linearised state-variable models is given by voltage equations system
written in stationary reference frame:
u = L px + R x
(792)
where: x is the selected set of state-variables and represent also the output of the model, u is the
previously detailed input vector, L is the coefficients matrix (it can be formed by reactance values,
or non-dimensional elements) for multiplying the time derivative of the state-variables, R is the
coefficients matrix (it can be formed by resistances and reactances values or non-dimensional
elements) for multiplying the state-variables.
Finally, the elements for the three-phase induction machine mathematical model in linearised
form will be:
px (t ) = A x (t ) + B u(t ) + E z (t )
(793-794)
y (t ) = C x (t ) + D u(t )
where:
x( t ) = x 1q
x 1d
x2q
x2d
u( t ) = uqs
uds
uqr
udr TL
(795)
(796)
and:
A = M 1 S
B = M 1
L
M=
l 21
0(4,1)
r12
R
1 ; S =
0 (1,4) 0
2H
l 21 = K x2d0
x2q0
x1d0
(797-798)
x1q0
T
1 0 0 0 0
0 0 0 0 0
0 1 0 0 0
0 0 0 0 0
C = 0 0 1 0 0 = I 5 D = 0 0 0 0 0 = 05
0 0 0 1 0
0 0 0 0 0
0 0 0 0 1
0 0 0 0 0
The perturbations vector can be extracted from the inputs vector as:
z (t ) = TL
and:
(799-800)
(801)
138
E = 0 0 0 0
(802)
2H
The eigenvalues can be determined from the characteristic equation. The other way of
determining the roots of this equation is by using the motional matrix impedance:
det [ Z m ( ) ] = det [ M + S ] = 0
(803)
In Fig. 6.1 is illustrated the small-signal equivalent circuit of the three-phase induction machine in
stationary reference frame:
r
s
L
ls
L'
lr
'
0 dr
+
r'
'
dr0
+
u
+
i'
i qs
L
M
qs
qr
u'
rs
Lls
L'lr
'
0
qr
+
-
r' r
'
qr0
+
+
i
qr
ds
L
M
i'
dr
ds
-
u' dr
-
(805)
Te = K x1q x2d kx1d x2q
k
where K denotes a constant corresponding to the selected set of state-variables. For example:
P x
K = m if the stator and rotor flux linkages are selected as state-variables.
2 k b D
If we apply the small-displacement approach to the state-variables, it results:
I) Flux linkages models or currents models:
139
(806)
II) Mixed flux linkages and currents models (x 1 current space vector, x 2 flux linkage space
vector):
1
1
(807)
Te = K x1q0 x2d + x2d0 x1q kx1d0 x2q kx2q0 x1d
k
k
For a complete description of the linearised induction machine model, we have to include also the
mechanical equation, expressed in small-signal form:
r
P
p
=
( Te TL )
(808)
b
2J
Following are the final elements , which describe the linearised model of the single-phase induction
machine:
px (t ) = A x (t ) + B u(t ) + E z (t )
(809-810)
y (t ) = C x (t ) + D u(t )
where:
x( t ) = x 1q
u( t ) = uqs
x 1d
uds
x2q
uqr
x2d
udr TL
(811)
(812)
A = M 1 S
B = M 1
L
M=
l21
0 (4,1)
r12
R
2 J ; S =
0 (1,4) 0
P
l 21 = K x2d0
x2q0
x1d0
(813-814)
1
x1q0 or l 21 = K x2d0
k
kx2qo
kx1d0
x1q0
k
r12 = 0 0
dr0 ( x1d0 , x2d0 ) k qr0 ( x1q0 , x2q0 ) ;
k
1 0 0 0 0
0 0 0 0 0
0 1 0 0 0
0 0 0 0 0
C = 0 0 1 0 0 = I 5 D = 0 0 0 0 0 = 05
0 0 0 1 0
0 0 0 0 0
0 0 0 0 1
0 0 0 0 0
The perturbations vector can be extracted from the inputs vector as:
(815-816)
P
z (t ) = TL and E = 0 0 0 0
(817-818)
2J
The eigenvalues can be computed either from the characteristic equation or by using the motional
impedance matrix of the machine:
det [ Z m ( ) ] = det [ M + S ] = 0
(819)
In Fig. 6.2 is illustrated the small-signal equivalent circuit of the single-phase induction machine in
stationary reference frame:
140
r
m
L
lm
L'
lr
0
-
'
+
+
u
r'
i'
i qs
L
M
qs
'
dr0 (1/k)
dr (1/k)
qr
u'
ra
Lla
L'lr
'
k
0
qr
+
-
+
i
u
qr
'
k
qr0
+
2
k r'
ds
2
k L
M
i'
r
dr
ds
-
u' dr
-
141
References
[1] Adkins, B. 1957. "The General Theory of Electrical Machines", Chapman & Hall Ltd, London
[2] Alger, P.L. 1970. "Induction Machines. Their Behaviour and Uses", Second Edition, Gordon
and Breach Science Publishing, New York
[3] Blaschke, F. 1971. "Das Prinzip der Feldorientierung die Grundlage fur die Transvektor Regelung von Diehfeldmaschinen", Siemens-Zeitschrift 45, heft 10, pp. 757-760
[4] Blaschke, F. 1973. "Das Verfahren der Feldorientierung zur Reegelung der
Drehfeldmaschine", Ph.D. dissertation, University of Braunschweig, Braunschweig, Germany
[5] Blaschke, F. 1972. "The Principle of Field Orientation as Applied to the New Transvector
Closed Loop Control for Rotating Machines", Siemens Rev., Vol. 39, No. 5, pp. 217-220
[6] Bose, B.K. 1986. "Power Electronics and Drives", Prentice-Hall, Englewood Cliffs, New
Jersey
[7] Chin, T-H. 1994. "Approaches for Vector Control of Induction Motor Without Speed Sensor",
in Conf. Rec. IECON'94, Vol. 3, pp. 1616-1620
[8] Clarke, E. 1943. "Circuit Analysis of Power Systems - Vol. I, Symmetrical and Related
Components", John Wiley & Sons, New York
[9] Correa, M.B.R., et al. 1998. "Field Oriented Control of a Single-Phase Induction Motor
Drive", in Conf. Rec. Power Electronics Specialists, PESC'98, Fukuoka, Japan, Vol. II, pp.
990-996
[10] Correa, M.B.R., et al. 1999. "Single-phase Induction Motor Drives Systems", in Conf. Rec.
Annual Power Electronics Conf., APEC'99, Anaheim, USA, pp. 403-409
[11] DeDoncker, R., Novotny, D.W. 1988. "The Universal Field Oriented Controller", in Conf.
Rec. IEEE-IAS Annual Meeting, pp. 450-456
[12] DeDoncker, R., et al. 1995. "Comparison of Universal Field Oriented (UFO) Controllers in
Different Reference Frames", in IEEE Trans. Power Electronics, Vol. 10, No. 2. pp. 205-213
[13] Diana, G., Harley, R.G. 1989. "An Aid for Teaching Field Oriented Control Applied to
Induction Machines", in IEEE Trans. Power Systems, Vol. 4, No. 3, pp. 1258-1262
[14] Erdman, W.L., Hoft, R.G. 1990. "Induction Machine Field Orientation along Airgap and
Stator Flux", in IEEE Trans. Energy Conv., Vol. 5, No. 1, pp. 115-121
[15] Fitzgerald, A.E., et al. 1990. "Electric Machinery" Fifth Ed., McGraw-Hill Publishing, New
York
[16] Fortescue, C.L. 1918. "Method of Symmetrical Co-ordinates Applied to the Solution of
Polyphase Network", AIEE Transactions, Vol. 37, pp. 1027-1115
142
[17] Franklin, G.F., et al. 1997. "Digital Control of Dynamic Systems", 3rd Edition, AddisonWesley, New York
[18] Hasse, K. 1969. "Zur Dynamic Drehzahlgeregelter Antriebe Mit Stromrichter gespeisten
Asynchron-Kuzschlublaufermaschinen", Ph. D. dissertation, Technische Hochschule
Darmstadt, Darmstadt, Germany
[19] Holtz, J. 1995. "The Representation of AC Machine Dynamics by Complex Signal Flow
Graphs", in IEEE Trans. Ind. Appl., Vol. 42, No. 3, pp. 263-271
[20] Holtz, J. 1998. "Sensorless Position Control of Induction Motors - an Emerging Technology",
in Conf. Rec. AMC'98, Coimbra, Brazil, pp. 1-14
[21] Joetten, R., Maeder, G. 1982. "Control Methods for Good Dynamic Performance Induction
Motor Drives Based on Current and Voltage as Measured Quantities", in Conf. Rec. IEEE Int.
Semiconductor Power Converter Conf., Orlando, pp. 397-407
[22] Kelemen, A., Imecs, M. 1987. "Vector Control of AC Drives", G.J. Ritter, Budapest
[23] Kovacs, P.K., 1984. "Transient Phenomena in Electrical Machines", Elsevier Science
Publihsers, Amsterdam
[24] Kovacs, P.K., Racz, L. 1959. "Transiente Vorgnge in Wechselstrommaschinen", Ungarischen
Akademie der Wissenschaften, Budapest
[25] Krause, P.C., et al 1995. "Analysis of Electric Machinery", IEEE Press, New York
[26] Krause, P.C. 1965. "Simulation of Unsymmetrical 2-phase Induction Machines", in IEEE
Trans. Power Apparatus and Systems, Vol. 84, No. 11, pp.1025-1037
[27] Krause, P.C., Thomas, C.H. 1965. "Simulation of Symmetrical Induction Machinery", in IEEE
Trans. Power Apparatus and Systems, Vol. 84, No. 11, pp. 1038-1053
[28] Kuznetov, S.B. 1981. "Brushless Asynchronous Induction Machines", Ph.D. dissertation,
University of London, London, UK
[29] Lai, Y-S. 1999. "Modelling and Vector Control of Induction Machines- a New Unified
Approach", in Conf. Rec. Power Engineering Soc. Winter Meeting, Vol. I, pp. 47-52
[30] Lemaire-Semail, B., et al. 1991. "Modelling of Vector Controlled Cage Induction Motor with
FEM", in IEE Proceedings B, Vol. 138, No. 6, pp. 297-302
[31] Leonhard, W., 1985. "Control of Electric Drives", Springer Verlag, New York
[32] Levi. E. 1984. "Polyphase Motors", Wiley, New York
[33] Lorenz, R.D., et al. 1994. "Motion Control with Induction Motors", in Proceedings of the
IEEE, Vol. 82, No. 8, pp. 1215-1238
143
[34] Murata, T. et al. 1990. "Vector Control for Induction Machine on the Application of Optimal
Control Theory", in IEEE Trans. Ind. Electronics, Vol. 37, No. 4, pp. 282-290
[35] Nabae, A. et al 1980. " An Approach to Flux Control of Induction Motors Operated with
Variable-Frequency Power Supply", IEEE Trans. Ind. Appl. Vol.IA-16, No. 3, pp. 342-349
[36] Novotny, D.W., Lipo, T.A. 1996. "Vector Control and Dynamics of AC Drives", Oxford
University Press, New York
[37] Ong, C-M. 1998. "Dymanic Simulation of Electric Machinery - Using Matlab/Simulink",
Prentice Hall, Upper Saddle River, New Jersey
[38] Ohtani, T., et al. 1992. "Vector Control of Induction Motor without Shaft Encoder", in IEEE
Trans. Ind. Appl., Vol. 28, No. 1, pp. 157-164
[39] Park, R.H. 1929. "Two-Reaction Theory of Synchronous Machines Generalised Method of
Analysis - Part I", AIEE Transactions, Vol. 48, pp. 716-727
[40] Popescu, M., Navrapescu, V. 2000. "Modelling in Stationary Frame of Single and Two-phase
Induction Machines Including the Effect of Iron Loss and Magnetising Flux Saturation" - in
Proceedings of International Conference of Electrical Machines, ICEM 2000, 28-30 August,
Espoo, Finland, Vol. I, pp. 407-411
[41] Popescu, M. 2000. "Analysis and Modelling of Single-phase Induction Motor with External
Rotor for Domestic Applications" - in Proceedings of IEEE-IAS Annual Meeting, 8-12
October, Rome, Italy, Vol. I, pp. 463-470
[42] Rajashekara, K., et al. 1996. "Sensorless Control of AC Motor Drives" IEEE Press, New York
[43] Rogers, G.J. 1965. "Linearized Analysis of Induction Motor Transients", Proc. IEE, Vol. 112,
pp. 1917-1926
[44] Rosenbrock, H.H. 1963. "Some General Implicit Processes for the Numerical Solution of
Differential Equations", Computer Journal 5, pp. 329-330
[45] Salvatore, L., et al. 1998. "Vector Control of Induction Motors by Using EKF and CNF", in
Conf. Rec. PEVD'98, Power Electronics and Variable Speed Drives, Seventh International
Conference, London, UK, pp. 504-509
[46] Schauder, C. 1992. "Adaptive Speed Identification for Vector Control of Induction Motors
without Rotational Transducers", in IEEE Trans. Ind. Appl. Vol. 28, No. 5, pp. 1054-1061
[47] Slemon, G.R. 1992. "Electric Machines and Drives", Addison Wesley, Reading
[48] Slemon, G.R. 1994. "Electrical Machines for Variable-Frequency", in Proceedings of the
IEEE, Vol. 82, No. 8, pp.1123-1138
[49] Slemon, G.R. 1989. "Modelling of Induction Machines for Electric Drives", in IEEE Trans.
Ind. Appl., Vol. 25, No. 6, pp. 1126-1131
144
[50] Stanley, H.C. 1938. "An Analysis of the Induction Motors", AIEE Transactions, Vol. 57
(Supplement), pp. 751-755
[51] Thiringer, T. 1996. "Measurements and Modelling of Low-Frequence Disturbances in
Induction Machines" Ph. D. dissertation, Chalmers University of Technology, Gteborg,
Sweden
[52] Trzynaldowski, A.M. 1994. "The field Orientation Principle in Control of Induction Motors",
Kluwer Academic Publishers, Amsterdam
[53] Vainio, O., et al. 1992. "A Digital Signal Processing Approach to Real-Time AC Motor
Modeling", in IEEE Trans. Ind. Appl., Vol. 39, No. 1, pp. 36- 45
[54] Vas, P. 1990. "Vector Control of AC Machines", Oxford University Press, Oxford
[55] Vas, P. 1992. "Electrical Machines and Drives", Oxford University Press, Oxford
[56] Vas, P: 1998. "Sensorless Vector and Direct Torque Control", Clarendon Press, Oxford
[57] Veinott, C.G. 1959. "Theory and Design of Small Induction Machines", McGraw Hill, New
York
[58] Xu, X., et al. 1988. "A Stator Flux Oriented Induction Machine Drive", in Conf. Rec. Power
Electronics Specialists, PESC'88, Kyoto, Japan, pp. 870-876.
[59] Xu, X., Nowotny, D.W. 1990. "Implementation of Direct Stator Flux Oriented Control on a
Versatile DSP Based System", in Conf. Rec. IEEE-IAS Annual Meeting, pp. 437-443
[60] Xu, X. Nowotny, D.W. 1992. "Selection of the Flux Reference for Induction Machines in
Field Weakening Regin", in IEEE Trans. Ind. Appl., Vol. 28, No.6, pp. 1353-1358
[61] Zhao, Y., Lipo, T.A. 1996. "Modeling and Control of a Multi-phase Induction Machine with
Structural Unbalance" Part I, in IEEE Trans. Energy Conv., Vol. 11, No. 3, pp. 570-577
[62] Zhao, Y., Lipo, T.A. 1996. "Modeling and Control of a Multi-phase Induction Machine with
Structural Unbalance" Part II, in IEEE Trans. Energy Conv., Vol. 11, No. 3, pp. 578-584
[63] *** 1999. "Vector Control of AC Induction Motors Using the ADMC 331" Whitepaper,
Analog Devices Inc., USA