EP3437955B1

EP3437955B1 - Train disembarking passenger number prediction system, congestion visualization and evaluation system, and riding capacity calculation system

Info

Publication number: EP3437955B1
Application number: EP17773617.0A
Authority: EP
Inventors: Rui NING; Manabu Katou; Masayasu Fujiwara
Original assignee: Hitachi Ltd
Current assignee: Hitachi Ltd
Priority date: 2016-03-30
Filing date: 2017-01-31
Publication date: 2021-04-28
Anticipated expiration: 2037-01-31
Also published as: JP6640988B2; EP3437955A4; WO2017169068A1; JPWO2017169068A1; EP3437955A1

Description

Technical Field

The present invention relates to a system that provides visualization and prediction information on a congestion situation.

Background Art

In railway stations, an increase of congestion often occurs due to transportation trouble of transportation or the like in addition to occurrence of daily congestion in a commuting time zone or the like. There are concerns about crowd accidents such as an increase of train delay caused by an increase in train getting on and off time and fall from a platform due to congestion. Thus, it is important to timely grasp a congestion situation in a station and to take appropriate guidance and countermeasures. In order to timely grasp the congestion situation in the station, it is necessary to timely grasp the number of persons flowing into the station. The inflow to the station can be divided into customers entering from outside the station and alighting customers from a train, but it is difficult to directly measure the number of customers alighting from the train.
PTL 1 discloses a vehicle congestion rate prediction system that includes a counting device, which receives data including alighting stations read by an automatic ticket checker from tickets passing through the automatic ticket checker on an entrance side of each station and counts the number of persons alighting at each alighting station; and means for predicting a vehicle on which a user passing through the automatic ticket checker rides by referring to a storage device, which stores statistical data in which an alighting station of a ticket is associated with each ride ratio of users, who have tickets designating the alighting station, riding on each vehicle, and calculating the number of persons riding on each vehicle and the number of persons alighting from each vehicle based on the prediction. SELBY COXON ET AL: "Testing the Efficacy of Platform and Train Passenger Boarding, Alighting and Dispersal Through Innovative 3D Agent-Based Modelling Techniques" URBAN RAIL TRANSIT, vol. 1, no. 2, 1 June 2015 (2015-06-01), pages 87-94 proposes the building of a boarding and alighting simulator using crowd behavior modelled in 3D animated figures. JP 2016-7906A proposes a train selection assistance system for assisting boarding of the train by a passenger.

Citation List

Patent Literature

PTL 1: JP 2004-178358 A

Summary of Invention

Technical Problem

However, there are the following problems in estimation of a train alighting person count according to the method disclosed in PTL 1.
In PTL 1, it is necessary to record passage of an automatic ticket checker at each station in order to calculate the train alighting person count. Thus, it is necessary to acquire information of all the stations even in a case where it is desired to acquire a congestion situation of a single station. It is difficult to obtain the information on all the stations in a complicated rail network due to direct operation between railway operators and the like.
In addition, it is necessary to make it possible to timely acquire records of passage of the automatic ticket checkers of all the stations in order to timely acquire the train alighting person count, so that it is necessary to make a great deal of investment for the automatic ticket checkers and the accompanying systems.
An object of the present invention is to make it possible to predict the number of alighting persons from a train using information that can be acquired at a single station.

Solution to Problem

A train alighting person count prediction system according to claim 1 is provided.

Advantageous Effects of Invention

It is possible to predict a train alighting person count from a train using information that can be acquired at a single station.

Brief Description of Drawings

[FIG. 1] FIG. 1 is a diagram illustrating an example of a configuration of an alighting person count prediction device of the present invention.
[FIG. 2] FIG. 2 is a diagram illustrating an example of an installation position of a sensor of a measurement unit.
[FIG. 3] FIG. 3 is a view illustrating an example of a data structure of person count measurement information.
[FIG. 4] FIG. 4 is a view illustrating an example of a data structure of departure/arrival time information.
[FIG. 5] FIG. 5 is a view illustrating an example of a data structure of alighting person count information.
[FIG. 6] FIG. 6 is a view illustrating an example of a data structure of train interval information.
[FIG. 7] FIG. 7 is a view illustrating an example of a data structure of prediction model information.
[FIG. 8] FIG. 8 is a flowchart illustrating an example of a database creation process.
[FIG. 9] FIG. 9 is a flowchart illustrating an example of an alighting person count prediction process.
[FIG. 10] FIG. 10 is a schematic graph illustrating an example of a method of calculating an alighting person count.
[FIG. 11] FIG. 11 is a flowchart illustrating an example of processing of the alighting person count calculation unit.
[FIG. 12] FIG. 12 is a schematic view illustrating an example of a train interval calculation method.
[FIG. 13] FIG. 13 is a schematic graph illustrating an example of a train alighting person count prediction model.
[FIG. 14] FIG. 14 is a flowchart illustrating an example of processing of a prediction model creation unit.
[FIG. 15] FIG. 15 is a schematic graph illustrating an example of the train alighting person count prediction model using a delay rate.
[FIG. 16] FIG. 16 is a flowchart illustrating an example of processing of an alighting person count prediction unit.

Description of Embodiments

An embodiment of a train alighting person count prediction device of the present invention will be described hereinafter with reference to the drawings.

First Embodiment

FIG. 1 is a diagram illustrating an example of a configuration of a train alighting person count prediction device of the present invention. The train alighting person count prediction device is a device that timely predicts the number of persons alighting from a train at a railway station, and includes a measurement unit 100, an arithmetic unit 200, a recording unit 300, and an output unit 400. The measurement unit 100, the arithmetic unit 200, the recording unit 300, and the output unit 400 can communicate with each other and operate on one or a plurality of interconnected computers.
The measurement unit 100 includes a person count measurement unit 101 that measures a passer-by count inside a station and a train departure/arrival detection unit 102 that detects departure and arrival of a train in the station.
The arithmetic unit 200 includes: an alighting person count calculation unit 201 that estimates a past train alighting person count; a train interval calculation unit 202 that calculates an interval between arrival times of a target train and a train which has arrived immediately previously on the same track; a prediction model creation unit 203 that creates a prediction model of a train alighting person count based on statistical information on a train interval and a alighting person count of the past; and an alighting person count prediction unit 204 that predicts a train alighting person count with an input of a train interval of a train at the time of arrival of the train.
The recording unit 300 is a database that holds, as data, person count measurement information 301 which is a detection result of the passer-by count, departure/arrival time information 302 which is departure/arrival time of a train, alighting person count information 303 which is an estimated value of an alighting person count for each train, train interval information 304 which is an arrival interval for each train, and prediction model information 305 which is a prediction model for predicting a train alighting person count based on a train interval.
The output unit 400 outputs a result of prediction of the train alighting person count.

Next, a function of each constituent element and data to be used will be described.
First, functions of elements constituting the measurement unit 100 will be described.
The person count measurement unit 101 is a sensor device capable of measuring a local passer-by count in the station for each direction of movement, and outputs the passer-by count as the person count measurement information 301 for each time zone and direction. The person count measurement unit 101 is realized, for example, by using a surveillance camera installed in the station as a sensor and measuring the number of persons by image processing. In the present embodiment, it is assumed that the sensors are installed in stairs, escalators and the like connecting a platform and a ticket gate floor in order to estimate the past train alighting person count. For example, the sensors are installed at the positions of a camera 701 and a camera 702 in FIG. 2 to measure each passer-by count at points 711 and 712.
The train departure/arrival detection unit 102 is a sensor device capable of detecting departure and arrival of a train, and detects arrival or departure of a train, records the time thereof, and outputs a result of the detection as the departure/arrival time information 302. The train departure/arrival detection unit 102 is realized, for example, by using a surveillance camera installed on the platform such as the camera 703 in FIG. 2 as a sensor and detecting departure and arrival of a train by image processing.
Next, functions of elements constituting the arithmetic unit 200 will be described.
The alighting person count calculation unit 201 receives inputs of the past person count measurement information 301 and the past departure/arrival time information 302 and allocate the measured passer-by count to trains to estimate the number of persons alighting from each train, and outputs the estimated count as the alighting person count information 303.
The train interval calculation unit 202 calculates an arrival interval time of a train arriving at the same track and outputs the calculated time as the train interval information 304.
Based on data obtained by associating the past alighting person count information 303 with the past train interval information 304, the prediction model creation unit 203 creates a model for predicting the train alighting person count based on the train interval and outputs the created model as the prediction model information 305.
When arrival of a train is detected, the alighting person count prediction unit 204 predicts the number of persons alighting from the train using the prediction model information 305 with the train interval output by the train interval calculation unit 202 as an input, and outputs the predicted count.
Next, a data structure used in the recording unit 300 will be described.
The person count measurement information 301 is data obtained by recording the measurement result of the person count measurement unit 101, is data constituted by a position ID which specifies a sensor installation position, a measured date, measurement start time and end time, a direction ID which specifies a movement direction of a pedestrian to be measured, and the number of measured persons as illustrated in FIG. 3, and is held as a database in the recording unit 300.
The departure/arrival time information 302 is data obtained by recording the detection result of the train departure/arrival detection unit, is data constituted by a track ID which specifies an arrival track of a target train, a date and time when detecting the train, and a type for distinguishing whether the detected train is of arrival or departure as illustrated in FIG. 4, and is held as a database in the recording unit 300.
The alighting person count information 303 is data obtained by recording the alighting person count for each train, is data constituted by a date when detecting a train, a track ID, an arrival time, and the alighting person count as illustrated in the drawing, and is held as a database in the recording unit 300. The date, the track ID, and the arrival time are information configured to uniquely specify a train, and data in which a train ID is associated with an alighting person count by attaching the train ID for each train may be used.
The train interval information 304 is data obtained by recording the train interval with an immediately previous train for each train, is data constituted by a date when detecting a train, a track ID, an arrival time, and a train interval as illustrated in FIG. 6, and is held as a database in the recording unit 300. The date, the track ID, and the arrival time are information configured to uniquely specify a train, and data in which a train ID is associated with an alighting person count by attaching the train ID for each train may be used.
The prediction model information 305 is data obtained by recording the model for predicting the train alighting person count from the train interval and is constituted by a time zone, an attribute, a track ID and a model formula as illustrated in FIG. 7. In the present embodiment, the prediction model is recorded as the model formula, but the model is not limited to the formula. For example, the model may be held in the form of a table in which a delay time and an alighting person count are associated for each condition.

Next, an example of the entire processing flow of the train alighting person count prediction device will be described, and then, an example of a processing flow of each unit constituting the train alighting person count prediction device will be described. The processing of the train alighting person count prediction device can be divided into a database creation process and an alighting person count prediction process.
First, a processing flow of the database creation process will be described using a flowchart of FIG. 8. Hereinafter, a "step" will be abbreviated as "S". For example, Step 4001 will be denoted as S4001.
S4001: A passer-by count at a predetermined position in the station is measured using the person count measurement unit 101, and a measurement result is saved as the person count measurement information 301 in the recording unit 300, thereby creating the database of the person count measurement information 301.
S4002: A departure or arrival time of a train departing or arriving at the station is detected using the train departure/arrival detection unit 102, and the detection result is saved as the departure/arrival time information 302 in the recording unit 300, thereby creating the database of the departure/arrival time information 302.
S4003: The passer-by count recorded in the person count measurement information 301 is divided by the train arrival time recorded in the departure/arrival time information 302, a passer-by count from an arrival time of each train to an arrival time of a train arriving subsequently to the corresponding train is allocated to the train to calculate the number of persons alighting from the train in the alighting person count calculation unit 201, and the calculated count is saved as the alighting person count information 303 in the recording unit 300, thereby creating the database of the alighting person count information 303.
S4004: When the train departure/arrival detection unit 102 detects arrival of a train and the departure/arrival time information 302 is output, the train interval calculation unit 202 calculates an interval with respect to an arrival time of a train which has immediately previously arrived on the same track as the train from the departure/arrival time information 302, which has been recorded in the recording unit 300, as a train interval and is saved the train interval information 304 in the recording unit 300, thereby creating a train interval information database.
S4005: The prediction model creation unit 203 associates the alighting person count information 303 recorded in the recording unit 300 with the train interval information 304 to be classified for each condition such as the track, the time zone, and the like, then a relational expression between a train interval and an alighting person count is calculated for each condition and saved as the prediction model information 305 in the recording unit 300, thereby creating a prediction model information database.
As above, it is possible to create all the databases recorded in the recording unit 300. The database is updated by repeating the above-described process timely or periodically in accordance with a measurement result of the measurement unit 100. "Periodically" refers to updating, for example, on a daily basis.
Subsequently, a processing flow of the alighting person count prediction process at the time of arrival of a train will be described using a flowchart of FIG. 9.
S4101: When the train departure/arrival detection unit 102 detects an arriving train, the departure/arrival time information 302 is output and the alighting person count prediction processing is started.
S4102: The train interval calculation unit 202 calculates an arrival interval between the train and a train which has immediately previously arrived on the same track based on the departure/arrival time information 302.
S4103: The alighting person count prediction unit 204 acquires the prediction model information 305 conforming to a condition at the time of arrival of the train from the recording unit 300, and inputs the train interval to the prediction model to calculate a predicted value of the alighting person count under the condition.
S4104: The output unit 400 outputs the predicted value of the alighting person count. For example, the predicted value of the alighting person count is output to a known pedestrian simulator device capable of estimating a congestion situation of a predetermined space by simulating movement of pedestrians with the number of pedestrians as an input and visualizing and evaluating the congestion situation, thereby realizing visualization and prediction of the congestion situation in the station. In addition, in the case of including means for measuring or estimating the number of passengers in a train by a known method, it is possible to calculate a passenger count after alighting by subtracting the predicted value of the alighting person count from the number of passengers, and it is possible to calculate the number of persons that can get on a train by subtracting the passenger count after alighting from the capacity of the train. As a result, it is possible to visualize and predict more precisely the number of persons staying on the platform.
As above, the description of the processing flow of the alighting person count prediction process has ended.
Subsequently, an example of a processing flow of each unit in the present embodiment will be described.
The measurement unit 100, the recording unit 300, and the output unit 400 are realized using a known sensing technique, a known database technique, and a known data transfer technique, respectively, and thus, the description of the processing flows thereof will be omitted.
An example of the processing flow of the alighting person count calculation unit 201 will be described with reference to FIGS. 10 and 11.
First, an outline of processing of the alighting person count calculation unit 201 will be described with reference to FIG. 10. Symbols 1001 to 1006 represent passer-by counts in the respective time zones by extracting the number of passers-by of target track and direction from the person count measurement information 301. Symbols 1011 and 1012 are train arrival times of the target track extracted from the departure/arrival time information 302. As illustrated in FIG. 10, the alighting person count calculation unit 201 divides the passer-by count by the train arrival time and allocates the passer-by count to the immediately previous train, thereby calculating the number of persons alighting from the immediately previous train. For example, the symbols 1001 to 1003 are the number of persons who has passed by stairs on the platform toward a ticket gate floor between arrival of a train 1011 until arrival of a train 1012, and can be estimated as the number of persons alighting from the train 1011.
Subsequently, an example of the processing flow of the alighting person count calculation unit 201 will be described using a flowchart of FIG. 11.

S5001: The person count measurement information 301 of a track ID of a target track for calculating an alighting person count and a direction ID of a direction of movement from the platform to the ticket gate floor is extracted from the recording unit 300.
S5002: The departure/arrival time information 302 of the track ID of the target track for calculating the alighting person count is extracted from the recording unit 300.
S5003: The extracted person count measurement information is divided by the train arrival time out of the extracted departure/arrival time information.
S5004: A total value of the number of persons in the person count measurement information divided by the train arrival time is taken as the number of persons alighting from a train which has immediately previously arrived.
S5005: The alighting-person count is output as the alighting person count information 303 and recorded in the recording unit 300.

Next, an example of the processing flow of the train interval calculation unit 202 will be described using a flowchart of FIG. 12.

S5101: When the train departure/arrival detection unit 102 detects arrival of a train, the departure/arrival time information 302 at arrival of a train which has immediately previously arrived on the same track as the train is extracted from the recording unit.
S5102: A difference between a train arrival time of the extracted departure/arrival time information 302 and an arrival time of the train is calculated as a train interval of the train.
S5103: The train interval is output as the train interval information 304 and recorded in the recording unit 300.

Subsequently, an example of the processing flow of the prediction model creation unit 203 will be described with reference to FIGS. 13 and 14.
First, an outline of processing of the prediction model creation unit 203 will be described with reference to FIG. 13. FIG. 13 is a scatter diagram in which the horizontal axis represents a train interval and the vertical axis represents a train alighting person count. Each of ranges 1101 to 1103 represents dispersion of data distinguished by an attribute, a track, a time zone, and the like. Each of curves 1111 to 1113 is a relational expression of a train interval and a train alighting person count corresponding to the data of the ranges 1101 to 1103. For example, the relational expression is calculated by regression analysis using the train interval as an explanatory variable and the train alighting person count as an objective variable. The prediction model creation unit 203 creates a relational expression for each condition as a prediction model and outputs the created model as the prediction model information 305.
Subsequently, an example of the processing flow of the prediction model creation unit 203 will be described using a flowchart of FIG. 14.
S5201: The train interval information 304 and the alighting person count information 303 are associated depending on the date, the track ID, and the arrival time.
S5202: The associated data is classified and distinguished in accordance with the condition such as the date, the track ID, the time zone of the arrival time, and the like.
S5203: A relational expression between a train interval and a train alighting person count is calculated for data of each condition and set as a model formula. For example, the relational expression is calculated by regression analysis using the train interval as an explanatory variable and the train alighting person count as an objective variable.
S5204: The model formula is output as the prediction model information 305 and saved in the recording unit 300.
A method for creating the prediction model in the prediction model creation unit 203 is not limited to the above-described one. For example, the model formula may be created using a relational expression between a delay time and a train alighting person count using each mode or average value of train intervals and train alighting person counts aggregated for each condition as standard train interval and train alighting person count under the condition and using a difference between the input train interval and the standard train interval as the delay time. In addition, the model formula may be created using a relational expression between a delay rate, which is a ratio of the delay time relative to a standard delay time, and an alighting person count change rate which is a ratio of the train alighting person count relative to the standard train alighting person count as illustrated in FIG. 15 With such normalization as the ratios, it is not always necessary to create the model formula for every condition (time zone). In addition, it is possible to use a model formula of a station for another station where statistical information is not sufficient.
Next, an example of the processing flow of the alighting person count prediction unit 204 will be described using a flowchart of FIG. 16.

S5301: The alighting person count prediction unit receives an input of the train interval calculated by the train interval calculation unit 202 at arrival of a train.
S5302: The prediction model information 305 conforming to conditions of the train is extracted from the recording unit 300.
S5303: A train alighting person count is predicted by substituting the input train interval to the extracted model formula.
S5304: The predicted train alighting person count is output.

The processing of the alighting person count prediction unit 204 is changed depending on a method of creating a prediction model. For example, when a prediction model is created using a relational expression between a delay rate and a train alighting person count increase rate, the delay rate is calculated from a train interval, a change rate of a train alighting person count is calculated using the relational expression, and the change rate of the train alighting person count and a standard train alighting person count are multiplied to obtain the train alighting person count.

With the train alighting person count prediction device of the present embodiment, it is possible to calculate the train interval from the train arrival time at a stage of detecting the arrival of the train using only the information obtained from the single station and statistically predict the number of persons alighting from the train which has arrived based on the train interval. As a result, it is possible to visualize and predict the congestion situation in the station including train alighting customers in real time with only single station information by inputting the alighting person count to a known pedestrian simulator device at the time of arrival of the train. It is also possible to realize grasp of the congestion situation only with the information obtained from the single station.
Since it is possible to timely predict the train alighting person count using the information that can be acquired at the single station, it is possible to timely grasp the congestion situation in the station at low cost. Incidentally, "timely" refers to an arrival stage of a train before the alighting customers actually starts alighting.

Reference Signs List

100: measurement unit
200: arithmetic unit
201: alighting person count calculation unit
202: train interval calculation unit
203: prediction model creation unit
204: alighting person count prediction unit
300: recording unit
301: person count measurement information
302: departure/arrival time information
303: alighting person count information
304: train interval information
305: prediction model information
400: output unit
701: camera
702, 703: camera
711: point
712: point

Claims

A train alighting person count prediction system comprising:
an alighting person count calculation unit (201) that measures or estimates the number of persons alighting from a train which has arrived;

a train departure/arrival detection unit (202) that detects arrival of a train;

a train interval calculation unit that calculates a train interval which is an interval between arrival times of a first train and a second train;

a recording unit (300), in which a plurality of numbers of alighting persons measured or estimated by the alighting person count calculation unit are recorded in association with respective train intervals; and

an arithmetic unit (200) which creates a prediction model from statistical information of the numbers of alighting persons recorded in the recording unit

wherein the alighting person count calculation unit is configured to predict the number of persons alighting from a train arriving in the future from the previously measured or estimated numbers of alighting persons, the calculated train interval, and the prediction model.
The train alighting person count prediction system according to claim 1, wherein
the arithmetic unit is configured to create the prediction model using a delay time which is a difference from a past standard train interval, or a delay rate which is a ratio of the delay time relative to the standard train interval.
A congestion visualization and evaluation system comprising:
the train alighting person count prediction system according to claim 1; and

a pedestrian simulator that is configured to receive an input of the number of alighting persons predicted by the train alighting person count prediction system and to estimate a congestion situation of a space by simulating movement of pedestrians.
An acceptable passenger count calculation system comprising:
the train alighting person count prediction system according to claim 1; and

means for measuring or estimating a passenger count in a train,

wherein the acceptable passenger count calculation system is configured to calculate number of persons that can get on a train by subtracting the number of passengers after alighting, calculated by subtracting the number of alighting persons output by train alighting person count prediction system from the passenger count, from a capacity of the train.