Complete modeling and analysis of high-speed rail

transit synchronous linear motor traction system

Guo Jiling Liu Li Lu Ke
school of Electrical Engineering school of Electrical Engineering school of Electrical Engineering
Southwest Jiaotong University Southwest Jiaotong University Southwest Jiaotong University
Chengdu , China Chengdu , China Chengdu , China
[email protected] [email protected] [email protected]

Abstract—Aiming at the linear traction of high-speed rail model test bench drive system: In a short period (≤1s) and
transit, a complete model is put forward taking the power within a limited distance (50m), the model train(220kg)
supply, converter, linear synchronous motor, vehicle and motor achieves a high speed (≥112m/s). There are two model trains
control into account. Especially the analysis
is conducted focusing on the influence of the electromagnetic
pass each other in the meeting section from the eastern line
parameters of linear motor and the converter control and western line respectively. As shown in Fig.1, each line
parameters onto the electromagnetic thrust. With comparison traction system is composed by 4 parts: the energy and power
2021 13th International Symposium on Linear Drives for Industry Applications (LDIA) | 978-1-7281-7210-1/21/$31.00 ©2021 IEEE | DOI: 10.1109/LDIA49489.2021.9505828

of the simulation and test results, the model could be optimized subsystem, the model vehicle, the SLM and the converter.
so as to provide reference and guideline in real application.

Keywords—synchronous linear motor, high-speed, modeling

I. INTRODUCTION
The long primary synchronous linear motor (SLM) is
widely used in the field of the high speed maglev
transportation because it has the merits of high power factor
and high power density compared to the linear induction
motor (LIM) [1]. Many scholars have carried out some research
on a variety of types and structures of the linear synchronous
motor, and analyzed and optimized the motor parameters, Fig.1 composition of the system
motor magnetic field, thrust and so on. [2][3][4] The objectives to construct the complete modeling are:
Most literatures focus on the structural optimization and • Simulate the whole operation process of all parts, and
electromagnetic improvement of the linear motor, or propose analyze the dynamic movement;
some new motor control algorithm. Factually the linear motor
is only a core part of the linear traction system from the • Get overall index of the system including total power,
total efficiency, energy utilization rate and the
perspective of traction drive system. We can not only focus on changing effect of the key parameters in each part;
the motor’s own electromagnetic performance, but carry out
the research on the converter, vehicle, power source and • Find out the optimization method to obtain multi
control strategy of the whole linear traction system. Some objects for real application.
literatures take more parameters including actual values or III. COMPLETE MODELING
virtue ones calculated by finite element (FE) simulation, e.g.
the inductance value[5] [6]. In recent years, some authors have In order to simplify the system model, only the main
parameters of the energy storage system and the rotor are
conducted modeling analysis by combining the motor model
selected to participate in the model calculation. And the SLM,
and control scheme, for example, a CO simulation of the FE converter and control algorithm are chosen to be the core parts
model of the linear motor and the converter control algorithm of the system model.
simulation model in MATLAB is conducted in paper [7].
In this paper, a complete modeling method of SLM A. Energy and Power Subsystem
traction system of high-speed rail transit is proposed based on • Subsystem composition
the "Drive system multi-functional high-speed train-model The energy and power subsystem provide energy for the
test bench". With key parameters of each subsystem of the traction system. Due to the large energy release demand in a
bench selected, the complete model composed by the power short time, a large inertia flywheel energy storage mode is
supply, the converter, the SLM, the vehicle and the motor selected based on a pulse synchronous generator.
control is constructed. The model calculation is carried out
according to the actual parameters as far as possible, which
help both the model and the real bench be optimized each
other by comparing the simulation and experimental results.
The model can be used to guide the experiments of the whole
traction system, and provide system-level optimization
solutions for some important indexes.
Fig.2 composition of the energy and power subsystem
II. COMPOSITION OF THE DRIVE BENCH As shown in Fig.2 ， a 28.5MW pulse synchronous
The objective of the multi-functional high-speed train- generator is driven to speed nN by a same shaft 3-phase

978-1-7281-7210-1/21/$31.00 ©2021 IEEE

Authorized licensed use limited to: Universidad Nacional de Colombia (UNAL). Downloaded on November 29,2024 at 20:52:30 UTC from IEEE Xplore. Restrictions apply.
 induction motor, and its rotor is exited with a DC current If ejection process is not considered.
before the vehicle ejection. Then the AC power generated is
converted into a DC voltage Udc through the rectifier and Fem – μ1(FN + mmg) – μ2mvg = (mm+mv)a (2)
finally connected to the DC bus of 12-phase SLM converter.
• Modeling of the DC bus voltage C. Synchronous Linear Motor
• FE simulation modeling of SLM
The likelihood is that Udc will decrease at the moment of
ejection start or SLM rotor is at high speed, when more output As a core component of the model vehicle traction system,
power is needed. Regardless of reasons which cause the DC a double-side 12-phase synchronous linear motor is used. The
voltage drop, we just introduce a decreasing coefficient purpose of bilateral structure is to reduce the normal force.
named Kg, and analyze its influence on electromagnetic thrust. The advantage of multi-phase structure is to reduce the phase
current when the voltage is fixed, reduce the thrust ripple, and
The DC bus voltage provided by the energy storage system
improve the reliability when the open-phase fault occurs. And
is proportional to the speed and excitation current of the pulse
there are 8 air-core electrically-excited poles in the rotor. A
generator, so formula (1) could be obtained.
set of complex excitation power supply devices are needed,
including a super-capacitor pack, some electric brushes on
Udc = Kg DIf nN (1) the rotor and two copper bars arranged along the motion track.
Compared with the permanent magnetic poles, the air-core
B. Model vehicle and SLM rotor electrically-excited ones are lighter and the MMF (Magnetic
• Subsystem composition Motive Force) is easier to be adjusted.

TABLE I. FE SIMULATION PARAMETERS OF SLM

nph number of phases 12 np number of poles 4
number of slots
q 1 hF height of rotor 50cm
per phase belt
number of
τ pole pitch 360mm Nff 50
exciting turns
Ws width of slot 6mm Iff exciting current 103A
Ds depth of slot 20mm g length of air gap 5mm
thickness of iron given phase
hFe 50mm Ig 780A
core current value

(a) SLM rotor

（a）FE model

（b）Flux

(b) Motion Cross Section

Fig.2 SLM for model vehicle driving

The rotor structure and motion cross-section of the linear (c)Fem and FN with different offset distance from the center line
motor are shown in Fig. 3 (a) and (b) respectively. The Fig.4 SLM FE simulation result
moving load mass includes the rotor (mass mr), model
The simulation parameters are listed in Table. I. Fig.4 (a)
vehicle (mass mv shown in Fig.9). As for the moving
shows the 2D FE SLM model, and its mechanical dimensions
resistance, only the friction between rotor sliding blocks and
are shown in Tab.II. Fig.4 (b) shows the flux density
the frame support (friction coefficient μ1), and the friction
distribution at v=100m/s and the mean value of air gap flux
between the copper bar of DC power for rotor poles
density (BmN) is about 1T. The thrust with average value
excitation and electric brushes (friction coefficient μ2), are
75kN shown in Fig. 4 (c) can be obtained by applying 12
taken into consideration.
symmetric square-wave currents with amplitude 780A. As
• Newtonian dynamic model the offset distance of the rotor from the center line increases
from 0mm to 2mm, the variation of the thrust value is very
According to Newton's law of motion, we can get formula small, but the double-side synthetic normal force (FN)
(2), in which the friction force frf1 is proportional to the increases dramatically due to the bigger deviation between
normal force FN. And the friction force between the copper the left and right air gaps, which could increase the friction
bar and the brushes, frf2,is regard as a constant resistance force. force. In addition, the phase inductance (La), is about 0.8mH
For simplicity, the dynamic balance of the vehicle during the by FE simulation calculation.

Authorized licensed use limited to: Universidad Nacional de Colombia (UNAL). Downloaded on November 29,2024 at 20:52:30 UTC from IEEE Xplore. Restrictions apply.
 • Simplified SLM modeling • Control scheme
The main objective of simplified SLM modeling is to Aiming at the requirement of a high speed of 112m/s, it is
obtain the electromagnetic thrust expression (3) according to of importance to guarantee a fast and accurate driving of
the basic motor parameters including structure size, power switching devices. The control block diagram of one
electromagnetic parameters, some FE calculation parameters, power unit is given in Fig.7. The "software + hardware"
and ignoring the influence of end effect. control mode has the advantages of simple algorithm, fast and
12
accurate response.
Fem = 2 Lef (2 p ) N y  Bmi I ai Software aspect: The converter controller uses high-
i =1 (3) performance MCU chip to calculate and sends the reference
where Bmi is the flux density of each phase, as shown in (4) phase current Igi and pulse blocking/opening enable
F fi + Fai command to every power unit;
Bmi =BmN
FfN Hardware aspect: Every power unit is equipped with a
(4)
PWM generation and distribution board, which determines
where FfN =Nf If , is the nominal value of the rotor
switches’ on-off status in the H-bridge according to the signal
excitation MMF. transmitted from the rotor position detection device. The on-
off time is controlled by the hardware hysteresis comparator
circuit, where the input signals are given phase current value
(Igi) from the controller and the actual phase current value (Iai)
from sensors.

Fig.5 Ff , Fa and Em of stator phase coils

Since q=1, the position of each phase winding can be Fig.7 “Software + Hardware” controlling scheme
determined by the position of the slot where it is embedded.
In Fig.5, Ffi, which is distributed as a trapezoidal shape, could • Photoelectric rotor position detection
be constructed by eight points (lp1,lp2…lp8). At the same time, To satisfy the control scheme requirement, a photoelectric
Fai, which is relative with the magnitude and direction of the position detection system is designed, which includes the
current in the corresponding phase winding, can be obtained photosensitive belt adhered to the surface of the rotor and
according to the converter control scheme. Finally, the air- many reflective photoelectric sensors installed at position
gap flux density at every positon along the moving direction points corresponding every slot along the motion track.
is the result of the superposition of these two MMFs.
Based on Bmi, the phase back EMF (Electromotive Force),
Emi could be expressed as (5)

Emi = Ke Bmiv (5)

D. Converter and control scheme

The converter is mainly composed by 3 parts: the main
circuit, the controller and position detection devices.
• Topology of the main circuit

Fig.8 Schematic diagram of position detection device based on reflective

photoelectric sensors

As illustrated in Fig. 8, when the rotor is in position (a), the

sensors representing S1, N2 ~ N11 will output high-level signal,
which means that the 1st power unit should be “reverse on”
(T2 & T3 on, T1 & T4 off, Udc_st_i = -Udc) and power units from
No.2 to No.11 should be “forward on” (T1 & T4 on, T2 & T3
Fig.6 12 phase H-Bridge topology of the main circuit off, Udc_st_i = +Udc). Similarly, in the position (b), sensors
representing N11, N12, S1 ~ S10 output high level signals,
Corresponding to 12 phase stator windings, the main which means that the power units from No.1 to No.10 should
circuit topology of the converter is two-level 12 phase H- be “forward on”, and the 11th and 12th power unit should be
bridge mode. Every H-bridge could be regarded as one “reverse on”.
independent power unit, in which 4 IGBTs with 3300V
voltage and 1200A current capacity are selected. The leading or lagging of the phase current commutation

Authorized licensed use limited to: Universidad Nacional de Colombia (UNAL). Downloaded on November 29,2024 at 20:52:30 UTC from IEEE Xplore. Restrictions apply.
 can be realized by changing the position of black-and-white
strip in the photosensitive belt. For example, although the
position of rotor in (c) is the same as that of (a), the white
reflective belt is one slot ahead, which is called “leading
commutation”, the offset commutation distance dm=+3 mm.
On the contrary, dm<0 when “lagging commutation” occurs.
dm means phase angle difference between Fai and Ffi, which
then affects Emi.
nr, the number of power units devoted in operation actually, (b) Ua & Ia (3rd phase)
can be reduced by shortening the length of white part in the
belt. Take (d) as an example, nr = 9.
• Modeling of the converter and controlling scheme
The purpose of the modeling is mainly to design the
hysteresis comparator and change rate of the phase-current.
Hysteresis comparator: As for every power unit shown
as in Fig.7, the reference phase current Igi ’s direction signal
is given according to the relative position between the N or S
polarity of the rotor pole and each phase stator winding; and
the amplitude value is calculated according to the traction
demand. The actual phase current Iai takes into account the (c) Ig & Ia（3rd phase & 4th phase）
constant time TRC of RC filter in the PWM generation &
distribution board, where the Band-Band hysteresis width ΔIh
is also given according to the circuit hardware parameters. It
is worth noting the necessity of avoiding the narrow pulse
output (tkmin) in the PWM driving signal.
Change rate of the phase current:
U dc _ st _ i − Emi − I ai Ra
dI ai =
La （6）
dIai influences Iai to follow Igi. As shown in equation (6), （d）Ff，Fa & Em (3rd phase)
factors affecting dIai include the DC bus voltage Udc_ st_ i, the
back EMF Emi and resistance Ra and inductance La of the Fig.9 Simulation results in typical working condition
phase winding. Fig.9 (a) illustrates that the model vehicle runs to velocity
120 m/s with a constant acceleration in 0.675s, and the Fem is
IV. SIMULATION AND EXPERIMENT RESULTS
kept at about 72 kN. Fig.9 (b) shows the change of the phase
A. Simulation Experiment voltage and current of 3rd phase winding during ejection
• Simulation in typical working condition process. Fig. 9 (c) shows the current waveforms of the 3rd and
the 4th phase at v≈110m/s. Figure 9 (d) shows the waveforms
TABLE II Typical working condition simulation parameters of Ff, Fa and Em of 3rd phase winding at v≈93 m/s.
Sub System Simulation Parameters
Energy & Power nN=3000r/min; Kg=1; IfN=500A; Udc=1250V • Parameter influence analysis
Model Vehicle &
mr=200kg; mv=200kg; μ1=0.2; μ2=0.1
Mechanical devices The main function of complete-model is to analyze the
nph=12; τ=360mm; δ=5mm; q=1;Ra=0.16Ω; influence of parameter changes on the system, so as to
LSM
La=0.8mH; np=4; Nff =50; Iff =1050A; BmN =1T provide reference for guiding the test or optimizing the design.
Δih=40A; dm= +0.045m; nr=12; TRC=1e-5s;
Converter & Control
tkmin=50us; Igi=780A When Udc decreases (kg = 0.8), there is an obvious decline
Simulation parameters are listed in Tab.II. With the ending of thrust in Fig.10 (a) caused by a reduced dIa in Fig.10 (b)
condition motion distance s >40.32m (traction distance of compared to the normal one at high speed.
eastern line of experiment bench) or vmax >120m/s, simulation
Fig.11 is obtained when dm=0 which means there is no
results are shown as Fig.9.
lagging or leading commutation for phase current. Compared
with Fig.9, the phase angle difference between Fai and Ffi is
postponed and the thrust is also reduced in the high speed
region.

(a) Fem
(a) Fem & v

Authorized licensed use limited to: Universidad Nacional de Colombia (UNAL). Downloaded on November 29,2024 at 20:52:30 UTC from IEEE Xplore. Restrictions apply.
 Ia3

750A/grid
Ia4

5ms/grid
（b）Ig & Ia（3rd phase & 4th phase） (b) Ia3 & Ia4 at v≈92m/s
Fig.10 Simulation results when Kg=0.8
Fig.13 Experiment results
Fig. 13 shows the current waveform of adjacent two phases.
In Fig.13(a), the value of phase current is about 700A±80A
and the frequency is about 67Hz, which means v ≈ 48m/s; In
Fig.13(b), the current frequency is about 128Hz, which
means v ≈ 92m/s, and the wave shape changes dramatically
(a) Fem
caused by a large back EMF and the decline of the Udc at high
speed region. The current waveform is similar to the
numerical and simulation results, which demonstrates the
feasibility and correctness of the system complete model.
V. CONCLUSION
Through the construction and analysis of the complete
model, the whole operation processes of all parts is analyzed
clearly, which give us an effective reference or guidance for
high speed traction experiments and symmetrical
optimization. Next step is optimizing overall index of the
system including total power, total efficiency, energy
（b）Ff，Fa & Em (3rd phase) utilization rate, and so on.
Fig.11 Simulation results when dm = 0
ACKNOWLEDGMENT
• Experiment platform and high-speed Experiment
This work was supported in part by the Fundamental
Research Funds for the Centrale Universities under Grant
2682021ZT035, 2682021ZT036 and 2682021ZT037.
REFERENCES
[1] Hyung-Woo Lee, et al., "Improvement of Thrust Force Properties of
Linear Synchronous Motor for an Ultra-High-Speed Tube Train, "
IEEE Transactions on Magnetics, vol. 47, no. 11, pp. 4629-4634,Nov
(a)pipeline (b) SLM & model vehicle 2011.
Fig.12 Multifunctional model vehicle test platform [2] KOU B Q, et al.,"The Thrust Characteristics Investigation of Double-
Side Plate Permanent Magnet Linear Synchronous Motor for EML,"
Fig.12 demonstrates the linear drive system of multi- IEEE Transactions on Magnetics, vol.45, no.1,pp:501-505,Jan 2009.
function high-speed model train in low pressure pipeline. [3] Yufei Wang, et al., "Electromagnetic Emission of Maglev Vehicle by
The ejection test was conducted in conditions shown in Tab.II. Linear Synchronous Motor, " IEEE 3rd International Electrical and
Energy Conference (CIEEC),pp.1507-1512, in Beijing, 2019.
Along the eastern-line track, the rotor accelerated the model
[4] Han-Wook Cho, et al.,"Design and Characteristic Analysis on the
train to the maximum test-speed 116m/s in 40.32m, then the Short-Stator Linear Synchronous Motor for High-Speed Maglev
rotor entered the eddy current brake section, while the model Propulsion, " IEEE Transactions on Magnetics,vol.44, no.11,pp.4369-
train entered the passing-each-other section by inertia. 4372,Nov 2008.
[5] Narayan C. Kar, et al., "Measurement of the Saturation Characteristics
Ia3 in the Quadrature Axis of Synchronous Machines,"IEEE Transaction
on Energy Conversion,vol. 21, no.3, pp.690-698,Sep 2006.
[6] Fernandez F, et al.," Determination of parameters in interior permanent
magnet synchronous motors with iron losses without torque
750A/grid

measurement".IEEE Transactions on Industry Applications,

vol.37,no.5,pp.1265-1272, Mar 2000.
Ia4 [7] Yang Li, et al., "Real-time Simulation of Linear Synchronous Motor
Drives with AC-DC-AC Converters,” IEEE 6th Conference on
Industrial Electronics and Applications, pp.680-685, in Beijing, Dec
2011.

5ms/grid
(a) Ia3 & Ia4 at v≈48m/s

Authorized licensed use limited to: Universidad Nacional de Colombia (UNAL). Downloaded on November 29,2024 at 20:52:30 UTC from IEEE Xplore. Restrictions apply.