Note: Descriptions are shown in the official language in which they were submitted.
1
SYSTEM AND METHOD FOR OPTIMIZING OPERATIONS OF TRAINS TRAVELLING
ALONG A RAILWAY LINE
Field of the invention
The present invention concerns a system and a method for optimizing the
operations
of trains travelling along a railway line.
The system and method according to the present invention are particularly
suitable
to be used in urban or metro rail lines, and they will be described
hereinafter by making
specific reference to such applications, without intending in any way to limit
their possible
application to other type of railway networks.
Background of the invention
As it is known, railway transportation systems, and in particular urban rail
transit
systems, are widely and increasingly used worldwide.
During service, the various trains have to perform their operations based on a
pre-
established timetable generated by a Centralized Control Center ("CCC"); the
timetable
can be impacted by some events on the railway line that influences the real
operating
times of each train.
In particular, on-board Automatic Train Operation (ATO) systems might not be
able
to meet a desired headway due to the variable dwell time at each station, i.e.
the time
needed for completing boarding/deboarding of passengers at the station
platforms.
In fact, the dwell time depends on several variables and can vary over the
time at
each station and from station to station; for example, such variables include
the peak/off-
peak hours/seasons, the historic/tourist/industrial importance of places near
each station,
current climatic conditions, et cetera.
Hence, from one side it is difficult to respect timetables based on nominal
dwell
times uniform for all stations; on the other side, it is often difficult to
generate timetables at
centralized control centers with different dwell times due to uncertainties
and variables
varying anyhow in real time, as above mentioned.
Summary of the invention
The present invention is aimed at facing such issues, and in particular at
providing a
system and a method for optimizing operations of trains travelling along a
railway line,
offering substantial improvements over known solutions, especially as regard
to the
capability of properly taking into account the dwell time at each station in
relation to the
real time operating conditions.
This aim is achieved by a system for optimizing the operations of a train
travelling
along a railway line, characterized in that it comprises at least:
Date Recue/Date Received 2021-01-19
2
- one or more cameras which are installed at a platform of a first station
located
along the railway line and are adapted to capture images of the train relative
to said
platform and of passengers over the platform moving for boarding on/de-
boarding from
the train;
- an image processing system configured at least to measure the actual dwell
time
and to calculate an optimum dwell time for the train at the platform of the
first station
based on the image data received from said one or more cameras;
- a first data elaboration system which is configured to calculate a
predicted dwell
time for the train at the platform of an upcoming station along the railway
line, based on
one or more parameters selected from the group comprising a calculated number
of
actual passengers inside the train, number of passengers at the platform of
the upcoming
station, measured actual dwell time and calculated optimum dwell time at the
platform of
the upcoming station for a preceding train travelling along the railway line,
data related to
actual environmental conditions along the railway line or parts thereof, data
indicative of
passengers traffic-related characteristics for a calendar date or part
thereof.
The above mentioned aim is also achieved by a method for optimizing the
operations of a train travelling along a railway line, characterized in that
it comprises at
least the following steps:
- (a): capturing images of the position of the train relative to a platform
of a first
station and of passengers over the platform moving for boarding on/de-boarding
from the
train;
- (b): based on the images captured, measuring the actual dwell time and
calculating an optimum dwell time for the train at the platform of the first
station;
- (c): calculating a predicted dwell time for the train at the platform of
an upcoming
station along the railway line, based on one or more parameters selected from
the group
comprising one or more of a calculated number of actual passengers inside the
train,
number of passengers at the platform of the upcoming station, measured actual
dwell time
and calculated optimum dwell time at the platform of the upcoming station for
a preceding
train travelling along the railway line, data related to actual environmental
conditions along
the railway line or parts thereof, data indicative of passengers-related
traffic
characteristics for a calendar date or part thereof.
Brief description of drawings
Further characteristics and advantages will become apparent from the
description of
some preferred but not exclusive exemplary embodiments of a system and a
method
according to the present disclosure, illustrated only by way of non-limitative
examples with
the accompanying drawings, wherein:
Date Recue/Date Received 2021-01-19
3
Figure 1 is a block diagram schematically representing a system for optimizing
the
operations of a train travelling along a railway line according to the
invention;
Figure 2 is a flow chart illustrating a method for optimizing the operations
of a train
travelling along a railway line, according to the invention;
Detailed description of preferred embodiments of the invention
It should be noted that in the detailed description that follows, identical or
similar
components, either from a structural and/or functional point of view, have the
same
reference numerals, regardless of whether they are shown in different
embodiments of the
present disclosure.
It should be also noted that in order to clearly and concisely describe the
present
disclosure, the drawings may not necessarily be to scale and certain features
of the
disclosure may be shown in somewhat schematic form.
Further, when the term "adapted" or "arranged" or "configured" or "shaped", is
used
herein while referring to any component as a whole, or to any part of a
component, or to a
combination of components, it has to be understood that it means and
encompasses
correspondingly either the structure, and/or configuration and/or form and/or
positioning.
In particular, for electronic and/or software means, each of the above listed
terms
means and encompasses electronic circuits or parts thereof, as well as stored,
embedded
or running software codes and/or routines, algorithms, or complete programs,
suitably
designed for achieving the technical result and/or the functional performances
for which
such means are devised.
In figure 1 there is schematically illustrated an exemplary system for
optimizing the
operations of trains travelling along a railway line 10, according to the
present invention,
indicated by the overall reference number 100.
The railway line 10 can be for example an urban rail system, e.g. a subway,
and
comprises a plurality of line stations located along the railway line 10 at a
certain distance
from each other.
In particular, for the sake of simplicity, in figure 1 the railway line 10 is
represented
by a portion of a track and as including only two line stations, namely a line
station 3
referred to in the following also as the "first station 3", and a further line
station 4, referred
to in the following also as the "upcoming station 4", which follows the first
station 3 with
reference to a train, schematically represented in figure 1 by the reference
number 1,
moving along the line 10 in the direction indicated by the arrow "A".
Clearly, the number and type of line stations, as well as the type and
composition of
each train, can vary according to the specific applications.
Date Recue/Date Received 2021-01-19
4
Each station has one platform 2 associated to a track of the line 10 where
trains,
while performing their duty service, stop for on/off boarding of passengers.
As schematically illustrated in figure 1, the system 100 according to the
present
invention comprises one or more cameras 110 which are installed at the
platform 2 of the
first station 3 and are adapted to capture images of the train 1, and in
particular of its
actual position relative to the platform 2, and also images of passengers 5
over the
platform 2 moving for boarding on/de-boarding from the train 1.
According to a possible embodiment, the system 100 comprises at least a first
camera 111, placed at the platform 2, e.g. at the entrance thereof, with its
field of vision
oriented to capture image data of the train 1 entering into or leaving the
platform 2, and at
least a second camera 112, preferably a plurality of second cameras 112,
mounted at the
platform 2, e.g. at the roof or at a side thereof, with its/their field of
vision oriented for
example towards the expected position of the doors of the train 1 once the
train has
stopped at the platform 2, to capture image data of passengers moving into/out
from the
train 1.
The second cameras 112 can be properly positioned to cover the door-to-door
distance of the train 1, and eventually the entire platform 2, possibly
without any gap.
The first camera 111 can be for example a dual head video camera, with the
front
head facing the entrance of the platform 2, and the rear head facing the last
car of the
train as shown in figure 1; such camera 111 can trigger for instance an
automatic
recording of video images by the second cameras 112 based on the arrival and
departures of a train 1.
For the sake of ease of illustration, only the platform 2 of the first station
3 is shown
equipped with the one or more cameras 110; clearly, more stations, preferably
all stations
located along the railway line 10, are provided with the various cameras 110
at each
relevant platform 2.
As illustrated, the system 100 further comprises an image processing system
120
which is configured at least to measure the actual dwell time and to calculate
an optimum
dwell time for the train at the platform 2 of the first station 3, based on
the image data
received from the one or more cameras 110.
For example, the image processing system 120 comprises or is constituted by a
video analytics server that is in operative communication with the one or more
cameras
110, via respective communication devices 125.
According to a possible embodiment, the processing system or video analytics
server 120 is centralized, namely it is located for example in a centralized
control room
devised to supervise the entire railway line 10, and is in operative
communication and
Date Recue/Date Received 2021-01-19
5
receive all image data from all cameras 110 installed in the various stations
of the line 10
itself.
In practice, for each station, a virtual machine is identified to process the
corresponding dedicated video data, and the system 120 performs video
analytics to
extract the count of passengers 5 at a platform 2 for boarding/de-boarding the
train 1
based for example on the direction of movement of passengers, and also
measures the
actual dwell times and calculates the optimal dwell times.
For instance, since the video images are initially taken just before the train
arrival
time, all the passengers 5 waiting on the platform can be counted as the
number of
passengers boarding the train; based on the velocity of passengers 5 moving
towards the
platform, the de-boarding passengers can be identified and using the top view
of heads,
hairs and noses side pointing, the number of de-boarding passengers from the
train 1 is
calculated.
The "actual dwell time" is for example measured from the video data based on
the
time interval between the doors opening and the doors closing automatically;
the "optimal
dwell time" is calculated from the video data based on the time interval
between the doors
opening and there is no movement of passengers before the doors closing
automatically.
The "optimal dwell time" can be shorter or equal to the "actual dwell time".
According to the invention, a first elaboration system or unit 130,
operatively
connected to the image processing system 120, is configured to calculate a
predicted
dwell time for the train 1 at the platform 2 of an upcoming station 4 along
the railway line
10, based on one or more parameters selected from the group comprising a
calculated
number of actual passengers inside the train 1, number of passengers at the
platform 2 of
the upcoming station 4, measured actual dwell time and calculated optimum
dwell time at
the platform 2 of the upcoming station 4 for a preceding train travelling
along the railway
line 10 in the same direction A, data related to actual environmental
conditions along the
railway line 1 or parts thereof, data indicative of passengers traffic-related
characteristics
for a calendar date or part thereof.
According to an embodiment, the first elaboration system 130 comprises, or is
constituted by, a neural-network-based first server 130 that is in operative
communication
with the video analytics server 120, and is also centralized, i.e. it is
placed in a centralized
control room for the entire line 10 and is unique for the various stations
along the line 10
itself.
Since the number of passengers boarding/de-boarding a train depends also on
the
location of the station, a kind virtual machine is identified at the system
130 for each
station i.e. a neural network-based model is defined for each station.
Date Recue/Date Received 2021-01-19
6
In practice, according to this embodiment, the neural-network-based first
server 130
is pre-trained with various relevant data for each station and using suitable
weights for the
various parameters inputs above indicated, namely the number of passengers 5
inside a
relevant train 1, number of passengers 5 at the platform 2 of the upcoming
station 4,
measured actual dwell time and calculated optimum dwell time at the platform 2
of the
upcoming station 4 for a preceding train travelling along the railway line 10
in the same
direction A, environmental or climatic data such as temperature, wind speed,
rain, snow et
cetera, data related to calendar dates, such as public holidays and festival
days, vacation
periods, which can be input by operators, peak/non-peak times which can be
assigned
automatically or input manually.
According to a possible embodiment, the environmental or climatic data such as
the
ones above indicated, are provided by one or more sensors of the system 100;
these
sensors are placed at selected positions along the railway line 10 for
detecting actual data
of one or more corresponding environmental parameters to be supplied to the
first server
130 for calculating the predicted dwell time for the train 1 at the platform 2
of an upcoming
station 4. For example, such one or more sensors, schematically represented in
figure 1
by the reference number 115 only for the first station 3, can be outdoor
sensors mounted
at each station.
Once the training of the elaboration system 130 is completed, it is validated
using
one or more set of real-time data, and then the elaboration system 130 is put
in real time
operations for predicting the dwell time at the various stations.
Since the whole system 100 processes and collects data throughout every day, a
re-
learning/re-calculation of the system 130 can be performed periodically, for
example every
night at a fixed time. If the weights previously used for whatever reason
diverge
substantially from a tolerable limit, new weights are uploaded into the system
130. This
ensures the continuous learning of dwell time prediction over the period and
improves the
accuracy in the prediction.
As illustrated in figure 1, the system 100 according to the invention
comprises a
second elaboration system or unit 140, operatively connected to the first
elaboration
system or unit 130, which is configured to calculate, for the train 1 leaving
the platform 2
of the first station 3, first the total travel time needed to reach and leave
the platform 2 of
the upcoming station 4 and then to generate, based on the calculated total
travel time, at
least one optimized driving profile to be followed by the train 1 from the
platform 2 of the
first station 3 up to stopping at the platform of the upcoming station 4.
In practice, the total travel time is calculated as the arrival time from the
first station 3
at the upcoming station 4 plus the dwell time at the upcoming station 4
itself.
Date Recue/Date Received 2021-01-19
7
The headway between two trains can be calculated, for instance automatically,
based on the arrival time at the same station using images captured by the
dual head
camera 111.
According to an embodiment, the second elaboration system or unit 140
comprises,
or is constituted, by an optimization server 140 (hereinafter also referred to
as the second
server 140) that is in operative communication with the first elaboration
system or unit 130
and is also centralized, i.e. it is placed in a centralized control room for
the entire line 10
and is unique for the various stations along the line 10 itself.
In particular, according to a possible embodiment, the second elaboration
system or
server 140 is configured first to generate a plurality of optimized driving
profiles based on
the calculated total travel time, and then to select one driving profile among
the plurality of
driving profiles generated to be followed by the train 1 based on a selectable
operative
target desired to be achieved.
The selectable operative target can be for example punctuality, e.g. the
respect of a
prefixed timetable, or alternatively energy consumption, for example to
achieve certain
savings; of course, other operative targets and/or combinations thereof could
be selected.
Accordingly, in one possible embodiment of the system 100, the second server
140
is configured to select among the plurality of generated driving profiles, one
driving profile
to be followed by the train 1 based on a predefined travelling time to be
respected by the
train 1, i.e. according to the operative target of punctuality, if the
predicted dwell time for
the train 1 is longer than or equal to a predefined nominal dwell time at the
platform 2 of
the upcoming station 4; alternatively, the driving profile to be followed by
the train 1 can be
selected based on a desired level of energy savings to be achieved by the
train 1 (i.e. the
operative criteria of consumption is chosen) if the predicted dwell time for
the train 1 is
shorter than a predefined nominal dwell time at the platform 2 of the upcoming
station 4.
The nominal dwell time is for example the typical one taken into consideration
by the
centralized control center, schematically represented in figure 1 by the
reference number
150, when generating the overall timetable for the trains travelling over the
railway line 10.
In practice, for each train, a virtual machine is identified to perform the
optimization
calculations based on various inputs, namely data from train and track
database, such as
data related to maximum acceleration/deceleration rates for each train,
permanent speed
restrictions for sections of tracks, et cetera. A suitable optimization
algorithm of the
second server 140 provides a set of optimized driving profiles wherein, for
example, in
one scenario, either acceleration-cruising (AC) or acceleration-cruising-
coasting (ACC)
times are optimized. Accordingly, a train can have different modes such as
Motoring-
Cruising-Coasting-Braking (MCCB) or Motoring-Cruising-Braking (MCB). Hence,
the
Date Recue/Date Received 2021-01-19
8
optimization algorithm provides the set of optimized driving profiles, i.e. an
optimized time
for AC/ACC and an optimized mode of operation for MCCB/MCB along with the
output of
"optimized total time" to be traveled (punctuality) and predicted energy
consumption. The
decision logic of the second server 140 selects the best-optimized driving
profile from the
set of generated profiles based on the chosen objective punctuality or energy
saving, and
transmits it to the on-board Automatic Train Operation (ATO) system of the
train 1 for
execution.
A method for optimizing the operations of a train 1 traveling along a railway
line 10,
according to the invention, will be now described with reference to figure 2.
In particular, the method, indicated by the overall reference number 200,
comprises
at least the following steps:
- 210: capturing, for example by means of the one or more cameras 110 which
are installed at a platform 2 of the first station 3 located along the railway
line 10, images
of the train 1 relative to the platform 2, and of passengers 5 over the
platform 2 moving for
boarding on/de-boarding from the train 1;
- 220: based on the images captured by the one or more cameras 110,
measuring
the actual dwell time and calculating an optimum dwell time for the train 1 at
the platform 2
of the first station 3, for example by means of the image processing system
120;
-
230: calculating, for example by means of the first data elaboration system
130,
a predicted dwell time for the train 1 at the platform 2 of an upcoming
station 4 along the
railway line 10, based on one or more parameters selected from the group
comprising a
calculated number of actual passengers inside the train 1, number of
passengers at the
platform 2 of the upcoming station 4, measured actual dwell time and
calculated optimum
dwell time at the platform 2 of the upcoming station 4 for a preceding train
travelling along
the railway line 10 in the same direction, data related to actual
environmental conditions
along the railway line 1 or parts thereof, data indicative of passengers-
related traffic
characteristics for a calendar date or part thereof.
In one possible embodiment, the method 200 further comprises the following
steps:
-
240: calculating for the train 1 leaving the platform 2 of the first
station 3, for
example by means of the second elaboration system 140, a total travel time to
reach and
leave a platform 2 of the upcoming station 4; and then
-
250: generating, based on the calculated total travel time, and still for
example
by means of a second elaboration system 140, at least one optimized driving
profile to be
followed by the train 1 from the platform 2 of the first station 3 up to the
platform 2 of the
upcoming station 4.
Date Recue/Date Received 2021-01-19
9
According to an embodiment, the step 250 of generating at least one optimized
driving profile comprises a first sub-step 252 of generating a plurality of
optimized driving
profiles based on the calculated total travel time, and a second sub-step 254
of selecting,
among the plurality of generated driving profiles, one driving profile to be
followed by the
train 1 based on a selectable operative target, such as in particular
punctuality or energy
consumption.
More in particular, in one embodiment, the second sub-step 254 of selecting
comprises selecting among the plurality of generated driving profiles, one
driving profile to
be followed by the train 1 based on a predefined travelling time to be
respected by the
train 1 if the predicted dwell time for the train 1 is longer than or equal to
a predefined
nominal dwell time at the platform 2 of the upcoming station 4; or
alternatively, the one
driving profile is selected based on a level of energy savings to be achieved
by the train 1
if the predicted dwell time for the train 1 is shorter than a predefined
nominal dwell time at
the platform 2 of the upcoming station 4.
In one embodiment, the method 200 comprises a step 260 of supplying with
actual
data of one or more corresponding environmental parameters detected by one or
more
sensors 115 placed along the railway line 10, the first data elaboration
system 130 for the
calculation at step 230 of the predicted dwell time for the train 1 at the
platform 2 of the
upcoming station 4.
In practice, when a train 1 starts its service and enters the first station 3
of the line
10, the number of passengers inside the train can be considered equal to zero.
Once the
train 1 arrives within the field of view (FOV) of the front head of first
camera 111, the first
camera 111 sends the alert to the second cameras 112 to start the recording.
Based on
this alert, recording of the video images of crowd at the platform 2 is
started to count a
number of passengers 5 ready for boarding the train 1. Complete stop of the
train 1 at the
platform 2 is detected using the rear head of the camera 111, and dwell time
counter
starts based on the trigger from the data of the rear head of the camera 111.
At the same
time, the second cameras 112 capture moments of people de-boarding the train
1. The
video recording continues until the train 1 starts the movement leaving the
platform 2 after
the doors of the train 1 have been closed. The train movement is detected
using the rear
head of the first camera 111; this completes the calculation of the "actual
dwell time" by
the video analytics server 120. To this end, data are transmitted to the video
analytics
server 120 via the communication devices 125 and using for example high
bandwidth
wireless communication. Using the analyzed video data, the actual time at
which
passenger boarding is completed (irrespective of the actual dwell time) is
calculated and
represents the "optimum dwell time". When used, outdoor sensors 115 send
Date Recue/Date Received 2021-01-19
10
environmental data to the first data elaboration system 130 via respective
communication
devices 125, using for example low bandwidth wireless connection. Based on the
video
data and sensors data at the upcoming station 4, the first data elaboration
system 130
predicts the "optimum dwell time" using the neural network weights specific
for the
upcoming station 4. If the predicted dwell time is longer than the typical
nominal dwell
time, then the running profile for the train 1 is selected such that train 1
reaches the
upcoming station few seconds ahead of the scheduled time of arrival. This
gives
additional time for the passengers to board/de-board the train 1. As the early
time of
arrival is utilized for dwell time, the overall timetable is not affected. The
early arrival can
be dynamically updated into the centralized control enter 150. If the
predicted dwell time is
substantially equal to the typical nominal dwell time, then the running
profile is selected
such that for example the scheduled time is strictly followed. In this case,
punctuality is
given primary importance over energy consumption. If the predicted dwell time
is shorter
than the typical nominal dwell time, then the running profile is selected such
that energy
consumption and hence savings is given more importance than punctuality. This
may
result in a profile with more coasting time, and hence can result in a slight
delay for arrival
time. For fewer passengers, a shorter dwell time is sufficient. Thus, only the
additional
unused dwell time is exploited in favour of energy savings. This will not
affect the
timetable but still, the delayed arrival by few seconds can be communicated to
the control
center 150 via the second elaboration system 140. The optimized profiles for
each
scenario are generated by the optimization server 140 using the train-track
data and any
constraint for the respective scenario, i.e. the trip time for punctuality,
and a percentage of
savings for energy consumption. Based on the optimized profiles, the driving
profile is
chosen and forwarded to train 1. The selected driving profile includes, inter
alia,
information about optimum motoring time-cruising time - coasting time or
motoring time-
cruising time. The on-board automatic train operation system drives the train
1 as per the
Motoring-Cruising-Coasting-Braking (MCCB) or Motoring-Cruising-Braking (MCB)
modes,
based on the respective profile received from the optimization server 140. The
various
steps above are repeated at each station and for each train autonomously
wherein, in
sequence, the upcoming station 4 would represent the first station 3 once the
train 1
arrives there, and the following station along the line 10 would represent the
new
upcoming station 4.
Hence, it is evident from the foregoing description that the system 100 and
method
200 according to the present invention achieve the intended aim since they
allow to
optimize the operations of trains along a railway line, in particular by
taking into account,
in real time, the dwell time at each station and then by dynamically adjusting
the actual
Date Recue/Date Received 2021-01-19
11
driving profile of each train. Accordingly, it is possible to obtain flexibly
either energy
savings and/or punctuality based on the selected driving profiles.
The system 100 and method 200 thus conceived are susceptible of modifications
and variations, all of which are within the scope of the inventive concept, as
defined in
particular by the appended claims; for example, the various systems or units
120, 130,
140 previously described can be preferably positioned in a unique operative
room where
also the central control room 150 is located, each and any of them can
comprise an
assembly of HW and SW components, such as one or more workstations and related
control displays, elaboration units or processor based devices, et cetera, as
schematically
represented by the graphical symbols depicted in figure 1.
Date Recue/Date Received 2021-01-19
