CN110516866B - Real-time estimation method for urban rail transit train crowding degree - Google Patents

Real-time estimation method for urban rail transit train crowding degree Download PDF

Info

Publication number
CN110516866B
CN110516866B CN201910774323.9A CN201910774323A CN110516866B CN 110516866 B CN110516866 B CN 110516866B CN 201910774323 A CN201910774323 A CN 201910774323A CN 110516866 B CN110516866 B CN 110516866B
Authority
CN
China
Prior art keywords
passenger flow
time
train
current
characteristic day
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Active
Application number
CN201910774323.9A
Other languages
Chinese (zh)
Other versions
CN110516866A (en
Inventor
胡华
赵源
邓紫欢
刘志钢
郝妍熙
刘秀莲
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Shanghai University of Engineering Science
Original Assignee
Shanghai University of Engineering Science
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Shanghai University of Engineering Science filed Critical Shanghai University of Engineering Science
Priority to CN201910774323.9A priority Critical patent/CN110516866B/en
Publication of CN110516866A publication Critical patent/CN110516866A/en
Application granted granted Critical
Publication of CN110516866B publication Critical patent/CN110516866B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F17/00—Digital computing or data processing equipment or methods, specially adapted for specific functions
    • G06F17/10—Complex mathematical operations
    • G06F17/18—Complex mathematical operations for evaluating statistical data, e.g. average values, frequency distributions, probability functions, regression analysis
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/04—Architecture, e.g. interconnection topology
    • G06N3/045—Combinations of networks
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/08—Learning methods
    • G06N3/084—Backpropagation, e.g. using gradient descent
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q10/00—Administration; Management
    • G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
    • G06Q50/40—
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V20/00—Scenes; Scene-specific elements
    • G06V20/50—Context or environment of the image
    • G06V20/52—Surveillance or monitoring of activities, e.g. for recognising suspicious objects
    • G06V20/53—Recognition of crowd images, e.g. recognition of crowd congestion
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04W—WIRELESS COMMUNICATION NETWORKS
    • H04W4/00—Services specially adapted for wireless communication networks; Facilities therefor
    • H04W4/30—Services specially adapted for particular environments, situations or purposes
    • H04W4/40—Services specially adapted for particular environments, situations or purposes for vehicles, e.g. vehicle-to-pedestrians [V2P]
    • H04W4/42—Services specially adapted for particular environments, situations or purposes for vehicles, e.g. vehicle-to-pedestrians [V2P] for mass transport vehicles, e.g. buses, trains or aircraft

Abstract

The invention belongs to the technical field of urban rail transit intelligent management, and discloses a real-time estimation method for the congestion degree of an urban rail transit train, which is used for predicting the congestion degree of different running sections of the same train on a line, wherein a time period between T1 time after the starting time when a certain number of trains pass through a certain running section and T2 time before the ending time is taken as a statistical period, and a certain peak time period of the current characteristic day is divided into N statistical periods; calculating WiFi passenger flow and video passenger flow of the whole vehicle in each statistical period on the current characteristic day, wherein the last characteristic day corresponds to AFC section passenger flow in the statistical period; and step three, utilizing a neural network, taking WiFi passenger flow and video passenger flow in the current statistical period of the current characteristic day, and AFC section passenger flow in the statistical period corresponding to the last characteristic day as input, and estimating the actual passenger flow of the whole train in the current statistical period of the current characteristic day in real time, thereby finishing the real-time estimation of the train congestion degree.

Description

Real-time estimation method for urban rail transit train crowding degree
Technical Field
The invention belongs to the technical field of urban rail transit intelligent management, and particularly relates to a real-time estimation method for the congestion degree of an urban rail transit train.
Background
Along with the continuous development of national economy of China, the living standard of urban residents is continuously improved, urban traffic which is one of four functions of a city is more and more emphasized by people, wherein trains are favored by urban residents due to the characteristics of safety, convenience and punctuality, but the local crowding phenomenon can occur to passenger flow on subway trains due to the fact that the passenger flow has the distribution characteristics of time and space, and therefore the problems that passengers are difficult to get on the train, the driving efficiency is low, the passenger traffic risk is large and the like are caused. Therefore, the method can accurately judge the train crowding degree in real time, and has important significance for adjusting the subway train passenger transportation scheme, ensuring the subway train running safety, improving the train operation energy utilization rate and improving the train service level.
In recent years, the informatization and intelligentization construction of trains is greatly promoted in all cities, and automatic passenger flow monitoring technologies such as AFC, video, WiFi and Bluetooth are continuously put into use on subway trains in order to obtain the real-time congestion degree of the subway trains. Through investigation and analysis, the existing train passenger flow detection technology and the defects thereof are as follows:
(1) the intelligent video analysis technology comprises the following steps: shaking is generated in the running process of the train to influence the video identification precision; and the video monitoring range is repeated or has blind areas, so that the video monitoring range is not suitable for passenger flow identification or estimation in the whole train global range;
(2) WiFi probe technology: the sampling rate is high, the duplicate removal is easy through the unique MAC address, and the method is suitable for estimating the passenger flow of the whole train under a certain sampling rate; but the WiFi sampling rate may fluctuate under the condition of high passenger flow density;
(3) AFC passenger flow collection technology: the passenger AFC station entrance and exit card swiping or two-dimensional code data is utilized, 15min (minimum time granularity, 1 hour and the like) section passenger flow is obtained through a clearing model, the passenger flow belongs to full-sample data, and the sampling data can be supplemented; however, real-time cross section passenger flow data cannot be obtained, and passenger flow data of a certain train cannot be accurately obtained, and statistical processing is required;
(4) bluetooth positioning technology: the sampling rate is extremely low, the requirement of the lowest sampling rate of passenger flow statistical analysis cannot be met, and the method cannot be applied to real-time passenger flow judgment of trains at present.
Disclosure of Invention
The invention provides a real-time estimation method for the congestion degree of an urban rail transit train, and solves the problems of poor effectiveness, low accuracy and the like of the conventional congestion degree calculation method.
The invention can be realized by the following technical scheme:
a real-time estimation method for the congestion degree of an urban rail transit train is used for predicting the congestion degree of different running sections of the same train passing through a line and comprises the following steps:
step one, taking a time period between T1 time after the train in a certain operation direction of a certain number line passes through the starting time of a certain operation section and T2 time before the ending time as a statistical period, and dividing a certain peak time period of a current characteristic day p into N statistical periods;
step two, calculating WiFi passenger flow of the whole vehicle in each statistical period of the current characteristic day p
Figure GDA0002326666260000021
Video passenger flow volume
Figure GDA0002326666260000022
AFC section passenger flow in statistical period corresponding to last characteristic day p
Figure GDA0002326666260000023
Step three, utilizing the neural network to count WiFi passenger flow in the period at present according to the present characteristic day p
Figure GDA0002326666260000024
Video passenger flow volume
Figure GDA0002326666260000025
AFC section passenger flow in statistical period corresponding to last characteristic day p
Figure GDA0002326666260000026
As input, the actual passenger flow Y of the whole vehicle in the current statistical period of the current characteristic day p is estimated in real time(p)So as to complete the column in the current statistical period for the current characteristic day pEstimating the degree of congestion of the vehicle in real time;
and step four, repeating the step two to the step three to finish the real-time estimation of the train crowding degree in N statistical cycles at a certain peak period.
Further, each carriage on the train is divided into a plurality of grids, a WIFI probe is arranged at the center of each grid, the MAC address of the passenger mobile terminal is used as a detection object, and the WiFi passenger flow of the whole train in the current characteristic day p within the current statistical period is calculated
Figure GDA0002326666260000027
Further, screening the detection data of all WIFI probes in each carriage, removing MAC address data with detection frequency lower than twice, reserving only one same MAC address data, and summing the number of the residual MAC addresses in all carriages to serve as the WiFi passenger flow of the whole vehicle in the current statistical period of the current characteristic day p
Figure GDA0002326666260000031
Further, the time period corresponding to the T1 time is smaller than the time period corresponding to the T2 time.
Further, the following equation is utilized to calculate AFC section passenger flow of the whole train in a statistical period corresponding to the last characteristic day p
Figure GDA0002326666260000032
Figure GDA0002326666260000033
Further, one or more cameras are arranged in each carriage and used for collecting passenger videos in the carriages, the shooting ranges of the cameras in the adjacent carriages are not overlapped, the passenger flow in each carriage is calculated by utilizing a video people counting algorithm combining a convolutional neural network and ridge regression, and the sum of the passenger flow in each carriage is taken as the video passenger flow in the current characteristic day p in the current counting period
Figure GDA0002326666260000034
Further, the following equation is used for calculating the train congestion degree D in the current statistical period of the current characteristic day p
Figure GDA0002326666260000035
The beneficial technical effects of the invention are as follows:
the real-time estimation method integrates WiFi probe detection, video people number analysis and AFC sorting data system to carry out fusion estimation on the passenger flow, reduces the limitation of a single detection method as much as possible, enlarges the data acquisition range, realizes advantage complementation, improves the passenger flow detection precision, further improves the real-time estimation precision of the train crowding, and simultaneously saves a large number of manpower and material resources, the method provides a data basis for platform staff to relieve peak congestion, provides a parameter basis for analyzing traffic characteristics of passengers in stations in peak hours, simulating and predicting space-time distribution of the passengers in the stations, optimizing a train station passenger transportation organization scheme, starting a station large passenger flow plan and the like, and has important effects on improving the management level of the train station large passenger flow and guaranteeing the travel safety of the passengers.
Drawings
FIG. 1 is a schematic overview of the process of the present invention;
fig. 2 is a schematic diagram of WIFI probe distribution at the gateway of the present invention;
FIG. 3 is a schematic flow chart of a video people counting algorithm of the present invention;
FIG. 4 is a schematic diagram of the structure of the neural network of the present invention;
FIG. 5 is a data diagram of a training sample of the present invention;
fig. 6 is a schematic diagram of the comparison of the predicted passenger flow volume and the actual passenger flow volume using the method of the present invention.
Detailed Description
The following detailed description of the preferred embodiments will be made with reference to the accompanying drawings.
As shown in figure 1, the invention provides a real-time estimation method for the congestion degree of an urban rail transit train, which is used for predicting the congestion degree of different running sections of the same train on a line, takes the time period of a certain number of trains passing through the running sections as a statistical period, and calculates the WiFi passenger flow of the whole train in the current statistical period on the current characteristic day p
Figure GDA0002326666260000041
Video passenger flow volume
Figure GDA0002326666260000042
AFC section passenger flow in statistical period corresponding to last characteristic day p
Figure GDA0002326666260000043
And the actual passenger flow Y of the whole vehicle in the current statistical period of the current characteristic day p is estimated in real time by taking the estimated passenger flow as the input of the neural network(p)Therefore, the real-time estimation of the train congestion degree in the current statistical period of the current characteristic day p is completed, and when the train congestion degree condition of a certain operation section in the current statistical period is known, a worker can properly plan the passenger flow of the next operation section or the arrival condition of the train, so that the congestion condition is relieved as much as possible, and the intelligent level of urban rail transit management is improved.
The method specifically comprises the following steps:
step one, taking a time period between T1 after the starting time of a certain number of line trains passing through a certain operation section and T2 before the ending time as a statistical period, and dividing a certain peak time period of a current characteristic day p into N statistical periods. In order to avoid the influence on the data acquisition of the WiFi probe and the camera on the train under the conditions of stop, deceleration at the arrival station and acceleration at the departure station of the train, the time from the beginning to the end of the train during the train passing through a certain operation section is selected as a statistical period, the train does not get on or off the train during the period, the passenger flow is stable and unchanged, the data acquisition of the camera and the WiFi probe is facilitated, a good environment is provided for the accuracy of data acquisition, the time period corresponding to the time T1 is less than the time period corresponding to the time T2, more time is left for data processing and transmission as far as possible, the real-time estimation is completed within the time when the train passes through the whole operation section, and a basis is provided for improving the congestion degree of the subsequent operation section for workers earlier.
Step two, calculating WiFi passenger flow of the whole vehicle in each statistical period of the current characteristic day p
Figure GDA0002326666260000051
Video passenger flow volume
Figure GDA0002326666260000052
AFC section passenger flow in statistical period corresponding to last characteristic day p
Figure GDA0002326666260000053
For WiFi passenger flow
Figure GDA0002326666260000054
Dividing each carriage on the train into a plurality of grids, arranging a WIFI probe at the center of each grid, and calculating the WiFi passenger flow of the whole train in the current statistical period of the current characteristic day p by taking the MAC address corresponding to the passenger as a detection object as shown in figure 2
Figure GDA0002326666260000055
The number of grids can be determined according to the actual performance of the WIFI probe and the actual volume of the train, and the detection range of the WIFI probe can cover the whole train.
In order to improve the accuracy of subsequent calculation, the detection data of all WIFI probes in each carriage needs to be screened.
(1) In order to avoid the influence of WiFi equipment data in the ground and non-trains on the overhead line, MAC addresses with the frequency of being detected less than twice in a statistical period are screened out;
(2) data deduplication: and (4) screening out the data repeatedly collected in the statistical period, namely only one MAC address is reserved for the same MAC address.
And finally, calculating the sum of the number of the residual MAC addresses in all the carriages as the WiFi passenger flow of the whole train in the current statistical period of the current characteristic day p
Figure GDA0002326666260000056
For video passenger flow
Figure GDA0002326666260000057
The camera is arranged in each carriage and used for collecting passenger videos in the carriages, and shooting ranges of the cameras in the adjacent carriages are not overlapped. Aiming at a certain frame of video image in the statistical period, a video people counting algorithm combining a convolutional neural network and ridge regression is utilized, as shown in fig. 3, the central point of a head in the frame of video image is regressed through the convolutional neural network to obtain a crowd density distribution characteristic diagram, then the crowd density distribution characteristic diagram is analyzed by using a ridge regression model to obtain the number of people corresponding to the frame of video image, finally, a selected camera in each carriage of the train is taken to analyze to obtain the sum of the number of the heads, namely the sum of the number of the video passengers in the current statistical period of the current characteristic day p
Figure GDA00023266662600000611
Passenger flow for AFC section
Figure GDA0002326666260000061
Because AFC clearing data has time delay, the data of the whole network entering and leaving station can be uniformly cleared in the next morning, namely, only the statistical time interval corresponding to the historical characteristic day can be obtained through the AFC clearing dataThe average value of the passenger flow of the AFC clearing section on the corresponding operation section is used as the passenger flow of the AFC section, and the passenger flow of the section is predicted by the average value of the passenger flow of the AFC clearing section on the corresponding operation section
Figure GDA0002326666260000062
The calculation formula is as follows:
Figure GDA0002326666260000063
step three, utilizing the neural network, as shown in fig. 4, carrying out WiFi passenger flow in the current statistical period according to the current characteristic day p
Figure GDA0002326666260000064
Video passenger flow volume
Figure GDA0002326666260000065
AFC section passenger flow in statistical period corresponding to last characteristic day p
Figure GDA0002326666260000066
As input, the actual passenger flow Y of the train in the current statistical period of the current characteristic day p is estimated in real time(p)Therefore, the real-time estimation of the train crowding degree in the current statistical period of the current characteristic day p is completed.
Before prediction, the neural network needs to be trained, training data of the neural network is obtained by the calculation method, and WiFi passenger flow passing through different running sections of the same train in corresponding time periods of a predicted statistical cycle, such as peak time periods and off-peak time periods
Figure GDA0002326666260000067
Video passenger flow volume
Figure GDA0002326666260000068
AFC section passenger flow volume
Figure GDA0002326666260000069
As input, counting with artificial videoThe number of passengers getting on or off each door of each station is counted by video from the initial station of the train in a manual counting mode, and the actual passenger flow Y of the train passing each running section is calculated according to the following formulatAs an output, the neural network is trained. In order to ensure the prediction accuracy of the neural network, the number of samples for training data is not less than 120, and if the data in the current time period is insufficient, the data in the same time period of a characteristic day on the same train or a characteristic day corresponding to the current time period can be taken.
Yt∑ get on the bus- ∑ get off the bus
Calculating the train congestion degree D in the current statistical period of the current characteristic day p by using the following equation
Figure GDA00023266662600000610
The invention regards the train as serious congestion when the degree of congestion is more than 100% and less than or equal to 130%; when D is more than 80% and less than or equal to 100%, the congestion is considered; when D is more than 60% and less than or equal to 80%, the general crowding is considered; when D is less than or equal to 60%, it is considered not to be congested.
And step four, repeating the step two to the step three to complete the real-time estimation of the train crowding degree in N statistical cycles at a certain peak period.
The method of the present invention is described in detail by taking the train of train No. 9 in the sea subway as an example.
The first step is as follows: analyzing the time distribution characteristics of the passenger flow data of the Shanghai subway number 9 linear train, and dividing the prediction characteristic day and the prediction time interval into a working day peak time interval, a working day off-peak time interval and a non-working day time interval. Taking the predicted characteristic day as 12 months in 2018 and 28 days in friday, and taking the predicted characteristic time as 7 in the early peak time 7:00-9: 00: 00-7:02, at this time, train XXX passes through city station and tunnel jing station of Yangtze university at the running section.
The second step is that: the existing peanut WiFi probe on each carriage of the 9-line train is used for acquiring data, and a camera at a fixed congestion position is selected for video data analysis in each carriage of the train, so that a non-overlapping monitoring area is ensured.
The third step: according to train operation ATS data including the time from the 9 # line train to the departure time, a time period between 30s after the departure time and 50s before the end time of the 9 # line train on an operation section between a Yangtze university city station and a tunnel jing station is selected as the beginning and end time of a statistical period, namely the time period is used as a sample data statistical period T of WiFi probes, videos and AFC data. In this example, T is 1 min.
The fourth step: taking the number n of samples as 120, and screening the data acquired by the WiFi probe according to the following rules according to the statistical period of the data of each sample so as to obtain an effective WiFi probe data set:
(1) in order to avoid the influence of WiFi equipment data in non-trains on the ground and an overhead line, MAC addresses with detection frequency lower than twice in a sample data statistical period are screened out;
(2) and (5) data deduplication. And (4) screening out the data repeatedly collected in the statistical period, namely only one MAC address is reserved for the same MAC address.
According to the uniqueness of the MAC address of the equipment, the WiFi passenger flow of the train with the number of 9 and the number of XXX passing through different running sections in 2018, namely 12 months, 7 days, 14 days, 21 days (friday) 7:00-9:00
Figure GDA0002326666260000071
The collected partial data is shown in fig. 5, and the number of samples is taken as 120.
The fifth step: selecting a certain frame of a camera video in each carriage in a sample statistical period in the process that the train passes through a running section between a city station and a Jing station of the Yangtze university according to each sample data statistical period, utilizing a video people counting method combining a convolutional neural network and ridge regression, namely, regressing the center point of the head in a monitoring video image through the convolutional neural network regression train to obtain a crowd density distribution characteristic diagram, then analyzing the crowd density distribution characteristic diagram by using a ridge regression model to obtain the number of people corresponding to the frame video image, thereby obtaining the number of people of the 9-number line train with the number XXX, wherein the number of people is 12 months, 7 days, 14 days and 21 days (Friday) 7:00-9:00 in 2018, and the average value of the number of the heads, namely the video passenger flow volume is obtained by analyzing the selected cameras of each carriage and different running sections
Figure GDA0002326666260000081
The partial data collected are shown in fig. 5.
And a sixth step: according to the statistical period of each sample data, extracting and calculating the average value of the number of passengers of an AFC clearing train passing through different running sections by a train with the number XXX in 2018, wherein the number of the passengers of the AFC clearing train is the same between 11 and 30 days in 2018, 12 and 14 days in 12 and 7 days in 12 (friday) in 15min between 7:00 and 9:00, and the average value is used as the passenger flow of the AFC section
Figure GDA0002326666260000082
The partial data collected are shown in fig. 5. The calculation formula of AFC section passenger flow is as follows:
Figure GDA0002326666260000083
the seventh step: taking the number n of samples as 120, combining with train operation ATS data, manually arranging trains from a train starting station through video frequency 9 line for 12 months, 7 days, 14 days and 21 days (Friday) 7:00-9:00 of passengers getting on or off each station and each door in 2018, and calculating the passenger flow Y of the actual operation section train with XXX train passing through different operation sections according to the following formulatThe collected partial data are shown in FIG. 5:
Yt∑ get on the bus- ∑ get off the bus
Eighth step: WiFi passenger flow volume for establishing Shanghai subway number 9 line by adopting BP neural network
Figure GDA0002326666260000084
Video passenger flow volume
Figure GDA0002326666260000085
AFC section passenger flow volume
Figure GDA0002326666260000086
To the actual running section train passenger flow YtRefer to fig. 4. And learning and training the BP neural network by using 120 groups of collected sample data to establish X meeting the training precision requirementt→YtRefer to fig. 4.
The ninth step: inputting WiFi passenger flow of a running section between 12 months and 28 days (Friday) in 2018, 7:00-7:02 days (Friday) in 28 months in 2018 and between 30s after the initial time and 50s before the final time of a train passing through a city station of Yangtze university and a tunnel jing station
Figure GDA0002326666260000087
Video passenger flow volume
Figure GDA0002326666260000088
AFC section passenger flow volume
Figure GDA0002326666260000089
Predicting to obtain the actual number Y of passengers on the section of the train in the current statistical period of 7:00-7:02p. Fig. 6 shows a comparison graph of the predicted number of passengers and the actual number of passengers in the predicted characteristic day and the predicted characteristic time zone for the shanghai subway 9 train. As can be seen from the graph, the coincidence degree of the predicted value curve and the actual value curve is high, the maximum absolute deviation value of the predicted number of the train passengers and the actual number of the train passengers is 178, and the average absolute relative error rate is 5.1%.
The tenth step: and converting the number of passengers on the actual train section predicted by the prediction model into the congestion degree D of the actual train section by using the following formula:
Figure GDA0002326666260000091
judging the grade of the crowdedness of the train section according to the following method:
when the congestion degree of the section of the actual train is more than 100% and less than or equal to 130%, the train is regarded as serious congestion; when D is more than 80% and less than or equal to 100%, the congestion is considered; when D is more than 60% and less than or equal to 80%, the general crowding is considered; when D is less than or equal to 60%, it is considered not to be congested.
Although particular embodiments of the present invention have been described above, it will be understood by those skilled in the art that these are by way of example only and that various changes or modifications may be made to these embodiments without departing from the spirit and scope of the invention and, therefore, the scope of the invention is to be defined by the appended claims.

Claims (4)

1. A real-time estimation method for the congestion degree of an urban rail transit train is characterized by being used for predicting the congestion degree of different running sections of the same train on a line, and comprising the following steps of:
step one, taking a time period between T1 time after the train in a certain operation direction of a certain number line passes through the starting time of a certain operation section and T2 time before the ending time as a statistical period, and dividing a certain peak time period of a current characteristic day p into N statistical periods;
step two, calculating WiFi passenger flow of the whole vehicle in each statistical period of the current characteristic day p
Figure FDA0002445786430000011
Video passenger flow volume
Figure FDA0002445786430000012
AFC section passenger flow in statistical period corresponding to last characteristic day p
Figure FDA0002445786430000013
Step three, utilizing the neural network to count WiFi passenger flow in the period at present according to the present characteristic day p
Figure FDA0002445786430000014
Video passenger flow volume
Figure FDA0002445786430000015
AFC section passenger flow in statistical period corresponding to last characteristic day p
Figure FDA0002445786430000016
As input, the actual passenger flow Y of the whole vehicle in the current statistical period of the current characteristic day p is estimated in real time(p)Thereby completing the train counting for the current characteristic day p in the current counting periodEstimating the degree of congestion in real time;
step four, repeating the step two to the step three to complete the real-time estimation of the train crowding degree in N statistical cycles at a certain peak period;
dividing each carriage on the train into a plurality of grids, arranging a WIFI probe at the center of each grid, taking the MAC address of a passenger mobile terminal as a detection object, and calculating the WiFi passenger flow of the whole train in the current statistical period of the current characteristic day p
Figure FDA0002445786430000017
Calculating AFC section passenger flow of the whole train in a statistical period corresponding to the last characteristic day p by using the following equation
Figure FDA0002445786430000018
Figure FDA0002445786430000019
One or more cameras are arranged in each carriage and used for collecting passenger videos in the carriages, the shooting ranges of the cameras in the adjacent carriages are not overlapped, the passenger flow in each carriage is calculated by utilizing a video people counting algorithm combining a convolutional neural network and ridge regression, and the sum of the passenger flow is taken as the video passenger flow in the current characteristic day p current counting period
Figure FDA00024457864300000110
2. The real-time estimation method for the degree of congestion of urban rail transit train according to claim 1, characterized in that: screening the detection data of all WIFI probes in each carriage, removing MAC address data with detection frequency lower than twice, reserving only one same MAC address data, and summing the number of the residual MAC addresses in all carriages to obtain the WiFi passenger flow of the whole vehicle in the current characteristic day p in the current statistical period
Figure FDA0002445786430000021
3. The real-time estimation method for the degree of congestion of urban rail transit train according to claim 1, characterized in that: the time period corresponding to the T1 time is smaller than the time period corresponding to the T2 time.
4. The real-time estimation method for the degree of congestion of urban rail transit train according to claim 1, characterized in that: calculating the train congestion degree D in the current statistical period of the current characteristic day p by using the following equation
Figure FDA0002445786430000022
CN201910774323.9A 2019-08-21 2019-08-21 Real-time estimation method for urban rail transit train crowding degree Active CN110516866B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201910774323.9A CN110516866B (en) 2019-08-21 2019-08-21 Real-time estimation method for urban rail transit train crowding degree

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201910774323.9A CN110516866B (en) 2019-08-21 2019-08-21 Real-time estimation method for urban rail transit train crowding degree

Publications (2)

Publication Number Publication Date
CN110516866A CN110516866A (en) 2019-11-29
CN110516866B true CN110516866B (en) 2020-06-26

Family

ID=68626105

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201910774323.9A Active CN110516866B (en) 2019-08-21 2019-08-21 Real-time estimation method for urban rail transit train crowding degree

Country Status (1)

Country Link
CN (1) CN110516866B (en)

Families Citing this family (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN111372196A (en) * 2020-02-20 2020-07-03 杭州海康威视系统技术有限公司 Data processing method and device, electronic equipment and machine-readable storage medium
CN112215409B (en) * 2020-09-24 2024-01-30 交控科技股份有限公司 Rail transit station passenger flow prediction method and system
CN112231640A (en) * 2020-10-14 2021-01-15 中国民航信息网络股份有限公司 Civil aviation passenger quantity prediction method, device and equipment and readable storage medium
CN112766950A (en) * 2020-12-31 2021-05-07 广州广电运通智能科技有限公司 Dynamic path cost determination method, device, equipment and medium
CN113068215B (en) * 2021-03-12 2022-01-11 华扬联众数字技术(深圳)有限公司 Weight and filtering-based multi-WiFi probe MAC address dynamic line algorithm
CN114785943B (en) * 2022-03-31 2024-03-05 联想(北京)有限公司 Data determination method, device and computer readable storage medium

Family Cites Families (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20060198302A1 (en) * 2005-03-03 2006-09-07 Sofman Lev B Traffic dimensioning in a metro area with IPTV architecture
CN105279572B (en) * 2015-09-16 2017-02-08 北京城建设计发展集团股份有限公司 City track traffic passenger flow density index calculating and releasing system
CN107360590A (en) * 2017-06-29 2017-11-17 上海工程技术大学 Track station part congestion points passenger flow condition judgement method

Also Published As

Publication number Publication date
CN110516866A (en) 2019-11-29

Similar Documents

Publication Publication Date Title
CN110516866B (en) Real-time estimation method for urban rail transit train crowding degree
CN110493816B (en) Real-time prediction method for rail transit subway station passenger flow volume
CN109871876B (en) Expressway road condition identification and prediction method based on floating car data
CN108346292B (en) Urban expressway real-time traffic index calculation method based on checkpoint data
CN111724595B (en) Highway section flow estimation method based on charging data
CN108322891B (en) Traffic area congestion identification method based on user mobile phone signaling
CN111210612B (en) Method for extracting bus route track based on bus GPS data and station information
WO2019047905A1 (en) Road traffic analysis system, method and apparatus
CN107958031B (en) Resident travel OD distribution extraction method based on fusion data
CN112512020B (en) Traffic state weak signal perception studying and judging method based on multi-source data fusion
CN112288131B (en) Bus stop optimization method, electronic device and computer-readable storage medium
CN103984994B (en) Method for predicting urban rail transit passenger flow peak duration
CN108665178B (en) AFC-based passenger flow prediction method for stairs in subway station
CN110390349A (en) Bus passenger flow volume based on XGBoost model predicts modeling method
CN109255972B (en) Optimization method of ground public transport fixed line timetable based on big data
CN105390013A (en) Method for predicting bus arrival time based on bus IC card
CN102521965A (en) Effect evaluation method of traffic demand management measures based on identification data of license plates
CN111144281B (en) Urban rail transit OD passenger flow estimation method based on machine learning
CN106650976A (en) Travel analysis and forecasting method and system, and travel analysis and forecasting method and system based on IC card
CN111476494A (en) Method for accurately analyzing geographic distribution of public transport population based on multi-source data
CN113593220A (en) Road network bearing capacity estimation method based on macroscopic basic graph
CN114596700A (en) Real-time highway section flow estimation method based on portal data
CN110096804A (en) The layout method and system of Expressway Service based on mobile terminal data
CN115995149A (en) Multi-source data-based parking supply and demand characteristic dynamic evaluation method and system
CN112669595B (en) Network taxi booking flow prediction method based on deep learning

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
CB03 Change of inventor or designer information
CB03 Change of inventor or designer information

Inventor after: Hu Hua

Inventor after: Zhao Yuan

Inventor after: Deng Zihuan

Inventor after: Liu Zhigang

Inventor after: Hao Yanxi

Inventor after: Liu Xiulian

Inventor before: Hu Hua

Inventor before: Deng Zihuan

Inventor before: Liu Zhigang

Inventor before: Hao Yanxi

Inventor before: Liu Xiulian

GR01 Patent grant
GR01 Patent grant