Download as pdf or txt
Download as pdf or txt
You are on page 1of 16

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

net/publication/321739522

Train-Network Interactions and Stability Evaluation in High-Speed Railways--


Part I: Phenomena and Modeling

Article  in  IEEE Transactions on Power Electronics · December 2017


DOI: 10.1109/TPEL.2017.2781880

CITATIONS READS

112 765

6 authors, including:

Haitao Hu Haidong Tao


Southwest Jiaotong University Southwest Jiaotong University
74 PUBLICATIONS   1,697 CITATIONS    13 PUBLICATIONS   310 CITATIONS   

SEE PROFILE SEE PROFILE

F. Blaabjerg Xiongfei Wang


Aalborg University KTH Royal Institute of Technology
3,292 PUBLICATIONS   143,766 CITATIONS    517 PUBLICATIONS   16,040 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

AI-Power: Artificial Intelligence for Next-Generation Power Electronics View project

Reliable Power Electronic Based Power System. View project

All content following this page was uploaded by Haitao Hu on 12 December 2017.

The user has requested enhancement of the downloaded file.


This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TPEL.2017.2781880, IEEE
Transactions on Power Electronics
> REPLACE THIS LINE WITH YOUR PAPER IDENTIFICATION NUMBER (DOUBLE-CLICK HERE TO EDIT) < 1

Train-Network Interactions and Stability Evaluation


in High-Speed Railways--Part I: Phenomena and
Modeling
Haitao Hu, Member, IEEE, Haidong Tao, Student Member, IEEE, Frede Blaabjerg, Fellow, IEEE,
Xiongfei Wang, Senior Member, IEEE, Zhengyou He, Senior Member, IEEE, Shibin Gao

Abstract—This paper presents an impedance-based model to not only deteriorate the supply environment of the traction
systematically investigate the interaction performance of multiple network, adverse also affect the onboard traction drive, auxil-
trains and traction network interaction system, aiming to evaluate iary and their control systems.
the serious phenomena, including low-frequency oscillation As the train is a typical harmonic and disturbance source
(LFO), harmonic resonance and harmonic instability. The
train-network interaction mechanism is therefore studied and one
operating in the traction network, the large power demand and
presents a detailed coupling model for investigating the three disturbance will degrade the voltage or power quality of the
interactive phenomena and their characteristics, influential fac- traction network, which can further deteriorate the stability and
tors, analysis methods and possible mitigation schemes. In the safety of the traction drive system. Obviously, as shown in Fig.
Part I of the two-part paper, the measured waveforms of such 1, The train and traction network essentially interact with each
three phenomena are firstly characterized to indicate their fea- other. Three-phase 220kV power source is stepped down to
tures and principles. A unified framework of the train-traction
network system for investigating the three problems is then pre-
feed the catenary system through the traction transformer in a
sented. In order to reveal the interaction mechanism, the traction substation. The traction network not only supplies the
all-frequency impedance behaviors of the electric trains and trac- energy to the running electric trains, but also absorbs the in-
tion network are equally modeled. In which, an impedance-based jected harmonics and regenerated power from the trains. Due to
input behavior of the train is fully investigated with considering different control modes, numbers, positions and loading con-
available controllers and their parameters in DQ-domain. While, ditions of multiple electric trains, the train-subsystem of the
the entire traction network, including traction transformer, ca-
tenary, supply lines, is represented in a frequency-domain nodal
interaction system show different behaviors interacting with the
matrix. Furthermore, the impedance-frequency responses of both traction network.
electric train and traction network are measured and validated Seen from the traction network side, the electric trains run-
through frequency scan method. Finally, a generalized ning on the rails are usually represented by a constant current or
train-network simulation and experimental systems are proposed PQ model for harmonic and power flow studies [8,9]. Published
for verifying the theoretical results of the two-part paper. works have shown that the input admittance of the 4-quadrant
converters (4QC) of the high-speed train will affect the traction
Index Terms—Low-frequency oscillation; harmonic resonance;
harmonic instability; train-network interactions; high-speed network analysis and thereby arise oscillation or instability
railway. problem[10]. While, seen from the electric train side, the trac-
tion network is usually assumed to be a linear multi-port im-
pedance-based network. However, the traction network may
I. INTRODUCTION have inherent impedance behavior or resonance points. Almost
all power quality problems in the high-speed railways are
w ITH the fast development of worldwide high-speed rail-
way (HSR) along with the widely adoption of the
AC-DC-AC voltage source converter (VSC) based
caused by the different features of the scheduled trains as well
as the traction network.
high-speed trains (HSTs), the harmonic and interaction insta-
Utility PLL wt VC +
bility problems have been frequently appeared and reported in udcref
3 phase
220 kV

system _+ _
previous literatures [1-3]. Among the reported problems, har- Supply uac1 iac
PWM CC
monic resonances, LFO and harmonic instability have been line
ma
udc
given more attentions due to the dynamic interactions between Traction 4QC Inverter I
multiple trains and traction network, and serious impacts on transformer 1770V M
AC/DC DC/3AC
2x27.5 kV

system power quality and stability. Such their adverse impacts


1 phase

Supply in
line Ytrain Train in
Ytrain 2
in Up
This work was supported by National Natural Science Foundation of China
out
Z net Y train 1 track

(NSFC 51677154, 51525702). Traction


H. Hu, H. Tao, Z. He and S. Gao are with the School of Electrical Engi- in
Ytrain
neering, Southwest Jiaotong University, Chengdu 610031, China (e-mail:
network Down track

hht@swjtu.edu.cn, thdswjtu@163.com, hezy@swjtu.edu.cn


gao_shi_bin@126.com). Fig. 1 A typical HSR train-network interaction system.
F. Blaabjerg, and X. Wang are with the Department of Energy Technology
Aalborg University, Aalborg 9220, Denmark (Email: fbl@et.aau.dk, The standard EN50388-2012 [4], added some new phe-
xwa@et.aau.dk).

0885-8993 (c) 2017 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TPEL.2017.2781880, IEEE
Transactions on Power Electronics
> REPLACE THIS LINE WITH YOUR PAPER IDENTIFICATION NUMBER (DOUBLE-CLICK HERE TO EDIT) < 2

nomena from its previous version in 2005, described and clas- experimental results to assess and validate the influential fac-
sified the root causes of the overvoltage problems happening in tors, relationship among the three phenomena in a unified im-
the AC or DC electrified railways as oscillation (or instability) pedance-based framework.
and harmonic problems. Based on the previous literatures, the
LFO, harmonic resonance and harmonic instability phenomena II. FIELD PROBLEMS IN HSR
in electric railway systems can be overviewed as follows. In order to investigate the electrical characteristics of the
 Low-frequency oscillation, from 1 Hertz to several Hertz, three different phenomena (i.e., LFO, harmonic resonance and
often happens in a rail deport in transformer-based electric harmonic instability), various associate field data have been
railways when multiple trains are simultaneously ener- presented and discussed in this section.
gizing or operating [5]. Alternatively, the oscillation can
A. Field Tests of the Low-frequency Oscillation
also be caused by the rotatory frequency converter in some
European railway systems, such as Norway, Sweden [6]. There are two types of LFO scenarios dependent on the rotary
Consequently, the oscillatory voltage and current wave- frequency converter existing or not. Some electrified railways
forms contain a significant low-frequency component. in Norway, America, Sweden, Germany, and Switzerland have
adopted the rotary frequency converter as the power supply
 Harmonic resonance, often from a few hundred to kilo
solution [6,16,17]. The LFO has been early reported from those
Hertz, has been widely reported in many HSR systems due
railway systems. TABLE I surveys the LFO cases which
to the interaction between capacitive and inductive net- adopted a rotary converter in the electric railway systems over
work parameters. As a distributed RLC circuit, the cate- the world [18]. The typical oscillatory frequency fosc (f0-f) is
nary network can experience series or parallel resonances approximately 10-30% of the respective power system’s
at one or multiple specific frequencies that amplifies the fundamental frequency f0. The rotary converter, used for the
harmonic currents generated from electric trains and leads power supply mode variations in traction substation, is exciting
to the interference in adjacent communication lines and oscillations, i.e. given by active power changes in the 16⅔-Hz
the rail signaling system, overheating, and vibration at the ac railway system. The oscillation is critical stable and the
power capacitors, and mal-operation at the protections damping of the system is insufficient even negative.
[2,3]. TABLE I LOW FREQUENCY OSCILLATION CASES WITH ROTARY CONVERTER
 Harmonic instability, typically a few hundred Hertz, is
caused by the feedback loop of network-connected 4QCs No. f1/Hz f0/Hz Time Located
1 1.6 16.67 2007 Norway [6]
of electric trains, was firstly reported in Zürich (Switzer- 2 1.2 25 1997 America [16]
land) in 1995 [1] where the electric train emits excessive 3 1.6 16.67 1991 Germany [17]
harmonic currents due to instability. As the instability 4 1.9 16.67 1991 Germany [17]
originate from destabilization of poorly damped harmonic
resonances, this phenomenon is also known as resonance The second scenario of LFO phenomena mainly happen in a
instability or harmonic instability [7]. The harmonic rail depot, where multiple trains are prepared for operating and
instability may lead to continuous amplification of voltage located far from the traction substation. Multiple trains in the
waveforms with a high harmonic content, and lead to the rail depot have rose the pantographs to energize, leading to a
significant harmonic overvoltage either at the train loca- 0.6~7 Hz LFO of the catenary voltage. As a result, the 4QCs of
tion or in other positions along the traction network. electric trains were therefore locked and shut down by protec-
Therefore, in order to investigate the characteristics of the tion devices due to the transient fluctuated overvoltage. Alter-
natively, different types of electric trains in different railway
three phenomena in a unified framework, the input or output
systems have experienced similar LFOs in other rail depots,
equivalent “admittance” or “disturbance” in train-network
and the oscillatory frequencies were not the same. Some cases
interaction system should be fully considered. There are many of the LFO phenomena happened in a rail depot without the
published documents focusing on the modeling and analysis of rotary converter are listed in TABLE II.
the electrical characteristics in electric railway systems
TABLE II LFO CASES WITHOUT ROTARY CONVERTER [19]
[2,3,11-15]. However, they simplified the modeling of either
the electric train system or the traction network for harmonic or No. f1/Hz f0/Hz time Case
stability studies, and studied the three phenomena separately. 1 5 50 2008 Thionville, France [20]
2 7 50 2006 Siemens test, Germany [21]
Thus then, this paper and the companion one is organized to 3 3 25 2006 Washington, DC, USA [22]
investigate these problems. 4 5 16.67 1995 Zurich, Switzerland [23]
In this two-part paper, we aim to investigate and improve the 5 2~4 50 2008-01 Hudong Depot, HXD1
6 5 50 2010-01 Shanghai Nanxiang, CRH1
electrical interaction performance between the traction network 7 5 50 2010-09 Qingdao Depot, CRH5
and multiple electric trains. Part I of the two-part paper sys- 8 0.6~2 50 2015-12 Xuzhou Hub, HXD2B
tematically characterizes the phenomena of the harmonic res- 9 6-7 50 2011-11 Shanhaiguan Hub, HXD3B
onance, LFO and harmonic instability, focusing on phenomena Note: the default location of the listed cases is in China.
description and impedance modelling of the train-network Fig. 2 is the measured waveforms under LFO condition in
system. The field tests in electric railway systems are organized Xuzhou hub substation of Jiangsu Province, China. It repre-
in order to reveal the train-network interaction mechanism or sents a typical measured voltage and current waveforms under a
laws. Part II of the paper will present detailed simulations and LFO. The harmonic spectrum of the current waveform ab-

0885-8993 (c) 2017 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TPEL.2017.2781880, IEEE
Transactions on Power Electronics
> REPLACE THIS LINE WITH YOUR PAPER IDENTIFICATION NUMBER (DOUBLE-CLICK HERE TO EDIT) < 3

sorbed by the electric train is composed of two inter-harmonic Voltage (V)


AC Voltage DC Voltage Current (A) DC voltage (V)
50000 100 4000
components, which are symmetrically superimposed to the 75 3500
fundamental frequency f0, and corresponds to the frequency of 25000 50
25
3000
2500
the voltage excitation into consideration. 0 0
-25
2000
1500
Overhead line voltage / kV

-25000 -75 1000


50 -50 500
AC Current -100 0
-50000
0 100 200 300 400 500
25 Time / ms

0
(a)
-25 Voltage (V) Curruent (A) DC voltage (V)
50000 AC Voltage DC Voltage 100 4000
-50 0 0.4 0.8 1.2 1.6 2 f f0 75 3500
Time / s Frequency / Hz 25000 50 3000
25 2500
(a) 0 0 2000
-25
Train output current / A

1500
100 -25000 -75 1000
-50 500
50 AC Current
-50000 -100 0
0 100 200 300 400 500
Time / ms
0
-50 (b)
Voltage (V) Current (A) DC voltage (V)
-100 f f0 2f0-f
0 0.4 0.8 1.2 1.6 2 50000 AC Voltage DC Volt age 100 4000
Time / s Frequency / Hz 75 3500
(b) 25000 50 3000
25 2500
0 0 2000
Fig. 2 Measured waveforms of LFO: (a) voltage waveform and its spectrum; (b) -25 1500
current waveform and its spectrum. -25000 -75 1000
-50 500
AC Current
Fig. 3 is the measured waveforms of LFO in Cangkou traction -50000
0 100 200 300 400 500
-100 0
Time / ms
substation of Shandong Province, China. The measurements of (c)
the LFO in a railway deport with the increasing number of
Fig. 3 Measured voltage and current waveforms with different train numbers:
electric trains connected to the traction network are shown in (a) less than 5 trains; (b) 6 trains; (c) 7 trains.
Fig. 3, where the voltage and current waveforms are oscillating
when the 6th train is connecting. Moreover, the oscillatory 0.6~0.8 Hz low- 2 Hz low-Frequency No-regular
frequency oscillation oscillation oscillaiton
frequencies in Fig. 3(b) and Fig. 3(c) are different as the result 50 4

of one more train connected. It is worthy noted that the traction 3


25 2
lockout (which means that the main circuit breaker of the train
Voltage / kV

Current / kA
1
trips, the network current is going to be zero and the DC-link
0 0
voltage is also decreased) aroused in the tested train, owing to -1
the voltage and current fluctuated intensely at the period of -25 -2
350-500 ms. It indicates that more electric trains may easily -3
destabilize the traction network and change the oscillatory -50 -4
32 34 36 38 40 42 44 46 48 50 52
frequency. Time / s
Furthermore, in order to illustrate different oscillatory fre- (a)
quencies and modes, field tests have been taken in Xuzhou 40

railway deport in China in Jan. 2016. When 7 or 8 trains con- 20


nected to the traction network, the voltage and current wave-
forms of the traction network started to oscillate. Meanwhile, 0
P / MW

the oscillatory voltage and current waveforms measured in the -20


train are shown in Fig. 4.
-40
Obviously, they may have three different oscillatory modes
involved: 1) Periodical fluctuation around 0.6 Hz to 0.8 Hz and -60
32 34 36 38 40 48 44 46 48 50 52
the oscillation is in critical unstable mode; 2) Periodical fluc- Time / s
tuation around 2 Hz and the oscillation is highly damped; 3) (b)
Irregular fluctuation and the oscillation is getting unstable.
Fig. 4 Measured data of different oscillatory modes: (a) The test waveform of
Among them, irregular fluctuation contributes to the worst voltage and current; (b) The test oscillation of the active power.
current amplification (about 4000 A). The three oscillatory
modes and their oscillatory amplitudes can be found in Fig. 4(b). B. Field Problems of Harmonic Resonance
Experiences from China railway system show that the critical The current harmonics generated from a modern train may be
number of trains leading to a LFO is about 6~8 that dependents amplified by the network harmonic resonance at some specific
on the grid strength and train types. It should be mentioned that frequencies in the traction network [13,15]. The harmonic
the short circuit capacity of the utility system is usually below resonance is an inner nature of an industrial power system and
500 MVA and is usually regarded as a weak grid. A LFO may cause some serious power quality problems like volt-
happens easily under the condition of a weak supply network age/current distortion, overvoltage with capacitive devices, and
and a large number of the electric trains. interference problems with secondary measurement system and

0885-8993 (c) 2017 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TPEL.2017.2781880, IEEE
Transactions on Power Electronics
> REPLACE THIS LINE WITH YOUR PAPER IDENTIFICATION NUMBER (DOUBLE-CLICK HERE TO EDIT) < 4

electromagnetic interference with the communication system 60 (kV) (A) 80

Contact voltage of the train / kV

Injected current of the train / A


[14,15]. The resonance phenomena, which are surveyed in 40
60
TABLE III, have been widely reported around the world. 40
20
TABLE III HARMONIC RESONANCE FIELD CASES IN TPSS 20
0 0
No. Frequency / pu Time Location
-20
1 17~18 2008-05 Hewu HSR, China [11] -20
-40
2 50~55 2011-01 Jinghu HSR, China [24] -40
-60
3 50~55 2010 Boao HSR, China (s)
Shinchungju–Yongjung and -60 -80
4 20~25 2004 0 0.012 0.024 0.036
Pyongtaek–Maha HSR, Korea [2] Time / s
5 39 2011 Italy [14] (a)
6 17~21 2006 Britain [25] 16

IHDi of injected current / %


7 3~7 2009 Unknown, Spain [26] 14
12
10
The harmonic resonance phenomenon has been frequently
8
happened in many electric railway systems around the world, 6
and the parallel resonant frequency ranges from 11 pu to 55 pu 4
with the different traction network topologies and 2
corresponding parameters, which are listed in TABLE III. On 0
5 10 15 20 25 30 35 40 45 50 55 60
the other hand, L. Sainz and M. Caro in [26] reported the series Frequency / pu
resonance phenomenon with considering the connection of the (b)
Steinmetz circuit, and the series resonance frequency is about 3 Fig. 5 Harmonic distortions in the leading segment of JingHu HSR: (a) voltage
pu ~ 7 pu. In this paper, we mainly focus on the parallel and current waveforms of the CRH380AL electric train; (b) the individual
resonances due to the amplification of voltage distortion and current distortion of measured current.
their severe adverse impacts on system operations. Fig. 5 to Fig.
THD of traction bus voltage / %
24
7 measured some field waveforms under harmonic resonance
20
conditions in China HSR system. Those phenomena have been
tested in the same or similar supply conditions and they can be 16

illustrated by the following principles: 12

1) A serious harmonic distortion occurs on JingHu HSR as 8


shown in Fig. 5, it can be found obviously in Fig. 5(b) that the 4
resonance appears in the two times of the switching frequency 0
of the train (for CRH380AL, the switching frequency is 10:00 12:00 14:00 16:00
Time / hour
1250Hz). In Fig. 6, the harmonic resonance occurs periodically
Fig. 6 The total harmonic distortion of traction bus voltage in Qionghai section
in the Qionghai section post and this periodicity is closely post, Haidong Line of China.
related to the operation interval and the schedule of the rolling
Individual voltage distortion / %

25
stock. The measurements show that the harmonic resonance
Individual voltage distortion / %

The contact line voltage of 40 The contact line voltage of


phenomenon in electric railway systems is generated by the 20 Nanjing direction in Chang-Anji Nanjing direction in Chang-Anji
traction substation traction substation
traction network along with fixed positions of substations and 15 30
Urms=29.1kV, UTHD=23.9%
the flexible positions, and the numbers of the trains. Their 20
Urms=32.5kV, UTHD=56.0%
10
parameters matched specific operation conditions.
5 10
2) The same conditions of the traction network may excite
multiple different frequencies of the harmonic resonance as 0
4 12 20 28 36 44 52 60
0
4 12 20 28 36 44 52 60
shown in Fig. 7. It illustrates that the harmonic resonance is Frequency / pu Frequency / pu
corresponding to the electric parameters of the electric train. In (a) (b)
other words, the electric train is not the only excitation source Fig. 7 Harmonic distortions in Changanji TSS, HeWu HSR: (a) time 12:20,
of harmonic resonances but also the participative network resonance frequency is 17 pu; (b) time 14:19, resonance frequency is 18 pu.
component. The different positions, powers, numbers of the
C. Field Problems of Harmonic Instability
trains may change the system impedance and the associated
resonance frequencies. The harmonic instability phenomenon in the electric railway
3) From Fig. 5 to Fig. 7, the harmonic resonance is varying was first reported in Zürich, Switzerland in 1995 [1], due to the
with different traction networks and rolling stocks (e.g., train adoption of new type trains which used high-frequency PWM
types, organization, schedule). Many influential factors might converters, and this phenomenon was then defined in
affect the harmonic resonance, such as the parameters of cate- EN50388-2012 [4]. The PWM converter injects massive
nary network, traction transformer. Then the essence of har- harmonic currents and shows a negative damping to the traction
monic resonance is caused by the traction network matched network. This phenomenon has also been frequently reported in
with the train parameters. high-voltage direct current systems (HVDC) and VSC-based
renewable energy interactions. The typical scenarios of the
harmonic or harmonic instability problems in the power system

0885-8993 (c) 2017 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TPEL.2017.2781880, IEEE
Transactions on Power Electronics
> REPLACE THIS LINE WITH YOUR PAPER IDENTIFICATION NUMBER (DOUBLE-CLICK HERE TO EDIT) < 5

and electric railway system are listed in TABLE IV. interactive phenomena can be preliminarily summarized as
TABLE IV HARMONIC INSTABILITY IN POWER SYSTEMS AND TPSS
follows.
1) Low-frequency oscillation happens usually under the
No. Items Time Location weak supply system and a number of trains condition from our
1 Railway 1995 Zürich, Switzerland [1] field test. The simultaneously energization of multiple trains
2 HVDC 1965 Kingsnorth, England [27]
3 Wind generation 2012 Denmark, Poland [28] cause the voltage fluctuations and lead to a LFO. The
4 Solar panel 2004 Amersfoort, Netherlands [29] oscillation is mainly dependent on the equivalent impedance
5 Railway 2015 Chengdu, China ratio between the traction network (corresponding to the supply
distance and system capacity) and the train system
Fig. 8 presents the measured unstable voltage waveforms of (corresponding to the connection number of the trains and their
the harmonic instability in Rongchang traction substation of control parameters).
Chongqing, China. The harmonic components are sustained Generally, it is assumed that the voltage oscillation is a
amplified. There is one or multiple resonance points resulting critical condition of the normal operation. Most of the time, the
from the interaction between the traction system and a long trains are disconnected the power supply by the software
power cable. However, the harmonic resonance is a stable protection, which detects an outage of fluctuated pantograph
harmonic amplification in a passive network. In contrast, the voltage or a false pantograph bouncing. Moreover, due to the
unstable harmonic distortion involves active devices with faster power fluctuation, the upstream power system may arise flicker
controllers, such as the PWM converter of HSTs. These con- effect, which forces the power plant to disconnect the railway
verters can easily be unstable under a network-side resonance substation as well as causes a shutdown of the traffic.
situation with a poorly damped network. The controllers (e.g., 2) Harmonic resonance happens under the following
current controller, voltage controller and phase-locked loop) of conditions in electric railways: a) the inductive system with
the 4QC are interacting with the passive components in the distributed capacitance gives one or more resonance frequen-
traction network and then arise harmonic instability problems cies; b) the harmonic spectra of the HST cover one or more
[29,30]. resonance frequencies; c) on the basis of a) and b), the resulted
50 resonances are poorly damped so that the peak driving
30
impedance is relatively steep. Note that the equivalent
impedance of the onboard transformer in a train can change the
Voltage / kV

10
Zoom in
impedance behavior of the traction network seen from the
-10 pantograph. Therefore, the equivalent impedance modeling of
both HST and traction network in harmonic domain is required.
-30
The total capacitance of the traction network, which is mainly
-50 composed of the primary supply line (overhead lines from
1.5 1.6 1.7 1.8 1.9 2
Time / s upstream substation to a traction substation), secondary supply
(a) line (power cables from traction substation to catenary system,
50 about 50~200 m) and catenary network (overhead lines), is
small so that the system resonance frequency may be up to
30
several kHz (which means high-order harmonic resonance,
Voltage / kV

10 exist typically 1~3 kHz).


-10
3) Harmonic instability. Since the harmonic resonance
happens in a linearized traction network, the network voltage is
-30 highly distorted but still guarantees stable. However, the har-
-50 monic instability can cause overvoltage problems with
1.7 1.72 1.74 1.76 1.78 1.8
Time / s
capacitive devices and arresters [4], the gradually amplified one
(b) or more of harmonic components will cause the arrester and
protector to explode, electrical equipment overheating, motor
Fig. 8 Measured supply voltage of HST under harmonic instability: (a) voltage
waveform of the traction substation; (b) zoom-in view of the turning point. failure. With the rapid development of the HSTs, this
phenomenon have been a critical power quality issue for
D. Summary electrical railways especially the high-speed trains. The har-
The mentioned phenomena might deteriorate and destabilize monic instability, which is different from the LFO mainly
both the traction network and the utility grid. Moreover, the affected by the low frequency characteristics of the voltage
electric trains would easily be shut down by the protective controller (VC) and current controller (CC), is an interactive
equipment because of the LFO, harmonic resonance or har- high-frequency stability problem. Because the harmonic
monic instability. An overview of the typical oscillatory instability is mainly affected by the high-frequency
waveforms, the matched operating conditions, and their char- characteristics of the phase-locked loop (PLL) and CC of the
acteristics for the three phenomena are summarized in Fig. 9. electric train output behavior rather than the traction network
Meanwhile, the potential influential factors in terms of the characteristics. If the train-network close-loop system has high
train-network interaction system are involved in the analytical frequency unstable poles, the harmonic resonance in network
models in an impedance-based framework. side may increase gradually and lead to harmonic instability.
On the basis of the discussions in this section, the three

0885-8993 (c) 2017 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TPEL.2017.2781880, IEEE
Transactions on Power Electronics
> REPLACE THIS LINE WITH YOUR PAPER IDENTIFICATION NUMBER (DOUBLE-CLICK HERE TO EDIT) < 6

Phenomena Waveforms Formation conditions Charateristics Influential Factors Model


50 1
With/without a rotating With rotary converter Train
Low-frequency

Current / kA
Voltage / kV
25 0.5 converter 1~2 Hz
oscillation Onboard transformer Active or Passive
0 0 Without rotary Norton model
Weak supply network
Near instable -25 -0.5 converter 0.6~7 Hz 4QC control parameters
Low-freq. -50 -1 More than 6~8 trains Protection malfunction (CC, VC, PLL)
0 1 2 Yin ( s )
Instable 50 40 Capacitive&inductive Resonance frequency Power consumption
High-freq. 2 2
17~55pu
Voltage / kV
parameters matched Positions & number

Current / A
5 0
Harmonic 0 0 Series resonance
Harmonic source
resonance frequency 3~7 pu
-25 -20 covers the resonance Network Impedance via
Harm to capacitive
Active -50 -40 frequencies frequency
0 0.02 0.04 devices Utility power system
80
Nodal matrix
Passive
Easily happened under Primary & Secondary
Current / A

40 Dependent on the
Harmonic resonance conditions supply lines Zout ( s )
0 resonance frequency
instability
-40 Negative damping of Traction transformer
Out of train control
-80 train control system
0 0.08 0.16 Catenary (length & multiple)
Time / s

Fig. 9 An overview of the field problems in electric railway systems.

probabilistic method. However, the spectrum distribution of the


III. MODELING FOR THE HIGH SPEED TRAIN harmonic current would be influenced by the railway infra-
Considering the traction units in a high-speed train as a single structure, traction network and operation modes of the train.
electrical entry, the most effective way is to formulate re- Thus, this method of simplified modeling cannot reflect the
strictions on the input admittance, including the impacts from coupled relationship between the 4QCs of HSTs and the trac-
all the controllers in the 4QC of a HST. It can be done either by tion network, and the output admittance and the participation
defining allowed regions for magnitude and phases, or better feature were ignored.
for real and imaginary parts of the output admittance. The real The Norton model shown in Fig. 10 is usually employed to
part of the output admittance directly determines the damping represent the linear characteristic of the HST in harmon-
and/or excitation contribution. It reflects that no instability ic-domain [36]. As the resonance frequencies usually beyond
sources are presented, yet that the instability is always a prob- the switching frequency of the 4QC, it is reasonable to use the
lem of the interaction behavior [4]. passive model of the 4QC for representing the electric train in
studying harmonic resonance. The equivalent admittance has
A. Norton Equivalent Passively-based Model of the HST passive and active behaviors in terms of considering the effects
The train-network system behaves differently at different of the control system. The injected harmonic currents of the
frequency bandwidths. Mostly, recent publications regarding train can be presented as,
harmonic resonance problem have been concentrated on ana-  I s (1)   ys (1) 0 0 0   Vs (1)   ys (1)Vac (1) 
lyzing the former condition without considering the non-linear  I (2)   0 ys (2) 0 0  Vs (2)   ys (2)Vac (2) 
loads (NLLs) model or modeling them as a set of ideal current  s   
harmonics, which are not comprehensive enough. Many re-    0 0 0     
searchers found that the harmonic interaction between network        
 I s (n)   0 0 0 ys ( n)  Vs ( n )   ys (n )Vac (n ) 
supply voltages and injected current harmonics of NLLs was (1)
presented. Reference [31] described several uncoupled or cou-
where ys (h) , Vs (h) are the harmonic equivalent admittance and
pled NLL models such as Norton, full model [31], crossed
frequency admittance matrix (CFAM) [32] and frequency harmonic voltage on the secondary side of the traction trans-
coupled matrix (FCM) [33], and aimed to present the output former respectively, Vac ( h) and I ac ( h ) are the hth order har-
behaviors of the NLLs for harmonic studies. The comparison monic voltage and current of the grid-side of the train.
effects of different types of NLL models on the resonance
Is
behavior have also been taken into account in [34].
These models based on analytical and measured data can I ac ,2 I ac ,n
I ac ,1
satisfy major of harmonic and harmonic resonance studies in
the power system as well as the electric railway system. Vs y h 
However, dealing with the stability and oscillation which in-
volved the conv the comprehensive control system of the power
converters, these passively-based NLL models are not appli-
cable. One example is that the electric train model was usually
Fig. 10 Norton equivalent model of a train for harmonic resonance analysis.
replaced by the 1 pu ~ 50 pu harmonic current source with 1 pu
magnitude in [2] for the analysis of the harmonic propagation B. Detailed Actively-Based Model of the HST
and amplification. Alternatively, the harmonic current model of For passive harmonic analysis, Ytrain (s) is the admittance of
the electric train was established using the measured data and
Double Fourier series [35]. Reference [36] pointed out that the the onboard transformer based on an assumption that the sys-
spectrum of the HST harmonic current has strong randomness, tem is stable and the effect of the control system is neglected.
and need to calculate the random harmonic current with a However, the passivity Norton-equivalent model shown in Fig.
10 is also a limitation for investigating the instability or oscil-

0885-8993 (c) 2017 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TPEL.2017.2781880, IEEE
Transactions on Power Electronics
> REPLACE THIS LINE WITH YOUR PAPER IDENTIFICATION NUMBER (DOUBLE-CLICK HERE TO EDIT) < 7

lation problems. Therefore, a detailed model of the train in- The input admittance seen from the electric train in
volving variable controllers is considered in this subsection and low-frequency range affects the low-frequency voltage
is also called the actively-based model. variation just as the harmonic resonance and instability in
The traction drive system in a HST consists of the onboard high-frequency range. Based on the multi-time scale
transformer, single-phase 4QC, DC link, three-phase inverters, characteristics of the VSC control system [39]. The 4QC
and induction motors [37]. The front-end power converter of system adopts current controller with about 50~1000Hz
HSTs usually adopts a single-phase 4QC to achieve high power bandwidth, excited voltage or/and power controllers with about
factor, low current harmonics, and also bidirectional 10~100 Hz bandwidth. The dynamic characteristics of the
power-flow capability. The PWM of the 4QC and its control different controllers are listed in TABLE V. Note that the
system are of the primary concern for interaction studies and dynamic characteristics in different frequency bands are closely
the detailed control system is shown in Fig. 11. related to the corresponding controller parameters.
AC voltage measurement Synchronous rotating controllers TABLE V DIFFERENT CONTROLLER FREQUENCY BANDS
u ud
uac 
SOGI u uq 0~10Hz 100Hz~ kHz Several kHz
dq Controller
Low-fre. Medium high-fre. High-fre.
Voltage controller √ × ×
1 PLL iqref  0A
uq ud Current controller √ √ √
AC current measurement
i  id udcref Phase-locked loop
√ √ √
SOGI i iq idref controller
CC VC idcm Feed-forward low
dq × × √
uqref udref udc pass filter

1 dq
LPF LPF
1) Phase-Locked Loop
 DC The PLL acts as a synchronization control loop whose main
iac
ref
uPWM measurement function is to make the network voltage and current in same
idc idcm phase, which means that the power factor is close to the per unit.
1~
+ Lin Rin + The synchronization angle 1 from the PLL is used for
Transformer vac udc
uac PWM Cd  / dq transformation in 4QC control system. It can be found
- - that there are two synchronous rotating frames, i.e., network dq
=
Motor frame and 4QC dq frame. In this paper, the subscript c and g is
Fig. 11 The control block diagram of the 4QC.
adopt for distinguishing the 4QC dq frame and network dq
frame, for instance the voltage vector U g and current vector
The active power flow is controlled by the VC, keeping the
I g belong to the network dq frame and the U c and I c are in
DC-link voltage udc to its reference value u dcref by computing
4QC dq frame. The block diagram of the PLL is shown in Fig.
the reference active current idref for the CC. Based on the active
12.
and reactive current references idref and iqref , and the network U Ud
current iac and voltage uac , the CC gives the reference 0
U  /dq U q + 1
1
voltage u ref
for the inverter to pulse-width modulate udc into PI
PWM

an AC voltage vac . This voltage is referred to as the actual


- s

voltage since it is this voltage that the vehicle actually controls.


The phase of uac is estimated by the PLL, into the angle 1 . Fig. 12 Block diagram of the phase-locked loop.

Notice that for an AC power system, due to the sinusoidal There is the following relationship between the stationary
inputs of AC grids and nonlinear controllers, the input signal  frame and dq frame [40]:
may still has the time-varying behavior when using the average U  e j U g , U  e j U c ,   1  
1
(2)
model. And these nonlinearities cannot be removed
immediately with the standard linearization processes because The error angle  is zero in the steady state. There will be a
a well-defined operating point cannot be easily identified. phase-locked error owing to occur a small disturbance at
Therefore, a similar impedance-based model is not directly steady-state operating point, giving
applicable due to their sinusoidal behavior [38]. In order to U c  e  j  U g (3)
address the time-varying behavior, the dq transformation is Equation (3) can be linearized as,
used to obtain a nonlinear model with well-defined operating U c  (1  j  )(U d 0  U g )  U d 0  U g  jU d 0 
points in the dq frame. Then the system stability can be (4)
addressed by the generalized Nyquist stability criterion (GNC).  Im  U c   Im  U g   U d 0 
In addition, for the single-phase ac system, the second-order where the U q 0 is equal to zero in steady-state for unit power
generalized integrator (SOGI) is used to construct a virtual dq
component. Therefore, the single-phase AC system with dq factor. From Fig. 11, the phase-locked error can be expressed
decoupling is feasible to apply the small-signal impedance as,
analysis.

0885-8993 (c) 2017 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TPEL.2017.2781880, IEEE
Transactions on Power Electronics
> REPLACE THIS LINE WITH YOUR PAPER IDENTIFICATION NUMBER (DOUBLE-CLICK HERE TO EDIT) < 8

s1 = 0  FPLL (s) Im  U c   Fig. 13 Block diagram of the current controller.
(5) The output voltage signal of the CC is the input of the PWM
 s =0  s1 =FPLL (s) Im  U c 
controller. If the PWM delay and computation delay are con-
where FPLL ( s)=KpPLL  KiPLL / s is the transfer function of the sidered, we have
PI controller in the PLL. 0 is the fundamental angular fre- vdc  udref  1
quency, 0 =s . Combining from (2) to (5), one can obtain the  c   Gd ( s )  ref  , Gd ( s )  (12)
 vq  uq  Td s  1
following equation. where the time delay Gd ( s ) is first-order process, Td  1.5Ts
FPLL ( s )
s  w0  FPLL ( s ) s   1     Im  U g  (6) and Ts is sampling time [41].
s  U d 0 FPLL ( s )
Combining (9) to (12), the small-signal based transfer func-
GPLL ( s )
tion of the CC in dq-frame can be expressed as
where FPLL ( s)=KpPLL  KiPLL / s is the transfer function of the  idc   idref   udc 
PI controller in the PLL. Inserting (6) into (4). The transfer  c  G CC ( s )  ref   YCC ( s )  c (13)
function in matrix form of the PLL can be obtained by  iq   iq   uq 
 U dc   1 0   U d 
g where G CC ( s ) and YCC ( s) are the transfer function and input
 c
   g (7) admittance of the CC, respectively.
 U q   0 1  U d 0 GPLL ( s )   U q 
 1  d
G ( s ) FCC ( s ) 0 
GuPLL ( s ) G CC ( s )  H dq  
  0 G d ( s ) F CC ( s ) 
Correspondingly, the transfer function of the current syn-
 1  G ( s ) 0 
chronization system can be expressed as  1 d
(14)
 YCC ( s )  H dq 
 I dc   I dg   0 I q 0 GPLL ( s )   U dg    0 1  Gd ( s ) 
 c   g  g (8)  Gd ( s ) FCC ( s )  Rin  sLin (0 (1  Gd )) Lin 
 I q   I q   0 I d 0 GPLL ( s )   U q   H dq
1
 
GiPLL ( s ) 
   (  0 (1  G d )) Lin G d ( s ) F CC ( s )  Rin  sLin 

3) Voltage Controller
where I d 0 =P0 / U d 0 and I q 0 =Q0 / U d 0 are the steady-state
The VC needs to provide the active current reference signal
active and reactive currents. P0 and Q0 are output active I dref for the CC by power conservation principle, which called
power and reactive power of the 4QC, respectively. double voltage and current closed-loop control. The voltage
2) Current Controller vector and current vector of different coordinate systems are
Applying the law of Kirchhoff in Fig. 11, the input voltage of equally in steady state, i.e., U g =U c and I g =I c . Setting U g
the 4QC is given by
as a phase reference, if there is a small disturbance around the
Vg  U g  I g ( Rin  j0 Lin ) (9)
steady-state operating point, the network voltage and current
Equation (9) can be transformed to the 4QC dq frame, fol- are giving by
lowing the conversion relationship between the 4QC dq frame

 U g  U d 0  U d  j U q
g g
and network dq frame in (2), yields:  (15)
 I g  I d 0  I d  j ( I q 0  I q )
g g
vdc  udc   Rin  sLin 0 Lin  idc  
 c   c   (10)
 vq  uq   0 Lin Rin  sLin  iqc  Thus then, the instantaneous active and reactive powers at the
network side of the 4QC can be expressed as
The decoupling is added to the control loop in the form of an
 P  Re  U g I g   U d 0 I d 0  I d 0 U dg  I q 0 U qg  U d 0 I dg
inner positive-feedback loop with gain w0 Lin . From Fig. 13, the 
 P0 P
ref
voltage reference matrix U can be derived as  (16)
Q  Im  U g I g   U d 0 I q 0  I d 0 U q  I q 0 U d  U d 0 I q
g g g

u  u 
ref
d
c
d
i  idc   0 ref
d 0 Lin  idc  
      FCC ( s)  c
   (11)  Q0 Q
ref
u   u 
c
i  iq  0 Lin
ref
0  iqc 
q q q
Assuming that the switching process of the 4QC is ideal and
where FCC ( s)=KpCC  KiCC / s is the transfer function of the lossless, the powers between AC side and DC side satisfy the
CC. following relationship,
2
c 1 dudc u2
+ i ref
+ ud udg idg  u qg iqg 
2
 Cd
dt
 dc
RL
(17)
- d
-
PI
+ The VC is a simple proportional integral (PI) control, the
udref
idc main task of the VC is to keep the DC-link voltage udc fol-
0 L
0 L lowing its reference value u dcref . The transfer function of the VC
c
iq - uqref is
PI - idref  FVC ( s )  (udc
ref
 udc ) (18)
- +ref + uc
iq  0 q where FVC ( s)=KpVC  KiVC / s is the transfer function of the VC.
Lennart Harnefors in [40] introduced the modeling method of

0885-8993 (c) 2017 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TPEL.2017.2781880, IEEE
Transactions on Power Electronics
> REPLACE THIS LINE WITH YOUR PAPER IDENTIFICATION NUMBER (DOUBLE-CLICK HERE TO EDIT) < 9

the VC. Based on this method, after linearizing (17) and (18), traction network mainly contains the equivalent circuit model,
the small-signal transfer function of the VC can be derived as generalized symmetrical component model, and chain-circuit
 idref   Gdcd ( s ) Gdcq ( s )   u dg  model [42]. The equivalent circuit model with resistance–
 ref     g (19) inductance is widely used for tuning the control of inverters and
 iq   0 0   u q  studying the control performance [43]. For a single-phase ac
GVC ( s ) system, the voltage drop over a resistor Rs and an inductor
where the elements of GVC ( s ) can be expressed as Ls can be described in (23).
 P0 P0   di (t )
u S (t )  uac (t )  Rs  i (t )  Ls
 YCC (1,1)  2  2 G CC (1,1)  FVC ( s )  dt (23)
U U P
Gdcd ( s )    i (t )  id cos(0 t )  iq sin(0 t )
d0 d0
 02 (20) 
sCd  G CC (1,1) FVC ( s ) Ud0
where us (t ) is the output voltage of the traction substation,
Q0  U d20 YCC (1, 2)  FVC ( s )
G (s)   2
q
(21) uac (t ) is the input voltage of the electric train, i (t ) is the
U d 0  sCd  G CC (1,1) FVC ( s ) 
dc
network current. Transform (23) into the dq reference frame,
Combing (7), (8), (13) and (19), based on the above analysis, the network voltage dynamic is given in (24).
the small signal linearization block diagram of train’s 4QC  u dg ( s )   Rs +sLs 0 Ls   idg ( s ) 
 g    (24)
control system can be obtain in Fig. 14.
 u q ( s )   0 Ls Rs +sLs   iqg ( s ) 
U g U c I c I g Zs ( s)
GePLL YCC + + where  denotes the small deviation of the respective variable
I ref + + from the equilibrium point, s is the Laplace operator d/dt. 0
GVC GCC is the fundamental angular frequency, Rs and Ls are the
VC CC equivalent resistance and inductance of the network,
respectively. Z s ( s) is the required equivalent impedance of
GiPLL
the traction network. In order to investigate the interaction
PLL 4QC of a HST behaviors of the traction network and HST system in a same
Fig. 14 Small signal linearization block diagram of the 4QC control system of a voltage level, one can convert network voltage and impedance
HST. at the primary-side of the 4QC in the HST to its secondary side,
The input admittance matrix of the electric train including the i.e., Z S =Z S / k 2 and k is the onboard transformer ratio.
CC, VC and PLL is expressed as, B. Detailed Traction Network Model
I g ( s )=  G CC G VC  YCC G uPLL  G iPLL  U g ( s ) The essential elements of the traction power supply system
are traction substation (TSS), auto-transformers (ATs), traction
 idg ( s )  Ydd ( s ) Ydq ( s )   u dg ( s )  lines (including contact wires (CWs), messenger wires (MWs),
 g  (22)
 g 
 iq ( s )  Yqd ( s ) Yqq ( s )   u q ( s )  feeders, protect wires (PWs), integrated ground lines (IGLs),
Ytrain ( s ) rails). The utility power system supplies 220kV source to the
TSS that are step down to 2×27.5 kV by the V/x traction
where G uPLL ( s) and G iPLL ( s ) represent the voltage and current transformers in the TSS. The traction lines have doubled
transfer function of the PLL, respectively. YCC ( s) and 2×27.5 kV AT-fed traction network and connected the same
G CC ( s ) are the input admittance and transfer function in the type wires though crossing wires (CW) in ATs and section
posts (SPs) which installed approximately every 10~15 km
CC, respectively. GVC ( s) represent the transfer function of the along the track.
VC. 1) Traction Transformers
The model of the traction transformer installed in the traction
IV. TRACTION NETWORK MODELING substation has been widely presented in previous work [3]. V/x
Different from the analysis method of the LFO and harmonic connection transformer is the most popular type in China
instability, the harmonic resonance depends more information high-speed railways for the its simple connection,
about the network topology and its parameters, and also high-capacity utilization and the better solution of the unbal-
represent the train as a passive harmonic model. Therefore, a ance problem. The model of this transformer can be found in
detailed network model including all network components is [3]. Alternatively, reference [2] also proposed a simulation
built in this section for analyzing the influential factors of the model for Scott-connection transformer widely adopted in
harmonic resonance. Korea. The numerous models of representative traction trans-
formers, such single-phase, V/v, Scott and Balanced trans-
A. Equivalent Traction Network Model for LFO and Har- former, can be found in [44]. Moreover, the modeling for the
monic Instability Analysis AT is similar to the traditional autotransformers [45]. The
The all-parallel AT-fed catenary system is generally adopted aforementioned electrical modeling of different traction electric
in China high-speed railway. The mathematical model of the devices is combined in a unified system in terms of a nodal
admittance matrix, which is used to study the low-frequency

0885-8993 (c) 2017 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TPEL.2017.2781880, IEEE
Transactions on Power Electronics
> REPLACE THIS LINE WITH YOUR PAPER IDENTIFICATION NUMBER (DOUBLE-CLICK HERE TO EDIT) < 10

and high-frequency electrical characteristics. Z1n Z2n Z3n


2) Capacitance and Impedance of the Traction Conductors Y1n Y1n Y2n Y2n Y3n Y3n
The traction line that contains the up and down tracks is an- Node: 3n+1~4n jn+1~(j+1)n
1~n n+1~2n
alyzed by a 5-conductors model. It contains up contact wires,
up feeders, rails, down contact wires and down feeders. The
construction of the catenary conductors is different with vari- MTL MTL MTL
ous geographical conditions, leading to different parameters
and algorithms in the traction conductors model. The self- and
Slices:
mutual impedance of conductors can be calculated with com- TSS ATS Trains SP
plex depth equation in (25) [46,47].
 2h Segment 1: l1 Segement 2: l 2 Segment i: li
Z ii  j 0 ln i  2  Rii  j X ii 
2 ri Fig. 16 Schematic of traction MTL segments.
, (25)
 D
Z ij  j 0 ln ii  2 Rij  j X ij
2 d ij
  Each segment is represented by a Π-equivalent circuit matrix
with 2n×2n elements. The segments of the catenary network are
where the first item in (25) is line impedance in the case of ideal resulted from the slices of the dynamic positions of running
earth. The second item is correction item when the earth is not trains and fixed positions of AT substations, section posts and
ideal. Additionally, the correction item is an infinite integral traction substations. Moreover, these catenary conductors can
from in the traditional Carson model. be represented as parallel transmission lines with self- and
As to the self- and mutual- admittance of conductors, poten- mutual- impedances and admittances and these parameters are
tial coefficient can be obtained by (26) frequency dependent due to skin effect. With some assumptions
Dij in [48], the MTL can be obtained by a set of partial different
1 2h 1
pii  ln i (F/m) 1 , pij  ln (F/m) 1 (26) equations in time-domain (see (27)).
2 0 req 2 0 d ij
  
where pii and pij are the self- and mutual- potential coeffi- V  x, t    RI  x , t   L I  x , t  
dx t 
cients, respectively. And  0 is the permittivity of vacuum,  (27)
 
I  x, t   GI  x, t   C V  x, t  
 0  8.854188  1012  F / m  . The structural parameters in (25) x t 
and (26) are illustrated in Fig. 15. The capacitance matrix can where V is the voltage vector between the catenary conduc-
be derived from the inverse matrix of the potential coefficient tors and the earth, I is the current vector through the catenary
1
matrix P , given as C  P . The admittance matrix of the conductors, and R , L , G and C are the 2n×2n matrices of
traction network conductors is Y =j 2 f  C . line parameters per unit length. Based on the eigenvalues de-
composition, one can get
ri ZY = T1ΛT (28)
hij rj where, Z and Y are the impedance matrix and admittance
matrix of an MTL, respectively, and Z  R +jhL ,
hi Dij hj Y  G +jhC , h is the harmonic order and  is the funda-
Ground mental angular frequency.
Each section can be equivalent to a multi-conductor Π-type
hj circuit that Z L represents impudence matrix and YL / 2 rep-
hi x ij Mirror
resents admittance matrix, shown in Fig. 16. Z L and YL / 2 ,
2n×2n complex symmetric matrices, can be obtained by (29)
through the phase-mode transformation
 Z L  T  Λ 1/ 2 sinh  Λ 1/ 2  l   T 1Y 1
Fig. 15 Conductor geographical structure.   
 (29)
 YL  2  Z T  Λ tanh  Λ  l / 2   Y
1 1/ 2 1/ 2 1
3) Multiple Transmission Line (MTL) Model for the Catenary
Network where l is the length of a MTL segment.
The catenary network is another important component in As a consequent, the whole traction network can be modeled
studying the train-network interaction. The multiple catenary in a unified modal matrix with jointing all segments. Each
network conductors can be regarded as one symmetrical linear section cut by these slices could be regarded as a passive and
and passive multi-port network. A MTL model is widely ac- homogeneous multi-conductor transmission line. Therefore,
cepted for modeling different segments of the catenary network the equivalent node-admittance matrix can be described by (30)
conductors which is illustrated in Fig. 16. through a series of MTL matrices in Fig. 16.

0885-8993 (c) 2017 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TPEL.2017.2781880, IEEE
Transactions on Power Electronics
> REPLACE THIS LINE WITH YOUR PAPER IDENTIFICATION NUMBER (DOUBLE-CLICK HERE TO EDIT) < 11

 Y1n  Z1n
1 1
 Z1n 
 1 1 1 
  Z1n Z  Y2n  Z
1n 2n 
Ynet =   (30)
 1 1 1

 Z (N  2) n  Y(N 1) n  Z (N 1) n  Z (N 1) n 
 1
 Z (N 1 
 1) n Z (N 1) n  YNn 

The nodal injection current of the catenary network is com- conductors are mainly capacitive in higher frequency region.
posed of these slices. The detailed traction conductors’ parameters and their types
I net = I1 I2 I N-1 I N h
T
(31) can be found in Fig. 18. In order to practically assess the ap-
proximate effects of the main electrical elements in harmonic
The injection current vectors change with the state of muta- resonance, LFO and harmonic instability, we listed the typical
tional slices. Take the location of a train as an example, if the impedance values and their ratios of the total impedance in
slice i represents the location of a HST, the current vector in TABLE VI. The impedance ratio represents the ratio of the
this slice is not zero vector. Otherwise, if the HST located on impedance of each component to the total impedance of the
T
the up track, then I i    I itrain 0 I itrain 0 0 h , else network. Practically, the utility grid, primary supply line,
T
transformer, secondary supply line and catenary lines for a
I i  0 0 I itrain  I itrain 0 h . conventional railway (110 kV, 600 MVA), will be about 8.3%,
4) Impedance Ratio of each Network Components: A Simple 4.6%, 41.3%, 0.0% and 45.8% and for the HSR (220 kV, 4.5
Case GVA), where the impedance ratio is about 2.2%, 9.3%, 33.0%,
The traction network subsystem includes the utility power 0.1% and 55.4%, respectively. TABLE VI summarizes these
grid, primary/secondary supply lines, traction transformer, parameters and they can be practical for assessing and
catenary lines and autotransformers. Majority of the traction mitigating the power quality phenomena. The impedance ratios
network elements is resistive-inductive in low-frequency re- of different elements also affect their influential degrees on the
gion, while the presence of secondary supply lines and catenary harmonic resonance and oscillations.
TABLE VI TYPICAL IMPEDANCE RATIO OF EACH NETWORK COMPONENTS

Classification Total impedance / Ω Utility Primary line Transformer Secondary line Catenary
Conventional railway 30.52 8.3% 4.6% 41.3% 0.0% 45.8%
High speed railway 15.14 2.2% 9.3% 33.0% 0.1% 55.4%
Note: the length of primary line, secondary line and catenary are set as 5 km, 100 m and 25 km, respectively.

TABLE VII DYNAMICS OF THE 4QC UNDER DIFFERENT FREQUENCY WIDTH

Frequency width Low-frequency Medium high-frequency High-frequency


Problems Low frequency oscillation Harmonic instability Harmonic resonance
Characteristic equation G ol (s)  I  nYtrain (s)ZS (s) G ol (s)  I  nYtrain (s)ZS (s) ×
Generalized Nyquist criterion: Generalized Nyquist criterion:
Criterion ×
nZ S ( s)Ytrain ( s) nZ S ( s)Ytrain ( s)
Dynamics part VC, PLL, CC PLL, CC, Td ×
Note: VC is the outer voltage controllers, CC is the inner current controller, PLL is the phase-locked loop and Td is time delay.

current controller, outer voltage/power controller, PLL and


V. TRAIN-TRACTION NETWORK INTERACTION MODEL related parameters into consideration. In addition, according to
Many power delivery and harmonic distortion problems in the characteristic of multi-time scale characteristic in the
traction system are due to interactions between the traction control system of inverter [39], it is necessary to consider the
network and the electric trains. Fig. 17 shows the low-frequency, medium high-frequency, high-frequency and
impedance-based diagram of train-network system. the corresponding dynamic components are shown in TABLE
VII.
Network-side Train-side
From TABLE VII, there are four factors need to be
I +
considered in the detail impedance model.
Primary Secondary Catenary
supply line TSS supply line network 1) the total time delay Td, i.e., computation-plus-PWM
US U 2) the phase-locked loop,
ZS =RS  sLS 3) the inner current control dynamics, and
I train -1 Ztrain -1 I trainn Ztrain -n
- 4) the dynamics of the outer power/voltage controllers, i.e.,
the voltage control and power control, provided that such are
Fig. 17 Impedance-based analysis of the network system and train system. used, for the reactive power or the point-of-common-coupling
In this study, the LFO and instability problem is paid attention (PCC) voltage magnitude.
to from the point of the system. It is therefore only taken inner Causes 1), 2) and 3) affect the harmonic instability, whereas

0885-8993 (c) 2017 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TPEL.2017.2781880, IEEE
Transactions on Power Electronics
> REPLACE THIS LINE WITH YOUR PAPER IDENTIFICATION NUMBER (DOUBLE-CLICK HERE TO EDIT) < 12

causes 3) and 4) affect the LFOs. Train uac udc


Ls Rs DSP
iac idc
A. Generalized Simulation Model for Case Studies +
Cs Train1 Train2 1~
A generalized simulation model, consisting of power utility e* Lin
Cd RL

240V
-S

20V
PWM
system, traction system, and trains, is presented for evaluating
the interactions between trains and traction network. The pa- Network =

rameters are listed in Fig. 18 and TABLE VIII~Ⅹ in Appendix (b)


A. Fig. 19 Experimental platform of the multi-train and network model for LFO
and harmonic instability studies: (a) circuit diagram; (b) platform diagram.
B. Generalized Experiment Platform for Case Studies
C. Impedance-Based Train-Network Model Validation
Experimental platform of a small-power multiple trains and
network system have been conducted. The hardware picture is In order to validate the mathematical model, a generalized
shown in Fig. 19. Where the Ls and Rs represent the equivalent simulation model and an experimental platform have been
impedance between the first train and power source. Cs is conducted. The impedance behaviors of the traction network
network equivalent capacitance, which is used to construct the and HSTs have been compared with the simulation results.
resonant point. The main circuit and controller structures of Here, the frequency-scanning method is used for the computing
trains are the same with the simulation model, but the param- impedance of the simulation results.
eters of experimental system in Fig. 19 have different values 1) Traction Network Impedance Validation
with the simulation example. In the experimental system, the Fig. 20 shows the utility test results of traction network im-
source is AC 240V, the switching frequency is set at 1250Hz, pedance versus frequency through the frequency-scanning
and the control system of 4QC is implemented into DSP method. In this case, the traction network modeled in subsec-
(TMS320F28335). The high-speed 12bit A/D board, tion IV(B) consists of power system, primary supply line, V/x
DSO-X3034A, is used for sampling the voltage and currents, transformer, secondary supply line and multiple transmission
and the digital wave-form output board, TX-DA962D4, is used lines. This structure is similar to the simulation model in Fig. 18,
for the PWM generation. In addition, a custom docking board and the parameters listed in TABLE VIII~X. Meanwhile, the
with voltage and current sensing was designed. test frequency is from 50 Hz to 3 kHz and the step is 50 Hz. The
impedance via frequency curve of traction network is shown in
Power system Fig. 20, where the red solid line is from calculated and the blue
Capacity: 4.5GVA circle ‘○’ are from detailed measured. The measured results
A X/R ra tio: 14 Primary su pply lin e
Z=0.078 75+j0.414 3Ω/km
matched well with the mathematical method. It is valid that the
B
C=8.6nF/km simulation system can represent the critical resonance and can
C
V/x transformer (TSS) be applied for the case studies. The experiment system includes
220kV Capacity: 31.5/20/20 MVA a simple RLC-circuit for representing the critical resonance
Ratio: 220/27.5X2 kV
Sho rt circu it volta ge: 10.42%
points and low-frequency impedance. It is approximated yet
Second ary suppl y line reliable and valid to perform the works and case studies of this
27.5kV
Z=0.047+j0.083Ω/km paper.
C=179nF/km
1600
Tα Tβ Zn Calculation
Measurement
Magnitude / Ω

Rα Rβ
Yn Yn 1200 Experiment

Fα TSS
Fβ 800
Catenary Trains (Table XI)
(Table IX & X)

Fig. 18 Sketch of the train-network simulation model.


400

0
Voltage 0 500 1000 1500 2000 2500 3000
regulator Frequency / Hz
Fig. 20 Impedance via frequency curve of the traction network. Red solid line:
Host calculation results; Blue circle: measurement results; Green fork line: experi-
Oscilloscope
computer ment results.
Train 1 Train 2
2) Train Impedance Validation
In order to validate the proposed calculated impedance model
of the 4QC of the HST, the frequency-scanning method is
applied. The impedance frequency responses of the 4QC is
measured from 25 Hz to 1000 Hz and the step is 25 Hz. Fig. 21
Network depicts the measured impedance frequency responses
over-plotted with calculated results using (22) for unit power
Controller Controller
factor control. The parameters are listed in TABLE XI and the
(a)
load resistance of train is set as RL =5  . In Fig. 21, the red
solid line is from calculated and the blue circle ‘○’ are from

0885-8993 (c) 2017 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TPEL.2017.2781880, IEEE
Transactions on Power Electronics
> REPLACE THIS LINE WITH YOUR PAPER IDENTIFICATION NUMBER (DOUBLE-CLICK HERE TO EDIT) < 13

simulation measured. The results show good accuracy of the and up to 20 S/km . The Z1  6.915  j160.1 ,
proposed impedance model of the train. As a result, the
train-network simulation and experiment systems can be able to Z 2  Z 3  0.268  j 3.94 ; the impedance of the AT is
studying the harmonic and instability problems mentioned Z 2  0.667  j1.607 .
earlier.
40
TABLE VIII MAJOR PARAMETERS OF POWER SYSTEM, TRACTION
30
TRANSFORMER AND AT TRANSFORMER
20
Magnitude / dB

30

Magnitude / dB
20
10 Components Parameters Values
0 Nominal capacity /MVA 31.5/20/20
10
-10
V/x transform- Ratio / kV/kV 220 / 2×27.5
0 -20
180 180 er Impedance voltage 10.42 %
90 Load loss /kW 141.76
Phase / deg

Phase / deg

90
0
Nominal capacity /MVA 16
0 Ratio / kV/kV 55 / 27.5
-90
Impedance voltage 1.7 %
-90 1 -180 1 AT transformer
10 10
2
10
3
10 10
2
10
3
Load loss /kW 56.412
Frequency / Hz Frequency / Hz
(a) (b) No-load loss /kW 11.313
25 40 No-load current 0.071 %
Magnitude / dB

30
Magnitude / dB

20
20 TABLE IX TRACTION NETWORK IMPEDANCE AT 50 HZ (Ω / KM)
15
10
T1 F1 R T2 F2
10 0 0.0781 0.0501 0.0497 0.0484 0.0482
180 180 T1
+0.6203i +0.3276i +0.3193i +0.3357i +0.3002i
0.0501 0.0956 0.0539 0.0482 0.0485
F1
Phase / deg

Phase / deg

90 90
+0.6376i +0.7361i +0.3185i +0.3002i +0.2862i
0.0497 0.0539 0.0568 0.0497 0.0539
0 0 R
+0.3193i +0.3185i +0.3991i +0.3193i +0.3185i
0.0484 0.0482 0.0495 0.0781 0.0501
-90 1 -90 1 T2
10 10
2
10
3
10 10
2
10
3
+0.3357i +0.3002i +0.3193i +0.6203i +0.3276i
Frequency / Hz Frequency / Hz 0.0482 0.0485 0.0485 0.0501 0.0956
(c) (d) F2
+0.3002i +0.2862i +0.3184i +0.3276i +0.7361i
Fig. 21 Calculated and measured impedance via frequency curve of the train.
Red solid line: calculated results. Blue circle: measured results. (a) Zdd; (b) Zdq;
(c) Zqd; (d) Zqq. TABLE X TRACTION NETWORK CAPACITANCE AT 50HZ (NF /KM)

T1 F1 R T2 F2
VI. CONCLUSION T1 12.02 1.14 3.64 1.86 0.37
This paper presented the field measurements for F1 1.14 8.75 3.74 0.37 0.14
demonstrating the train-network interaction phenomena R 3.64 3.47 67.23 3.64 3.47
including the LFO, harmonic resonance and harmonic T2 1.86 0.37 3.64 12.02 1.14
instability in high-speed railway systems. An impedance-based F2 0.37 0.14 3.47 1.14 8.75
models of multiple trains and traction network interactions
system considering available components and parameters is
TABLE XI PARAMETERS OF THE TRAIN-NETWORK SYSTEM FOR LFO AND
presented to investigate the formation mechanism and HARMONIC INSTABILITY ANALYSIS
impedance behaviors of the three phenomena. The detailed
modeling of the trains and traction network have been Simulation Experiment
Symbol Description
value value
represented to construct a unified train-network numerous and Secondary side voltage of
experimental system. The low-, medium- and high- frequency e S* 1770 V 240 V
the onboard transformer
input admittance behavior of the electric train is the primary f0
Network voltage fre-
50 Hz 50 Hz
task as well as the linear traction network to investigate quency
train-network interactions. The output impedance of the k Onboard transformer ratio 15.54 12
25 mH for LFO 30 mH for LFO
traction network and train are obtained by both theoretical Network equivalent and 12.5 mH for and 16.7 mH for
analysis and measurements. The results can be effective for LS
inductance harmonic insta- harmonic insta-
investigating the traction system compatibility for modern bility bility
electric train system. Finally, the generalized simulation model 12.6 mΩ for
LFO and 6.3
and experiment are developed where all available factors or RS
Network equivalent
mΩ for har- 0Ω
parameters are included. resistance
monic instabil-
ity
APPENDIX A Network equivalent
C S 2.8 F 2.1 F
capacitance
The rail-to-earth conductance G is 2 S/km (typical value) Lin
Leakage inductance of the
6 mH 14 mH
onboard transformer

0885-8993 (c) 2017 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TPEL.2017.2781880, IEEE
Transactions on Power Electronics
> REPLACE THIS LINE WITH YOUR PAPER IDENTIFICATION NUMBER (DOUBLE-CLICK HERE TO EDIT) < 14

Leakage resistance of the sion in high-speed traction system," Proc. Asia-Pacific Power and Energy
Rin 0.01 Ω 0Ω
onboard transformer Eng. Conf., pp. 1-4, 2011.
1000 Ω for LFO 1000 Ω for LFO [13] M. Brenna, F. Foiadelli and D. Zaninelli, "Electromagnetic model of high
and 50 Ω for and 50 Ω for speed railway lines for power quality studies," IEEE Trans. Power Syst.,
RL Load resistance
harmonic insta- harmonic insta- vol. 25, no. 3, pp. 1301-1308, Mar. 2010.
bility bility [14] A. Dolara, M. Gualdoni and S. Leva, "Impact of high-voltage primary
Cd DC-link capacitance 9 mF 4.4 mF supply lines in the 2 x 25 kV-50 Hz railway system on the equivalent im-
pedance at pantograph terminals," IEEE Trans. Power Del., vol. 27, no. 1,
Td Time delay 1.2 ms 1.2 ms pp. 164-175, Oct. 2012.
Switching frequency of [15] M. Brenna, A. Capasso, M. C. Falvo, F. Foiadelli, R. Lamedica and D.
fS 1250 Hz 1250 Hz
the 4QC Zaninelli, "Investigation of resonance phenomena in high speed railway
U dc DC voltage of the 4QC 3600 V 40 V supply systems: Theoretical and experimental analysis," Electric Power
Systems Research, vol. 81, no. 10, pp. 1915-1923, Oct. 2011.
d channel voltage of the
Ud0 2503 V 20 V [16] M. A. Eitzmann, J. J. Paserba, J. M. Undrill and C. Amicarella, " Model
4QC development and stability assessment of the Amtrak 25 Hz traction system
U q0 q channel voltage of the from New York to Washington DC," Proc. Railroad Conf., Boston, MA,
0V 0V
4QC
USA, pp. 21-28, 1997.
Kp_VC Proportional gain of the [17] H. B. P. Schmidt, "Die dezentrale Bahnenergieversorgung von 162/3-Hz
1 0.1
4QC voltage controller Einphasen-wechselstrombahnen über synchron-synchorn-umformer," El-
Ki_VC Integrator gain of the
0.5 0.025 ektrische Bahnen, vol. 11, no. 89, pp. 179-181, 1991.
4QC voltage controller [18] S. Danielsen," Electric traction power system stability: Low-frequency
Kp_ PLL Proportional gain of the interaction between advanced rail vehicles and a rotary frequency con-
2 1
4QC phase-locked loop
verter,". Ph.D. dissertation, Radiat. Lab., Norwegian Univ. Sci. Technol.,
Ki_ PLL Integrator gain of the Trondheim, Norway, 2010.
60 30
4QC phase-locked loop [19] S. Danielsen," Electric traction power system stability,". Trondheim:
3 for LFO and 7
Proportional gain of the Norwegian University of Science and Technology, 2010.
Kp_ CC 0.8 for harmonic
4QC current controller [20] D. Frugier and P. Ladoux, " Voltage disturbances on 25kV-50 Hz railway
instability
lines – Modelling method and analysis," Proc. of Speedam, pp.
Ki_ CC Integrator gain of the
2 0.15 1080-1085, 2010.
4QC current controller
[21] B. S. M. Pröls, "Stabiltätskriterien für Wechselwirkungen mit Um-
richteranlagen in Bahnsystemen," Elektrische Bahnen, vol. 11, no. 104, pp.
542-552, 2006.
REFERENCES [22] M. Debruyne, " Low frequency instability in the Amtrak's 12kV-25Hz
network," Proc. of Interaction Workshop 2006 Thun, Switzerland, pp.,
[1] E. Mollerstedt and B. Bernhardsson, "Out of control because of harmon- 2006.
ics-an analysis of the harmonic response of an inverter locomotive," IEEE [23] M. M. Menth S, "Low frequency power oscillations in electric railway
Control Systems, vol. 20, no. 4, pp. 70-81, Aug. 2000. systems," Elektrische Bahnen, vol. 104, no. 4, pp. 216-221, 2006.
[2] H. Lee, C. Lee, G. Jang and S. H. Kwon, "Harmonic analysis of the Korean [24] W. Song, S. Jiao, Y. W. Li and J. Wang, "High-frequency harmonic res-
high-speed railway using the eight-port representation model," IEEE onance suppression in high-speed railway through single-phase traction
Trans. Power Del., vol. 21, no. 2, pp. 979-986, Mar. 2006. converter with LCL filter," IEEE Trans. Transp. Electrif., vol. 2, no. 3, pp.
[3] Z. He, H. Hu, Y. Zhang and S. Gao, "Harmonic resonance assessment to 347-356, Jun. 2016.
traction power-supply system considering train model in China high-speed [25] R. Kadhim and D. Kelsey, " 25 kV Harmonic resonance modelling on the
railway," IEEE Trans. Power Del., vol. 29, no. 4, pp. 1735-1743, Oct. channel tunnel rail link," Institution of Engineering and Technology Sem-
2013. inar on EMC in Railways, Birmingham, England, pp. 137-151, Mar. 2006.
[4] EN 50388 Standard - Power supply and rolling stock - Technical criteria [26] L. Sainz, M. Caro and E. Caro, "Analytical study of the series resonance in
for the coordination between power supply and rolling stock to achieve power systems with the Steinmetz circuit," IEEE Trans. Power Del., vol.
interoperability, 2012. 24, no. 4, pp. 2090-2098, Sep. 2009.
[5] H. Wang, M. Wu and J. Sun, "Analysis of low-frequency oscillation in [27] J. D. Ainsworth, "Harmonic instability between controlled static conver-
electric railways based on small-signal modeling of vehicle-grid system in tors and a.c. networks," Proc. Inst. of Elect. Eng., vol. 114, no. 7, pp.
dq frame," IEEE Trans. Power Electron., vol. 30, no. 9, pp. 5318-5330, 949-957, 1967.
Jan. 2015. [28] L. H. Kocewiak, J. Hjerrild and C. L. Bak, "Wind turbine converter control
[6] S. Danielsen, M. Molinas, T. Toftevaag and O. B. Fosso, "Constant power interaction with complex wind farm systems," IET Renewable Power
load characteristic's influence on the low-frequency interaction between Generation, vol. 7, no. 4, pp. 380-389, Jun. 2013.
advanced electrical rail vehicle and railway traction power supply with [29] J. H. R. Enslin and P. J. M. Heskes, "Harmonic interaction between a large
rotary converters," Electromotion, vol. 17, no. 1, pp. 61, Jan. 2010. number of distributed power inverters and the distribution network," IEEE
[7] L. Harnefors, A. Yepes, A. Vidal and J. Doval-Gandoy, "Passivity-based Trans. Power Electron., vol. 4, no. 6, pp. 1586-1593, Nov. 2004.
controller design of grid-connected VSCs for prevention of electrical [30] J. Rocabert, A. Luna, F. Blaabjerg and P. RodríGuez, "Control of power
resonance instability," IEEE Trans. Ind. Electron., vol. 62, no. 2, pp. converters in AC microgrids," IEEE Trans. Power Electron., vol. 27, no.
702-710, Feb. 2015. 11, pp. 4734-4749, May. 2012.
[8] K. Song, M. Wu, V. G. Agelidis and H. Wang, "Line current harmonics of [31] A. B. Nassif, J. Yong and W. Xu, "Measurement-based approach for
three-level neutral-point-clamped electric multiple unit rectifiers: analysis, constructing harmonic models of electronic home appliances," Proc. IEE
simulation and testing," IET Power Electronics, vol. 7, no. 7, pp. Gener. Transmiss. Distrib., vol. 4, no. 3, pp. 363-375, 2010.
1850-1858, Jul. 2014. [32] M. Fauri, "Harmonic modelling of non-linear load by means of crossed
[9] G. W. Chang, H. W. Lin and S. K. Chen, "Modeling characteristics of frequency admittance matrix," IEEE Trans. Power Syst., vol. 12, no. 4, pp.
harmonic currents generated by high-speed railway traction drive con- 1632-1638, Nov. 1997.
verters," IEEE Trans. on Power Del., vol. 19, no. 2, pp. 766-773, Mar. [33] Y. Sun, G. Zhang, W. Xu and J. G. Mayordomo, "A harmonically coupled
2004. admittance matrix model for AC/DC converters," IEEE Trans. on Power
[10] X. Wang, F. Blaabjerg and W. Wu, "Modeling and analysis of harmonic Syst., vol. 22, no. 4, pp. 1574-1582, Oct. 2007.
stability in an AC power-electronics-based power system," IEEE Trans. [34] M. Esmaili, H. A. Shayanfar and A. Jalilian, "Modal analysis of power
Power Electron., vol. 29, no. 12, pp. 6421-6432, Feb. 2014. systems to mitigate harmonic resonance considering load models," Energy,
[11] L. Guo, Q. Li and Y. Xu, " Study on harmonic resonance of traction line in vol. 33, no. 9, pp. 1361-1368, Sep. 2008.
electrified high-speed traction system," Proc. Int. Conf. Sustain. Power [35] K. Song, M. Wu, V. G. Agelidis and H. Wang, "Line current harmonics of
Gen. Supply, pp. 1-4, 2009. three-level neutral-point-clamped electric multiple unit rectifiers: analysis,
[12] H. Hu, M. Zhang, C. Qian, Z. He and L. Fang, " Research on the harmonic simulation and testing," IET Power Electron., vol. 7, no. 7, pp. 1850-1858,
transmission characteristic and the harmonic amplification and suppres-

0885-8993 (c) 2017 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TPEL.2017.2781880, IEEE
Transactions on Power Electronics
> REPLACE THIS LINE WITH YOUR PAPER IDENTIFICATION NUMBER (DOUBLE-CLICK HERE TO EDIT) < 15

Jul. 2014. Frede Blaabjerg (S’86–M’88–SM’97–F’03) was


[36] H. Hu, Z. He, X. Li and K. Wang, "Power-quality impact assessment for with ABB-Scandia, Randers, Denmark, from 1987 to
high-speed railway associated with high-speed trains using train timetable 1988. From 1988 to 1992, he got the PhD degree in
—Part I: methodology and modeling," IEEE Trans. Power Del., vol. 2, no. Electrical Engineering at Aalborg University in 1995.
31, pp. 693-703, Oct. 2015. He became an Assistant Professor in 1992, an
[37] S. Danielsen, O. B. Fosso, M. Molinas, et al, "Simplified models of a Associate Professor in 1996, and a Full Professor of
single-phase power electronic inverter for railway power system stability power electronics and drives in 1998. From 2017 he
analysis — development and evaluation," Electric Power Systems Re- became a Villum Investigator.
search, vol. 80, no. 2, pp. 204-214, Feb. 2010. His current research interests include power
[38] S. Lissandron, L. D. Santa, P. Mattavelli and B. Wen, "Experimental electronics and its applications such as in wind
validation for impedance-based small-signal stability analysis of sin- turbines, PV systems, reliability, harmonics and adjustable speed drives. He has
gle-phase interconnected power systems with grid-feeding inverters," published more than 500 journal papers in the fields of power electronics and its
IEEE J. Emerg. Sel. Topics Power Electron., vol. 4, no. 1, pp. 103-115, applications. He is the co-author of two monographs and editor of 6 books in
Sep. 2015. power electronics and its applications.
[39] M. Zhao, X. Yuan, J. Hu and Y. Yan, "Voltage dynamics of current control He has received 24 IEEE Prize Paper Awards, the IEEE PELS Distinguished
time-scale in a VSC-connected weak grid," IEEE Trans. on Power Syst., Service Award in 2009, the EPE-PEMC Council Award in 2010, the IEEE
vol. 31, no. 4, pp. 2925-2937, Nov. 2016. William E. Newell Power Electronics Award 2014 and the Villum Kann
[40] L. Harnefors, M. Bongiorno and S. Lundberg, "Input-admittance calcula- Rasmussen Research Award 2014. He was the Editor-in-Chief of the IEEE
tion and shaping for controlled voltage-source converters," IEEE Trans. TRANSACTIONS ON POWER ELECTRONICS from 2006 to 2012. He has
Ind. Electron., vol. 54, no. 6, pp. 3323-3334, Nov. 2007. been Distinguished Lecturer for the IEEE Power Electronics Society from
[41] X. Wang, L. Harnefors and F. Blaabjerg, "A unified impedance model of 2005 to 2007 and for the IEEE Industry Applications Society from 2010 to
grid-connected voltage-source converters," IEEE Trans. Power Electron., 2011 as well as 2017 to 2018.
vol. PP, no. 99, pp. 1, Apr. 2017. He is nominated in 2014, 2015, 2016 and 2017 by Thomson Reuters to be
[42] Z. He, Z. Zheng and H. Hu, "Power quality in high-speed railway systems," between the most 250 cited researchers in Engineering in the world. In 2017 he
International Journal of Rail Transportation, vol. 4, no. 2, pp. 71-97, Feb. became Honoris Causa at University Politehnica Timisoara (UPT), Romania.
2016.
[43] Y. Wang, X. Wang, F. Blaabjerg and Z. Chen, "Harmonic instability Xiongfei Wang (S’10-M’13-SM’17) received the
assessment using state-space modeling and participation analysis in in- B.S. degree from Yanshan University, Qinhuangdao,
verter-fed power systems," IEEE Trans. Ind. Electron., vol. 64, no. 1, pp. China, in 2006, the M.S. degree from Harbin Insti-
806-816, Jul. 2016. tute of Technology, Harbin, China, in 2008, both in
[44] T. H. Chen and H. Y. Kuo, "Network modelling of traction substation electrical engineering, and the Ph.D. degree in
transformers for studying unbalance effects," Proc. IEE Gener. Transmiss. energy technology from Aalborg University, Aal-
Distrib., vol. 142, no. 2, pp. 103-108, Mar. 1995. borg, Denmark, in 2013. Since 2009, he has been
[45] R. Cella, G. Giangaspero, A. Mariscotti and A. Montepagano, "Meas- with the Aalborg University, Aalborg, Denmark,
urement of AT electric railway system currents at power-supply frequency where he is currently an Associate Professor in the
and validation of a multi-conductor transmission-line model," IEEE Trans. Department of Energy Technology. His research interests include modeling and
Power Del., vol. 21, no. 3, pp. 1721-1726, Jun. 2006. control of grid-connected converters, harmonics analysis and control, passive
[46] M. Wu, "Physical interpretation of impedance formulas for conductors and active filters, stability of power electronic based power systems.
enclosed in a cylindrical tunnel," IEEE Trans. Power Del., vol. 26, no. 3, Dr. Wang serves as an Associate Editor for the IEEE Transactions on Power
pp. 1354-1360, Apr. 2011. Electronics, the IEEE Transactions on Industry Applications, and the IEEE
[47] M. Wu and Y. Fang, "Numerical calculations of internal impedance of Journal of Emerging and Selected Topics in Power Electronics. He is also the
solid and tubular cylindrical conductors under large parameters," Proc. Guest Editor for the Special Issue “Grid-Connected Power Electronics Systems:
IEE Gener. Transmiss. Distrib., vol. 151, no. 1, pp. 67-72, 2004. Stability, Power Quality, and Protection” in the IEEE Transactions on Industry
[48] M. Brenna, F. Foiadelli and D. Zaninelli, "Electromagnetic model of high Applications. He received the second prize paper award and the outstanding
speed railway lines for power quality studies," IEEE Trans. on Power Syst., reviewer award of IEEE Transactions on Power Electronics in 2014 and 2017,
vol. 25, no. 3, pp. 1301-1308, Aug. 2010. respectively, the second prize paper award of IEEE Transactions on Industry
Applications in 2017, and the best paper awards at IEEE PEDG 2016 and IEEE
Haitao Hu (S’13-M’14), received the B.S. degree PES GM 2017.
from Zhengzhou University, Zhengzhou, China, in
2010, and the Ph.D. degree from Southwest Jiaotong Zhengyou He (M’10-SM’13), received the B.Sc.
University, Chengdu, China, in 2014, all in electrical degree and M. Sc. degree in Computational Mechanics
engineering. From 2013 to 2014, He worked as a from Chongqing University, Chongqing, China, in
visiting doctoral scholar at University of Alberta, 1992 and 1995, respectively. He received the Ph.D.
Edmonton, Canada. Currently, he is an Associate degree in the School of Electrical Engineering from
Professor with School of electrical engineering at Southwest Jiaotong University, Chengdu, China, in
Southwest Jiaotong University. 2001. Currently, he is a Professor in the School of
His main research interests are power quality and Electrical Engineering at Southwest Jiaotong Univer-
stability of the electric traction system. sity.
His research interests include signal process and
information theory applied to electrical power system, and application of
Haidong Tao (S’17), received the B.S. degree in wavelet transforms in power system.
electrical engineering from Southwest Jiaotong
University, China, in 2015. He is currently pursuing Shibin Gao, received B.S degree, M.S. degree and
the Ph.D. degree in electrical engineering at South- Ph.D. in electrical engineering from Southwest Jiao-
west Jiaotong University. tong University, China, 1986, 1999, 2006, respectively.
His research interests include modeling and con- Currently, he is a professor in Southwest Jiaotong
trol of high-speed trains and grid-connected convert- University.
ers. Harmonics analysis and control, stability of His research interests are protection and control of
power electronic based power systems. electric traction system.

0885-8993 (c) 2017 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
View publication stats

You might also like