2. 628 IEEE TRANSACTIONS ON INTELLIGENT TRANSPORTATION SYSTEMS, VOL. 14, NO. 2, JUNE 2013
of both operators and passengers are very difficult to predict.
It has been demonstrated that many disastrous urban rail ac-
cidents were caused by human factors, such as the fire in the
Daegu Metropolitan Subway in 2003, the derailment accident
in Amagasaki in 2005, etc. Faced with these human-induced
accidents, it is natural and important to study how to model and
analyze human factors appropriately [35]. Unfortunately, there
is no existing effective method for solving such a problem.
Based on the given description, one perceives that the design
of ERSs is a central issue for the operation of rail systems in
accidental situations. Traditionally, designing an ERS is based
on the possible consequences of a specific accident, where
the possible consequences are deduced by human experts
based on their experience and prior knowledge [18]. Therefore,
the design process itself inevitably includes the uncertainty
brought by incomplete knowledge and imprecise judgment of
the experts. Considering this, the involvement of human factors
in the strategy design of emergency response can possibly
make the corresponding strategy unreliable and incomplete
[21]. Consequently, how to design a comprehensive, scientific,
and objective strategy for emergency response has become an
urgent yet challenging issue in the development of modern
urban rail transport systems.
The main difficulty in emergency management of urban rail
systems is that the potential causes or incident scenarios of
rail accidents are impossible to repeat in real-life experiments
[7], [30]. Fortunately, computerized simulation can provide
us with an outlet for the evaluation of the effectiveness of a
proposed emergency management strategy [19]. Nowadays, a
large number of excellent simulation software has been devel-
oped and applied to the planning, design, and management of
urban rail construction. Typical platforms include VISION for
analyzing the duration of trains and capacity of lines, LOGSIM
for train scheduling and traffic control, and OpenTrack for
optimization of train scheduling. Other typical rail simulators
include MEDYNA, SIMPACK, NUCARS, ADAMS/rail, etc. It
should be mentioned that each of the aforementioned simula-
tion platforms focuses on one specific aspect of the rail system,
such as train dynamics, rail networks, or signaling systems.
However, emergency response is a concern of every facet of the
rail system; hence, the construction of an emergence response
platform should take the high complexity of the rail system into
consideration. Furthermore, such a system should be designed
to cope with the uncertainty of human factors induced by
operators and passengers. Considering these requirements, the
development of such a platform is a very challenging task [10].
During the past decade, many different methods have been
proposed for emergency response during disasters [23], [26],
[31]–[34], but they are not very capable of coping with the high
complexity and human factors of real systems.
Integrating artificial systems, computational experiments,
and parallel execution (ACP) is a recently emerging approach
to modeling, simulating, and intervening real complex sys-
tems [1], [2], [4], [5], [37]. In 2004, Wang proposed a basic
framework of the ACP method [2]. The implementation of
this method can be divided into three different stages: 1) con-
structing an artificial system that comes from the real system;
2) experimenting on this artificial system to obtain related
Fig. 1. Framework of the proposed artificial system.
information on the potential consequences of the triggering
events; 3) combining the obtained information with the real
system to evaluate the real system and to make improvements
on related strategies. By repeating the three steps iteratively,
the plans to be executed can be optimized to realize its best
performance, reaching multiple optimization indexes such as
energy, time, and cost savings [9]. ACP has been applied to
many real engineering projects [13], [14]. In 2011, Ning et al.
introduced the method of parallel control to the management
of high-speed railway systems [9] and urban rail transportation
systems [10]. Following these existing works, the main purpose
of this paper is to apply the ACP method to the emergency
response of urban rail transport systems.
Traditionally, the evaluation of ERSs is based on the ana-
lytical hierarchy process (AHP) method [37], [38]. The basic
idea of an AHP is as follows: Given an evaluation object, fac-
torize this object into different subobjects. Furthermore, each
subobject can be also factorized into smaller factors. Based on
the given factorization, a hierarchical tree can be constructed.
Consequently, the evaluation of the ERS can be performed from
top to bottom and layer by layer, making the evaluation process
organized and efficient.
The given AHP method can be improved into different ver-
sions, such as the method given in [37]. However, the evaluation
process will inevitably involve the uncertainty of the experts,
which is derived from their decision inaccuracy, subjective
prejudice, and incomplete knowledge. Furthermore, the cause-
and-effect relationship between the accidental events and the
final scenario is too complicated to be precisely predicted by
human deduction and experience. Therefore, a question arises
as how to eliminate the benevolent human factors in the process
of evaluation. Fortunately, the ACP method can provide an
optimal solution for such a problem.
This paper is organized as follows: Section II introduces the
structure of the emergency response platform for urban rail
transport systems based on parallel control and management;
Sections III–V elaborate upon the proposed framework sepa-
rately from three different scales: Points, Lines, and Networks;
and Section VI concludes this paper and points out some
potential applications of the emergency response platform.
3. DONG et al.: EMERGENCY MANAGEMENT OF URBAN RAIL TRANSPORTATION BASED ON PARALLEL SYSTEMS 629
TABLE I
COMPARISON OF THE ACP METHOD AND THE TRADITIONAL SIMULATION METHOD
II. FRAMEWORK OF THE PROPOSED METHOD
The basic framework of the proposed method is demon-
strated in Fig. 1. As shown, the construction of the artificial
system is divided into three interdependent aspects: Points,
Lines, and Networks.
Points represent the modeling of urban rail stations. Rail
stations are the basic components of the rail system and, hence,
play the most important role for the successful operation of
the rail system. In the construction of the artificial system,
several critical aspects of the rail station will be considered,
such as the station architectural structure, the characteristics of
passenger flow, the arrival and departure of trains, the condition
of the station equipment, etc. Based on the aforementioned
emergency-related aspects, an artificial system can be virtually
established. Therefore, for emergency situations in the rail
station, a lot of virtual experiments can be carried out based
on the artificial system. Furthermore, the experimental results
can be used to improve the management of the rail station or to
rectify the defects of the given ERS.
Lines are the abstract urban rail connections between dif-
ferent stations, together with the operation of trains along the
rail lines. Generally speaking, the operation of trains relates to
the following subsystems: automatic train surveillance (ATS),
automatic train protection (ATP), and automatic train operation
(ATO) [11]. Based on the basic functions of the aforementioned
subsystems, the modeling of Lines will consider the following
related factors: velocity and position surveillance, excessive
speed protection, headway control, temporary velocity restric-
tion, signaling control, etc. In case of emergency, the congestion
of the whole network usually originates from the accident in
some specific links. Hence, the modeling of Lines will illustrate
why a given rail link is disconnected and how to restore the
connectivity of this link. The aforementioned two questions are
critical for the emergency response of the rail system. Above
all, the modeling of Lines will design the scheduling strategy
between the failed station and its adjacent normal stations, the
resolution of which will provide helpful information for the
evacuation of passengers in case of emergency.
Networks can be viewed as the network composed of all
the urban rail lines of the transportation system. It should be
pointed out that the difference between Lines and Networks is
that Lines focus on the microscopic modeling of the rail con-
nections, whereas Networks focus on macroscopic modeling.
The main purpose of Networks modeling is to explain how
local failure of some specific stations leads to the congestion
or breakdown of the whole rail system. The resolution of such
an issue will help us understand the mechanism of failure
spreading over networks. In detail, Networks modeling will
apply the theory of complex networks, network flows, and
graph theory into the analysis of the rail system. Furthermore,
based on the experiment on the artificial network, the stations
and lines of the rail system will be classified according to
their importance in the process of emergency response, giving
potential guidelines for the design of ERS.
It should be noted that the operation of the rail system is
always strongly interconnected with the behavior of humans,
for example, the operation of train drivers, the dynamics of
pedestrians, and the directing of train scheduling by dispatchers
[7]. In normal cases, it is not difficult to obtain the characteris-
tics of human behavior. However, in emergency cases, the hu-
man behavior characteristics will abruptly change, making the
evolution of the rail system hard to predict. Take the behavior
of pedestrians for example. In the case of a fire, the velocity
of the pedestrians will be much faster than that in the normal
case, and the density of pedestrians will be relatively higher in
some regions, such as the exits. Based on this discussion, a key
problem for emergency response is how to combine the human
factors into the construction of the artificial system. For the
aforementioned three aspects of the artificial system, the human
factors mainly include pedestrian properties of the Points, the
interaction of train scheduling and passenger flows of the Lines,
and the priori knowledge of passengers of the Networks.
By using the ACP method, an artificial system of an urban
rail system can be constructed, describing both the macroscopic
and microscopic properties of a real urban transportation sys-
tem. Basically, the advantages of the ACP method over the
traditional simulation method can be summarized as in Table I.
The ACP method can also provide a good solution to the ERS
of urban rail systems. Based on the artificial rail system, a series
of disastrous scenarios can be simulated beforehand. Therefore,
it is also possible to integrate the ERS into the scenario simula-
tion to observe the potential consequences under the conditions
of the given strategy. In this case, the evolution process of the
accidental scenario is perspicuous and, hence, can be easily
evaluated. After the execution of each computational exper-
iment, the response strategy can be further improved based
on the simulation result. The process is repeated again and
again, allowing the final ERS to be optimized, catering to the
requirements on the response efficiency, property safety, and
execution costs, etc.
III. FIRST MODELING ASPECT: POINTS
The rail station is the basic component of the rail transport
system and is also the place in which most of the passengers
stay. It has been demonstrated that most rail accidents originate
from stations, and the emergency response of rail stations is
quite concerned with the personal security of passengers. In the
framework of the ACP method, the first step in the successful
design of an ERS for rail stations is the construction of an
artificial system for the stations.
The constructed artificial system for stations is composed of
the following modules: urban rail station architecture, scene
library, and pedestrian dynamics simulator.
4. 630 IEEE TRANSACTIONS ON INTELLIGENT TRANSPORTATION SYSTEMS, VOL. 14, NO. 2, JUNE 2013
Fig. 2. Planar structure of the Xuanwumen Subway Station.
A. Urban Rail Station Architecture
The architecture of the urban rail station has a close relation-
ship with the design of strategies for emergency response. The
proposed artificial system contains the architecture graphs of
many rail stations of the Beijing Subway, which is the busiest
subway network in China.
Take the Xuanwumen Subway Station, Beijing, as an exam-
ple. As shown in Fig. 2, the basic components of the station
are provided, including platforms, tunnels, waiting zones, en-
trances and exits, etc.
Designing the station architecture should facilitate the evacu-
ation of passengers in case of emergency. However, the resolu-
tion of such a problem is not easy since it is nearly impossible to
conduct a real experiment on real station facilities. Luckily, the
artificial system can provide an outlet for such a problem. Based
on the constructed rail station models and the related pedestrian
dynamics module (which will be introduced in Section III-C),
evacuation experiments can be realized by using corresponding
algorithms, making it possible to assess the safety of related
facilities.
B. Scene Library
To cope with different accidents or disaster scenarios, it
is necessary to build a corresponding scene library for these
events. Based on the scene library, a series of computational
experiments can be executed, generating a great amount of
useful information for designing ERSs or improving the ex-
isting strategies. Generally speaking, the disasters in the sta-
tion can be categorized as follows: natural disasters such as
earthquakes; operational accidents such as train crashes and
fires; public hygiene accidents such as gas poisoning; and mass
disturbance, which is usually caused by political reasons, such
as terrorist attacks. Among the given accidents, an operational
accident is the most common accident and needs to be seriously
addressed [28].
The designed emergency response platform can provide
computational experiments for rail stations under different sce-
narios, such as a fires or terrorist attacks. As shown in Fig. 3,
the circular region represents an explosion area. In the case of
an explosion, the passengers located in that circle will escape
from it as soon as possible; therefore, pedestrian dynamics will
Fig. 3. Three-dimensional scene graph of explosion.
Fig. 4. Three kinds of forces between passengers.
be vastly different from those in the normal case. In detail, these
differences mainly come from passengers’ velocities, densities,
and rationality indexes. To describe the main property of the
passenger dynamics in the case of a fire, the platform provides
an interface to adjust those human-related parameters, such as
velocity and rationality index.
C. Pedestrian Dynamics Simulator
As was pointed out in the previous section, the pedestrian
dynamics in an emergent scenario are quite different from
normal. Generally speaking, the characteristics of passenger
flow in the emergent case include high velocity in movement,
unevenness of distribution, and irrationality in decision-making
[39]–[41]. The constructed simulator for pedestrian dynamics
has taken the aforementioned characteristics into consideration.
The basic idea for the construction of the simulator is to take
each passenger as an autonomous agent and to have any two
adjacent agents interact via some specified local rules [42].
Moreover, the simulator also considers the interaction between
the agents and the walls of the stations, making the simulation
similar to real cases.
As shown in Fig. 4, the existing model for the pedestrian
dynamics includes three different local forces: repulsion, align-
ment, and attraction. It should be pointed out that the repulsion
5. DONG et al.: EMERGENCY MANAGEMENT OF URBAN RAIL TRANSPORTATION BASED ON PARALLEL SYSTEMS 631
TABLE II
EVACUATION TIME OF LINE 2, BEIJING SUBWAY
TABLE III
EVACUATION TIME OF LINE 2, BEIJING SUBWAY
region in the case of emergency is very small relative to the nor-
mal case due to human survival instinct. Therefore, governed
by this human psychology, the passenger density in the station
will be quite unevenly distributed: Most of the population will
center around the entrances or exits of the station, leaving most
of the remaining station areas vacant. However, in the normal
case, most of the passengers are distributed along the outer
region of the platforms.
Calculation of the evacuation time is an important aspect of
emergency response [36]. The artificial system provides two
basic methods for computing the evacuation time: the Chinese
standard algorithm and the U.S. standard algorithm.
The Chinese standard algorithm calculates the evacuation
time based on the following formula:
TCH =
Q1 + Q2
0.9[A1b(N − 1) + A2B]
+ I
where
T evacuation time in the platform, min;
Q1 number of passengers in the train, which varies with the
types of trains;
Q2 total number of passengers waiting for the trains and the
operators in the station in peak hours;
A1 passage capacity of elevators, person/(min-m);
A2 passage capacity of stairs, person/(min-m);
N number of elevators;
B, b width of the rung of an elevator, m;
I response time of a person;
0.9 practical passage capacity of stairs, where moving stair-
cases are reduced to 90%.
It is required by Chinese standard GB50157-2003 that the
width of the ladders and the evacuation aisles should guarantee
the evacuation of all the passengers in the platform and all the
related workers within 6 min in case of a fire during peak hours.
By using the Chinese standard algorithm as an example,
the evacuation time of Line 2, Beijing Subway, is given in
Table II.
The evacuation time obtained by the U.S. standard algorithm
can be given by the following formula:
TUS = T + Wp +
n
i=1
Wi
where
T walking time spent on the road to outlet, min;
Wp waiting time of the outlet of the platform, min;
Wi waiting time of the other moving zone, min.
The details for the calculation of the given items are omitted
here due to space restrictions. By using the U.S. standard
algorithm, the evacuation time of Line 2, Beijing, is given in
Table III.
IV. SECOND MODELING ASPECT: LINES
The operation of a train from one station to another concerns
two different aspects: One is the train itself, which contains
the related control, operation, and communication systems; the
other is the rail track between two stations, which contains the
related signaling and detection systems. Therefore, the artificial
system of lines mainly focuses on the aforementioned two
aspects.
The constructed artificial system for lines has the following
typical properties: accurate modular structure and fast genera-
tion of the train diagram.
6. 632 IEEE TRANSACTIONS ON INTELLIGENT TRANSPORTATION SYSTEMS, VOL. 14, NO. 2, JUNE 2013
Fig. 5. Multilayer structure of the artificial system.
Fig. 6. Structure of the artificial ATP system.
A. Modular Structure
The operation of trains includes three interdependent sys-
tems: ATS, ATP, and ATO. Since each of the aforementioned
systems contains a multitude of components, the modular struc-
ture is beneficial for the design of the artificial system.
The artificial system of ATS has a multilayer structure with
multifunctions. As shown in Fig. 5, the software structure of
ATS is composed of three layers: user layer, processing layer,
and data layer. The user layer processes the interfaces for
human–machine interaction. The processing layer integrates
various logic rules and equipment principles, receiving the
commands from the user layer, whereas the data layer stores the
information obtained by the processing layer and also provides
related information for the processing layer when necessary.
The ATP system focuses on security-related tasks, such as
position and velocity detection, velocity surveillance, excess
speed protection, control of trans–trains distance, control of
doors, accident recording, etc.
Based on the basic function of ATP, the artificial ATP system
is designed with the structure as shown in Fig. 6, which is
divided into the following major modules:
1) Train and environment. This module is used to initialize
the related states of trains and lines.
Fig. 7. Different surveillance curves of the artificial system.
2) Trajectory computation of trains. This module calculates
the positions and velocities of trains to provide data for
excessive speed protection.
3) Excessive speed protection. This module provides a real-
time protection curve based on the information of trains
and lines.
4) Emergency braking. This module provides the strategy of
emergency braking based on the braking model and the
conditions of trains and lines.
5) Information recording and analysis. This module inte-
grates and analyzes the information of the aforemen-
tioned modules.
According to the description of the aforementioned modules,
the artificial system centers around the protection of trains in
the case of overspeed. In fact, there are many requirements for
overspeed protection of trains, i.e., it is forbidden to exceed the
maximal secure velocities of the train, the rail tracks, and the
current operational mode. The artificial system integrates three
different surveillance curves. The principle of these curves is
shown in Fig. 7.
B. Fast Generation of Train Diagram
A train diagram is an auxiliary method for the description of
the schedules of trains. Specifically, the generation of a train
diagram has several advantages in emergency situations, for
example, transparency, by which it is easy to obtain the related
operational information from the diagram; maneuverability, by
which the reschedule design of trains is relatively easy by using
the diagram; and detectability, by which the design errors and
faults can be easily detected from the diagram.
Generally, the real numerical data of train operation are not
beneficial for the intuitive analysis of the train schedule. The
artificial system provides a method for quick transformation of
the numerical data into a visible diagram, which is given in
Fig. 8. According to Fig. 8, one can easily get the operation
7. DONG et al.: EMERGENCY MANAGEMENT OF URBAN RAIL TRANSPORTATION BASED ON PARALLEL SYSTEMS 633
Fig. 8. Train diagram.
information on the uplink and downlink of Line 5, Beijing
Subway.
In fact, the design of the train schedule has a close rela-
tionship with the operational safety of trains. During the past
decade, many mathematical models have been proposed to
design the timetable of metrolines. It should be noted that op-
erational safety is an important aspect in such a design process.
In general, the operational safety of the trains is quantized by
several factors, such as the train headway, the highest velocity,
the braking distance, etc. Therefore, the involvement of these
safety factors in the design of timetables is quite helpful for
accident avoidance in rail systems.
V. THIRD MODELING ASPECT: NETWORKS
Networks represent the topology formed by all the rail
connections, which is also the topmost structure of the urban
rail system. According to the extensive data of rail accidents,
many of them are caused by Networks-level reasons, such as
incorrect scheduling of trains, unpredicted rail congestion, etc.
However, most of the Networks-level reasons are indeed human
induced, which can be avoided by serious preplanning and strict
organization. This section mainly introduces some Networks-
level modeling of the rail system, providing possible solutions
to avoid the Networks-level accidents.
The constructed artificial system for networks is concerned
with the following aspects: rescheduling of trains, searching
for the shortest path in cases of emergency, and the impact of
passengers’ prior knowledge on the choice of lines.
A. Search of Candidate Paths
In normal cases, a major objective of rail operation is to
optimize the passenger flow of the rail network. However, in
an emergency situation, the abnormal flow in one station or
rail link may lead to the breakdown of the whole network.
Therefore, it is necessary to reschedule the trains to meet the
needs of emergency response. The framework for the design of
Networks is shown in Fig. 9.
As shown in Fig. 9, the core component of this framework
is the algorithm library, which is composed of four classes
of algorithms. These are the generation algorithm of avail-
able pathways, the matching algorithm between passengers
and trains, the adjustment algorithm of programs in case of
emergency, and the optimization algorithm of a train diagram.
The passengers always make their decisions based on some
intuitive criteria, such as the minimum number of transfers, the
shortest riding time, or the shortest distance from the starting
point to the destination. Therefore, to model the passenger flow
of the rail network, the artificial rail system should provide
different generation strategies based on these human-induced
criteria. As shown in Fig. 10, the developed artificial system
provides the transfer strategies of the Beijing Subway based on
the criteria of shortest path and minimal transfers.
In case of an emergency, the preference and choices of
passengers will be abruptly changed. Qualitatively speaking,
passengers will become more irrational in choosing candidate
paths, making the traditional algorithm ineffective. Based on
this fact, it is necessary to quantitate the emotional factors in
both normal and abnormal cases.
B. Impact of Passengers’ Prior Knowledge
Faced with different candidate pathways, passengers do not
make their choices randomly; the choice of the final path
is usually dependent on prior knowledge of the passengers.
Consequently, an interesting question arises: How great is the
probability that the passenger will choose a given candidate
path.
It should be noted that the choice of paths is a very sub-
jective issue, and different passengers usually have different
requirements. To evaluate the different requirements of the
passengers, a new concept named generalized expenditure is
usually considered in the mathematical modeling of passenger
distribution. By generalized expenditure, we mean the total cost
from one station to another, including traveling time, comfort
level, transfer times, etc.
A logit model is a famous mathematical model for the cal-
culation of the choosing probability among several given paths.
Let ωf be the weight of passenger familiarity on a specific path,
which is defined as
θ = f(ωf ) = tan
πωf
2ωfmax
, ωf ∈ [0, ωfmax
].
According to the logit model, the distribution ratio of the
candidate paths is given as follows:
Pw
m =
exp {θf (Tw
m/Tw
min)}
n exp {θf (Tw
m/Tw
min)}
where the denominator is the sum over all the n candidate
paths, Pw
m denotes the choosing probability of the mth path,
Tw
m represents the time spent on the candidate path m, and
Tw
min represents the minimal travel time of the candidate paths.
It is easy to verify that when ωf → ωfmax
, there is Pw
m →
1; when ωf → 0, there is Pw
m → 1/n. Based on the given
calculation, the choosing probability of the four candidate paths
in the condition of different familiarity indexes is shown in
Fig. 11.
8. 634 IEEE TRANSACTIONS ON INTELLIGENT TRANSPORTATION SYSTEMS, VOL. 14, NO. 2, JUNE 2013
Fig. 9. Framework for the designing of Networks.
Fig. 10. Transfer strategies provided by the artificial system.
VI. CONCLUSION
This paper has proposed an ACP framework for emergency
response of urban rail systems. The proposed framework is
composed of three parts: Points, Lines, and Networks, repre-
senting urban rail stations, railtrack-related facilities, and rail
networks, respectively.
For each of the three components, related emergency cases
are correspondingly discussed. Points concentrate on the mod-
eling of accidental or disastrous events, such as a fire, a terrorist
attack, and a public hygiene problem. Lines focus on the model-
ing of events such as rail congestion, trains crashing, and signal-
ing failure. Networks focus on network-induced accidents, such
as train rescheduling, large-scale passenger evacuation, and the
spread of an epidemic.
Since all of the three components have been integrated into
the artificial system, every aspect of the accident scenario can
be simulated on both macroscopic and microscopic scales. As
Fig. 11. Choosing probability varies with familiarity.
a result, computing experiments on the artificial system can
provide accurate and guiding information for the consequence
of a given accident, which is impossible to implement in the
traditional simulation platform of rail systems.
Based on the constructed artificial system, the ERS can be
evaluated and improved via repeatedly computational experi-
ments and by integrating the modified strategies, making the
emergency response fast, easy, and efficient.
REFERENCES
[1] F. Y. Wang, “On the abstraction of conventional dynamic systems: From
numerical analysis to linguistic analysis,” Inf. Sci., vol. 171, no. 1–3,
pp. 233–259, Mar. 2005.
[2] F. Y. Wang, “Parallel system methods for management and control of
complex systems,” Control Decision, vol. 19, no. 5, pp. 485–489, 2004.
[3] F. Y. Wang, “Computational experiments for behavior analysis and de-
cision evaluation of complex systems,” J. Syst. Simul., vol. 16, no. 5,
pp. 893–897, May 2004.
[4] F. Y. Wang, “Toward a paradigm shift in social computing: The ACP
approach,” IEEE Intell. Syst., vol. 22, no. 5, pp. 65–67, Sep./Oct. 2007.
9. DONG et al.: EMERGENCY MANAGEMENT OF URBAN RAIL TRANSPORTATION BASED ON PARALLEL SYSTEMS 635
[5] F. Y. Wang, L. Li, X. Huang, and Y. Zou, “A discussion of fundamental
theory of long period continuous production emphasizing effectiveness,
safety and energy saving,” Comput. Appl. Chem., vol. 24, no. 12, pp. 171l–
1713, Dec. 2007.
[6] F. Y. Wang, “Parallel control and management for intelligent transporta-
tion system: Concepts, architectures, and applications,” IEEE Trans.
Intell. Transp. Syst., vol. 11, no. 3, pp. 630–638, Sep. 2010.
[7] L. F. Li, H. Zhang, X. F. Wang, W. Lu, and Z. P. Mu, “Urban transit
coordination using an artificial transportation system,” IEEE Trans. Intell.
Transp. Syst., vol. 12, no. 2, pp. 374–383, Jun. 2011.
[8] B. Ning, T. Tang, Z. Y. Gao, F. Yan, F. Y. Wang, and D. Zeng, “Intelligent
railway system in China,” IEEE Intell. Syst., vol. 21, no. 5, pp. 80–83,
Sep./Oct. 2006.
[9] B. Ning, T. Tang, H. Dong, D. Wen, D. Liu, S. Gao, and J. Wang, “An
introduction to parallel control and management for high-speed railway
systems,” IEEE Trans. Intell. Transp. Syst., vol. 12, no. 4, pp. 1473–1483,
Dec. 2011.
[10] B. Ning, H. Dong, D. Wen, L. Li, and C. Cheng, “ACP-based control
and management of urban rail transportation systems,” IEEE Intell. Syst.,
vol. 26, no. 2, pp. 84–88, Mar./Apr. 2011.
[11] H. R. Dong, B. Ning, B. G. Cai, and Z. S. Hou, “Automatic train con-
trol system development and simulation for high-speed railways,” IEEE
Circuit Syst. Mag., vol. 10, no. 2, pp. 6–18, Second Quart., 2010.
[12] H. R. Dong, B. Ning, G. Qin, Y. Lü, and L. Li, “Urban rail emergency
response using pedestrian dynamics,” IEEE Intell. Syst., vol. 27, no. 1,
pp. 52–55, Jan./Feb. 2012.
[13] G. Xiong, K. F. Wang, F. H. Zhu, and C. Cheng, “Parallel traffic man-
agement for the 2010 Asian Games,” IEEE Intell. Syst., vol. 25, no. 3,
pp. 81–85, May/Jun. 2010.
[14] G. Xiong, S. Liu, X. Dong, F. Zhu, B. Hu, D. Fan, and Z. Zhang, “Parallel
traffic management system helps 16th Asian games,” IEEE Intell. Syst.,
vol. 27, no. 3, pp. 74–78, May/Jun. 2012.
[15] O. Masahiro, “The book review (digested) the railway system of Japan
coming after 20 years,” Jpn. Railway Eng., vol. 49, no. 1, pp. 485–642,
2009.
[16] J. Eberhard, “Railway infrastructure and the development of high-speed
rail in Germany,” Railway Techn. Rev., vol. 2, pp. 43–51, 2005.
[17] J. A. Rosser, “High-speed planning intensifies,” Elect. Railway, vol. 53,
no. 218, pp. 184–185, Nov. 2008.
[18] P. Kramer, A. Aul, B. Vock, and C. Frank, “Emergency response manage-
ment near the tracks of the public railway network,” Anaesthesist, vol. 59,
no. 11, pp. 1021–1028, Nov. 2010.
[19] M. L. Yeh and C. T. Chang, “An automata based method for online
synthesis of emergency response procedures in batch processes,” Comput.
Chem. Eng., vol. 38, no. 5, pp. 151–170, Mar. 2012.
[20] K. Kepaptsoglou, M. G. Karlaftis, and G. Mintsis, “Model for planning
emergency response services in road safety,” J. Urban Plan. Dev.—ASCE,
vol. 138, no. 1, pp. 18–25, Mar. 2012.
[21] X. Liu, W. Li, Y. L. Tu, and W. J. Zhang, “An expert system for an
emergency response management in Networked Safe Service Systems,”
Expert Syst. Appl., vol. 38, no. 9, pp. 11 928–11 938, Sep. 2011.
[22] C. H. Chien, S. N. Yu, Y. Y. Huang, and F. C. Chong, “An efficient
framework of emergency response to facilitate disaster recovery for fire-
damaged medical equipment—Case study at a large medical centre after
a fire,” Safety Sci., vol. 49, no. 5, pp. 727–734, Jun. 2011.
[23] S. Arkoulis, D. E. Spanos, S. Barbounakis, A. Zafeiropoulos, and
N. Mitrou, “Cognitive radio-aided wireless sensor networks for emergency
response,” Meas. Sci. Technol., vol. 21, no. 12, p. 124 002, Dec. 2010.
[24] M. H. Zhong, C. L. Shi, T. R. Fu, L. He, and J. H. Shi, “Study in
performance analysis of China Urban emergency response system based
on Petri net,” Safety Sci., vol. 48, no. 6, pp. 755–762, Jul. 2010.
[25] P. S. Georgiadou, I. A. Papazoglou, C. T. Kiranoudis, and N. C. Markatos,
“Multi-objective evolutionary emergency response optimization for major
accidents,” J. Hazard. Mater., vol. 178, no. 1–3, pp. 792–803, Jun. 2010.
[26] M. G. Lenne, T. J. Triggs, C. M. Mulvihill, M. A. Regan, and
B. F. Corben, “Detection of emergency vehicles: Driver responses to
advance warning in a driving simulator,” Hum. Factors, vol. 50, no. 1,
pp. 135–144, Feb. 2008.
[27] J. B. Sheu, “An emergency logistics distribution approach for quick re-
sponse to urgent relief demand in disasters,” Transp. Res. E—Logist.
Transp. Rev., vol. 43, no. 6, pp. 687–709, Nov. 2007.
[28] D. F. Brown and W. E. Dunn, “Application of a quantitative risk as-
sessment method to emergency response planning,” Comput. Oper. Res.,
vol. 34, no. 5, pp. 1243–1265, May 2007.
[29] A. Dantas, E. Seville, and D. Gohil, “Information sharing during emer-
gency response and recovery—A framework for road organizations,”
J. Transp. Res. Board, no. 2022, pp. 21–28, 2007.
[30] D. Mendonca, G. E. G. Beroggi, D. van Gent, and W. A. Wallace, “De-
signing gaming simulations for the assessment of group decision support
systems in emergency response,” Safety Sci., vol. 44, no. 6, pp. 523–535,
Jul. 2006.
[31] N. S. Grigg, “Ready or not? Disaster preparedness and emergency re-
sponse in the water industry,” J. Amer. Water Works Assoc., vol. 98, no. 2,
pp. 242–255, Mar. 2006.
[32] D. Hinton, T. E. Klein, and M. Haner, “An architectural proposal for
future wireless emergency response networks with broadband services,”
Bell Labs Techn. J., vol. 10, no. 2, pp. 121–138, 2005.
[33] K. Lorincz, D. J. Malan, T. R. F. Fulford-Jones, A. Nawoj, A. Clavel,
V. Shnayder, G. Mainland, M. Welsh, and S. Moulton, “Sensor networks
for emergency response: Challenges and opportunities,” IEEE Pervasive
Comput., vol. 3, no. 4, pp. 16–23, Oct./Dec. 2004.
[34] J. Schwartz, “Emergency preparedness and response: Compensating
victims of a nuclear accident,” J. Hazard. Mater., vol. 111, no. 1–3,
pp. 89–96, Jul. 2004.
[35] F. Al-Qurashi, “New vision of emergency response planning,” Process
Safety Progr., vol. 23, no. 1, pp. 56–61, Mar. 2004.
[36] R. Könnecke and V. Schneider, “Evacuation from underground rail-
way stations—Available and required safe egress time for different sta-
tion types and general evaluation criteria,” Pedestrian Evac. Dyn. 2005,
pp. 363–368, 2007.
[37] Y. Wang, C. Zhang, L. Zhu, and Q. Sun, “Application of improved AHP
in the evaluation of railway emergency plans,” in Proc. Int. Conf. Netw.
Sens. Control, 2010, pp. 564–569.
[38] N. Sun, “Research on comprehensive evaluation of subway emergency
response capacity based on fuzzy centralization statistical theory,” in
Proc. Int. Symp. Emergency Manage., 2010, pp. 197–201.
[39] J. Zhang, G. Zhu, H. Li, and L. Zhang, “The evaluation of personnel
evacuate safely in commercial pedestrian street,” Procedia Eng., vol. 11,
pp. 675–681, 2011.
[40] G. Chu and J. Wang, “Study on probability distribution of fire scenarios
in risk assessment to emergency evacuation,” Reliab. Eng. Syst. Safety,
vol. 99, pp. 24–32, Mar. 2012.
[41] A. Borrmann, A. Kneidl, G. Köster, S. Ruzika, and M. Thiemann, “Bidi-
rectional coupling of macroscopic and microscopic pedestrian evacuation
models,” Safety Sci., vol. 50, no. 8, pp. 1695–1703, Oct. 2012.
[42] C. Shi, M. Zhong, X. Nong, L. He, J. Shi, and G. Feng, “Modeling and
safety strategy of passenger evacuation in a metro station in China,” Safety
Sci., vol. 50, no. 5, pp. 1319–1332, Jun. 2012.
Hairong Dong (SM’12) received the B.S. and M.S.
degrees in automatic control and basic mathematics
from Zhengzhou University, Zhengzhou, China, in
1996 and 1999, respectively, and the Ph.D. degree
in general and fundamental mechanics from Peking
University, Beijing, China, in 2002.
She is currently a Professor with the State Key
Laboratory of Rail Traffic Control and Safety,
Beijing Jiaotong University. She was a Visit-
ing Scholar with the University of Southampton,
Southampton, U.K., in 2006; The University of
Hong Kong, Pokfulam, Hong Kong, in 2008; the City University of
Hong Kong, Kowloon, Hong Kong, in 2009; The Hong Kong Polytechnic
University, Kowloon, in 2010; and the KTH Royal Institute of Technology,
Stockholm, Sweden, in 2011. In 2007, she served as a Project Level-3 Expert
with the Department of Transportation for the Beijing Organizing Committee of
the Olympic Games. Her research interests include stability and robustness of
complex systems control theory, intelligent transportation systems, automatic
train operation, and parallel control and management for high-speed railway
systems.
10. 636 IEEE TRANSACTIONS ON INTELLIGENT TRANSPORTATION SYSTEMS, VOL. 14, NO. 2, JUNE 2013
Bin Ning (SM’12) received the B.S., M.S., and
Ph.D. degrees from Beijing Jiaotong University,
Beijing, China.
From 1991 to 1992, he was a Visiting Scholar with
Brunel University, London, U.K., and, from 2002 to
2003, with the University of California, Berkeley.
He is currently a Professor with the State Key Lab-
oratory of Rail Traffic Control and Safety, Beijing
Jiaotong University, and is also the President of
Beijing Jiaotong University. He has directed many
key national scientific projects in China. His sci-
entific interests include intelligent transportation systems, communication-
based train control, rail transport systems, system fault-tolerant design, fault
diagnosis, system reliability, and safety studies.
Dr. Ning is a Fellow of the Institution of Railway Signal Engineers and
the Institution of Engineering Technology (IET), a Senior Member of the
China Railway Society, and a member of the Western Returned Scholars
Association. He is the Chair of the Technical Committee on Railroad Systems
and Applications of the IEEE Intelligent Transportation Systems Society.
Yao Chen received the B.S. degree in mathemat-
ics from China Three Gorges University, Yichang,
China, in 2007 and the Ph.D. degree from the
Academy of Mathematics and Systems Science, Chi-
nese Academy of Sciences, Beijing, China, in 2012.
From November 2009 to March 2010, he was a
Research Assistant with the Department of Math-
ematics, City University of Hong Kong. From
November 2010 to November 2011, he was a Re-
search Assistant with the School of Electrical and
Computer Engineering, Royal Melbourne Institute of
Technology, Melbourne, Australia. He is currently a Postdoctoral Fellow with
the School of Electronic and Information Engineering, Beijing Jiaotong Univer-
sity. His current research interests include complex networks and applications,
emergent behavior of multiagent systems, and rail traffic control.
Xubin Sun received the B.S. degree in electrical
engineering and automation from Beijing Jiaotong
University, Beijing, China, in 2002 and the Ph.D.
degree in control theory and control engineering
from the Institute of Automation, Chinese Academy
of Sciences, Beijing, in 2007.
He is currently a Lecturer with the School of Elec-
tronic and Information Engineering, Beijing Jiaotong
University. His research interests include parallel
control and management of urban rail transit and
high-speed railway systems, emergency response op-
timization, and stochastic control.
Ding Wen (SM’99) is a Professor with the National University of Defense
Technology, Changsha, China, where he is also a Senior Advisor with the Re-
search Center for Computational Experiments and Parallel Systems. His main
research interests include behavioral operation management, human resource
management, management information systems, and intelligent systems. He
has extensively published and received numerous awards for his work in these
areas.
Yuling Hu received the B.S. degree in electrical en-
gineering and the M.S. degree in pattern recognition
and intelligent systems from Beijing University of
Technology, Beijing, China, in 1996 and 2004, re-
spectively. She is currently working toward the Ph.D.
degree with the State Key Laboratory of Manage-
ment and Control for Complex Systems, Institute of
Automation, Chinese Academy of Sciences, Beijing.
She is currently an Associate Professor with
Beijing University of Civil Engineering and Archi-
tecture. Her research interests include evacuation
strategies in high-rise building fires and emergency management.
Renhai Ouyang received the B.S. degree in 2011
from the School of Electronic and Information Engi-
neering, Beijing Jiaotong University, Beijing, China,
where he is currently working toward the Ph.D.
degree with the School of Electronic and Information
Engineering.
His current research interests include complex
networks and applications, emergency response for
urban rail transit networks, and parallel control and
management of urban rail transit.