CN114037158A - Passenger flow prediction method based on OD path and application method - Google Patents

Passenger flow prediction method based on OD path and application method Download PDF

Info

Publication number
CN114037158A
CN114037158A CN202111322651.9A CN202111322651A CN114037158A CN 114037158 A CN114037158 A CN 114037158A CN 202111322651 A CN202111322651 A CN 202111322651A CN 114037158 A CN114037158 A CN 114037158A
Authority
CN
China
Prior art keywords
time
passenger
data
passenger flow
station
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.)
Pending
Application number
CN202111322651.9A
Other languages
Chinese (zh)
Inventor
李洲
宋伟
刘瑜
逄淑荣
牛燕斌
王克非
张博
韩国奇
曹丽娟
范志峰
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.)
Beijing Bii Erg Transportation Technology Co ltd
Original Assignee
Beijing Bii Erg Transportation Technology Co ltd
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 Beijing Bii Erg Transportation Technology Co ltd filed Critical Beijing Bii Erg Transportation Technology Co ltd
Priority to CN202111322651.9A priority Critical patent/CN114037158A/en
Publication of CN114037158A publication Critical patent/CN114037158A/en
Pending legal-status Critical Current

Links

Images

Classifications

    • 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"
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
    • G06F16/24—Querying
    • G06F16/245—Query processing
    • G06F16/2458—Special types of queries, e.g. statistical queries, fuzzy queries or distributed queries
    • G06F16/2462—Approximate or statistical queries
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
    • G06F16/24—Querying
    • G06F16/245—Query processing
    • G06F16/2458—Special types of queries, e.g. statistical queries, fuzzy queries or distributed queries
    • G06F16/2465—Query processing support for facilitating data mining operations in structured databases
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
    • G06F16/24—Querying
    • G06F16/245—Query processing
    • G06F16/2458—Special types of queries, e.g. statistical queries, fuzzy queries or distributed queries
    • G06F16/2477—Temporal data queries
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F30/00—Computer-aided design [CAD]
    • G06F30/20—Design optimisation, verification or simulation
    • G06F30/27—Design optimisation, verification or simulation using machine learning, e.g. artificial intelligence, neural networks, support vector machines [SVM] or training a model
    • 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
    • G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
    • G06Q50/10—Services
    • G06Q50/26—Government or public services

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Business, Economics & Management (AREA)
  • General Physics & Mathematics (AREA)
  • Software Systems (AREA)
  • General Engineering & Computer Science (AREA)
  • Probability & Statistics with Applications (AREA)
  • Databases & Information Systems (AREA)
  • Computational Linguistics (AREA)
  • Data Mining & Analysis (AREA)
  • Tourism & Hospitality (AREA)
  • Economics (AREA)
  • Mathematical Physics (AREA)
  • Strategic Management (AREA)
  • Human Resources & Organizations (AREA)
  • Fuzzy Systems (AREA)
  • Marketing (AREA)
  • Evolutionary Computation (AREA)
  • Development Economics (AREA)
  • General Business, Economics & Management (AREA)
  • Primary Health Care (AREA)
  • Educational Administration (AREA)
  • Quality & Reliability (AREA)
  • Operations Research (AREA)
  • Entrepreneurship & Innovation (AREA)
  • Game Theory and Decision Science (AREA)
  • Health & Medical Sciences (AREA)
  • Artificial Intelligence (AREA)
  • Medical Informatics (AREA)
  • General Health & Medical Sciences (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Computer Hardware Design (AREA)
  • Geometry (AREA)
  • Train Traffic Observation, Control, And Security (AREA)

Abstract

The invention discloses a passenger flow prediction method based on an OD path, which comprises the following steps: acquiring sample data; inspecting and preprocessing sample data; and carrying out refined passenger flow prediction according to the preprocessed sample data, and predicting the outbound place and the outbound time of the passenger on the basis of the recent OD trip characteristics of different date types, different time periods and different time granularities and the real-time inbound card swiping data of the predicted current day. The invention also discloses an application method of the passenger flow prediction method based on the OD path. The invention solves the problems of low prediction efficiency and weak adaptability in the prior art.

Description

Passenger flow prediction method based on OD path and application method
Technical Field
The invention relates to the technical field of rail transit, in particular to a passenger flow prediction method based on an OD (station entrance to station exit) path and an application method.
Background
In the actual operation process, the track operation unit needs to obtain the real-time passenger flow data of each line to guide passenger flow control, driving organization and safety management. In recent years, with the rapid development of urban construction, individual differences become more obvious when passengers select a path, the change period of a path mode becomes shorter and shorter, and the research, development and perfection of a Dynamic Traffic Management System (DTMS) become a problem which needs to be solved urgently. Xiangming Yao of the Beijing university of transportation and transportation, Xinxinati university engineering, Hui Ren of the applied institute and the like propose a dynamic passenger flow distribution model in a transportation network running according to a planning diagram based on simulation, and AFC data of Beijing subway verifies that the simulation model can be used for judging the current network running state and predicting short-term future traffic tendency. Lijun Liu and the like of the editorial university of mansion adopt a prediction model of a deep learning method to carry out prediction modeling and verification on BRT station passenger flow. Florian torque et al propose using a Recurrent Neural Network (RNN) of a time Recurrent Neural network (LSTM) unit to predict a dynamic subway OD matrix, and compare the results with those of a conventional calendar model and a vector autoregressive model, which shows that the results are superior to those of the two methods. Haitao XU of computer academy of Hangzhou electronic science and technology university and the like propose a deep learning method based on a prediction stacked self-encoding (PSAE) model to predict passenger flow of a rapid bus station, a first layer adopts unsupervised learning training, the result is sent to the unsupervised learning of the next layer to train a self-encoding model, finally, a logistic regression prediction layer is constructed and trained, BP and GD are utilized to optimize the model, and the conclusion is that the prediction effect of the PSAE is superior to that of BPNN and SVM.
However, the prediction method in the prior art still has the following problems: time lag of OD (inbound outbound data) input; the path model and the parameter adjustment need time-consuming and labor-consuming manual following and are difficult to update in time; the passenger flow prediction is separately established for the normal state and the abnormal state and has no universality.
Disclosure of Invention
In order to solve the technical problems in the background art, the invention provides a passenger flow prediction method based on an OD path and an application method thereof, so as to solve the problems of low prediction efficiency and poor adaptability in the prior art.
The invention provides a passenger flow prediction method based on an OD path, which comprises the following steps:
101, acquiring sample data;
102, checking and preprocessing sample data;
and 103, carrying out refined passenger flow prediction according to the preprocessed sample data, and predicting the outbound place and the outbound time of the passenger on the basis of the recent OD trip characteristics of different date types, different time periods and different time granularities and the real-time inbound card swiping data on the predicted day.
The invention provides an application method of a passenger flow prediction method based on an OD path, which is characterized in that on the basis of the technical scheme, an early warning threshold value is set, and when the passenger flow prediction quantity exceeds the early warning threshold value, an alarm is given.
According to the invention, the historical travel characteristics of passengers in different time periods are considered, the historical travel track of a single passenger is fully analyzed, the maximum possible outbound place (namely D station) of the passenger is predicted in real time by combining real-time card swiping inbound data (namely O station), a complete OD path is formed, and other indexes such as inbound quantity, passenger capacity, transfer quantity, OD passenger flow quantity and section passenger flow quantity are carried out through a passenger flow clearing prediction model and a passenger flow simulation model; further, other passenger flow index data are considered, the subway real-time passenger flow conditions under the conditions of extreme weather, operation events and the like cannot be accurately mastered only by using historical passenger flow data as input data, and operation and scheduling under the condition of large passenger flow cannot be better guided; meanwhile, a machine learning algorithm is adopted to establish a model, self-learning and self-optimization capabilities of the machine learning algorithm are fully utilized, self-scheduling and self-training of the model are configured, data generated recently can be timely input into the model, the model is automatically trained periodically, continuous optimization of the model is achieved, and high prediction accuracy is kept.
Drawings
Fig. 1 is a flowchart of a passenger flow prediction method based on OD paths according to an embodiment of the present invention.
Detailed Description
The invention provides a passenger flow prediction method based on an OD path, which comprises the following steps as shown in figure 1:
step 101, obtaining sample data, including: transaction detail data for a successive year; swiping card inbound data in real time; external data such as holidays, weekdays, weekends, and the like; train plan operation chart, train actual operation chart, seal station and other operation information data.
Step 102, sample data is inspected and preprocessed. Quality inspection and pretreatment of sample data are important links in passenger flow prediction, and the passenger flow prediction accuracy is directly influenced. The method specifically comprises the following steps: and checking the integrity of the input sample data, namely whether missing values, abnormal values, extreme values, discrete values and the like exist, and if so, supplementing the input sample data by using the average value of the passenger flow data with the same or similar history and weather on the corresponding date.
103, carrying out refined passenger flow prediction according to the preprocessed sample data, wherein the passenger outbound prediction is based on OD trip characteristics of different date types, different time periods and different time granularities in the last two years, real-time inbound card swiping data (including inbound stations, inbound time and the like) on the prediction day and the like, and predicting the outbound place and the outbound time of the passenger; the date type: weekdays, weekends, holidays, and the like; time period: early departure, early peak, noon peak, late arrival, etc.; time granularity: 5 minutes, 10 minutes, 15 minutes, 30 minutes, etc.
Wherein, the refined passenger flow prediction specifically comprises the following steps:
step 1031, according to recent (for example, last two years) historical travel OD data, meteorological data, activity data, operation event data and the like, deeply analyzing the influence degree of external influence factors on subway OD passenger flows of each time period and each time granularity;
supposing that the passenger flow on a certain day is simultaneously influenced by severe weather, events and activities, taking the passenger flow in a time period under the condition of clear days without any influence factors as a reference value, and setting the passenger flow as xIs normalLet the passenger flow affected by severe weather be xWeather (weather)The passenger flow affected by the emergency operation event is xEvent(s)The passenger flow affected by a large activity is xMovement ofThen, the change rate of the passenger flow under the severe weather condition is:
Figure BDA0003346064570000041
the change rate of the passenger flow under the condition of the emergency operation event is as follows:
Figure BDA0003346064570000042
the rate of change of passenger flow under large activities is:
Figure BDA0003346064570000043
the sum of the influence degrees of all the influence events on the passenger flow is 1, and the influence degree of the weather on the passenger flow is beta within the time period tt, weatherThe degree of influence of passenger flow after an emergency operation event occurs is betat, eventThe degree of influence of passenger flow after large-scale activity is betat, activitiesThen there is
βt, weather+βt, event+βt, activities＝1
Figure BDA0003346064570000044
Figure BDA0003346064570000045
Figure BDA0003346064570000046
And establishing an influence factor weight matrix Q of passenger flows in different time periodst, factor
Figure BDA0003346064570000051
Wherein T is a time period, T is the number of the time periods, and T is more than or equal to 1 and less than or equal to T; taking 5 minutes as an example, dividing 24 hours into 288 segments according to 5 minutes, then T equals 288, and T is1In a time period of 00: 00-00: 05, t2The time period is 00: 05-00: 10, and so on.
Step 1032, deeply analyzing passengers entering each station in each time interval by using the historical 0D passenger flow data,
establishing an OD outbound distribution matrix Q according to the outbound distribution probability of each other station in the road networkt, outbound；
Setting the historical time-division OD passenger flow matrix as Qt，ODThen, then
Figure BDA0003346064570000052
Wherein n is the total number of all stations of the wire network, i is an inbound station, j is an outbound station, i is greater than or equal to 1 and less than or equal to n, j is greater than or equal to 1 and less than or equal to n, and t is a time period.
Let the number of passengers arriving from station i and departing from station j be Xi，jThen O isiDjThe proportion of the passenger data to the total number of the passengers entering the station i is alphai，jThen, then
Figure BDA0003346064570000053
Then the OD outbound distribution matrix
Figure BDA0003346064570000054
1033, deeply analyzing OD characteristic probability distribution of each passenger in each time period continuously in one year by using historical OD passenger flow data, and establishing a passenger travel OD probability distribution matrix Qt, passengers；
Setting the historical time-division OD passenger flow matrix as Qdate，t，ODThen, then
Figure BDA0003346064570000061
Wherein n is the total number of all stations of the wire network, i is an inbound station, j is an outbound station, i is more than or equal to 1 and less than or equal to n, and j is more than or equal to 1 and less than or equal to n; date is date code, and 1 ≦ date ≦ m, where m is the maximum value of date code of sample data, for example, if the sample data is one year, then m is 365; t is a time period.
Let the travel frequency of a single passenger on a certain OD in a certain period of the last year be Yi，jThen the passenger's trip probability on the OD in the period of the last year is γi，jThen, then
Figure BDA0003346064570000062
i. j is a fixed value
Then passenger trip OD probability distribution matrix
Figure BDA0003346064570000063
Step 1034, the OD is outbound distribution matrix Qt, outboundPassenger trip OD probability distribution matrix Qt, passengersInfluence factor weight matrix Q of passenger flows in different time periodst, factorHistorical passenger arrival card swiping data Qt, historical arrivalReal-time card swiping inbound data Qt, real timeEntering stationTime zone data Qt, time periodWeekday/weekend identification data QWeekday/weeklyHoliday data QHolidayAnd the later 5 matrixes are acquired in a mode of transaction detail data and label data and belong to source data, a deep neural network model is established, a probability matrix B of the station where the passenger gets out of the bus at this time is predicted, and the station with the highest probability is selected as the station where the passenger gets out of the bus at this time.
Set the sample data set of the deep neural network as { Zl，tl1, 2, L, wherein Z ═ Qt, out'Qt, passenger'Qt, factor'Qt, historical inbound'Qt, real-time inbound'Qt, period of time'QWeekday/weekend'QHolidayF (Z) of the output result matrix, then
Figure BDA0003346064570000071
Wherein β ═ β1，...，βL]TIs the output weight between the hidden layer (L nodes) and the output layer (m nodes, m is more than or equal to 1), and the calculation formula is beta ═ H + TH+Is a generalized inverse of matrix H, and H+＝(HT×H)-1×HT
T is an object matrix of the training data, an
Figure BDA0003346064570000072
H (z) is the output of the hidden layer, and has the formula h (z) ═ h1(Z)，...，hL(Z)]
Wherein h isl(Z) is the output of the l hidden layer node, and the formula is the hidden layer node parameter
hl(Z)＝g(wl，bl，Z)＝g(wlZ+bl)
Wherein g (w)l，bl，Z)(wlAnd blIs a hidden layer node parameter) is an activation function, and
Figure BDA0003346064570000073
wherein w and b are hidden layer node parameters and are randomly generated through random continuous probability distribution.
And 1035, after the station where the passenger takes the bus this time is determined, the station entering time, the trip path and the station exiting time of the passenger are obtained. The passenger outbound time prediction can know the passing route of a passenger k from an O station to a D station, the number of stations, the train running time and the like through a train running diagram, and because the passenger possibly has a plurality of optional paths from a 0 station to the D station, the maximum possible OD path of the passenger is determined as the current trip path of the passenger by adopting a shortest time length method, a shortest path (minimum trip mileage) method and least multiplication, so that the outbound time is determined.
In traffic flow distribution in urban road networks, the shortest mileage is generally employed to search for a route. However, in urban rail transit systems, the precise mileage length is a rather vague idea for most passengers, and the time taken for a trip is exactly what can be perceived. Investigation shows that more than 60% of passengers consider "shortest time" to be the primary factor for selecting a travel route of rail transit. Therefore, the method and the device determine the path by taking the generalized travel time as the impedance, and the path is closer to the reality and more reasonable.
The impedance in the invention refers to the traffic impedance of passengers passing through the gate of the station from station entrance to station exit in the rail transit network, and does not include the time overhead (except for transfer) of the passengers for station entrance and station exit. The trip impedance includes a section (i.e., section) impedance and a node (i.e., station) impedance. In the urban rail transit system, the road section impedance is expressed by the running time of the train in the section; the node impedance is the time spent by passengers at a station, and for passing through the station, the node impedance is the stop time of a train; and for the transfer station, the time taken includes transfer traveling time and transfer waiting time. In consideration of the same time, the transfer traveling and waiting process is longer in the psychological sense time of the passenger than the riding process, and therefore, the node impedance of the transfer station is represented by the transfer time multiplied by a transfer amplification factor α (when α ≧ 1, α ═ 1 indicates that the psychological sense of the passenger on the transfer is not considered), that is, the transfer time is converted into the riding time in the same sense by one transfer amplification factor.
1) Road section impedance Aij
The section impedance of the urban rail transit, namely the running time of the train on the section can be obtained by train section running time or train running time schedule, and t is usedijIs shown, then Aij＝tij
2) Node impedance Bk
The node impedance of rail transit can be divided into two cases:
the first method comprises the following steps: through the station. Passengers continue to ride the same train at the station, and the node impedance is equal to the stop time of the train at the node (k stations), denoted as tkThen B isk＝tk。
And the second method comprises the following steps: and (4) transferring stations. The passenger transfers another track traffic line at the station, and the node impedance is equal to the transfer time of the station multiplied by the transfer amplification factor alpha.
The transfer time comprises transfer traveling time and waiting time. The transfer traveling time is related to the distance of the transfer passage, the distance between two transfer stations can be divided by the ratio of the average walking speed for calculation, and the traveling time spent in transfer can be directly obtained through actual measurement for simple and convenient calculation; the waiting time is related to the departure interval of the transfer train, and the formula is as follows:
Figure BDA0003346064570000091
Figure BDA0003346064570000092
in the above formula, the first and second carbon atoms are,
Figure BDA0003346064570000093
the transfer time for transferring the line p to the line q at the k station is shown;
Figure BDA0003346064570000094
the travel time for the transfer from the k station platforms of line p to the k station platforms of line q,
Figure BDA0003346064570000095
to average the time to wait for a transfer from line p to line q at station k, one half of the departure interval for a train on line q may be taken.
Let the passenger arrive at the station as time tiThen the time t of the passenger's departurejIs composed of
Figure BDA0003346064570000096
Wherein, P is the route passed by the route with the shortest travel time of the passenger, and P is the total number of the routes passed by the route with the shortest travel time of the passenger; k is the transfer station through which the path with the shortest travel time of the passengers passes, and K is the total number of the transfer stations through which the path with the shortest travel time of the passengers passes.
And step 1036, predicting the arrival volume, the departure volume, the passenger transportation volume, the transfer volume and the cross-section passenger flow volume of the line network/line/station according to the passenger travel path and the departure time.
As can be seen from step 1035, the arrival time, the travel route (including the passing route, the transfer station, and the like), the departure time, and the like of each passenger are combined with the train running diagram to deduce the arrival amount, the departure amount, the passenger volume, the transfer amount, the cross-sectional passenger volume, and the like of the wire network/line/station in different time periods. For example, the station arrival amount is the sum of the number of passengers entering the station in a unit time period, and the section passenger flow amount is the sum of the number of all passengers passing through the section in the unit time period.
The invention relates to an application method of a passenger flow prediction method based on an OD (origin-destination) path, which is based on the passenger flow prediction method of the OD path, and an early warning threshold value is set, and if the prediction quantity exceeds the early warning threshold value, an alarm is given to ensure the operation safety. For example, the accurate passenger flow quantity in real time stations and vehicles of the whole line network is provided through passenger flow prediction, three-level early warning is realized through a threshold value, and current and future all passenger flow risk points are provided for passenger traffic management departments and stations for real-time monitoring and dynamic early warning information, so that passenger traffic organization is performed in a targeted manner, safety accidents are avoided, and operation safety is guaranteed.
In addition, on the basis of the passenger flow prediction method of the OD route, individual travel and service requirements of passengers are described and analyzed through passenger figures, and result data of passenger flow prediction are combined to provide personalized and accurate service of all-around and all-travel chains for the passengers, so that passenger travel experience is greatly improved, and passenger service capability and level of intelligent subways are improved.
In addition, on the basis of the passenger flow prediction method of the OD route, accurate passenger flow prediction lays a foundation for intelligent dynamic adjustment of transport capacity, and accurate passenger flow prediction information of a line network, each line and each station provides a reliable basis for intelligent dynamic adjustment of transport capacity according to requirements of a traffic dispatching department, and is a foundation for realizing green and efficient solid monitoring of an intelligent subway operation mode.
In addition, on the basis of the passenger flow prediction method of the OD path, quantitative evaluation can be carried out on emergency events, and data basis is provided for emergency disposal. For example, the passenger flow prediction provides reliable basis for the formulation and optimization of emergency treatment plans of emergencies through historical simulation evaluation and analysis, and provides instant decision support for emergency treatment of emergencies through real-time evaluation and analysis.
The above description is only for the preferred embodiment of the present invention, but the scope of the present invention is not limited thereto, and any person skilled in the art should be considered to be within the technical scope of the present invention, and the technical solutions and the inventive concepts thereof according to the present invention should be equivalent or changed within the scope of the present invention.

Claims (7)

1. A passenger flow prediction method based on OD paths is characterized by comprising the following steps:
101, acquiring sample data;
102, checking and preprocessing sample data;
and 103, carrying out refined passenger flow prediction according to the preprocessed sample data, and predicting the outbound place and the outbound time of the passenger on the basis of the recent OD trip characteristics of different date types, different time periods and different time granularities and the real-time inbound card swiping data on the predicted day.
2. An OD path based passenger flow prediction method as claimed in claim 1, wherein step 102 specifically comprises:
and checking the integrity of the input sample data, namely whether missing values, abnormal values, extreme values, discrete values and the like exist, and if so, supplementing the input sample data by using the average value of the passenger flow data with the same or similar history and weather on the corresponding date.
3. An OD path based passenger flow prediction method as claimed in claim 1, wherein step 103 specifically comprises:
1031, deeply analyzing the influence degree of external influence factors on the OD passenger flow of the subway in each time period and each time granularity according to recent historical travel OD data, meteorological data, activity data and operation event data;
1032, deeply analyzing passengers entering each station in each time period by using historical OD passenger flow data, establishing OD outbound distribution matrix Q according to outbound distribution probability of other stations in the road networkt, outbound；
1033, deeply analyzing OD characteristic probability distribution of each passenger in each time period in continuous year by using historical OD passenger flow data, and establishing a passenger travel OD probability distribution matrix Qt, passengers；
1034, outbound OD distribution matrix Qt, outboundPassenger trip OD probability distribution matrix Qt, passengersInfluence factor weight matrix Q of passenger flows in different time periodst, factorHistorical passenger arrival card swiping data Qt, historical arrivalReal-time card swiping inbound data Qt, real-time entering stationTime zone data Qt, time periodWeekday/weekend identification data QWeekday/weekendHoliday data QHolidayAs input data A, establishing a deep neural network model to predict the station where the passenger gets out of the station by taking a busThe probability matrix B is used for selecting a station with the highest probability as an outbound station for the passenger to take the bus;
1035, obtaining the arrival time, the travel path and the departure time of the passenger after determining the departure station of the passenger taking the bus;
1036, predicting the arrival, departure, transit, and cross-section traffic of the network/line/station according to the travel route and departure time of the passengers.
4. An OD path based passenger flow prediction method as in claim 3, wherein in step 1034, a deep neural network model is established, specifically:
set the sample data set of the deep neural network as { Zl，tl1, 2, L, wherein Z ═ Qt, out' Qt, passenger' Qt, factor'Qt, historical inbound' Qt, real-time inbound' Qt, period of time' QWeekday/weekend' QHolidayF (Z) of the output result matrix, then
Figure FDA0003346064560000021
Wherein β ═ β1，...，βL]TIs the output weight between the hidden layer (L nodes) and the output layer (m nodes, m ≧ 1), and its calculation formula is β ═ H + T
H+Is a generalized inverse of matrix H, and H+＝(HT×H)-1×HT
T is an object matrix of the training data, an
Figure FDA0003346064560000022
H (z) is the output of the hidden layer, and has the formula h (z) ═ h1(Z)，...，hL(Z)]
Wherein h isl(Z) is the output of the 1 st hidden layer node, and the formula is the hidden layer node parameter
hl(Z)＝g(wl，bl，Z)＝g(wlZ+bl)
Wherein g (w)l，bl，Z)(wlAnd blIs a hidden layer node parameter) is an activation function, and
Figure FDA0003346064560000031
wherein w and b are hidden layer node parameters and are randomly generated through random continuous probability distribution.
5. An OD path based passenger flow prediction method as claimed in claim 1, wherein the outbound time is obtained in step 1035, considering traffic impedance: road section impedance AijAnd node impedance Bk。
6. An OD path based passenger flow prediction method as claimed in claim 5, characterised by node impedance BkThe method comprises the following steps:
passing through the station where passengers continue to ride the same train, the node impedance is equal to the time that the train stops at the node (k stations), denoted as tkThen B isk＝tk；
And (3) transferring the station, wherein passengers transfer another track traffic line at the station, and the node impedance is equal to the transfer time of the station multiplied by a transfer amplification factor alpha.
7. An application method of a passenger flow prediction method based on an OD path is characterized in that on the basis of claims 1 to 6, an early warning threshold value is set, and when the passenger flow prediction amount exceeds the early warning threshold value, an alarm is given.
CN202111322651.9A 2021-11-09 2021-11-09 Passenger flow prediction method based on OD path and application method Pending CN114037158A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202111322651.9A CN114037158A (en) 2021-11-09 2021-11-09 Passenger flow prediction method based on OD path and application method

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202111322651.9A CN114037158A (en) 2021-11-09 2021-11-09 Passenger flow prediction method based on OD path and application method

Publications (1)

Publication Number Publication Date
CN114037158A true CN114037158A (en) 2022-02-11

Family

ID=80137023

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202111322651.9A Pending CN114037158A (en) 2021-11-09 2021-11-09 Passenger flow prediction method based on OD path and application method

Country Status (1)

Country Link
CN (1) CN114037158A (en)

Cited By (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114331234A (en) * 2022-03-16 2022-04-12 北京交通大学 Rail transit passenger flow prediction method and system based on passenger travel information
CN115759472A (en) * 2022-12-07 2023-03-07 北京轨道交通路网管理有限公司 Passenger flow information prediction method and device and electronic equipment
WO2023164983A1 (en) * 2022-03-03 2023-09-07 北京城建设计发展集团股份有限公司 Holographic subway station passenger flow real-time monitoring method and apparatus, and computer device
CN117669851A (en) * 2023-12-07 2024-03-08 重庆市凤筑科技有限公司 On-line decision support system oriented to public transport network optimization
CN117973640A (en) * 2024-03-29 2024-05-03 煤炭科学研究总院有限公司 Transfer subway station transfer personnel flow prediction method and device

Cited By (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2023164983A1 (en) * 2022-03-03 2023-09-07 北京城建设计发展集团股份有限公司 Holographic subway station passenger flow real-time monitoring method and apparatus, and computer device
CN114331234A (en) * 2022-03-16 2022-04-12 北京交通大学 Rail transit passenger flow prediction method and system based on passenger travel information
CN114331234B (en) * 2022-03-16 2022-07-12 北京交通大学 Rail transit passenger flow prediction method and system based on passenger travel information
CN115759472A (en) * 2022-12-07 2023-03-07 北京轨道交通路网管理有限公司 Passenger flow information prediction method and device and electronic equipment
CN115759472B (en) * 2022-12-07 2023-12-22 北京轨道交通路网管理有限公司 Passenger flow information prediction method and device and electronic equipment
CN117669851A (en) * 2023-12-07 2024-03-08 重庆市凤筑科技有限公司 On-line decision support system oriented to public transport network optimization
CN117973640A (en) * 2024-03-29 2024-05-03 煤炭科学研究总院有限公司 Transfer subway station transfer personnel flow prediction method and device

Similar Documents

Publication Publication Date Title
CN114037158A (en) Passenger flow prediction method based on OD path and application method
CN112598182B (en) Intelligent scheduling method and system for rail transit
Wang et al. Passenger-demands-oriented train scheduling for an urban rail transit network
CN111401614B (en) Dynamic passenger flow distribution method and system for urban rail transit
CN109886491A (en) A kind of routine bus system two stages Optimization Scheduling of magnanimity operation data driving
CN107358319A (en) Flow Prediction in Urban Mass Transit method, apparatus, storage medium and computer equipment
CN109508751B (en) Deep neural network model modeling method for high-speed railway train late time prediction
CN107194491A (en) A kind of dynamic dispatching method based on Forecasting of Travel Time between bus passenger flow and station
CN110084397B (en) Subway through line planning method
CN113222387B (en) Multi-objective scheduling and collaborative optimization method for hydrogen fuel vehicle
Wilson et al. The potential impact of automated data collection systems on urban public transport planning.
JPH07309546A (en) Traffic means controller
CN115527369B (en) Large passenger flow early warning and evacuation method under large-area delay condition of airport hub
CN113850417A (en) Passenger flow organization decision-making method based on station passenger flow prediction
CN114707709A (en) Safety early warning method and system for comprehensive passenger transport hub of railway
Bucak et al. Train timetabling for a double-track urban rail transit line under dynamic passenger demand
Pencheva et al. Evaluation of passenger waiting time in public transport by using the Monte Carlo method
CN112562377A (en) Passenger vehicle real-time scheduling method based on random opportunity constraint
CN109308539A (en) The method of passenger's retaining state of transfer stop in real-time estimation Metro Network
CN116109081A (en) Dynamic bus scheduling optimization method and device based on Internet of things
CN115860594A (en) Simulation system and method applied to intelligent bus scheduling
Liu et al. Understanding urban bus travel time: Statistical analysis and a deep learning prediction
Tiong et al. Analyzing factors contributing to real-time train arrival delays using seemingly unrelated regression models
CN109409563A (en) A kind of analysis method, system and the storage medium of the real-time number of bus operation vehicle
CN113408926A (en) Urban rail transit passenger flow distribution method under short-time interruption condition

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