DEGREE PROJECT IN ELECTRICAL ENGINEERING,

SECOND CYCLE, 30 CREDITS

STOCKHOLM, SWEDEN 2021

Field Oriented Current Control with

Harmonic Injection in a Six-Phase
Induction Machine

XINYUE SHAN

KTH ROYAL INSTITUTE OF TECHNOLOGY

SCHOOL OF ELECTRICAL ENGINEERING AND COMPUTER SCIENCE
Field Oriented Current
Control with Harmonic
Injection in a Six-Phase
Induction Machine

XINYUE SHAN

Date: October 29, 2021

Supervisor: Yixuan Wu
Examiner: Luca Peretti
School of Electrical Engineering and Computer Science
Swedish title: Fältorienterad strömreglering med harmonisk
injektion i en sex fas asynkronmaskin
© 2021 Xinyue Shan
 Abstract | i

Abstract
As a new type of machine, the multi-phase machine has the advantages of
higher power rating, higher fault tolerance capability and more degrees of
freedom to control compared with the traditional machine. Aiming at running
a multi-phase machine with different phase-pole configurations smoothly,
a new method called harmonic plane decomposition is presented. This
project focuses on applying harmonic plane decomposition to the modelling
and analysis of a six-phase machine. The parameters of the machine were
identified in both the first and fifth harmonic planes. The machine model
using harmonic plane decomposition method is built in Matlab/Simulink. A
simulation based on the indirect rotor field oriented control of the machine is
then performed. The simulation result shows that the machine could switch
between different phase-pole configurations without causing too much speed
ripple. Debugging and testing a prototype control board is conducted. The
control strategy of the machine is tested on the testbench. The machine could
operate as a six-phase machine successfully.

Keywords
Multi-phase machine, asymmetrical six-phase induction machine, harmonic
plane decomposition, indirect rotor field oriented control, parameter estima-
tion.
ii | Abstract
 Sammanfattning | iii

Sammanfattning
Som en ny typ har flerfasmaskinen fördelarna, inklusive högre effekt, högre
feltolerans och fler frihetsgrader att styra jämfört med den traditionella maski-
nen. Med sikte på att köra en flerfasmaskin med olika faspolskonfigurationer
smidigt presenteras en ny metod som kallas harmonisk plan sönderdelning.
Detta projekt fokuserar på att tillämpa harmonisk plan sönderdelning på mo-
dellering och analys av en sexfasmaskin. Maskinens parametrar identifierades
i både det första och det femte harmoniska planet. Maskinmodellen med
harmonisk plan sönderdelning är byggd i Matlab/Simulink. En simulering
baserad på den indirekta rotorfältorienterade styrningen av maskinen utförs
sedan, och resultatet visar att maskinen kan växla mellan olika faspolskon-
figurationer utan att orsaka oacceptabel hastighetsrippling. Felsökning och
testning av en prototypstyrplatta utförs också, och maskinens styrstrategi testas
på testbänken. Sammantaget kan maskinen fungera som en sexfasmaskin
framgångsrikt.

Nyckelord
Flerfasmaskin, asymmetrisk sexfasinduktionsmaskin, harmonisk plan sönder-
delning, indirekt rotorfältorienterad styrning, parameteruppskattning.
iv | Sammanfattning
 Acknowledgments | v

Acknowledgments
This master thesis project was conducted at the division of Electric Power and
Energy Systems in school of Electrical Engineering and Computer Science at
the KTH Royal Institute of Technology in Stockholm.
I would like to give special acknowledgement to my supervisor Yixuan Wu
for giving me this opportunity to do this interesting project. He is very patient
and gave many constructive suggestions to me in all respects.
A special thank to my examiner Luca Peretti who were always following
my progress and offering me support.
Particular thanks to all the professors, doctoral and master students in the
multi-phase machine group, Sjoerd Bosga, Gustaf Falk Olson, Konstantina
Bitsi, Omer Ikram Ul Haq, Jonas Hansson and Jiale Deng. We shared our
progress every week. You were always there and gave me many useful advice
on this project.
Also sincere thanks to the staff in the Sustainable Power Laboratory,
Patrick Janus, Jesper Freiberg, and Jelena Berg, you gave me a lot of help
when I was performing the experiment.

Stockholm, October 2021

Xinyue Shan
vi | Acknowledgments
 CONTENTS | vii

Contents

1 Introduction 1
1.1 Background . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.2 Research questions . . . . . . . . . . . . . . . . . . . . . . . 2
1.3 Research Method . . . . . . . . . . . . . . . . . . . . . . . . 3
1.4 Delimitations . . . . . . . . . . . . . . . . . . . . . . . . . . 3
1.5 Structure of the thesis . . . . . . . . . . . . . . . . . . . . . . 3

2 Background 5
2.1 Vector space decomposition and harmonic plane decomposition 5
2.1.1 Short introduction of vector space decomposition . . . 5
2.1.2 Introduction to harmonic plane decomposition . . . . 7
2.2 Indirect rotor field oriented current control . . . . . . . . . . . 12
2.2.1 Flux estimator . . . . . . . . . . . . . . . . . . . . . 13
2.2.2 Current controller . . . . . . . . . . . . . . . . . . . . 14
2.3 Modulation . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
2.4 Mechanical dynamics . . . . . . . . . . . . . . . . . . . . . . 16
2.5 Speed estimation . . . . . . . . . . . . . . . . . . . . . . . . 16
2.5.1 M method . . . . . . . . . . . . . . . . . . . . . . . . 16
2.5.2 T method . . . . . . . . . . . . . . . . . . . . . . . . 17
2.5.3 M/T method . . . . . . . . . . . . . . . . . . . . . . 18
2.5.4 Discussion on speed estimation methods . . . . . . . . 18

3 Theory for multiphase machines 21

3.1 Modelling of a six-phase induction machine . . . . . . . . . . 21
3.1.1 Dynamic Model . . . . . . . . . . . . . . . . . . . . 21
3.2 Indirect rotor field oriented current control of a six-phase
machine . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
3.2.1 Current controller . . . . . . . . . . . . . . . . . . . . 22
3.2.2 Field orientation . . . . . . . . . . . . . . . . . . . . 23
viii | Contents

3.2.3 Harmonic injection . . . . . . . . . . . . . . . . . . . 23

3.2.4 From vector space to harmonic plane . . . . . . . . . 28
3.3 Parameter estimation . . . . . . . . . . . . . . . . . . . . . . 28
3.3.1 Parameters in the first harmonic plane . . . . . . . . . 29
3.3.2 Parameters in the higher-order harmonic planes . . . . 29
3.4 Possible phase-pole configurations . . . . . . . . . . . . . . . 30

4 Results and Analysis 33

4.1 Simulation results . . . . . . . . . . . . . . . . . . . . . . . . 33
4.1.1 Harmonic injection strategies . . . . . . . . . . . . . 33
4.1.2 Results from the Simulink model-switching from dif-
ferent phase-pole configurations . . . . . . . . . . . . 34
4.2 Preliminary testing of the voltage source converter . . . . . . . 34
4.3 Parameter estimation . . . . . . . . . . . . . . . . . . . . . . 36
4.4 Indirect rotor field oriented current control of the machine . . 38
4.4.1 Current controller testing . . . . . . . . . . . . . . . . 39
4.4.2 Speed controller testing . . . . . . . . . . . . . . . . . 39

5 Discussion 43
5.1 No-load current and torque of the motor . . . . . . . . . . . . 43
5.2 Current difference between each phase set . . . . . . . . . . . 44

6 Conclusions and future work 45

6.1 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . 45
6.2 Future work . . . . . . . . . . . . . . . . . . . . . . . . . . . 45

References 47
 LIST OF FIGURES | ix

List of Figures

2.1 Transformation flow principle of the vector space decomposi-

tion (VSD) approach. . . . . . . . . . . . . . . . . . . . . . . 6
2.2 The practical reference frame of the six-phase machine (YY30)
and the fundamental reference frame [10]. . . . . . . . . . . . 7
2.3 αβ reference frame for a six phase machine m = 6 applying
the HPD with mb = 6. . . . . . . . . . . . . . . . . . . . . . 12
2.4 αβ reference frame for a three phase machine m = 3 applying
the HPD with mb = 6. The peak value is identified by the
dashed magenta line. . . . . . . . . . . . . . . . . . . . . . . 13
2.5 IRFOC vector control for a three-phase induction machine. . . 13
2.6 Block diagram of the current controller. . . . . . . . . . . . . 14
2.7 Principle of M method: M1 stands for the number of pulses
from the encoder while Tc is the sampling time period. . . . . 17
2.8 Principle of M method: M2 stands for the number of pulses of
the high frequency pulse, Tt is the sampling time period, f0 is
the frequency of high frequency pulse. . . . . . . . . . . . . 18

3.1 Optimal waveform with 1st , 3rd , 5th harmonics injected. . . . . 26

3.2 Six-phase voltage source converter and six-phase winding . . . 31
3.3 Phase-pole configuration with m = 3, p = 2. . . . . . . . . . . 31
3.4 Phase-pole configuration with m = 3, p = 4 (phases are
marked with colour), phase-pole configuration with m =
6, p = 2 (phases are marked in black) and two magnetic axes
that should belong to phase A (marked with dashed red circles). 32
3.5 Winding layout for the existing machine. . . . . . . . . . . . . 32

4.1 Minimum current peak harmonic injection. . . . . . . . . . . 34

4.2 Minimum voltage peak harmonic injection. . . . . . . . . . . 35
4.3 Speed and torque curve (switch from m = 6, p = 2 to m =
3, p = 2 at t=2 s). . . . . . . . . . . . . . . . . . . . . . . . . 35
x | LIST OF FIGURES

4.4 Testbench setup. . . . . . . . . . . . . . . . . . . . . . . . . . 36

4.5 Output voltage of one phase leg output Va1 when duty ratio
D = 0.5, switching frequency f = 10 kHz. . . . . . . . . . . 37
4.6 The amplitude of six-phase current in the fundamental har-
monic plane. . . . . . . . . . . . . . . . . . . . . . . . . . . . 38
4.7 The amplitude of six-phase current in the fifth harmonic plane. 39
4.8 Current id1 with a step reference 0.5 A at t = 0.01 s. . . . . . . 40
4.9 Current iq1 with a step reference 0.3 A at t = 0.01 s. . . . . . . 40
4.10 The amplitude of six-phase current in the fundamental har-
monic plane. . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
4.11 Frequency component analysis with different speed reference-
100 rpm, 150 rpm, 200 rpm. . . . . . . . . . . . . . . . . . . 41

5.1 The fundamental frame: Standard YY30 machine (in black)

and YY26.57 machine (in red). . . . . . . . . . . . . . . . . 44
 LIST OF TABLES | xi

List of Tables

2.1 Comparison between different PWM techniques. . . . . . . . 15

3.1 Desired injection ratios for six-phase machine calculated by

(3.8) with all odd order harmonics. . . . . . . . . . . . . . . . 25

4.1 Six-phase machine parameters in the first harmonic plane. . . 37

4.2 Six-phase machine parameters in the fifth harmonic plane. . . 37
xii | LIST OF TABLES
 List of acronyms and abbreviations | xiii

List of acronyms and abbreviations

CBPWM carrier-based PWM

DFT discrete Fourier transformation

FFT fast Fourier transform

HPD harmonic plane decomposition

IRFOC indirect rotor field oriented current control

MCU microcontroller unit

MPM multiphase machine
MTPA maximum torque per ampere
MTPV maximum torque per volt

PPR pulses per revolution

PWM pulse-width modulation

SVPWM space-vector PWM

VSD vector space decomposition

WICSC machine machine with wound independently-controlled stator coils

xiv | List of acronyms and abbreviations
 Introduction | 1

Chapter 1

Introduction

This chapter gives general background information of the thesis. Furthermore,

research questions, research method and delimitations are stated.

1.1 Background
Early in 1960s, the concept of a five-phase induction machine was proposed
[1]. Multiphase machines (MPMs) have attracted more attention due to the
developments in power electronics and modern control theory since 1990s.
There are several advantages of MPMs which can be beneficial from the
industry perspective:
• MPMs enable the increase of the power rating without exceeding the
voltage limit of an affordable power supply and the permissible current
density for the conductors. Thus, a MPM is suited for a propulsion
system such as electric locomotive, electric ship and other power
applications where the power supply voltage itself is limited [2].
• MPMs can be further applied in the industry for their higher fault
tolerance capability. Different control strategies can be applied in order
to keep the machines operating despite being faulty [3].
• The increase of the phase number produces more degrees of freedom
comparing with the traditional three-phase machine.
• The potential to operate the machine with different phase-pole config-
urations, which can be used as an electrical gear box. For example,
theoretically, a six-phase, two pole pairs (m = 6, p = 2) induction
machine can operate as a three-phase machine with four pole pairs
(m = 3, p = 4), here m is the phase number while p is the pole pair
number.
2 | Introduction

In order to take full advantage of MPMs, a suitable control strategy

becomes the key enabler for applying MPMs into different scenarios, e.g.
more degrees of freedom provide the potential to inject lower-order harmonics,
which can contribute to the torque. However, the control method for
a traditional three-phase machine can not achieve this. Therefore, the
conventional control approach should be adapted to a MPM.
In [4] a YY30 induction machine was analytically modeled using VSD
in three orthogonal vector spaces, where the traditional control strategy
can be applied. However, the number of vector spaces is fixed to m/2,
which brings challenges when changing the phase-pole configuration of the
MPM. Furthermore, other novel machine typologies are invented to utilise
the phase-changing capabilities such as a machine with wound independently-
controlled stator coils (WICSC machine) [5]. In [6], [7] a novel method
called harmonic plane decomposition (HPD) was proposed to keep the phase-
changing capabilities of the MPM. The main objective of this project is to
develop and implement a suitable control strategy for a six-phase induction
machine using the HPD approach.

1.2 Research questions

It is expected that the HPD is implemented for the six-phase induction machine
and on its basis, an indirect rotor field oriented current control (IRFOC)
is realised. This control should be universal for both possible phase-pole
configurations of the induction machine in this project. Moreover, it is
expected that the HPD theory is verified by parameter estimation for two
phase-pole configurations. Finally, an analysis about different operation
strategies is undertaken with a special focus on the transition of phase-pole
configurations.
The research questions are:
• How does a cascaded control work for multiphase induction machines,
such that the additional degrees of freedom are utilised?
• Using the HPD, is it possible to unify the control for all possible phase-
pole configurations of one multiphase induction machine, such that
smooth transitions are possible?
• What operation strategies are possible in multi-phase induction ma-
chines, e.g. maximum torque per ampere (MTPA) or maximum torque
per volt (MTPV)?
 Introduction | 3

1.3 Research Method

First, a literature review was conducted focusing on different aspects of MPMs.
A model in MATLAB/Simulink was implemented and the control strategy
was tested by simulation. When the model could be controlled successfully
in Simulink, then the control strategy was implemented in an microcontroller
unit (MCU) and tested on a testbench.

1.4 Delimitations
Deeper study into parameter estimation for different phase-pole configurations
using HPD will not be included in this project. Furthermore, the base case of
the HPD method in this work will be set as six phases and two pole pairs, which
means it cannot be applied directly in a MPM with a higher phase number.

1.5 Structure of the thesis

Chapter 1 gives a short introduction about this thesis. Chapter 2 presents
relevant background information about the multiphase machine. Chapter 3
states the relevant theories that should be used in this project. Chapter 4 gives
the results including some discussions about the them. Chapter 5 presents
further discussions on some worth studying issues. Chapter 6 concludes the
whole work and states some research worthy work that can be conducted in
the future.
4 | Introduction
 Background | 5

Chapter 2

Background

This chapter provides basic background information about the control strategy
of a MPM. Additionally, this chapter describes the HPD method. Basic theory
and related work will be introduced in this chapter. In addition, a discussion
about choosing the speed estimation methods in this project are presented.

2.1 Vector space decomposition and har-

monic plane decomposition
This section gives basic information about VSD and HPD. The limitations of
VSD, i.e. the motivation for proposing HPD is also discussed in this section.

2.1.1 Short introduction of vector space decomposi-

tion
In [4], VSD was first proposed and validated experimentally using space-
vector PWM (SVPWM). Additionally, the feasibility of the fault-tolerant
control of a MPM using VSD was also verified [8]. In general, VSD is a
state-of-the-art approach for modelling and analysis of MPMs [9].

Generalised transformations for multiphase machines

Both VSD and HPD share a similar transformation flow depicted in Figure 2.1.
Comparing them with the analysis and modelling of a three-phase machine,
VSD is an extension of the traditional transformations by adding dimensions.
The matrices used in the transformations are adapted to the number of phases
m with dimension m × m. The number of the two-dimensional vector spaces
6 | Background

Practical Tabc 123 Fundamental T123 0 Stator T 0dq 0 Rotor

reference frame reference frame reference frame reference frame
xabc x123 x 0 xdq 0

Figure 2.1: Transformation flow principle of the VSD approach.

is then limited to m/2. When m is an odd number, the configuration includes

(m + 1)/2 vector spaces. However, the last vector space only represents an
axis.
In addition, one transformation is added initially, i.e. from practical
reference frame to fundamental reference frame. It is shown in Figure 2.2, any
related quantity x, e.g. voltage and current, for a six-phase YY30 machine,
can be unified to a fundamental reference frame in [0, π) range by (2.1).

x1 1 0 0 0 0 0 xa1
     
x2  0 0 0 1 0 0   xb1 
 
   
x3  0 0 −1 0 0 0   xc1 
 = ·  (2.1)
x4  0
   0 0 0 0 −1 xa2 
 
x5  0 1 0 0 0 0   xb2 
x6 0 0 0 0 1 0 xc2
| {z }
Tabc→123

The practical reference frame sorts the windings depending on the phase
order, in contrast the fundamental reference frame sorts the windings depend-
ing on their physical location. As it is depicted in Figure 2.2(a), it describes a
machine with two sets of three-phase windings phase shifted by 30 electrical
degrees. In the fundamental reference frame, as it is shown in Figure 2.2(b),
the machine is defined in electrical angle π instead of 2π. Here, the phase
shift between the adjacent magnetic axes in fundamental reference frame is
π/m [10].
Any practical winding configuration can be transformed using this trans-
formation with different matrix elements depending on the configuration, to
a generic fundamental reference frame [10]. It enables a generalised analysis
method for a MPM. All type of m phase machines can be modelled and anal-
ysed in the stator reference frame using a generalised Clarke transformation
matrix with dimension m × m. Further information regarding the generalised
Clarke transformation will be presented in the next section.
 Background | 7

c2 x4
b1 c1 x5 x3
b2 a2 x6 x2

a1 a1
x1

a2 b2

c1 c2 b1

(a) Practical reference frame of an asymmet- (b) Fundamental reference frame

ric six-phase machine (YY30)

Figure 2.2: The practical reference frame of the six-phase machine (YY30)
and the fundamental reference frame [10].

Limitations
The existing MPM used in this project is a six-phase YY26.57 machine with
two pole pairs (m = 6, p = 2) [11]. Another phase-pole configuration can
be achieved, i.e., the machine can operate as a three-phase, two-pole-pair
machine (m = 3, p = 2) by controlling same current through two adjacent
phases (a1 , a2 , and b1 , b2 , and c1 , c2 ). Additionally, if the machine had
individually controlled solenoidal windings, this machine can also operate as
a three-phase four-pole-pair machine (m = 3, p = 4). Further discussion
about the feasibility of different phase-pole configurations will be presented
in Chapter 3. Shifting between different phase-pole configurations is difficult
when using VSD, since the dimension of the matrices need to change. This
kind of change brings discontinuity during the current control process [6], [7].
In [12], it shows that a transition from six-pole operation to two-pole operation
need approximately 0.5 s with VSD approach. However, using HPD method,
the transition time is smaller than this value according to the simulation work
in this project.

2.1.2 Introduction to harmonic plane decomposition

Base case
The limitations mentioned in section 2.1.1 can be overcome by introducing
another transformation approach called HPD [6], [7]. In this method, the
8 | Background

dimension of the matrices depends on a base case instead of the actual

phase number of the configuration. The base case will represent one phase-
pole configuration, which resolves the smallest winding unit of the machine.
Theoretically, the base case can be set as mb = Qs /2, pb = 1, where Qs is
the number of stator slots. In this situation, each coil can be considered as
an independent phase with its own magnetic axis. The modelling is based
on the physical structure of the machine instead of the electric quantities.
Therefore, all other configurations can be derived from the base case. With
this selection of the base case, machines like the WICSC machine can operate
within different phase-pole configurations freely.
The phase number of the base case is related to the possibility of
controlling the coils. The existing machine in this project is supplied by a
six-phase voltage source converter. Therefore, the phase number of base case
in this project is limited to six since only "six phase" voltages are controllable.

Generalised Clarke transformation

Quantities such as phase current and voltage are essential to the modelling
and control of the electrical machine. Similar to the analysis of a three-
phase machine, these quantities for a MPM can also be analysed in the stator
reference frame or rotor reference frame. For a YY30 induction machine,
the VSD approach generates three two-dimensional vector spaces, which are
perpendicular to each other. This can be achieved by a generalised Clark
transformation [10]. Field-oriented control strategies for a conventional three-
phase machine can be applied in each vector space.
The transformation between the stator reference frame and fundamental
reference frame, i.e. the generalised Clarke transformation, is given in
(2.2). K is the coefficient of the transformation, for an amplitude invariant
transformation, K = 1; for a power invariant transformation, K = 1/2. ξ is
the maximum odd number which is smaller than or equal to m. δ is the phase
shift between the adjacent magnetic axes in the 123 fundamental reference
frame. x stands for any interested space-vector quantity in different reference
frame.
 K
2
xαβ = Cx123
m
 1−K
2
x123 = CT xαβ
m
 Background | 9

 
1 cos(δ) cos(2δ) . . . cos((m − 1)δ)
0 sin(δ) sin(2δ)
 . . . sin((m − 1)δ) 
1 cos(3δ) cos(6δ)
 . . . cos((m − 1)3δ)
C = 0 sin(3δ) sin(6δ)
 . . . sin((m − 1)3δ) 
 .. .. .. .. ..

. . . . .


 
1 cos(ξδ) cos(2ξδ) . . . cos((m − 1)ξδ)
0 sin(ξδ) sin(2ξδ) . . . sin((m − 1)ξδ)
π
δ= (2.2)
m

Harmonic plane
VSD generates m/2 vector spaces for a m phase machine [4]. For a six-phase
machine, three vector spaces H = {ν1 , ν3 , ν5 } are created, while a three-phase
machine comprises of {ν1 , ν3 }. When using the HPD approach, the dimension
of the matrices will not change with different phase-pole configurations. Thus,
a notation called harmonics plane, h, is introduced. It can be seen as the vector
space created in the base case. Thus, for the base case, HPD is identical to
VSD.
With mb = 6, pb = 1 in (2.2), harmonic planes h ∈ H are generated. If
the machine is configured as m = 3, two vector spaces {ν1 , ν3 } will distribute
over three harmonic planes. This distribution can be expressed in (2.3). Here,
pb is the pole pair number of the base case, while p is real pole pair number of
the machine.
p
h = (νh ± mq) , ∀ νh = {1, 3, . . . , m}, q = {0, 2, 4, . . .} (2.3)
pb

The vector-space distribution in (2.3) gives the relation between vector

spaces νh of the actual phase configuration and the vector spaces for the
base case, i.e. harmonic planes h. However, this distribution is different
from harmonic mapping mentioned in [13]. Harmonic mapping describes the
relation between time harmonics and the vector spaces νh .

Generalised Clarke transformation & discrete Fourier transformation

The generalised Clarke transformation in (2.2) is identical to a discrete Fourier
transformation (DFT) of the space-vector quantity. In each harmonic plane h,
10 | Background

(2.2) with K = 1 can be reformulated by (2.4):

mb −1
2 X
xα,h (t) = xk+1 (t) cos(khδ)
mb k=0
mb −1
(2.4)
2 X
xβ,h (t) = xk+1 (t) sin(khδ)
mb k=0
T
x123 (t) = x1 (t) x2 (t) . . . xmb (t) can be seen as the instantaneous


value of quantity x at time instant t. Space vector X αβ,h can be represented

in complex form in (2.5). Using (2.4) and (2.5), (2.6) can be derived, and it is
equivalent to the Fourier transformation of x123 .

X αβ,h = xα,h + j · xβ,h = Xαβ,h ejϕh (2.5)

mb −1
2 X
 
jπkh
X αβ,h = xk+1 (t)e mb
(2.6)
mb k=0
Furthermore, from the polar form, there is a phase shift ϕh caused by the
DFT. This phase shift can be interpreted as the phase shift between the vector
space and the harmonic plane. This phase shift should be considered in the
Park transformation Tαβ→dq in (2.7) with base case mb = 6.

cos(θ1 ) sin(θ1 ) 0 0 0 0
 
− sin(θ1 ) cos(θ1 ) 0 0 0 0 
 
 0 0 cos(θ3 ) sin(θ3 ) 0 0 
Tαβ→dq =
 
 0 0 − sin(θ 3 ) cos(θ 3 ) 0 0 

 0 0 0 0 cos(θ5 ) sin(θ5 ) 
0 0 0 0 − sin(θ5 ) cos(θ5 )
θ1 = ν1 ωt + ϕ1
θ3 = ν3 ωt + ϕ3
θ5 = ν5 ωt + ϕ5
(2.7)

When the amplitude of the interested component x(t) is 1, the amplitude

of the complex factor Xαβ,h can be described in (2.8) [6], [7]. It is identical to
the winding factor of an AC machine [14]. It represents the scaling factor of
 Background | 11

the space vector that will appear in the harmonic plane h.

sin(hqs α2 )
Xα,β,h = kd,h = (2.8)
qs sin(h α2 )

Where
- qs = Qs /(2mp) is the number of slots per pole per phase;
- α = 2πp/Qs is the electrical angle per each slot.

Case study for a six-phase machine

In this section, the HPD is demonstrated by means of simulation for the two
configurations of the machine (three-phase and six-phase machine) with base
case mb = 6 under investigation in this project.
The existing machine is a six-phase machine, possible phase-pole configu-
rations are further discussed in section 3.4. In this project, the machine will be
operated as both six-phase and three-phase motor. For both configurations, a
set of phase harmonic currents with amplitude of 1 A for each odd harmonic is
applied. For a six-phase machine, the applied phase current for phases a1 , b1 ,
c1 , a2 , b2 , c2 can be expressed by (2.9). The fundamental frequency is chosen
as 50 Hz.
5
X
iν = sin(ν · (wt − φph ))
ν=1,3,5
  (2.9)
2π 4π π 5π 9π
φph = 0
3 3 6 6 6

However, for a three-phase machine, only fundamental harmonic is

applied. The phase current for phases a, b, c therefore becomes:

i1 = sin(wt − φph )
(2.10)
 
2π 4π
φph = 0
3 3

Figure 2.3 shows the αβ reference frame for a six-phase machine with
mb = 6. The peak value of the current in each harmonic plane is 1 A. In
addition, there is also no phase shift caused by DFT. The result shows again
that for the base case, HPD and VSD are identical.
Figure 2.4 shows the αβ reference frame for a three-phase machine with
mb = 6. The peak value of the current in each harmonic plane is no longer
12 | Background

Harmonic plane h=1

1

Current [A]
0

-1
0 0.005 0.01 0.015 0.02 0.025 0.03 0.035 0.04
Time (s)
Harmonic plane h=3
1
Current [A]

-1
0 0.005 0.01 0.015 0.02 0.025 0.03 0.035 0.04
Time (s)
Harmonic plane h=5
1
Current [A]

-1
0 0.005 0.01 0.015 0.02 0.025 0.03 0.035 0.04
Time (s)

Figure 2.3: αβ reference frame for a six phase machine m = 6 applying the
HPD with mb = 6.

1 A, and the phase shift caused by DFT appears. The fundamental harmonic
appears not only in the first harmonic plane h = 1, but also in the fifth
harmonic plane h = 5 as a negative sequence, which is in coherence with
(2.3).

2.2 Indirect rotor field oriented current con-

trol
Indirect rotor field oriented current control (IRFOC) is a vector control method
where the flux estimator is implemented in dq reference frame [15]. MPMs
provide more degrees of freedom in comparison to three-phase machines.
Thus, the control strategy for MPMs is slightly different from traditional three-
phase machines. However, the control method for a three-phase machine is
a good start. Figure 2.5 describes the principle of IRFOC for a three-phase
induction machine stated in [15]. Current model is used for the flux estimator.
Furthermore, speed-sensored control is used in this project so that both current
and speed are measured in order to provide necessary data for controlling the
machine.
 Background | 13

Harmonic plane h=1

0.9659

Current [A]
0

-1
0 0.005 0.01 0.015 0.02 0.025 0.03 0.035 0.04
Time (s)
Harmonic plane h=3
1
Current [A]

-1
0 0.005 0.01 0.015 0.02 0.025 0.03 0.035 0.04
Time (s)
Harmonic plane h=5
1
Current [A]

0.2588
0

-1
0 0.005 0.01 0.015 0.02 0.025 0.03 0.035 0.04
Time (s)

Figure 2.4: αβ reference frame for a three phase machine m = 3 applying the
HPD with mb = 6. The peak value is identified by the dashed magenta line.

idref , iqref Vref dq Vrefs

Current 2 PWM
Controller  3
id , iq
dq
2 
IM
 3 ~
Flux 1
Estimator iabc

r

Figure 2.5: IRFOC vector control for a three-phase induction machine.

2.2.1 Flux estimator

The flux of an induction machine is difficult to measure directly. Therefore,
in this project, a flux estimator based on the rotor equation of the induction
machine is implemented in the dq frame. Assuming that rotor flux Ψ̂R is real,
i.e. perfect field orientation, Ψ̂R = ψ̂R , the equations of the current model
14 | Background

 

1/ k p
 
ki / s
ref
idq 
 e 
PWM

ref ref
vdq

vdq
kp 
, sat

E
idq
j Lˆ  Ra

Figure 2.6: Block diagram of the current controller.

in a three-phase machine are given by (2.11). Here, “hats” are introduced to

represent the estimated motor parameters, and estimated rotor flux.

dψ̂R R̂R
= R̂R id − ψ̂R
dt L̂M
(2.11)
R̂R iq
ω2 = ω1 − ωr =
ψ̂R

As it is depicted in Figure 2.5, current references iref

d and iq
ref
are used instead
of the measured current. This reduces noise as compared to the measured
currents.

2.2.2 Current controller

As shown in Figure 2.6, a PI controller is installed in order to control the
current in dq frame [15]. The block "PWM" includes the saturation caused by
the voltage limitation of the voltage source converter. Thus, an anti-windup
method so called "back calculation" is used to prevent the wind-up phenomena
in the controller integrator and improve the performance. Furthermore, the
active resistance Ra makes the controller faster by moving the pole location.
The feed-forwarding back EMF can improve disturbance rejection.
 Background | 15

Table 2.1: Comparison between different PWM techniques.

PWM techniques CBPWM SVPWM
Complexity for implementation simple medium
Extension to higher number of phases and level easy hard

The gain of the PI controller can be selected by:

kp = αc L̂σ
(2.12)
ki = αc R̂s + Ra

where αc = ln(9)/tr is the bandwidth of the controller. tr is the desired rising

time. The value of the active resistance can be set by Ra = αc L̂σ − R̂s .

2.3 Modulation
Carrier-based PWM (CBPWM) and SVPWM are two typical modulation
methods for electrical drives. Five different pulse-width modulation (PWM)
techniques, i.e. three CBPWM methods and two SVPWM methods for a five-
phase voltage source inverter were compared in [16] and [17] with simulations
and experiments, respectively. Although different approaches were used,
similar characteristics are depicted in Table 2.1. A two-level six-phase votlage
source converter is used in this project [11]. When using SVPWM, the
number of the switching states of the converter is 26 = 64. 64 voltage
vectors are simultaneously mapped into three planes. Selecting the vectors and
calculating the time of application of each selected vector brings complexity
when dealing with multiphase converters [9].
Similar to three-phase PWM, for CBPWM, the fundamental phase voltage
can also be increased by injecting zero-sequence harmonics. In [18], [19], both
CBPWM with fifth harmonic injection and SVPWM for a five phase voltage
source converter were developed. Same increase of 5.1% in the maximum
fundamental output voltage could be obtained with both methods.
CBPWM with harmonic injection can achieve a better utilisation of
voltage source converters without introducing extensive complexity. Here,
it is worth mentioning that the objective of injecting zero-sequence is to
increase the utilisation ratio of the voltage source converter, while harmonic
injection for torque enhancement is to inject harmonics that are related to the
electromagnetic energy conversion. Further discussion will be presented in
16 | Background

Chapter 3.

2.4 Mechanical dynamics

The following equation represents the mechanical dynamics of the machine:

dωr
τe = J + Bωr + τL (2.13)
dt
where
- J is the moment of inertia;
- ωr is the mechanical speed of the rotor;
- Bωr stands for the viscous friction torque;
- τL represents an external load torque;
- τe is the electrical torque.

2.5 Speed estimation

In most motor control applications, the accuracy of speed measurement plays
a vital role in the speed control loop. In the previous project [11], the
speed estimator did not work as expected. Therefore, the speed estimation
was rewritten in this project. The encoder used in this project is an
incremental encoder from LTN called GA3WSLDBI-3600-5N3-05R2 [11]. In
this section, basic information of different speed estimation methods including
M method, T method, M/T method based on photoelectric encoder for speed
measurement were introduced. Their advantages and drawbacks are presented
and compared. In the end, a suitable method for this application was chosen,
and the reason behind was stated.

2.5.1 M method
Figure 2.8 describes the principle of M method. This method measures the
number of pulses M1 from the encoder within a certain time period Tc to
calculate the speed [20].
Assuming that pulses per revolution (PPR) of the encoder is Nppr , the
speed n can be calculated by (2.14).

60M1
n= (2.14)
Nppr Tc
 Background | 17

Pulse from
the encoder

Sampling
pulse

Tc

Figure 2.7: Principle of M method: M1 stands for the number of pulses from
the encoder while Tc is the sampling time period.

The maximum error of M method is one pulse, so the maximum error

can be calculated using (2.14) with M1 = 1. Therefore, the error rate of this
method is:
60
Nppr Tc 1
e = 60M1 × 100% = × 100%. (2.15)
Nppr Tc
M1
The error rate of this speed measurement method is inversely proportional
to the number of pulses. The higher the speed, the greater the value of M .
When the speed is very low, the value of M is small and the error rate will
increase. Therefore, the M method is suitable for high-speed measurement.

2.5.2 T method
T method is to measure the time interval between two pulses of the encoder to
calculate the speed, also known as the period method [20]. In actual use, the
time interval Tt between two pulses of the encoder is calculated by the number
of high-frequency clock pulses M2 .
Therefore, the speed can be calculated by (2.16).

60 60f0
n= = (2.16)
Nppr Tt Nppr M2

The resolution Q of M method is a constant value, while the resolution of this

T method is related to the speed, as can be seen in (2.17). This method is
18 | Background

High
frequency
pulse

M2

Pulse from
the encoder
Tt  M 2 / f 0

Figure 2.8: Principle of M method: M2 stands for the number of pulses of the
high frequency pulse, Tt is the sampling time period, f0 is the frequency of
high frequency pulse.

suitable for low speed application.

60f0 60f0 Nppr n2

Q= − = (2.17)
Nppr (M2 − 1) Nppr M2 60f0 − Nppr n

2.5.3 M/T method

In order to balance the accuracy in both high speed and low speed range, the
M method and T method are often used in combination, which is called M/T
method. This method could give high accuracy at both high and low speed.

2.5.4 Discussion on speed estimation methods

The MCU receives three signals from the encoder, two orthogonal pulse
signals A and B and one index pulse signal Z. Generally the rising and falling
edges of signal A and signal B are collected. Therefore, in this project
Nppr = 4 ∗ 3600 = 14400.
If the sampling frequency of the M method is same as the switching
frequency of the converter 10 kHz, the resolution of M method will be
41.67 rpm. This value is relatively large and not suitable for a speed controller,
even for the high speed application-in this project the synchronous speed of
the machine is 300 rpm. This method has an error rate of 13.9% at this speed.
Therefore, in this project, the sampling frequency is 100 Hz, which could gives
a maximum error 0.416 rpm using M method.
In the real application, the MCU reads the number of the pulses every
PWM period, which is every 0.1 ms, and an array will store the number of
 Background | 19

pulses from the first one to the 101st . Then the number of pulses M1 in (2.14)
can be calculated by M1 = M101st − M1st . Similarly, the second M1 after
0.1 ms will be M1 = M102nd − M2nd .
As stated above, the resolution of M method is 0.416 rpm. If T method is
used, it can gives a extremely high resolution (smaller than 0.001 rpm) at low
speed, but an low resolution of 2.4 rpm at 300 rpm when the high frequency
pulse is the system clock of the MCU. The best way to get a lower error rate is
to use M/T method. However, in this project, the machine will not run at very
low speed constantly. Although this error may have some effect on dynamic
performance of the control system when starting the machine, it is acceptable.
Therefore, here M method is used.
20 | Background
 Theory for multiphase machines | 21

Chapter 3

Theory for multiphase machines

This chapter shows the procedure of modelling the six-phase induction

machine using HPD. The control scheme is therefore introduced based on
the HPD method. The difference between the control strategy of three-phase
and six-phase machine is emphasized. In addition, the possible phase-pole
configurations for the existing machine are stated.

3.1 Modelling of a six-phase induction ma-

chine
3.1.1 Dynamic Model
For a six-phase induction machine, VSD creates three vector spaces ν = 1,
ν = 3, and ν = 5. In the previous study of MPMs, normally, it is assumed
that the air gap flux is perfectly sinusoidal [2]. In this case, higher-order spatial
harmonics of the air gap flux do not exist. Therefore, the coupling between the
stator and rotor only exists in ν = 1 plane, where the torque is produced. The
currents in ν = 3 plane can be considered as stator-circulating currents, which
will not produce any torque and flux but generate copper losses and iron losses
in the stator. A single neutral point of the two sets (a1 b1 c1 , a2 b2 c2 ) allows
i0+ = i0− , which means that the zero-sequence equations of the machine will
appear.
This kind of assumption makes the model equations of a six-phase machine
in ν = 1 plane and the torque production equation identical as for a
traditional three-phase machine. However, it is only valid when both the
stator and the rotor windings are distributed sinusoidally with appropriate
design. If sinusoidal windings are assumed, there is no potential of the torque
22 | Theory for multiphase machines

enhancement of the MPM. In reality, the coupling between the stator and rotor
will appear in all harmonic planes. Thus, the inverse-Γ model for a six-phase
induction machine with HPD method (h = 1, 3, 5) is:

diss,h dΨsR,h
Lσ,h s s
= vs,h − Rs,h is,h − (3.1)
dt dt

dΨsR,h
 
RR,h
= RR,h iss,h − − jhωr ΨsR,h . (3.2)
dt LM,h
The electromagnetic torque is therefore contributed by all harmonic planes
instead of one:
mb X
τe = pb h={ΨsR,h ∗ iss,h } (3.3)
2 h

where
- mb is the phase number of the base case, here mb = 6;
- pb is the pole pair number of the base case, here pb = 2;
- h represents the order of the harmonic planes, here h = 1, 3, 5.

3.2 Indirect rotor field oriented current con-

trol of a six-phase machine
The control scheme for a six-phase induction machine is an extension and
adaption of the conventional control strategy. The relation between harmonic
planes and vector spaces in (2.3) will bring complexity to the optimisation
problem when computing the desired current/voltage references. Therefore,
this section starts with the changes to be made from the traditional IRFOC to
the adapted control scheme using VSD. In the end, the control scheme will be
transferred from the VSD to he HPD method.

3.2.1 Current controller

As stated before, in order to enhance the torque production of the machine,
current in higher-order harmonic planes should also be controlled. Therefore,
the current controller is adapted to control the currents in h = 1, 3, 5
planes. Three equal controllers described as in Figure 2.6 are used. The
controller parameters are different and depend on the machine parameters in
the corresponding harmonic planes.
 Theory for multiphase machines | 23

3.2.2 Field orientation

As it is stated in section 2.2, indirect rotor field orientation is achieved by
estimating the rotor flux linkage and slip using the current model. However,
when applying VSD, each vector space is independent from each other while
the synchronous speed of the components in higher-order vector space is
related to the fundamental vector space. This relation is stated in (3.4).

ω1,ν = ν · ω1
(3.4)
ω2,ν = ν · ω2

This relation reduces the degrees of the freedom of the flux estimator for a
MPM. In other words, the slip relation in (2.11) cannot be used directly for
higher-order vector spaces. The flux estimator becomes (3.5):

dψ̂R,ν R̂R,ν
= R̂R,ν id,ν − ψ̂R,ν
dt L̂M,ν
(3.5)
R̂R,ν iq,ν
ω2,ν = ν(ω1,1 − ωr ) = νω2,1 =
ψ̂R,ν

Here, it is assumed that the components in the fundamental plane are used for
slip estimation. The second equation (3.5) reveals the relation between iq,ν and
ω2,1 . The q-axis currents in higher-order vector spaces are not independent but
coupled with the components in the fundamental vector space and the d-axis
current.

3.2.3 Harmonic injection

As it is stated in section 3.1.1, when neglecting the coupling inductance in the
higher-order vector spaces, the energy conversion appears to be zero in higher-
order vector spaces. In order to minimise the copper losses in the stator, the
desired current reference values in ν = 3 vector space of a MPM are always
zero. However, if the coupling inductance in higher-order vector spaces are not
zero, currents there have the potential to take part in the torque production. By
setting the current references iref
d,ν , iq,ν accordingly, these potentials are used.
ref

Injection ratios
It is to note that here harmonic injection is to be distinguished from zero-
sequence injection in order to improve the DC bus utilisation of a voltage
24 | Theory for multiphase machines

source converter. For example, for a five-phase voltage source converter,

5th harmonic is injected as a zero-sequence harmonic to extend the linear
modulation range by 5.15% [16]. However, in [21], both fundamental and
third harmonic are injected to a five-phase machine for torque enhancement.
This kind of harmonic injection is what will be discussed later.
Although the aim of these harmonic injection approaches are different, the
same analysis method can be used to shape the component to a desired wave-
form. [19] gives an analytic solution for injecting zero-sequence component
in a voltage source inverter. The reference leg voltage for phase a is given as:
∗
Vao = M1 0.5Vdc cos(ωt) + M5 0.5Vdc cos(νωt). (3.6)

Here, M1 , M5 are the injection ratios, Vdc is the DC bus voltage. One
higher-order harmonic is injected to expand the linear modulation range,
i.e. increase M1 . Thus, the injection ratio M5 is chosen to ensure that the
maximum value of the reference voltage Vao ∗
occurs when the higher-order
harmonic reaches its negative peak value. This relation can be expressed by
(3.7).
∗
dVao
=0
d(ωt) (3.7)
∗
|Va0 (ωt)| = 0.5Vdc

By solving (3.7), a solution for the injection ratios with one higher-order
harmonic injection can be obtained. However, for a MPM with m phases, in
principle m/2 harmonics can be injected, for example, for the existing six-
phase machine, three harmonics (1st , 3rd , 5th ) can be injected. The injection
ratios are not analytically solvable. The first step for this harmonic injection
is to compute the injection ratios for an optimal waveform for minimum peak
injection. A optimal problem can be used to determine the ratios:
X
max Mν2
ν
s.t. 0 ≤ θ ≤ π/2
(3.8)
M1 ≥ 1
X
M1 cos(θ) + Mν cos(νθ) ≤ 1.
ν

Similar to (3.7), the constraints of this optimisation problem ensure that the
fundamental component is larger than 1, in the meantime, the peak value the
 Theory for multiphase machines | 25

Table 3.1: Desired injection ratios for six-phase machine calculated by (3.8)
with all odd order harmonics.
Order of harmonics Values
1st 1.2071
3 rd
-0.2330
5th 0.0614

component is smaller or equal to one. The goal of this optimisation is to

maximise the peak value of each harmonic component.
Table 3.1 and Figure 3.1 show the calculated injection ratios and desired
waveform of the component in time domain for a six-phase machine.
If the machine operates as a three-phase machine, only fundamental
component is possible (M1 = 1).

Minimum current peak optimisation

After having computed the injection ratios, the next step is to define the current
reference in dq frame for each vector space to fulfill the injection ratios.
In steady state, by substituting the flux ψd,ν with LM,ν and id,ν , the slip
relation becomes:
RR,1 iq,1 RR,1 iq,1
ω2,1 = = ν=1
ψR,1 LM,1 id,1
(3.9)
RR,1 LM,ν id,ν
iq,ν =ν iq,1 ∀ν > 1.
RR,ν LM,1 id,1

The voltage in dq reference frame for each vector space in steady state can
be expressed by (3.10).

vsd,ν = Rs,ν id,ν − ω1,ν Lσ,ν iq,ν

(3.10)
vsq,ν = Rs,ν iq,ν − ω1,ν Lσ,ν id,ν + ω1,ν LM,ν id,ν

Therefore, the input electrical power is:

m X
Pel,in = p (vsd,ν id,ν + vsq,ν iq,ν ). (3.11)
2 ν

The aim of the minimum current peak optimisation is to shape the stator
current to a desired curve, as it is depicted in Figure 3.1. Within the same
current peak limitation, the RMS value of the optimal waveform is larger
26 | Theory for multiphase machines

Figure 3.1: Optimal waveform with 1st , 3rd , 5th harmonics injected.

than a sine wave. The current reference values iref

d , iq
ref
are determined by
an optimisation problem (3.12):

min Pel,in
m X
s.t. Te = p LM,ν id,ν iq,ν
2 ν
i2d,ν + i2q,ν (3.12)
Mν2 = 2
id,1 + i2q,1
R̂R,1 L̂M,ν îd,ν
iq,ν = ν iq,1 .
R̂R,ν L̂M,1 îd,1

The objective of this optimisation problem is to minimise the input electrical

power, while fulfilling the torque requirement. The input variables of this
optimisation function are the desired electric torque Te and the rotor speed ωr .
The speed and torque give a fixed value of the output power. To minimise the
losses in the machine is equivalent to minimise the input power. The second
constraint gives the injection ratios of each harmonic component. The last
constraint describes the coupling of the q-axis current in higher-order and
fundamental vector space as derived in (3.5).
The synchronous dq reference frame is aligned with the rotor flux linkage.
In order to yield the optimal waveform in Figure 3.1, not only the injection
ratios, but also the relative angle between the harmonics space vectors must be
controlled. This is equivalent to controlling the phase angle of each harmonic
plane. In (3.13), the transformation angle for the Park transformation is given.
The fundamental current space vector is used here as a reference and all other
 Theory for multiphase machines | 27

harmonic space vectors are shifted such that they have the correct phase angle
depending on the sign of the according injection ratio.

0 π
θ1,ν = νθ1,1 − θ1,ν − (sgn(Mν ) − 1)
2 (3.13)
θ1,ν = atan2 (iq,ν , id,ν )

Here, atan2(x, y) returns the arc-tangent of y/x. The sign of the values of
y and x determines the correct quadrant. If both x and y are negative, then
atan2(x, y) = arctan(y/x) − π.
When using Park Transformation mentioned in (2.7), these calculated
values should be then added together after transforming the current reference
value from vector spaces to harmonic planes. Therefore, the phase shift
contains two parts: the phase shift caused by DFT using HPD and the the phase
shift caused by the harmonic injection. The last three equations describing the
phase angle in Park Transformation in (2.7) are modified by:
0
θ1 = ν1 ωt + ϕ1 + θ1,1
0
θ3 = ν3 ωt + ϕ3 + θ1,3 (3.14)
0
θ5 = ν3 ωt + ϕ5 + θ1,5 .

Minimum voltage peak optimisation

In the field weakening region, the higher speed will let the machine reach
its voltage limit. This limit is caused by DC bus voltage. Here, another
optimisation called minimum voltage peak optimisation is proposed. The
optimal waveform is applied to the voltage.
The optimisation problem for this method is formulated in (3.15). The
objective function remains the same to minimise the input power. The torque
production function and the slip relation also remain unchanged as (3.12). The
injection ratios are now applied to the dq frame voltages as stated before.

min Pel,in
m X
s.t. Te = p LM,ν id,ν iq,ν
2 ν
2 2
vd,ν + vq,ν (3.15)
Mν2 = 2 2
vd,1 + vq,1
R̂R,1 L̂M,ν îd,ν
iq,ν = ν iq,1
R̂R,ν L̂M,1 îd,1
28 | Theory for multiphase machines

Similarly, the compensated Park transformation angle caused by this

injection:

0 π
θ1,ν = νθ1,1 − θ1,ν − (sgn(Mν ) − 1)
2 (3.16)
θ1,ν = atan2 (vq,ν , vd,ν ) .

3.2.4 From vector space to harmonic plane

Two phase-pole configurations were tested in this project: six-phase two pole-
pairs (m = 6, p = 2) and three-phase two pole pairs (m = 3, p = 2).
Further information is provided in section 3.4. HPD is used to switch from
one phase-pole configuration to another. However, the harmonic injection
methods described in the previous sections are based on VSD. The calculated
d,ν , iq,ν are then transformed to id,h , iq,h . When the machine
reference values iref ref ref ref

is operated as a six-phase machine, the VSD generates three vector spaces

{ν1 , ν3 , ν5 }, which are identical to three harmonic planes with base case
mb = 6. According to (2.3), in m = 3 case, vector space ν1 distributes over
harmonic planes h = 1, 5. The component in the first vector space appears as
a negative sequence in the fifth harmonic plane h = 5. Therefore, the current
reference in HPD is given by (3.17).
(
iref = iref
d,ν d,h
ref when m = 6
iref
q,ν = i q,h
 ref

 id,h1 = kd1 · iref d,ν1 (3.17)
 iref = −k · iref

d5
d,h5
ref
d,ν1
ref when m = 3
i
 q,h1
 = k d1 · i q,ν1
 iref = −k · iref

q,h5 d5 q,ν1

Here kd1 = 0.9659, kd5 = 0.2588 are the winding factors from DFT stated
in (2.5).

3.3 Parameter estimation

Similar to a three-phase machine, the parameters of the machine can be
obtained through no-load and locked-rotor tests. However, in order to control
the machine in more than one harmonic plane, the parameters in higher-order
harmonic planes should also be identified.
 Theory for multiphase machines | 29

3.3.1 Parameters in the first harmonic plane

The machine parameters in the fundamental harmonic plane are identified with
a YY30 voltage with pure fundamental frequency of 10 Hz. This synchronous
frequency is stated in [11]. The phase voltage and current are then measured
and transformed into stator reference frame according to (2.1) and (2.2).
Performing no-load and locked-rotor tests, the parameters of the machine in
the first harmonic plane can be calculated by (3.18) (inverse-Γ model). Here,
subscripts "nl","lr" stand for no-load and locked-rotor test, respectively, while
superscript "s" represents that it is in the stator reference frame.
s
vnl,1
jω1 (L̂σ,1 + L̂M,1 ) + R̂S,1 = s
inl,1
s (3.18)
vlr,1
jω1 L̂M,1 + R̂S,1 + R̂R,1 = s
ilr,1

3.3.2 Parameters in the higher-order harmonic planes

The magnetising inductance in the higher-order harmonic plane is much
smaller than in the fundamental plane. Thus, in higher-order harmonic planes,
the magnetising inductance and the leakage inductance may be of the same
order of magnitude. In such situation, the magnetising inductance can not be
considered as open-circuit when calculating the parameters with locked-rotor
data.
A pure third harmonic voltage will not produce torque. In the meantime, in
this project, the machine only operated as three-phase two pole-pair machine
and six-phase two pole-pair machine. It is described in section 2.1, when
applying the HPD theory with mb = 6 to the components of a three-phase
machine, the fundamental component will distribute in harmonic plane h = 1
and h = 5. Thus, the parameters in third harmonic plane is neglected in this
project. In the fifth harmonic plane, the equation above becomes (3.19).
s
vnl,5
jω5 (L̂σ,5 + L̂M,5 ) + R̂S,5 = s
inl,5
s (3.19)
vlr,5
jω5 L̂σ,5 + R̂S,5 + jω5 L̂M,5 //R̂R,5 = s
ilr,5

The stator resistance should be the same in all harmonic planes. (3.19) can
30 | Theory for multiphase machines

be further simplified to (3.20).

R̂S,5 = R̂S,1
s
vnl,5
R̂S,5 = <{ s }
inl,5
s
vnl,5
X̂σ,5 + X̂M,5 = ={ s }
inl,5 (3.20)
2 s
R̂R,5 X̂M,5 vlr,5
R̂S,5 + = <{ s }
2
R̂R,5 2
+ X̂M,5 ilr,5
2 s
R̂R,5 X̂M,5 vlr,5
X̂σ,5 + = ={ s }
2
R̂R,5 2
+ X̂M,5 ilr,5

In the real experiment, a pure fifth harmonic voltage injection needs a

large starting current. Therefore, the experiment is conducted with both
fundamental and fifth harmonic injection in the same time. The fifth-order
components are then extracted from the data using fast Fourier transform
(FFT).

3.4 Possible phase-pole configurations

As it is shown in Figure 3.2, the existing machine is a six-phase two pole-
pair machine, which is controlled by a six-phase converter. Figure 2.2 depicts
the winding configuration of a YY30 induction machine in one pole-pair, i.e.
electrical angle in the range of [0, 2π]. In principle, the machine can operate
as other two configurations: three-phase two pole-pair (m = 3, p = 2) and
three-phase four pole-pair (m = 3, p = 4). Figure 3.3 describes how the
machine works as a three-phase two pole-pair machine with a star connection.
Comparing with Figure 2.2, two adjacent phases become "one phase". For the
conventional machine, the conductors of two adjacent phases are connected in
series so that the currents through two both phases are the same.
Figure 3.4 gives a possible winding re-arrangement solution for a three-
phase four pole-pair machine (m = 3, p = 4). In addition, this arrangement
is given within electrical angle in the range of [0, 4π] in order to have a clear
comparison with the previous ones. This configuration (m = 3, p = 4) is not
achievable by just controlling the current/voltage of each phase. According
to Figure 3.4, the currents through both phase a1 and phase −a1 should be
phase A current. As it is depicted in Figure 3.5 given in [11], the conductors
 Theory for multiphase machines | 31

ia1
ib1
ic1
ia 2
ib 2
ic 2

Figure 3.2: Six-phase voltage source converter and six-phase winding

iA

-C
B -C
N
B A

iA iB iC
-A A

-A -B A B C

C -B
C

Figure 3.3: Phase-pole configuration with m = 3, p = 2.

in e.g. slots 1 and 19 are connected in series. This is a pole pitch of γp = π.

It corresponds to phase a1 in Figure 3.4. In order to achieve the m = 3, p = 4
configuration, Figure 3.4 tells us that at the same slots 1 and 19 the current
flow must be equal, i.e. both magnetic axes belong to phase A, indicated by
the dashed circles. A pole pitch of γp = π2 is needed, which is not feasible
with this machine windings. Thus, the current flowing through these phases
can not be the same direction. This configuration(m = 3, p = 4) will not be
tested in this project.
32 | Theory for multiphase machines

-A
C -c
2 -c 1
B

-B b1 a 2
-C
b2 a1
A A
-a1 -b2

-C -b1 -B
-a2
c1 c2

B C
-A

Figure 3.4: Phase-pole configuration with m = 3, p = 4 (phases are marked

with colour), phase-pole configuration with m = 6, p = 2 (phases are marked
in black) and two magnetic axes that should belong to phase A (marked with
dashed red circles).

Figure 3.5: Winding layout for the existing machine.

 Results and Analysis | 33

Chapter 4

Results and Analysis

4.1 Simulation results

In this section, the results of different harmonic injection strategies including
minimum current peak optimisation and minimum voltage peak optimisation
are stated. Furthermore, it is shown that using the HPD method, the machine
can switch between different phase-pole configurations smoothly.

4.1.1 Harmonic injection strategies

Figure 4.1 shows the current and voltage waveform using minimum current
peak optimisation method when the current peak hits its maximum value of
5 A. This value is limited by the PCB board and the core saturation stated
in the previous project [11]. Under this current limitation, the machine could
operate at Te = 3.5 Nm, ωr = 290 rpm.
Figure 4.2 shows the current and voltage waveform using minimum voltage
peak harmonic injection, with the speed and torque stated above. Comparing
with Figure 4.1, the maximum value of the voltage is smaller while there is a
current peak large than 5 A.
The control strategy is chosen according to the permissible current
and voltage of the machine and the voltage source inverter. Aiming at a
smaller current peak with higher torque production, minimum current peak
optimisation should be considered. When the motor needs to run at a high
speed, minimum voltage peak optimisation can be used to increase the RMS
value of the voltage furthest without exceeding the maximum voltage limit.
34 | Results and Analysis

Minimum current peak injection

5
Current(A)

-5
0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4
Time(s)
10
Voltage(V)

-10
0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4
Time(s)

Figure 4.1: Minimum current peak harmonic injection.

4.1.2 Results from the Simulink model-switching from

different phase-pole configurations
Figure 4.3 gives the simulation result of IRFOC control using minimum
current peak injection. At t = 2 s, the phase-pole configuration changes from
m = 6, p = 2 to m = 3, p = 2. A smooth transition can be seen in this
result with an overshoot less than 3.3% of the previous speed value. The speed
transition time is less than 0.2 s.

4.2 Preliminary testing of the voltage source

converter
Before giving the voltage supply to the machine, the voltage source converter
was first tested with no-load. The DC link voltage is set to 20 V.
Figure 4.5 shows the voltage between the output pin of phase a1 and
DC- with no load. The duty ratio of PWM is set to 0.5, and the switching
frequency is 10 kHz. Undesired switching overvoltage was detected. This
phenomenon is a problem of capacitor selection of the PCB board. It is
 Results and Analysis | 35

Minimum voltage peak injection

5

Current(A)
0

-5
0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4
Time(s)
10
Voltage(V)

-10
0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4
Time(s)

Figure 4.2: Minimum voltage peak harmonic injection.

Torque curve Speed curve
5 70

60
4 Load torque Reference speed
Electrical torque Real speed
50
3
Torque (Nm)

Speed (rad/s)

40

2 30

20
1
10
0
0

-1 -10
0 1 2 3 4 0 1 2 3 4
Time (s) Time (s)

Figure 4.3: Speed and torque curve (switch from m = 6, p = 2 to m = 3, p =

2 at t=2 s).

shown in Figure 4.5 that this voltage peak value can be two times larger
than the DC link voltage. However, the PCB design is not included in this
project. Furthermore, although the voltage waveform is unrecognisable on
the oscilloscope because of this switching overvoltage, the required harmonic
components (1st , 3rd , 5th ) can still be obtained through the exported data from
36 | Results and Analysis

Figure 4.4: Testbench setup.

oscilloscope by doing FFT in Matlab. It was verified that the results were
as expected, i.e. in coherence with the value given in the code. Moreover,
the permissible maximum voltage of the MOSFETs, diodes are larger than
the overvoltage when applying maximum DC bus voltage to the machine.
Therefore, this phenomenon will not cause any safety problem. For the further
work of this project, a redesigned PCB is recommended.

4.3 Parameter estimation

The six-phase machine is connected with the output of the PCB. Figure 4.4
shows the testbench setup in the lab.
The parameter identification is conducted as it is stated in section 4.3.
The parameters in first and fifth harmonic planes with the inverse-Γ model
are presented in Table 4.1 and Table 4.2.
The parameters are verified by comparing the real phase current with the
values obtained by running the Simulink model with same voltage applied in
the no-load test. The results are shown in Figure 4.6 and Figure 4.7. As it is
stated in section 4.3 the parameter identification for the fifth harmonic plane
was proceeded with both fundamental and fifth harmonic voltage injected. The
comparison shows in Figure 4.7 is presented with only fifth-order components,
 Results and Analysis | 37

Output voltage of one phase leg when duty ratio is 0.5

50

40

Voltage from Va1 to DC- (V) 30

20

10

-10

-20
0 0.5 1 1.5 2 2.5 3
Time (s) 10-4

Figure 4.5: Output voltage of one phase leg output Va1 when duty ratio D =
0.5, switching frequency f = 10 kHz.

Table 4.1: Six-phase machine parameters in the first harmonic plane.

Parameter Value
RR,1 0.2038 Ω
RS,1 0.4673 Ω
LM,1 51.166 mH
Lσ,1 5.478 mH

Table 4.2: Six-phase machine parameters in the fifth harmonic plane.

Parameter Value
RR,5 0.0868 Ω
RS,5 0.4673 Ω
LM,5 1.2995 mH
Lσ,5 3.024 mH

i.e. the applied voltage in the simulation only contains fifth-order components.
The machine is designed in a previous project [11], and the phase angle
between two three phase sets is 26.57 ◦ . However, in this project, the machine
is controlled under a standard transformation flow for a YY30 machine stated
in Figure 2.1, i.e. in (2.2), δ is 30 ◦ . It can be seen in Figure 4.6, when applying
"YY30 voltage" to a YY26.57 machine, the current through each phase set are
different. The current through the second phase set (ia2 , ib2 , ic2 ) is roughly
33% smaller than the first phase set(ia1 , ib1 , ic1 ). However, in Simulink model,
38 | Results and Analysis

the machine is a standard YY30 machine modelled by HPD method, so this

problem will not appear.
It is worth mentioning that when only fundamental voltage is applied,
every measured phase voltage has almost same amplitude. However, the
amplitude of the fifth harmonic component of the measured phase voltage
shows difference, i.e. the phase voltage of second phase set (va2 , vb2 , vc2 ) is
smaller than the first phase set (va1 , vb1 , vc1 ). Thus, the error between the
simulation results and the real value of the second set in Figure 4.7 are smaller
than the value in Figure 4.6.
Neglecting the error caused by the above-mentioned unbalanced problem,
the parameters are verified as an acceptable parameter estimation. However,
further investigation of this phase difference between the applied voltage and
the phase displacement of each winding set should be conducted in the future.

The amplitute of six-phase current in the fundamental harmonic plane

7
Value in Simulink
Real value obtabined from oscilloscope
6

5
Current (A)

0
ia1 ia2 ib1 ib2 ic1 ic2

Figure 4.6: The amplitude of six-phase current in the fundamental harmonic

plane.

4.4 Indirect rotor field oriented current con-

trol of the machine
The motor can be controlled in the first and fifth harmonic planes with known
parameters in these two planes. However, the above mentioned unbalanced
problem in section 4.3 makes it difficult to control the two planes perfectly
at the same time. Thus, in this section, IRFOC stated in section 2.2 was
 Results and Analysis | 39

The amplitute of six-phase current in the fifth harmonic plane

1.2
Value in Simulink
Real value obtabined from oscilloscope
1

Current (A) 0.8

0.6

0.4

0.2

0
ia1 ia2 ib1 ib2 ic1 ic2

Figure 4.7: The amplitude of six-phase current in the fifth harmonic plane.

implemented only in the first harmonic plane, while the reference voltage in
the third and fifth harmonic planes were set to zero.

4.4.1 Current controller testing

The inner current control loop was tested first. In real application, the rising
time of the inner current controller loop is often less than 10 ms. However, in
order to have more data during the rising time period, the current controller is
set with rising time 15 ms for testing. Figure 4.8 and Figure 4.9 show id1 and
iq1 with a step reference to 0.5 A and 0.3 A at t = 0.01 s. It can be seen that
the rising time is approximately 18.5 ms. Although there is some noise from
the board, the mean values from 0.1 S to 0.15 s are 0.5024 A, and 0.3044 A.
They are in coherence with the reference values.

4.4.2 Speed controller testing

After implementing and testing the current controller, the speed controller in
the outer loop was then tested. The bandwidth αs is designed with rising
time 0.4 s. Figure 4.10 shows the speed step response with a speed reference
200 rpm at 0 s. The rising time is around 0.598 s.
A small oscillation appears even when the speed reaches its steady state (2 s
to 6 s). In order to further determine the source of this oscillation, the motor
was tested with different reference speed 100 rpm,150 rpm and 200 rpm. The
oscillation still exists in these three situations. By applying FFT to the data
in a certain period which can be seen as steady state, i.e., from 2 s to 6 s,
40 | Results and Analysis

0.7 0.4

0.35
0.6

0.3
0.5

90% of the steady state value 0.25 90% of the steady state value
0.4
id1 (A)

iq1 (A)
0.2

0.3
0.15

0.2
0.1

0.1 0.05 10% of the steady state value

10% of the steady state value

0 0
0 0.0106 0.0286 0.05 0.1 0.15 0 0.0104 0.0295 0.05 0.1 0.15
Time (s) Time (s)

Figure 4.8: Current id1 with a step Figure 4.9: Current iq1 with a step
reference 0.5 A at t = 0.01 s. reference 0.3 A at t = 0.01 s.
Speed curve with step reference r = 200rpm
200

180
90% of the steady state value
160

140
Speed (rpm)

120

100

80

60

40
10% of the steady state value
20

0
0.044 0.642 1 2 3 4 5 6
Time (s)

Figure 4.10: The amplitude of six-phase current in the fundamental harmonic

plane.

the frequency component of this oscillation can be identified. The result in

Figure 4.11 shows that the frequency components in different situations are
related to their speed. For example, in the first figure with 100 rpm, the
oscillating frequency (1.667 Hz) is equal to the the rotor speed (100 rpm).
Therefore, the oscillation in Figure 4.10 may be caused by a bad mounting
of the encoder, or a slight rotor misalignment.
 Results and Analysis | 41

Single-Sided Value Spectrum (SpeedRef = 100rpm)

3
dc component

|Speed1(f)|
2
X 1.66667
Y 1.07922
1 X 5.66667 X 11.3333
Y 0.32227 Y 0.355749

0
0 5 10 15 20 25
f (Hz)
Single-Sided Value Spectrum (SpeedRef = 150rpm)
3
dc component
|Speed2(f)|

2
X 17.3333
X 2.66667 X 8.66667 Y 0.939489
1 Y 0.512292 Y 0.607409

0
0 5 10 15 20 25
f (Hz)
Single-Sided Value Spectrum (SpeedRef = 200rpm)
3
dc component
|Speed3(f)|

2
X 3.33111 X 22.3184
Y 0.92661 Y 0.992621
1 X 11.3258
Y 0.454223

0
0 5 10 15 20 25
f (Hz)

Figure 4.11: Frequency component analysis with different speed reference-

100 rpm, 150 rpm, 200 rpm.
42 | Results and Analysis
 Discussion | 43

Chapter 5

Discussion

This chapter gives a further discussion about the issues in this project,
including the saturation problem and the unusual phase displacement of the
existing six-phase machine.

5.1 No-load current and torque of the motor

It is shown in Figure 4.6, the current peak value in the no-load test for the first
harmonic plane is more than 5 A. However, in [11], it is stated that the core
saturation starts when current peak value reaches 5 A and its nominal torque
is 3 Nm. Thus, the core is saturated sightly even there is no load.
The no-load torque of the motor is determined by various no-load losses
of the motor. They are composed of a series of friction losses, such as the
wind friction losses of the rotor, the friction losses of the motor bearing during
operation, and etc. In this project, no external load is added as its no-load
torque is even slightly larger than its nominal torque. Further improvement
may need to reduce the losses in the future. In the meantime, considering
the motor efficiency, the nominal speed will be smaller. Now the machine’s
synchronous speed is only 300 rpm. Thus, future work should also include
increasing the current and voltage limitation of the PCB board in order to have
a higher rated power.
44 | Discussion

5.2 Current difference between each phase

set
In (4.3), the result shows that the current difference between each phase
set reaches up to 33% when applying "YY30 voltage" to this "YY26.57"
machine. Figure 5.1 further explains this difference in the fundamental
reference frame. When applying (2.1) to one interested component of the
machine, the angle between each phase is no longer always 30 ◦ . The angle
between x1 and x2 is 26.67 ◦ , while the angle between x2 and x3 is 33.43 ◦ .
Currently, there is no related study about a six-phase machine displacement
angle other than 30 ◦ or 60 ◦ . In this project, this difference is considered
as an "error", so the transformation remains the same as it should be for a
YY30 machine. However, when applying the HPD method to control the
current in both first and fifth harmonic plane, undesired oscillation appears
due to the aforementioned imbalance. The tolerance of angle displacement of
this control method can be further investigated in the future. Another control
strategy that may solve this problem to some extend is also worth thinking and
testing.

x4 x4
x5 x3
x6 x2
x6
33.43
x2
26.57
x1

Figure 5.1: The fundamental frame: Standard YY30 machine (in black) and
YY26.57 machine (in red).
 Conclusions and future work | 45

Chapter 6

Conclusions and future work

The chapter gives an conclusion of this degree project. Furthermore, some

relevant topics for the future are stated.

6.1 Conclusions
The goal of this thesis work is to apply the HPD method to model and analyse
a six-phase induction machine, in order to investigate how the control strategy
works for a multiphase machine.
The simulation results show that using the HPD method, a two pole-pair
YY30 machine can achieve a smooth transition from m = 6, p = 2 to m =
3, p = 2. Additional degrees of freedom, i.e., the components in the fifth
harmonic plane, can be used for a six-phase machine.
The parameter estimation was conducted in both first and fifth harmonic
planes, and it was validated through simulation. Although the existing six-
phase machine cannot be controlled perfectly in higher harmonic planes
because of the unusual phase displacement, the drive is working properly when
control the machine in the first harmonic plane.

6.2 Future work

Some suggested future work is stated below. Further explanation can be seen
in both chapter 4 and chapter 5.
• Increase the current and voltage limit of the PCB and do some drive tests
with an external load.
46 | Conclusions and future work

• Redesign the PCB board to decrease the overvoltage caused by switching

MOSFETs.
• Further investigations on how does the alignment of the real phase
displacement of the machine and the phase displacement of the applied
voltage affect the performance of the machine.
• Run the machine in faulty operation using the HPD method.
 REFERENCES | 47

References

[1] E. Ward and H. Haerer, “Preliminary investigation of an invertor-fed 5-

phase induction motor,” Proceedings of the Institution of Electrical Engi-
neers), vol. 116, no. 6, pp. 991–996, 1969. doi: 10.1049/piee.1969.0182
[2] E. Levi, R. Bojoi, F. Profumo, H. Toliyat, and S. Williamson,
“Multiphase induction motor drives-a technology status review,” IET
Electric Power Applications, vol. 1, no. 5, pp. 643–656, 2007. doi:
10.1049/iet-epa
[3] A. Smith, S. Williamson, and C. Hodge, “High torque dense
naval propulsion motors,” in IEEE International Electric Machines
and Drives Conference, 2003. IEMDC’03., vol. 3, 2003. doi:
10.1109/IEMDC.2003.1210638 pp. 1421–1427 vol.3.
[4] Y. Zhao and T. A. Lipo, “Space vector pwm control of dual three-
phase induction machine using vector space decomposition,” IEEE
Transactions on Industry Applications, vol. 31, no. 5, pp. 1100–1109,
1995. doi: 10.1109/28.464525
[5] K. Bitsi, O. Wallmark, and S. Bosga, “An Induction Machine with
Wound Independently-Controlled Stator Coils,” 2019 22nd International
Conference on Electrical Machines and Systems, ICEMS 2019, pp. 1–5,
2019.
[6] Y. Wu, G. F. Olson, L. Peretti, and O. Wallmark, “Harmonic Plane
Decomposition: An Extension of the Vector-Space Decomposition - Part
I,” IECON Proceedings (Industrial Electronics Conference), vol. 2020-
Octob, pp. 991–996, 2020. doi: 10.1109/IECON43393.2020.9254279
[7] G. F. Olson, Y. Wu, L. Peretti, and O. Wallmark, “Harmonic Plane
Decomposition: An Extension of the Vector-Space Decomposition - Part
II,” IECON Proceedings (Industrial Electronics Conference), vol. 2020-
Octob, pp. 991–996, 2020. doi: 10.1109/IECON43393.2020.9254279
48 | REFERENCES

[8] Y. Zhao and T. A. Lipo, “Modeling and control of a multi-phase induction

machine with structural unbalance,” IEEE Transactions on Energy
Conversion, vol. 11, no. 3, pp. 578–584, 1996. doi: 10.1109/60.537028

[9] J. G. Webster, M. J. Duran, E. Levi, and F. Barrero, “Multiphase Electric

Drives: Introduction,” Wiley Encyclopedia of Electrical and Electronics
Engineering, no. i, pp. 1–26, 2017. doi: 10.1002/047134608x.w8364

[10] A. A. Rockhill and T. A. Lipo, “A generalized transformation methodol-

ogy for polyphase electric machines and networks,” Proceedings - 2015
IEEE International Electric Machines and Drives Conference, IEMDC
2015, pp. 27–34, 2016. doi: 10.1109/IEMDC.2015.7409032

[11] A. Bianchi, “Development of an experimental six-phase induction

machine drive,” 2020.

[12] M. P. Magill and P. T. Krein, “A dynamic pole-phase modulation

induction machine model,” in 2015 IEEE International
Electric Machines Drives Conference (IEMDC), 2015. doi:
10.1109/IEMDC.2015.7409030 pp. 13–19.

[13] A. G. Yepes, J. Doval-Gandoy, F. Baneira, D. Pérez-Estévez, and

O. López, “Current harmonic compensation for n -phase machines with
asymmetrical winding arrangement and different neutral configurations,”
IEEE Transactions on Industry Applications, vol. 53, no. 6, pp. 5426–
5439, 2017. doi: 10.1109/TIA.2017.2722426

[14] T. A. Lipo, Introduction to AC Machine Design, 2017. ISBN

9781119352167

[15] L.Harnefors, M. Hinkkanen, and O. Wallmark, “Control of Voltage-

Source Converters and Variable-Speed Drives,” 2014.

[16] O. Dordevic, M. Jones, and E. Levi, “A comparison of pwm techniques

for three-level five-phase voltage source inverters,” in Proceedings of the
2011 14th European Conference on Power Electronics and Applications,
2011, pp. 1–10.

[17] ——, “A comparison of carrier-based and space vector pwm techniques

for three-level five-phase voltage source inverters,” IEEE Transactions
on Industrial Informatics, vol. 9, no. 2, pp. 609–619, 2013. doi:
10.1109/TII.2012.2220553
 REFERENCES | 49

[18] A. Iqbal and E. Levi, “Space vector modulation schemes for a

five-phase voltage source inverter,” 2005 European Conference on
Power Electronics and Applications, vol. 2005, pp. 1–12, 2005. doi:
10.1109/epe.2005.219194

[19] A. Iqbal, E. Levi, M. Jones, and S. N. Vukosavic, “Generalised sinusoidal

PWM with harmonic injection for multi-phase VSIs,” PESC Record
- IEEE Annual Power Electronics Specialists Conference, 2006. doi:
10.1109/PESC.2006.1712206

[20] R. Petrella, M. Tursini, L. Peretti, and M. Zigliotto, “Speed mea-

surement algorithms for low-resolution incremental encoder equipped
drives: a comparative analysis,” in 2007 International Aegean Con-
ference on Electrical Machines and Power Electronics, 2007. doi:
10.1109/ACEMP.2007.4510607 pp. 780–787.

[21] H. Xu, H. Toliyat, and L. Petersen, “Rotor field oriented control of

five-phase induction motor with the combined fundamental and third
harmonic currents,” APEC 2001. Sixteenth Annual IEEE Applied Power
Electronics Conference and Exposition (Cat. No.01CH37181), vol. 1, pp.
392–398 vol.1, 2001.
For DIVA
{
"Author1": {
"Last name": "Shan",
"First name": "Xinyue",
"E-mail": "[email protected]",
"organisation": {"L1": "School of Electrical Engineering and Computer Science ",
}
},
"Degree": {"Educational program": ""},
"Title": {
"Main title": "Field Oriented Current Control with Harmonic Injection in a Six-Phase Induction Machine",
"Language": "eng" },
"Alternative title": {
"Main title": "Fältorienterad strömreglering med harmonisk injektion i en sex fas asynkronmaskin",
"Language": "swe"
},
"Supervisor1": {
"Last name": "Wu",
"First name": "Yixuan",
"E-mail": "[email protected]",
"organisation": {"L1": "School of Electrical Engineering and Computer Science ",
}
},
"Examiner1": {
"Last name": "Peretti",
"First name": "Luca",
"E-mail": "[email protected]",
"organisation": {"L1": "School of Electrical Engineering and Computer Science ",
}
},
"Other information": {
"Year": "2021", "Number of pages": "xiii,49"}
}
TRITA-EECS-EX-2021:781

www.kth.se