Download as pdf or txt
Download as pdf or txt
You are on page 1of 6

An Efficient Route Control Model of the Train-centric Control System

Qi Wang1 , Ming Chai2 *, Hongjie Liu2 , Jidong Lv2 , Xiwang Guo3


1. Department of Electronic Engineering, Beijing Jiaotong University, Beijing, China
2. The National Engineering Research Center of Rail Transportation Operation and Control System, Beijing Jiaotong University,
Beijing, China
3. Computer and Communication Engineering College, Liaoning Shihua University, 113001, Fushun, China

Abstract: The Train-centric control system integrates the main functions of the whole system into the vehicle, so the
train can control the route resources independently. In this system, the route control method which is mainly reflected in
the control of the switch is the root to ensure driving safety. This paper focuses on the route control method of the Train-
centric system, to find ways to assure the safety of the switch control process and compare track capacity of different
control methods. Firstly, we discuss the structural and compositional differences between the Train-centric control system
and the traditional Communication Based Train Control (CBTC) system, present the route control process of the new
system, as route control method matters the security of the whole system, one of the key problem in this process is the
logic to add resource lock to switches, we propose a model to calculate safe distance, and propose two route control
approaches, using Simulink/Stateflow hybrid modeling approach to model for both of them, get the visual curve, then
build math model to measure and compare the efficiency of the models.
2021 33rd Chinese Control and Decision Conference (CCDC) | 978-1-6654-4089-9/21/$31.00 ©2021 IEEE | DOI: 10.1109/CCDC52312.2021.9602827

Key Words: Train-control, T2T communication, train autonomous, route control, Simulink/Stateflow

1 INTRODUCTION the T2T communication has been improved, it gradually


becomes one of the industry’s future development direc-
Transporting pressure is a long-existing challenge to the tions[17,18]. The Train-centric control system based on
rail transit system, for this typical safety-critical system[1], T2T communication is one of the research directions[19],
it is especially important to consider efficiency and safety it is based on CBTC but changes in structure and function
as a whole. Timetable and train operation strategy opti- allocation, the composition of the Train-centric control sys-
mization approaches are used to reduce traction energy, tem is shown in Figure 1. In this system, most functions are
increase transport capacity[2,3]. Flexible regulation ap- integrated on the vehicle, it removes some complex way-
proaches are also used to reduce the train dwell time[4]. side facilities, Zone Controller (ZC) and ground interlock,
In recent years, control decision and optimization algo- uses Object Controller (OC) to collect states of switches,
rithm has become an important research topic in the con- switch machines, etc, and provide these states to trains, in
trol domain[5,6,7]. The Train-centric train control system the meanwhile, it accepts commands from trains to lock the
is a hot issue that contains Train-Train (T2T) communi- switch resources so the train can control it’s route directly.
cation and train autonomous techniques[8]. Although the To replace the ground interlock, the Train-centric control
existing Communication Based Train Control (CBTC) sys- system use on-board interlock, combined with the electric
tem[9] is logical and has tight principles, the disadvantages map which saves route information, to manage and arrange
of it cannot be ignored: the train-wayside-train communi- route, and offer the vehicles with route condition. Fur-
cation leads to more delay time and more complicated in- thermore, the new system has Dynamic Capacity Decision
terfaces, the ground-centric control method make wayside (DCD) as its core ground facility, it can make plans and
subsystem heavy and complicated, the building and oper- send them to trains, supervise the trains, to assist dispatcher
ating cost will be high. To solve these problems, we adopt to manage trains of the line, the ground Trian Management
the Simulink/Stateflow hybrid modeling approach[10,11] Unit (TMU) carry on functions of the original ZC, it can
to build models. Many countries and institutions imple- register for the train, send an electronic map and transmits
mented project like the European Next Generation Train commands from DCD to the registered train. The on-board
Control project[12], the Shift2Rail project[13], the Ameri- TMUs can communicate with each other and the ground
can Positive Train Control project[14], signal manufactures TMU, transmits the traffic data of each vehicle. Overall, the
like ALSTOM has Urbalis Fluence system[15], Japanese biggest difference between traditional CBTC and the Train-
Next-Generation Railway Operation System[16]. With the centric control system is the onboard function[20,21]. The
development of communication technology, the stability of train can arrange route, get the information about obsta-
cles ahead initiatively and calculate Movement Authority
The work was partially supported by the National Key R&D Program
of China (2018YFB1201500), the Beijing Natural Science Foundation (MA) itself, in this process, trains communicate with OC
(L201004), the Fundamental Research Funds for the Central Universities and control the switch directly, if several trains have to use
(2019JBM002) and the Beijing Laboratory of Urban Rail Transit.

978-1-6654-4089-9/21/$31.00 2021
c IEEE 264

Authorized licensed use limited to: Chaitanya Bharathi Institute of Tech - HYDERABAD. Downloaded on March 13,2024 at 06:37:20 UTC from IEEE Xplore. Restrictions apply.
the value we adopt is 600ms. Based on the typical safe
brake model in IEEE standard for CBTC, we build the safe
distance calculation model, the model will be presented in
section 3.
2.2 Proposed Two Different Route Control Method
In the Train-centric control system, trains can communi-
cate with OC and control the switch directly, several trains
may have to use the same switch and send commands to
Figure 1: Composition of the Train-centric Control System.
OC at the same time, so if the trains are going in different
directions, the switch will be ”deadlock”. In order to solve
this problem, the Train-centric control system has to con-
the same switch, the corresponding OC might be reserved
strain the trains to add lock in some ways. When several
by several trains at the same time, if they’re going the same
trains have to use the same switch as they approaching it,
direction, they can through the switch normally, but if not,
the leading train will firstly add a lock to the switch, after
the OC will be reserved for several different statuses, then
it through the switch, it will release it, so the switch can
the switch will ”deadlock”, result in an unnecessary fault,
operate with the command of the next train, in this process,
therefore how to avoid this problem is one of the key tasks
we discussed two scenarios:
in the course of Train-centric control system study.
In recent years, the model-based test is widely used in The first situation is that the following trains can add a lock
train control technology[22] research, the hybrid model- to switch over the leading train in some cases, the logic
ing approach Simulink/Stateflow is one of the most popu- is to compare driving plans at first, if they have the same
lar tools. In the Train-centric control system, Simulink can driving plans, both the leader and the follower can add the
describe the dynamic behavior of the system, and State- same resource lock to the switch just like add one lock,
flow based on finite-state machine theory[23,24], can be we call it Comparing Add Lock Method (CALM), as it is
used to model the transition of status during the run time. shown in Figure 3 (a), if they have different driving plans,
This paper proposes two route control methods and uses the follower cannot add resource lock over the leader, as it
Simulink/Stateflow to model for the system of both of is shown in Figure 3 (b).
them, according to the resulting curve, figure out the differ- The second is that whether the follower has the same plans
ence between them. In order to get a specific efficiency dif- as the leader or not, it cannot add resource lock to the
ference between them, we defined a parameter to describe switch over the leader, it can only wait for the release of
driving efficiency and compared it of the two methods. the leader to add it, we call it Directly Add Lock Method
(DALM), the logic of adding resource lock is shown in Fig-
2 ROUTE CONTROL OF THE TRAIN- ure 3 (c).
CENTRIC CONTROL SYSTEM
2.1 Overview of the Train-centric Control System’s
Route Control Process
In the Train-centric control system, the DCD centre de-
mands the vehicle to through a switch according to the driv-
ing plan, for example, in Figure 2, the train needs the switch
to lock in the reverse position, firstly, the vehicle builds
communication with the corresponding OC, then send lock
command to it, OC operates as the command and lock the
switch to reverse position, feedback status to the vehicle
and maintain communication with it.

Figure 3: Logics of adding resource lock.

The main control difference of these two methods is


whether to compare driving plans between trains and
Figure 2: The Vehicle Directly Control Switch.
whether the following train can add resource lock over the
leading train, the CALM is more effective than DALM
In order to ensure that the vehicle can through the switch when the two trains have the same plans, because in this
safely and take efficiency into consideration, the on-board case, if the two trains cannot add a lock to OC at the same
train control device have to calculate the appropriate loca- time, the follower has to wait for the leader to through, af-
tion to build communication, after the vehicle through the ter the switch has released, it can add lock then. As for the
switch, it takes a while to release the lock to this switch, tracking distance between trains, the DALM needs the fol-

2021 33rd Chinese Control and Decision Conference (CCDC) 265

Authorized licensed use limited to: Chaitanya Bharathi Institute of Tech - HYDERABAD. Downloaded on March 13,2024 at 06:37:20 UTC from IEEE Xplore. Restrictions apply.
lowing train to reserve extra distance because of the extra vbuild = vcoast + a1 tbuild (8)
time-delay, the efficiency analysis will be present in section Where tbuild is the required time to build service brake, and a1
4. is the average braking rate of this process. Dbraking is the travel
3 SAFE DISTANCE CALCULATION MODEL distance of the brake process:
2
Figure 4 shows the safe distance calculation model of the vbuild
Dbraking = (9)
Train-centric control system. The train can traverse barriers 2B
Where B is the emergency braking rate. Dsaf e is the safety mar-
gin.
The above is the whole process of the calculation of safe distance,
the train can use this model to calculate the safe distance to the
barrier ahead, as for two tracing trains through the switch, if we
adopt the DALM to lock switch, the follower has to add an extra
distance because the switch needs an interval to be released, thus
will cause tracing distance to extend, the safe distance D is as
follows:  tgap
D = D + vleader dt (10)
0
Where vleader is the speed of the leading train and tgap is the
time-delay of DALM.
Figure 4: Safe Distance Calculation Model.
4 SIMULINK OF TWO ROUTE CONTROL
ahead from the head of itself, as it approaches the switch, METHOD AND EFFICIENCY ANALYSIS
this model calculates the safe distance to communicate with In order to vividly see the tracing process and tiny difference be-
the corresponding OC, safe distance D is calculated using tween the two methods, we use Simulink/Stateflow to model the
the typical safe brake model the calculation of it is as fol- system, the specific route is shown in Figure 6, in this figure,
lows: Train2 is the leader and Train1 is the follower, the distance to the
switch ahead are 0.5km and 1km respectively. Our focus is the
D = Dunc + Ddelay + Dresponse + Dcut + Dcoast on-board functions and the route control logic, so we simplified
+ Dbuild + Dbraking + Dsaf e (1) the whole system and omit the ground centre devices, the struc-
ture of the system in this situation is shown in Figure 5. The major
Dunc is the train location uncertainty considered in the variables are displayed in TABLE 1.
measure process, Ddelay is the travel distance in the stage
of communication delay: Table 1: Variables
Variable Names Meanings
Ddelay = vmax tdelay (2) Switch Status Switch occupancy status
Train1 OC Occupy Occupation request to the switch
Where tdelay is the extra time-delay of this method and Train1 Release Release request to the switch
vmax is the maximum speed of the line. Dresponse is the Train1 Distance Travelled distance of the train
travel distance in the on-board response process, consider Train1 Direction Driving direction in the plan
the worst-case stopping scenario, the train run at maximum
tracking acceleration rather than service braking rate, Dcut
The system contains two tracing trains, OC, and switch, in the
is the travel distance in the traction cut off process:
modeling process, Simulink is used to describe the dynamic be-
Dresponse + Dcut = vmax (tresponse + tcut ) havior of the system, based on the braking model, calculate ATP
1 curves to protect the train, Stateflow is used to describe state tran-
+ a0 (tresponse + tcut )2 (3) sitions of the system.
2
The model we adopt in the calculation of speed monitor is the tar-
Where a0 is the maximum traction acceleration, tresponse is the get speed model and the actual running behavior is according to
response time of on-board devices, tcut is the time to cut of trac- Newton’s second law, at any time between (t1, t2), the relation-
tion. Dcoast is the travel distance after the relay drop and before ship between train speed v, initial speed v0 , acceleration a and
the brake build: displacement s is as follows:
Dcoast = vcoast tcoast (4)
⎧  t2
vcoast = vresponse + a0 tcut (5) ⎨v = v0 + t1 a dt

vresponse = vmax + a0 tresponse (6) (11)

⎩  t2
Where vresponse is the initial speed of the coasting process, s= t1
v dt
vcoast is the maximum speed in the coasting process used to cal- Firstly we built a model for the CALM situation that Train1 can
culate coasting distance. tcoast is the service brake response time. add resource lock over Train2, and then slightly adjust it to model
Dbuild is the travel distance during the service braking setup pro- for the DALM situation that Train1 cannot add resource lock over
cedure, the braking rate will accelerate to service braking rate dur- Train2. To avoid any braking of Train1, we adjust the departure
ing this period: time interval and find the critical value. For CALM, it is 43.9s,
1 for DALM, it is 83.4s, these are the minimum departure inter-
Dbuild = vcoast tbuild + a 1 t2 (7) val time of each method. The time-speed curves of CALM and
2

266 2021 33rd Chinese Control and Decision Conference (CCDC)

Authorized licensed use limited to: Chaitanya Bharathi Institute of Tech - HYDERABAD. Downloaded on March 13,2024 at 06:37:20 UTC from IEEE Xplore. Restrictions apply.
Figure 5: The Structure of the Train-centric Control System.

Figure 6: The Specific Route for Simulink.

DALM are shown in Figure 7 (a) and Figure 7 (b). It can be seen
that the tracking time of CALM is obviously less than DALM. We
then calculate the departure of CALM and DALM in 30 minutes,
the departure status of them are shown in Figure 8 (a) and Figure
8 (b). In 30 minutes, 43 trains can depart if we use CALM, while
23 trains can depart if we use DALM. The tracking time and dis-
tance of the two methods have a gap, the CALM lead to closer
safe tracing interval time and distance, to get a better measure of
trafficability capacity and efficiency, a parameter Epass is defined
as:
k
Epass = (12)
ttrack
Where k is a constant related to different system’s specific cir-
cumstance and ttrack is the minimum tracking interval time of
(a) Time-Speed Curve of CALM. trains, therefore, the efficiency of the above two methods can be
measured as follow respectively:
k
Epass1 = (13)
ttrack
k k
Epass2 = = (14)
ttrack  ttrack + tgap
Where tgap is the extra time to add lock in DALM. The efficiency
ratio of them is:
Epass1 ttrack + tgap tgap
= =1+ (15)
Epass2 ttrack ttrack
The ratio of the efficiency in unit time can be calculated refer to
the train number. In this case, the ratio is :
(b) Time-Speed Curve of DALM.
Epass1 43
Figure 7: Speed-Distance Curve = = 1.87 (16)
Epass2 23
As a result, the logic to add lock has an impact on the efficiency
of the Train-centric system, as the tgap increases, the efficiency
will decline.

2021 33rd Chinese Control and Decision Conference (CCDC) 267

Authorized licensed use limited to: Chaitanya Bharathi Institute of Tech - HYDERABAD. Downloaded on March 13,2024 at 06:37:20 UTC from IEEE Xplore. Restrictions apply.
REFERENCES
[1] H. F. Wang, N. Zhao, B. Ning, T. Tang, M. Chai, Safety mon-
itor for train-centric CBTC system, IET Intelligent Transport
Systems, Vol.12, No.8, 931-938, 2018.

[2] S. Su, X. Li, T. Tang, A subway train timetable optimization


approach based on energy-efficient operation strategy, IEEE
Transactions on Intelligent Transportation Systems, Vol.14,
No.2, 883-893, 2013.

[3] S. Su, X. K. Wang, Y. Cao, An energy-efficient train opera-


tion approach by integrating the metro timetabling and eco-
driving, IEEE Transactions on Intelligent Transportation Sys-
tems, Vol.21, No.10, 4252-4268, 2019.

[4] S. Su, T. Tang, J. Xun, Design of running grades for energy-


efficient train regulation: a case study for beijing yizhuang
(a) Departure status of CALM. line, IEEE Intelligent Transportation Systems Magazine,
2019.

[5] X. W. Guo, M. C. Zhou, A. Abusorrah, F. Alsokhiry,


K. Sedraoui, Disassembly sequence planning: A survey,
IEEE/CAA Journal of Automatica Sinica, 1-17, 2020.

[6] X. W. Guo, M. C. Zhou, S. X. Liu, L. Qi, Multi-resource Con-


strained Selective Disassembly with Maximal Profit and Min-
imal Energy Consumption, IEEE Transactions on Automa-
tion Science and Engineering, 1-13, 2020.

[7] Z. Zhao, S. Liu, M. Zhou, D. You, X. W. Guo, Heuristic


Scheduling of Batch Production Processes Based on Petri
Nets and Iterated Greedy Algorithms, IEEE Transactions on
Automation Science and Engineering, 1-11, 2020.

[8] T. Chen, H. F. Wang, B. Ning, Y. Zhang, T. Tang, K. C. Li,


Architecture Design of a Novel Train-centric CBTC System,
2018 International Conference on Intelligent Rail Transporta-
(b) Departure status of DALM. tion (ICIRT), Singapore, 1-5, 2018.
Figure 8: Departure status [9] C. Schifers, G. Hans, IEEE standard for communications-
based train control (CBTC) performance and functional re-
quirements, Vehicular technology conference proceedings,
VTC, 581-1585, 2000.
5 CONCLUSION
[10] The Mathworks. Simulink User’s Guide Version 7, 2009.

In this paper, we proposed two different logics to add resource [11] The Mathworks. Stateflow User’s Guide Version 6, 2006.
lock to the switch, built a model to calculate safe distance, then
[12] J. Schuette, Next Generation Train Control (NGTC) Ap-
use Simulink/Stateflow tool to model for them and get straight
plicability for APMs, Automated People Movers and Auto-
forward difference, finally, defined a parameter to measure the ef-
mated Transit Systems 2016, 187-196, 2016.
ficiency of the Train-centric system and compared the two meth-
ods in the efficiency level. The result shows that if the following [13] É. Masson, C. Gransart, Cyber security for railways-a huge
train can add resource lock over the leading train, the efficiency challenge-Shift2Rail perspective, International workshop on
of transportation will get improved when the two tracking trains communication technologies for vehicles, 97-104, 2017.
have same plans.
[14] Vantuono, C. William, Pushing Forward PTC: Railroads
There are several interesting topics for future work. Firstly, we
and Suppliers Have Picked Up the Pace in Deploying Next-
have a model for two trains in the tracking scenario, but in re-
Generation Train Control, Railway Age, Vol.207, No.11,
ality, there can be more trains in a segment, for this reason, we
2006.
want to build an extended model to simulink for more trains. Sec-
ondly, we defined a formula to assess operating efficiency, in this [15] D. Briginshaw, Alstrom’s simplified CBTC technology to
formula, the value of parameter k is unknown so we can only debut in lille, International Railway Journal, Vol.53, No.6,
compare the efficiency of a different method, in order to make it 2003.
possible to evaluate efficiency separately, we are going to collect
real data and externalize the value. Last but not least, to verify [16] M. Korekoda, H. Kawasaki, Y. Tanaka, Considerations for a
the feasibility and security of the system, it is necessary to build a Next-generation Railway System in the Tokyo Metropolitan
timed automaton of the system. Area, JR East Technical Review, No.19, 2011.

268 2021 33rd Chinese Control and Decision Conference (CCDC)

Authorized licensed use limited to: Chaitanya Bharathi Institute of Tech - HYDERABAD. Downloaded on March 13,2024 at 06:37:20 UTC from IEEE Xplore. Restrictions apply.
[17] H. Chen, L. J. Liu, T. Novlan, J. D. Matyjas, B. L. Ng, J.
Z. Zhang, Spatial spectrum sensing-based device-to-device
cellular networks, IEEE Transactions on Wireless Communi-
cations, Vol.15, No.11, 7299-7313, 2016.

[18] N. Mastronarde, V. Patel, J. Xu, L. J. Liu and M. van der


Schaar, To relay or not to relay: Learning device-to-device
relaying strategies in cellular networks, IEEE Transactions
on Mobile Computing, Vol.15, No.6, 1569-1585, 2015.

[19] X. X. Wang, L. J. Liu, L. Zhu, T. Tang, Joint security and


QoS provisioning in train-centric CBTC systems under sybil
attacks, IEEE Access, Vol.7, 91169-91182, 2019.

[20] B. Bu, F. R. Yu, T. Tang, Performance improvements of


communication-based train control (CBTC) systems with un-
reliable wireless networks, Wireless networks, Vol.20, No.1,
53-71, 2014.

[21] L. Zhu, F. R. Yu, T. Tang, An integrated train-ground com-


munication system using wireless network virtualization: Se-
curity and quality of service provisioning, IEEE Transactions
on Vehicular Technology, Vol.65, No.12, 9607-9616, 2016.

[22] J. D. Lv, X. L. Zhu, K. C. Li, Model-Based Test Case Au-


tomatic Generation of CTCS-3 Train Control System, Jour-
nal of Southwest Jiaotong University Vol.50, No.5, 917-927,
2015.

[23] R. Alur, A. Kanade, S. Ramesh, K. C. Shashidhar, Proceed-


ings of the 8th ACM international conference on Embedded
software, 89-98, 2008.

[24] A. Kanade, R. Alur, F. Ivančić, Generating and analyzing


symbolic traces of Simulink/Stateflow models, International
Conference on Computer Aided Verification, 430-445, 2009.

2021 33rd Chinese Control and Decision Conference (CCDC) 269

Authorized licensed use limited to: Chaitanya Bharathi Institute of Tech - HYDERABAD. Downloaded on March 13,2024 at 06:37:20 UTC from IEEE Xplore. Restrictions apply.

You might also like