CN110428117B - Passenger flow accurate induction method and system under urban rail transit multi-scene - Google Patents

Passenger flow accurate induction method and system under urban rail transit multi-scene Download PDF

Info

Publication number
CN110428117B
CN110428117B CN201910753308.6A CN201910753308A CN110428117B CN 110428117 B CN110428117 B CN 110428117B CN 201910753308 A CN201910753308 A CN 201910753308A CN 110428117 B CN110428117 B CN 110428117B
Authority
CN
China
Prior art keywords
path
passenger
time
representing
congestion
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
CN201910753308.6A
Other languages
Chinese (zh)
Other versions
CN110428117A (en
Inventor
许心越
夏霖琪
刘军
张亚敏
赵若愚
李建民
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 Jiaotong University
Original Assignee
Beijing Jiaotong University
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 Jiaotong University filed Critical Beijing Jiaotong University
Priority to CN201910753308.6A priority Critical patent/CN110428117B/en
Publication of CN110428117A publication Critical patent/CN110428117A/en
Application granted granted Critical
Publication of CN110428117B publication Critical patent/CN110428117B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

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"
    • G06Q10/047—Optimisation of routes or paths, e.g. travelling salesman problem
    • 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

  • Business, Economics & Management (AREA)
  • Human Resources & Organizations (AREA)
  • Engineering & Computer Science (AREA)
  • Strategic Management (AREA)
  • Tourism & Hospitality (AREA)
  • Economics (AREA)
  • General Business, Economics & Management (AREA)
  • General Physics & Mathematics (AREA)
  • Marketing (AREA)
  • Theoretical Computer Science (AREA)
  • Development Economics (AREA)
  • Physics & Mathematics (AREA)
  • Educational Administration (AREA)
  • Health & Medical Sciences (AREA)
  • General Health & Medical Sciences (AREA)
  • Primary Health Care (AREA)
  • Game Theory and Decision Science (AREA)
  • Entrepreneurship & Innovation (AREA)
  • Operations Research (AREA)
  • Quality & Reliability (AREA)
  • Train Traffic Observation, Control, And Security (AREA)
  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)

Abstract

The invention provides a passenger flow accurate induction method and system under multiple scenes of urban rail transit, and belongs to the technical field of urban rail transit train operation control. The method comprises the steps of establishing a passenger path selection behavior model with the effectiveness maximization as a target under multi-scene induction information; calculating the time when the passenger passes through each section and the transfer station on the effective path, and quantitatively evaluating the overall congestion level of the path based on the congestion condition of each section and the corresponding time of the transfer station to reflect the congestion degree; selecting a behavior model based on the passenger path, and sequencing the feasible paths; and aiming at the sorted feasible paths, optimizing the parameters of the path selection behavior model by combining a Q-learning algorithm to obtain the optimal induced path. The invention combines the congestion degree, time, transfer times and other relevant factors to sort the feasible paths, and the feasible paths are recommended to passengers as the guidance information; and finally, optimizing parameters of the path selection behavior model by combining a reinforcement learning method, and effectively improving the informatization level and the service quality of urban rail transit.

Description

Passenger flow accurate induction method and system under urban rail transit multi-scene
Technical Field
The invention relates to the technical field of urban rail transit train operation control, in particular to a passenger flow accurate induction method and system under multiple scenes of urban rail transit.
Background
In recent years, the urban rail transit operation mileage and traffic volume in China are in the forefront of the world, and the urban rail transit has the characteristics of networking, large passenger flow, high density and the like, and the rail transit becomes a key support for guaranteeing urban operation. However, the capacity of partial sections of lines and stations is insufficient in the peak hours, the full section rate and the station platform density are continuously too high in the peak hours, which become the normal state of subway operation, and guidance and management on the traveling process of passengers in the peak hours are urgently needed. Although the induction theory has certain research progress at home and abroad, the method mainly focuses on the field of road traffic, and has less research on the induction direction of rail traffic passenger flow; in the aspect of practical application of an induction technology, the application of an induction system of a city in China is too simple and cannot meet the requirement of diversified multi-scene passenger information, so that the method is of great importance for accurate induction of passenger flow under multiple scenes of urban rail transit.
Disclosure of Invention
The invention aims to provide a method and a system for accurately inducing passenger flow under multiple scenes of urban rail transit, which can truly and effectively assist operators in releasing induction information and provide certain help for benign operation of the urban rail transit so as to solve the technical problems in the background technology.
In order to achieve the purpose, the invention adopts the following technical scheme:
on one hand, the invention provides a passenger flow accurate induction method under multiple scenes of urban rail transit, which comprises the following steps:
step S110: establishing a passenger path selection behavior model with the maximum utility as a target under multi-scene inducing information;
step S120: calculating the time when the passenger passes through each section and the transfer station on the effective path, and quantitatively evaluating the overall congestion level of the path based on the congestion condition of each section and the corresponding time of the transfer station to reflect the congestion degree;
step S130: selecting a behavior model based on the passenger path, and sorting the feasible paths according to the generalized utility;
step S140: and aiming at the sequenced feasible paths, optimizing parameters of a path selection behavior model by combining a Q-learning algorithm to obtain an optimal induced path.
Preferably, the step S110 specifically includes:
the passenger path selection behavior model is a discrete selection model taking effectiveness maximization as a target;
the train running time is calculated as follows:
Figure BDA0002167935210000021
wherein,
Figure BDA0002167935210000022
representing the total train travel time of OD for path k of a- > b,
Figure BDA0002167935210000023
representing the train's run time in the interval i-j in the path k,
Figure BDA0002167935210000024
representing the stop time of the train at the station m in the path k;
the waiting time of passengers is as follows:
Figure BDA0002167935210000025
wherein, TwaitThe time of the second waiting time is shown,
Figure BDA0002167935210000026
the train departure interval of the line l is represented;
the passenger transfer time is:
Figure BDA0002167935210000027
wherein,
Figure BDA0002167935210000028
represents the total transfer time of OD to the passenger on the path k in a- > b, tau represents the penalty coefficient of the passenger caused by transfer, nkRepresents the total number of transfers of the path k, represents the penalty coefficient of the number of transfers,
Figure BDA0002167935210000029
indicating the i-th transfer of the slave line1To l2The running time of (2);
the congestion perception coefficient of the passenger is as follows:
Figure BDA0002167935210000031
wherein, CijRepresenting the perceptual coefficients of the intervals i-j,
Figure BDA0002167935210000032
is the congestion coefficient, mijIs the full load of the interval i-j, m0Grading the threshold for the first full load, m2Grading a threshold for a second fullness; o isIRepresenting the set of affected regions, KYRepresenting a set of affected passengers;
the generalized travel cost of the passengers is as follows:
Figure BDA0002167935210000033
wherein,
Figure BDA0002167935210000034
represents the travel cost, ξ, of OD for path k for a- > bj,kIs a subordinate mark and indicates whether the section/station is on the path k or not, and xi isj,k＝{0,1}；
In the discrete selection model, the total utility is composed of two parts, one part is fixed utility, and the value is not changed as long as the passenger is determined and the related attribute is determined, and is matched with a certain passenger or a certain class of travelers, so that the generalized cost of the passenger for traveling is provided; but because of the perturbations in the real environment, random utility is used to represent the perturbation relationship. It is generally assumed that the two parts exhibit a linear relationship, as follows:
Figure BDA0002167935210000035
in the formula:
Figure BDA0002167935210000036
represents the total utility of path k for a- > b,
Figure BDA0002167935210000037
representing the random utility portion.
Constructing a Logit model, and discretely selecting the model:
Figure BDA0002167935210000038
in the formula: p is more than or equal to 0ki≤1,
Figure BDA0002167935210000039
Figure BDA00021679352100000310
Representing the probability of the passenger selecting a route;
Figure BDA00021679352100000311
representing the minimum generalized cost of a- > b.
Preferably, the step S120 specifically includes:
calculating the congestion degree of the path by taking the high-fullness interval proportion and the fullness distribution entropy as indexes;
the high-full-load-rate interval proportion refers to the ratio of the number of the high-full-load-rate intervals to the number of all the intervals of the urban rail transit network:
Figure BDA0002167935210000041
wherein Z represents the proportion of the high-loading-rate interval within the evaluation time; l represents the number of all intervals in the road network; l ishRepresenting the number of high-fill-rate intervals within the evaluation period.
Combining the interval full-load rate and the information entropy to construct a full-load rate distribution entropy index:
Figure BDA0002167935210000042
wherein H represents the full-load distribution entropy within the evaluation time; g represents the interval full load rateThe scattered value should be accurate to 10% accuracy, and G is 0, 1.., G; l isgL represents the proportion of the section with the full load rate g to all the sections of the road networkg≠0。
Preferably, the step S130 specifically includes:
under normal operation conditions, the passenger waiting time, the transfer time and the perception of the crowding degree establish utility functions, calculate the generalized travel cost of passengers in each path, and calculate the generalized travel cost of passengers according to Vk a,bSequencing the feasible paths and inducing passengers to reasonably select travel paths;
under the condition of train delay, knowing an estimated delay range and an estimated delay duration, after a passenger selects a starting point and sets a starting time, considering passenger waiting time, transfer time and perception of congestion degree when calculating a utility function aiming at a feasible path influenced by delay, and increasing and displaying waiting time caused by delay, wherein the delay waiting time has two conditions: the first existing path is not feasible, i.e. the waiting time is infinite; the second existing path is feasible and has a small latency (within one hour). And finally, comprehensively sequencing the paths, thereby inducing passengers to reasonably select travel paths.
Preferably, the step of the Q-learning algorithm comprises:
step S141: initializing a Q-table matrix, wherein 'row' represents the same proportion of the recommended path and the actual path as a state, and 'column' represents
Figure BDA0002167935210000043
And
Figure BDA0002167935210000044
the congestion coefficient is used as an active set, and each proportion corresponds to a Q-table matrix;
step S142: acquiring states of the recommended path and the actual path in the same proportion, wherein the state set comprises three states: inconsistent, more consistent, and very consistent;
step S143: selecting the one corresponding to the maximum Q value in the Q-table matrix by a greedy algorithm
Figure BDA0002167935210000051
And
Figure BDA0002167935210000052
a path selection behavior of;
step S144: calling positioning data to obtain an actual path of the passenger;
step S145: computing
Figure BDA0002167935210000053
And
Figure BDA0002167935210000054
a varying reward function;
step S146: calculating new state states of the passenger recommended path and the actual path in the same proportion after the next system induction;
step S147: updating the state sum of the recommended path after induction and the actual path of the passenger in the same proportion by using a Bellman Equation iterative formula
Figure BDA0002167935210000055
And
Figure BDA0002167935210000056
the Q-table matrix of (1);
step S148: judging whether the iteration times are reached, if not, corresponding
Figure BDA0002167935210000057
And
Figure BDA0002167935210000058
the new state states with the same proportion of the recommended path and the actual path are taken as the current state, and the step S143 is entered; if the iteration times are reached, terminating the iteration, and outputting the recommended paths corresponding to the actual paths of the passengers in the same proportion
Figure BDA0002167935210000059
And
Figure BDA00021679352100000510
preferably, the state set S represents the same proportion of the recommended route and the actual route of the passenger, and the activity set a represents the congestion coefficient
Figure BDA00021679352100000511
And
Figure BDA00021679352100000512
Figure BDA00021679352100000513
and
Figure BDA00021679352100000520
the selection probability of (2) is:
Figure BDA00021679352100000514
wherein, pi [ a | st]To represent
Figure BDA00021679352100000515
And
Figure BDA00021679352100000516
the greater the Q value, the greater the selection probability of (2)
Figure BDA00021679352100000517
And
Figure BDA00021679352100000518
the greater the probability of being selected.
Preferably, the reward function is a behavior model utility function selected by the passenger path under multiple scenes;
the Q function is:
Figure BDA00021679352100000519
Figure BDA0002167935210000061
the Bellman Equation is used to update the Q-table as follows:
Q(i,sn,a)＝(1-α)Q(i,sn,a)+α[Ri(sn,a)+γ·max Q(i,sn+1,a)]
in the formula: a is in [0,1]]The learning rate is represented, and the larger the learning rate is, the more remarkable the effect of the Q-table iterative updating is; gamma is belonged to 0,1]The discount factor is represented, and the larger the discount factor is, the larger the function of the future maximum reward function value is; r represents fixation
Figure BDA0002167935210000062
And
Figure BDA0002167935210000063
the value of the reward earned; MaxQ (i, s)n+1A) represents all of the Q-tables in the proportional state where the next recommended path is the same as the actual path of the passenger
Figure BDA0002167935210000064
And
Figure BDA0002167935210000065
the maximum value among the corresponding values.
On the other hand, the invention also provides a system for accurately inducing passenger flow under multiple scenes of urban rail transit, which comprises: the system comprises a road network state prediction module, a feasible path sorting module and a recommendation sorting optimization module;
the road network state prediction module is used for calculating corresponding technical indexes by taking road network original data as input, grading the corresponding technical indexes by using threshold values of the corresponding technical indexes, and performing space-time analysis on congestion of a road network by combining passenger ticket card information, a train schedule, train delay information and last bus information so as to predict the state of the road network;
the feasible path sequencing module is used for analyzing the sensibility of different passengers to the guidance information based on the influence of the guidance information on the passenger path selection, establishing a passenger path selection behavior model based on multi-scene guidance information and sequencing the feasible paths of the passengers in different scenes;
and the recommendation optimization module is used for optimizing the passenger path selection behavior model and parameters thereof by combining a reinforcement learning method after sequencing the feasible paths, thereby realizing the generation of the optimal induced path meeting personal preference.
Preferably, the feasible path ranking module comprises a passenger path selection calculating unit, a congestion degree calculating unit and a feasible path ranking unit;
the passenger path selection computing unit is used for establishing a passenger path selection behavior model with the effectiveness maximized as the target under the multi-scene inducing information;
the congestion degree calculating unit is used for calculating the time when the passenger passes through each section and the transfer station on the effective path, and quantitatively evaluating the overall congestion level of the path based on the congestion condition of each section and the corresponding time of the transfer station to reflect the congestion degree;
the feasible path sorting unit is used for selecting a behavior model based on the passenger path and sorting the feasible paths according to the generalized utility.
Preferably, the system further comprises: the system comprises a query module, an information push module, a last bus query module, a trip correction module and a data management module;
the query module is configured to: inquiring feasible paths and states thereof according to the appointed starting and stopping point and starting time input by the passenger, and displaying the feasible paths according to the principle of small crowding degree and short running time and less total station number for transfer to order and recommend the feasible paths to the passenger;
the information pushing module is used for: under a train delay scene, according to a common OD, a path and travel time set by a passenger, and in combination with a time-space influence range of delay, pushing guidance information for changing the path to the passenger affected by the delay for reminding;
the last bus query module is used for: displaying the reachable condition of the specified station going to other stations at the specified time and the reachable condition and the latest time of the travel path of the specified OD for the passenger;
the trip correcting module is used for: providing travel correction service according to the real-time positioning information of the passenger and combining the travel path of the passenger;
the data management module is used for: basic data support is provided for the system, and the basic data support comprises road network prediction data, passenger ticket card information, road network basic data, feasible path set data, a train schedule, an information template, train delay information, last-class train information and passenger actual travel path data.
The invention has the beneficial effects that: establishing a passenger path selection model by taking the effect maximization as a basic assumption, calculating a path congestion degree index based on indexes such as full load rate and the like, and sequencing feasible paths by combining relevant factors such as congestion degree, time, transfer times and the like to serve as guidance information to be recommended to passengers; and finally, optimizing parameters of the path selection behavior model by combining a reinforcement learning method, and effectively improving the informatization level and the service quality of urban rail transit.
Additional aspects and advantages of the invention will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of the invention.
Drawings
In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings needed to be used in the description of the embodiments are briefly introduced below, and it is obvious that the drawings in the following description are only some embodiments of the present invention, and it is obvious for those skilled in the art to obtain other drawings based on these drawings without creative efforts.
Fig. 1 is a flowchart illustrating steps of a passenger flow accurate induction method in multiple scenes of urban rail transit according to an embodiment of the present invention.
Fig. 2 is a schematic structural diagram of an accurate guidance system for passenger flow in multiple scenes of urban rail transit according to an embodiment of the present invention.
Fig. 3 shows the recommended path from daylily to ancestral temple and the congestion and current-limiting influence of real-time road conditions during normal operation of the train according to the embodiment of the present invention.
Fig. 4 shows the recommended route from daylily to ancestral temple during train delay and the congestion and current-limiting influence of real-time road conditions according to the embodiment of the present invention.
Fig. 5 is a diagram illustrating a modified route reminder pushed by the validation system for a passenger during a train delay according to an embodiment of the present invention.
Fig. 6 is a recommended route for last bus query from east-shan-mouth to west-village provided by the guidance system according to the embodiment of the present invention.
Fig. 7 illustrates a travel correction service provided by the guidance system for passengers according to the embodiment of the present invention.
Detailed Description
The embodiments described below with reference to the accompanying drawings are illustrative only for the purpose of explaining the present invention, and are not to be construed as limiting the present invention.
As used herein, the singular forms "a", "an", "the" and "the" are intended to include the plural forms as well, unless the context clearly indicates otherwise. It will be further understood that the terms "comprises" and/or "comprising," when used in this specification, specify the presence of stated features, integers, steps, operations, elements, and/or modules, but do not preclude the presence or addition of one or more other features, integers, steps, operations, elements, modules, and/or groups thereof.
It will be understood by those skilled in the art that, unless otherwise defined, all terms (including technical and scientific terms) used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs. It will be further understood that terms, such as those defined in commonly used dictionaries, should be interpreted as having a meaning that is consistent with their meaning in the context of the prior art and will not be interpreted in an idealized or overly formal sense unless expressly so defined herein.
For the convenience of understanding of the embodiments of the present invention, the following description will be further explained by taking specific embodiments as examples with reference to the drawings, and the embodiments are not to be construed as limiting the embodiments of the present invention.
It will be understood by those of ordinary skill in the art that the figures are merely schematic representations of one embodiment and that the elements or devices in the figures are not necessarily required to practice the present invention.
Examples
As shown in fig. 1, an embodiment of the present invention provides a method for accurately inducing passenger flow in multiple scenes of urban rail transit, including the following steps:
step S110: establishing a passenger path selection behavior model with the maximum utility as a target under multi-scene inducing information;
step S120: calculating the time when the passenger passes through each section and the transfer station on the effective path, and quantitatively evaluating the overall congestion level of the path based on the congestion condition of each section and the corresponding time of the transfer station to reflect the congestion degree;
step S130: selecting a behavior model based on the passenger path, and sorting the feasible paths according to the generalized utility;
step S140: and aiming at the sequenced feasible paths, optimizing parameters of a path selection behavior model by combining a Q-learning algorithm to obtain an optimal induced path.
Specifically, in step S110, the influence of the guidance information on passenger routing is studied, the sensitivities of different passengers to the guidance information are analyzed, and a passenger routing behavior model based on the multi-scenario guidance information is established.
In the embodiment, in the passenger routing, the generalized cost of the passenger for traveling is calculated mainly by considering whether the passenger receives the guidance information and the perception of the passenger on congestion and transfer, and the generalized cost is calculated as follows:
(1) train running time
The train operation time is mainly composed of the operation time and the stop time, as shown below
Figure BDA0002167935210000101
In the formula:
Figure BDA0002167935210000102
representing the total train travel time of OD for path k of a- > b,
Figure BDA0002167935210000103
representing the train's run time in the interval i-j in the path k,
Figure BDA0002167935210000104
representing the stop time of the train at station m in path k.
(2) Waiting time
Figure BDA0002167935210000105
In the formula: t iswaitThe time of the second waiting time is shown,
Figure BDA0002167935210000106
indicating the departure interval of the train on the route l.
(3) Transfer time
The transfer time of the passenger trip mainly comprises transfer walking time and waiting time, wherein the transfer time is corrected by a penalty factor in consideration of the sensitivity of the passenger to transfer, and the following steps are included:
Figure BDA0002167935210000107
in the formula:
Figure BDA0002167935210000108
represents the total time of transfer of OD to the passenger on path k in a- > b; wherein tau represents the penalty coefficient of the passenger caused by transfer; n iskRepresents the total number of transfers for path k; beta represents the punishment coefficient of the transfer times;
Figure BDA0002167935210000109
indicating the i-th transfer of the slave line1To l2The running time of (2).
(4) Congestion awareness
In the traveling process of passengers, traveling experience is also a factor considered, so that a congestion perception coefficient is introduced, and the coefficient has a direct relation with the full loading rate of a road network section as follows:
Figure BDA0002167935210000111
in the formula: cijRepresenting the perceptual coefficients of the intervals i-j,
Figure BDA0002167935210000112
is the congestion coefficient, mijIs the full load of the interval i-j, m0Grading the threshold for the first full load, m2Grading a threshold for a second fullness; o isIRepresenting the set of affected regions, KYRepresenting the affected passenger collection.
(5) Generalized trip cost for passengers
The generalized travel cost of the passengers obtained by combining the formula is as follows:
Figure BDA0002167935210000113
wherein,
Figure BDA0002167935210000114
represents the travel cost, ξ, of OD for path k for a- > bj,kIs a subordinate mark and indicates whether the section/station is on the path k or not, and xi isj,k＝{0,1}；
In this example, the passenger routing behavior is based on a discrete selection model with maximized utility. In the discrete selection model, the total utility is composed of two parts, one part is fixed utility, and the value is not changed as long as the passenger is determined and the related attribute is determined, and is matched with a certain passenger or a certain class of travelers, so that the generalized cost of the passenger for traveling is provided; but because of the perturbations in the real environment, random utility is used to represent the perturbation relationship. It is generally assumed that the two parts exhibit a linear relationship, as follows:
Figure BDA0002167935210000115
in the formula:
Figure BDA0002167935210000116
represents the total utility of path k for a- > b,
Figure BDA0002167935210000117
representing the random utility portion.
Constructing a Logit model, and discretely selecting the model:
Figure BDA0002167935210000118
in the formula:
Figure BDA0002167935210000121
Figure BDA0002167935210000122
representing the probability of the passenger selecting a route;
Figure BDA0002167935210000123
representing the minimum generalized cost of a- > b.
In the embodiment of the present invention, the numeric parameters of the model are shown in table 1:
Table 1 Model parameters values
Figure BDA0002167935210000124
specifically, in step S120, passenger flow indexes such as time-sharing full load rates of each section are analyzed based on the passenger flow prediction data, the time when the passenger passes through each section and the transfer station on the effective route is estimated prospectively, the overall congestion level of the route is quantitatively evaluated based on the congestion situation at the corresponding time of each position, and the congestion degree index is calculated.
The safety and stability of the road network operation state are measured from two aspects of the urban rail transit road network congestion range and the congestion distribution, the road network congestion range can be reflected by the number of high-full-load intervals according to the proportion of the high-full-load intervals, and the road network congestion distribution can be reflected by the full-load distribution entropy.
(1) Interval ratio of high full load rate
The proportion of the high-loading-rate intervals refers to the ratio of the number of the high-loading-rate intervals to the number of all the intervals of the urban rail transit network, wherein the high-loading-rate intervals refer to the intervals with the average loading rate of the intervals exceeding 80%, namely the load is greater than 80%. The proportion of the high-load-ratio interval can reflect the safety and the congestion range of the road network, and the value range is [0,1 ]. The larger the value of the proportion of the high-full-load rate interval is, the larger the number of the high-full-load rate intervals is, the larger the congestion range of the road network is, and the more potential safety hazards exist.
Figure BDA0002167935210000125
In the formula: z represents the proportion of the high-loading-rate interval in the evaluation time; l represents the number of all intervals in the road network; l ishRepresenting the number of high-fill-rate intervals within the evaluation period.
(2) Entropy of the full load rate distribution
Entropy, which is a physical quantity representing the state of a substance, can measure whether a system is chaotic or stable. The entropy of information is the average amount of information remaining after discarding redundant information. The method introduces the information entropy into the urban rail transit system, combines the full load rate of the interval with the information entropy, and constructs the full load rate distribution entropy index, which is shown in the following formula. The full-load rate distribution entropy reflects the congestion distribution of the road network and the stability of the road network. For a road network with G full-load rate intervals, the index value range is [0, lnG ], the larger the entropy value is, the more full-load rate levels of the road network intervals are, the more unbalanced the full-load rate distribution is, and the more unstable the road network state is.
Figure BDA0002167935210000131
In the formula: h represents the full-load rate distribution entropy in the evaluation time; g represents a discretized value of the interval full load rate, and the discretized value is accurate to 10%, wherein G is 0, 1. L isgL represents the proportion of the section with the full load rate g to all the sections of the road networkg≠0。
Specifically, in step S130, under the condition that the passenger selects the starting point and the ending point and sets the departure time, a utility function is established based on the passenger path selection behavior model under multiple scenes, in combination with the congestion degree, the time, the transfer times and other factors, and feasible paths are ranked according to the generalized utility and recommended to the passenger as guidance information.
Under normal operation conditions, the utility function is established by the waiting time, the transfer time and the perception of the crowding degree of the passengers, the generalized travel cost of the passengers in each path can be calculated, and the method is based on the principle that
Figure BDA0002167935210000132
And sequencing the feasible paths and inducing the passengers to reasonably select the travel paths.
Under the condition of train delay, knowing an estimated delay range and an estimated delay duration, after a passenger selects a starting point and sets a starting time, considering passenger waiting time, transfer time and perception of congestion degree when calculating a utility function aiming at a feasible path influenced by delay, and increasing and displaying waiting time caused by delay, wherein the delay waiting time has two conditions: the first existing path is not feasible, i.e. the waiting time is infinite; the second existing path is feasible and has a small latency (within one hour). And finally, comprehensively sequencing the paths, thereby inducing passengers to reasonably select travel paths.
Specifically, in step S140, after the recommended travel route is pushed to the passenger, the consistency between the recommended route and the actual route is compared, the guidance information distribution effect is evaluated, and the parameters of the route selection behavior model are optimized by combining the reinforcement learning method.
Q-learning algorithm basic elements:
the state set S represents the recommended path and the passenger entityProportional check with same interpath>90％，60％～90％，<60% }. Active set a representation
Figure BDA0002167935210000141
And
Figure BDA0002167935210000142
{0.12, 0.13, 0.14, 0.15, 0.16, 0.17, 0.18} and {0.17, 0.18, 0.19, 0.2, 0.21, 0.22, 0.23 }.
Figure BDA0002167935210000143
In the above formula, [ a ] st]To represent
Figure BDA0002167935210000144
And
Figure BDA0002167935210000145
the greater the Q value, the greater the selection probability of (2)
Figure BDA0002167935210000146
And
Figure BDA0002167935210000147
the greater the probability of being selected.
The reward function is a behavior model utility function selected by the passenger path under multiple scenes and is shown as the following formula. The invention reflects the generalized cost of passenger travel from the waiting time, the transfer time and the congestion perception, and further reflects the parameters
Figure BDA0002167935210000148
And
Figure BDA0002167935210000149
and (6) optimizing. The less generalized cost of passenger trip represents
Figure BDA00021679352100001410
And
Figure BDA00021679352100001411
the more accurate the parameters.
Figure BDA00021679352100001412
In the formula: r represents a reward function over an evaluation period; csA perceptual coefficient representing a section s;
Figure BDA00021679352100001413
represents the travel cost of OD to the path k of a- > b; tau represents the penalty coefficient of the passenger caused by transfer; n iskRepresents the total number of transfers for path k; beta represents the punishment coefficient of the transfer times;
Figure BDA00021679352100001414
indicating the i-th transfer of the slave line1To l2The running time of (2); t iswaitRepresenting waiting time; csA perceptual coefficient representing a section s;
Figure BDA00021679352100001418
representing the run time of the train in section s in path k,
Figure BDA00021679352100001415
representing the stop time of the train at station m in path k. Xij,kIs a subordinate mark and indicates whether the section/station is on the path k or not, and xi isj,k＝{0,1}。
Function of Q value:
Figure BDA00021679352100001416
Figure BDA00021679352100001417
the Q-learning algorithm firstly uses the recommended path and the passenger actual path which are the sameProportional state and
Figure BDA0002167935210000151
and
Figure BDA0002167935210000152
and forming a Q-table matrix table, and finally enabling the active set of the intelligent agent to tend to be optimal by continuously iteratively modifying the Q-table matrix table. During training, the Bellman Equation is used to update the Q-table as follows:
Q(i,sn,a)＝(1-α)Q(i,sn,a)+α[Ri(sn,a)+γ·max Q(i,sn+1,a)]
in the formula: a is in [0,1]]The learning rate is represented, and the larger the learning rate is, the more remarkable the effect of the intelligent agent Q-table iterative updating is; gamma is belonged to 0,1]The discount factor is represented, and the larger the discount factor is, the larger the function of the future maximum reward function value is; r represents fixation
Figure BDA0002167935210000153
And
Figure BDA0002167935210000154
the value of the reward earned; MaxQ (i, s)n+1A) represents all of the Q-tables in the proportional state where the next recommended path is the same as the actual path of the passenger
Figure BDA0002167935210000155
And
Figure BDA0002167935210000156
the maximum value among the corresponding values.
Q-learning algorithm flow:
the steps for solving the optimization of the path selection behavior model parameters based on the Q-learning algorithm are as follows:
step 1: initializing a Q-table matrix, wherein ' row ' represents the same proportion of the recommended path and the passenger's actual path as a state, and ' column ' represents
Figure BDA0002167935210000157
And
Figure BDA0002167935210000158
the congestion coefficient of (2) is used as an active set, and each proportion corresponds to a Q-table matrix.
Step 2: obtaining the states of the recommended path and the actual path of the passenger in the same proportion after the induction of the system, wherein the state set comprises three states: inconsistent, more consistent, and very consistent.
Step 3: to avoid local optima, greedy algorithms are used herein, which select the Q-value corresponding to the maximum Q value at the state in the Q-table matrix
Figure BDA0002167935210000159
And
Figure BDA00021679352100001510
action of congestion coefficient.
Step 4: calling the passenger flow accurate induction system under the multi-scene to obtain a passenger recommended path, calling APP positioning data to obtain a passenger travel track, and further deducing to obtain a passenger actual path.
Step 5: computing based on data
Figure BDA00021679352100001511
And
Figure BDA00021679352100001512
the changed reward function reward.
Step 6: and calculating the new state states of the passenger recommended path and the actual path in the same proportion after the next system induction.
Step 7: updating the state sum of the recommended path after induction and the actual path of the passenger in the same proportion by using a Bellman Equation iterative formula
Figure BDA0002167935210000161
And
Figure BDA0002167935210000162
congestion coefficient ofThe Q-table matrix of (1).
Step 8: judging whether the iteration times are reached, if not, corresponding
Figure BDA0002167935210000163
And
Figure BDA0002167935210000164
taking the new state states of the recommended path of the congestion coefficient and the actual path in the same proportion as the current state, and entering Step 3; if the iteration times are reached, terminating the iteration, and outputting the recommended paths corresponding to the actual paths of the passengers in the same proportion
Figure BDA0002167935210000165
And
Figure BDA0002167935210000166
and (4) congestion coefficient.
As shown in fig. 2, in the embodiment of the present invention, a passenger flow accurate guidance system under multiple scenes using the method is provided, where the system includes a prospective query module, an information push module, a last-class vehicle query module, a trip correction module, a background interaction module, and a data management module.
(1) A prospective query module: appointing a starting point and a finishing point and starting time according to passenger input, calculating feasible paths and states thereof in a prospective manner, wherein the feasible paths comprise reachable states, congestion indexes, time consumption, ticket prices, total station number of paths, transfer times, current-limiting stations and the like, and sequencing the feasible paths according to certain principles (no congestion, short time, less transfer and the like) and recommending the feasible paths to the passengers;
under the normal operation condition, passenger flow indexes such as time-sharing full load rate of each section are analyzed based on passenger flow prediction data, the time when passengers pass through each section and transfer station on an effective path is calculated in a prospective mode, and the overall congestion level of the path is quantitatively evaluated based on the congestion condition of the corresponding time of each position. After the passenger selects the starting and ending point and sets the starting time, the feasible routes are displayed to the passenger, the feasible routes comprise reachable states, congestion indexes, time consumption, ticket prices, total number of routes, transfer times, current-limiting stations and the like, the feasible routes are sequenced according to a certain principle, and the passenger is induced to reasonably select a travel route.
For example:
assuming that the passenger selects 8:00 to go from daylily on line six to ancestral temple one, the route pushed to the passenger is as shown in fig. 3, and the passenger can view the default recommended route or select a route with different attributes (less time, less transfer, not crowded, etc.).
Under the condition of train delay, knowing the estimated delay range and the estimated delay duration, after the passenger selects the starting point and the ending point and sets the starting time, calculating and increasing the waiting time caused by delay aiming at the feasible route influenced by delay, and inducing the passenger to reasonably select the trip route.
For example:
suppose an emergency occurs between the east mountain mouth station of the line six and the district station, which causes the departure interval of the trains starting from the district station to the direction of the peak sentry of the water section to be prolonged from two minutes to 10 minutes, and the duration is 7:30-8: 30. The passenger selects 8:00 to go from daylily sentry on line six to ancestry temple on line one, and the path and real-time road conditions pushed by the passenger are shown in fig. 4.
(2) The information pushing module: under a train delay scene, according to a common OD, a path and travel time set by passengers, and in combination with a delayed time-space influence range, a guidance information prompt for changing the path is prospectively pushed to the affected passengers;
the departure interval of the trains starting from the district and starting to the peak of the waterside is prolonged to 10 minutes from two minutes, and the duration is 7:30-8: 30. As shown in fig. 5, if the passenger sets a frequent travel OD, a route and a departure time in the APP, if the route passes through a delay range and the departure time is also within a delay time, the passenger is actively pushed a route change reminder.
(3) Last bus inquiry module: before the subway operation is finished, showing the reachable condition of the designated station going to other stations in the whole network at the designated time, the reachable condition of the travel path of the designated OD and the latest time for the passenger;
under the last car operation condition, by accessing each route reachable time period of each OD from the outside, as shown in fig. 6, the display of the reachable situation of the designated station heading to other stations in the whole network at the designated time is displayed when the passenger inquires, and the reachable situation and the latest reachable time of each travel route of the designated OD at the designated time are displayed at the same time.
(4) Go out and correct the module: according to APP real-time positioning information of passengers, a travel correction service is provided by combining travel paths of the passengers, and the problems of station passing, misdirection and the like of the passengers are avoided;
as shown in fig. 7, after the recommended travel path is pushed for the passenger, the travel track of the passenger is obtained by combining APP positioning information, and then the actual travel path of the passenger is estimated. And if the actual travel route of the passenger is not in the OD feasible route set, pushing the stop-passing and misdirection reminding for the passenger.
(5) A background interaction module: performing data interaction with a front-end application and a background database, and providing algorithm support;
the background interaction module comprises a road network state prediction module, a feasible path sorting module and a recommendation sorting optimization module.
In the road network state prediction module, the original data of three levels of road network points, lines and surfaces are used as input to calculate corresponding technical indexes, and the threshold values of the corresponding indexes are used for grading point, line and network evaluation indexes, so that the line capability can be evaluated. In addition, passenger ticket card information, a train schedule, train delay information, last bus information and APP data are combined to perform space-time analysis on congestion of the road network, so that road network state prediction is performed.
In the feasible path sequencing module, sensitivity of different passengers to induction information is analyzed by researching influence of the induction information on passenger path selection, and a passenger path selection behavior model based on multi-scene induction information is established by combining AFC, APP and congestion degree data of Guangzhou subway so as to sequence the feasible paths of the passengers in different scenes.
The feasible path sequencing module comprises a passenger path selection computing unit, a congestion degree computing unit and a feasible path sequencing unit;
the passenger path selection computing unit is used for establishing a passenger path selection behavior model with the effectiveness maximized as the target under the multi-scene inducing information;
the congestion degree calculating unit is used for calculating the time when the passenger passes through each section and the transfer station on the effective path, and quantitatively evaluating the overall congestion level of the path based on the congestion condition of each section and the corresponding time of the transfer station to reflect the congestion degree;
the feasible path sorting unit is used for selecting a behavior model based on the passenger path and sorting the feasible paths according to the generalized utility.
In the recommendation sequencing optimization module, after the feasible paths are sequenced, a passenger path selection behavior model and parameters thereof are optimized by combining a reinforcement learning method, so that the optimal induced path generation meeting the personal preference is realized.
(6) A data management module: basic data support is provided for the system, and the basic data support comprises road network prediction data, passenger ticket card information, road network basic data, feasible path set data, a train schedule, an information template, train delay information, last-class train information, APP data and the like, and meanwhile data updating and maintenance work is provided.
In summary, the method and the system provided by the embodiment of the invention are compared and analyzed with the actual operation guidance system of the subway network, and it is verified that the passenger flow accurate guidance method and the system provided by the invention can truly and effectively assist the issuing of guidance information to operators in multiple scenes of urban rail transit, and meanwhile, provide certain help for benign operation of urban rail transit.
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 changes or substitutions that can be easily conceived by those skilled in the art within the technical scope of the present invention are included in the scope of the present invention. Therefore, the protection scope of the present invention shall be subject to the protection scope of the claims.

Claims (9)

1. An accurate passenger flow induction method under multiple scenes of urban rail transit is characterized by comprising the following process steps:
step S110: establishing a passenger path selection behavior model with the maximum utility as a target under multi-scene inducing information;
step S120: calculating the time when the passenger passes through each section and the transfer station on the effective path, and quantitatively evaluating the overall congestion level of the path based on the congestion condition of each section and the corresponding time of the transfer station to reflect the congestion degree;
step S130: selecting a behavior model based on the passenger path, and sorting the feasible paths according to the generalized utility;
step S140: aiming at the sorted feasible paths, optimizing parameters of a path selection behavior model by combining a Q-learning algorithm to obtain an optimal induced path;
the step S110 specifically includes:
the passenger path selection behavior model is a discrete selection model taking effectiveness maximization as a target;
the train running time is calculated as follows:
Figure FDA0003252541710000011
wherein,
Figure FDA0003252541710000012
representing the OD versus the total train time for path k of a to b,
Figure FDA0003252541710000013
representing the train's run time in the interval i-j in the path k,
Figure FDA0003252541710000014
representing the stop time of the train at the station m in the path k;
the waiting time of passengers is as follows:
Figure FDA0003252541710000015
wherein, TwaitThe waiting time is shown as the time of waiting,
Figure FDA0003252541710000016
the train departure interval of the line l is represented;
the passenger transfer time is:
Figure FDA0003252541710000017
wherein,
Figure FDA0003252541710000021
represents the total transfer time of OD to the passenger on the path k from a to b, tau represents the penalty coefficient of the passenger due to transfer, and nkRepresents the total number of transfers of the path k, represents the penalty coefficient of the number of transfers,
Figure FDA0003252541710000022
indicating the i-th transfer of the slave line1To l2The running time of (2);
the congestion perception coefficient of the passenger is as follows:
Figure FDA0003252541710000023
wherein, CijRepresenting the perceptual coefficients of the intervals i-j,
Figure FDA0003252541710000024
is the congestion coefficient, mijIs the full load of the interval i-j, m0Grading the threshold for the first full load, m2Grading a threshold for a second fullness; o isIRepresenting the set of affected regions, KYRepresenting a set of affected passengers;
the generalized travel cost of the passengers is as follows:
Figure FDA0003252541710000025
wherein,
Figure FDA0003252541710000026
represents the travel cost, ξ, of the OD for path k from a to bj,kIs a subordinate mark and indicates whether the section/station is on the path k or not, and xi isj,k＝{0,1}；
In the discrete selection model, the total utility consists of two parts, wherein one part is a fixed utility and is a generalized cost for passengers to go out; the other part is random utility;
Figure FDA0003252541710000027
in the formula:
Figure FDA0003252541710000028
represents the total utility of path k from a to b,
Figure FDA0003252541710000029
representing a random utility portion;
constructing a discrete selection model:
Figure FDA00032525417100000210
in the formula:
Figure FDA0003252541710000031
Figure FDA0003252541710000032
representing the probability of the passenger selecting a route;
Figure FDA0003252541710000033
representing the minimum generalized cost of a through b.
2. The method for accurately inducing passenger flow in multiple urban rail transit scenes according to claim 1, wherein the step S120 specifically comprises:
calculating the congestion degree of the path by taking the high-fullness interval proportion and the fullness distribution entropy as indexes;
the high-full-load-rate interval proportion refers to the ratio of the number of the high-full-load-rate intervals to the number of all the intervals of the urban rail transit network:
Figure FDA0003252541710000034
wherein Z represents the proportion of the high-loading-rate interval within the evaluation time; l represents the number of all intervals in the road network; l ishRepresenting the number of high fullness intervals within the evaluation period;
combining the interval full-load rate and the information entropy to construct a full-load rate distribution entropy index:
Figure FDA0003252541710000035
wherein H represents the full-load distribution entropy within the evaluation time; g represents a discretized value of the interval full load rate, and the discretized value is accurate to 10%, wherein G is 0, 1. L isgL represents the proportion of the section with the full load rate g to all the sections of the road networkg≠0。
3. The method for accurately inducing passenger flow in multiple urban rail transit scenes as claimed in claim 2, wherein said step S130 specifically comprises:
under normal operation conditions, establishing a utility function according to the waiting time, the transfer time and the crowdedness of passengers, calculating generalized cost, and calculating the general cost according to the
Figure FDA0003252541710000036
Sequencing the feasible paths and inducing passengers to reasonably select travel paths;
under the condition of train delay, establishing utility function according to passenger waiting time, transfer time and crowding degree and waiting time caused by delay, calculating generalized costAccording to
Figure FDA0003252541710000037
And sequencing the feasible paths and inducing passengers to reasonably select travel paths.
4. The method for accurately inducing passenger flow in multiple scenes of urban rail transit according to claim 3, wherein the Q-learning algorithm comprises the following steps:
step S141: initializing a Q-table matrix, wherein a row represents the same proportion of a recommended path and an actual path as a state, and a column represents
Figure FDA0003252541710000041
And
Figure FDA0003252541710000042
the congestion coefficient is used as an active set, and each proportion corresponds to a Q-table matrix;
step S142: acquiring states of the recommended path and the actual path in the same proportion, wherein the state set comprises three states: inconsistent, more consistent, and very consistent;
step S143: selecting the one corresponding to the maximum Q value in the Q-table matrix by a greedy algorithm
Figure FDA0003252541710000043
And
Figure FDA0003252541710000044
a path selection behavior of;
step S144: calling positioning data to obtain an actual path of the passenger;
step S145: computing
Figure FDA0003252541710000045
And
Figure FDA0003252541710000046
a varying reward function;
step S146: calculating new state states of the passenger recommended path and the actual path in the same proportion after the next system induction;
step S147: updating the state sum of the recommended path after induction and the actual path of the passenger in the same proportion by using a Bellman Equation iterative formula
Figure FDA0003252541710000047
And
Figure FDA0003252541710000048
the Q-table matrix of (1);
step S148: judging whether the iteration times are reached, if not, corresponding
Figure FDA0003252541710000049
And
Figure FDA00032525417100000410
the new state states with the same proportion of the recommended path and the actual path are taken as the current state, and the step S143 is entered; if the iteration times are reached, terminating the iteration, and outputting the recommended paths corresponding to the actual paths of the passengers in the same proportion
Figure FDA00032525417100000411
And
Figure FDA00032525417100000412
5. the method for accurately inducing passenger flow under multiple scenes of urban rail transit according to claim 4, wherein the method comprises the following steps: the state set S represents the same proportion of the recommended path and the actual path of the passenger, and the activity set a represents the congestion coefficient
Figure FDA00032525417100000413
And
Figure FDA00032525417100000414
Figure FDA00032525417100000415
and
Figure FDA00032525417100000416
the selection probability of (2) is:
Figure FDA00032525417100000417
wherein, pi [ a | st]To represent
Figure FDA00032525417100000418
And
Figure FDA00032525417100000419
the greater the Q value, the greater the selection probability of (2)
Figure FDA00032525417100000420
And
Figure FDA00032525417100000421
the greater the probability of being selected.
6. The method for accurately inducing passenger flow under multiple scenes of urban rail transit according to claim 5, wherein the method comprises the following steps:
the reward function is a passenger path selection behavior model utility function under multiple scenes;
the Q function is:
Figure FDA0003252541710000051
Figure FDA0003252541710000052
the Q-table is updated using the Bellman Equation as follows:
Q(i,sn,a)＝(1-α)Q(i,sn,a)+α[Ri(sn,a)+γ·maxQ(i,sn+1,a)]
in the formula: alpha is belonged to 0,1]The learning rate is represented, and the larger the learning rate is, the more remarkable the effect of the Q-table iterative updating is; gamma is belonged to 0,1]The discount factor is represented, and the larger the discount factor is, the larger the function of the future maximum reward function value is; riShow fixation
Figure FDA0003252541710000053
And
Figure FDA0003252541710000054
the value of the reward earned; MaxQ (i, s)n+1A) represents all of the Q-tables in the proportional state where the next recommended path is the same as the actual path of the passenger
Figure FDA0003252541710000055
And
Figure FDA0003252541710000056
the maximum value among the corresponding values.
7. The utility model provides an accurate induction system of passenger flow under urban rail transit multi-scene which characterized in that includes: the system comprises a road network state prediction module, a feasible path sorting module and a recommendation sorting optimization module;
the road network state prediction module is used for calculating corresponding technical indexes by taking road network original data as input, grading the corresponding technical indexes by using threshold values of the corresponding technical indexes, and performing space-time analysis on congestion of a road network by combining passenger ticket card information, a train schedule, train delay information and last bus information so as to predict the state of the road network;
the feasible path sequencing module is used for analyzing the sensibility of different passengers to the guidance information based on the influence of the guidance information on the passenger path selection, establishing a passenger path selection behavior model based on multi-scene guidance information and sequencing the feasible paths of the passengers in different scenes;
the recommendation sequencing optimization module is used for optimizing a passenger path selection behavior model and parameters thereof by combining a reinforcement learning method after sequencing the feasible paths, so that the generation of an optimal induced path meeting personal preference is realized;
the process of the feasible path ordering module for establishing the passenger path selection behavior model based on the multi-scenario guidance information specifically comprises the following steps:
the passenger path selection behavior model is a discrete selection model taking effectiveness maximization as a target;
the train running time is calculated as follows:
Figure FDA0003252541710000061
wherein,
Figure FDA0003252541710000062
representing the OD versus the total train time for path k of a to b,
Figure FDA0003252541710000063
representing the train's run time in the interval i-j in the path k,
Figure FDA0003252541710000064
representing the stop time of the train at the station m in the path k;
the waiting time of passengers is as follows:
Figure FDA0003252541710000065
wherein, TwaitThe waiting time is shown as the time of waiting,
Figure FDA0003252541710000066
the train departure interval of the line l is represented;
the passenger transfer time is:
Figure FDA0003252541710000067
wherein,
Figure FDA0003252541710000068
represents the total transfer time of OD to the passenger on the path k from a to b, tau represents the penalty coefficient of the passenger due to transfer, and nkRepresents the total number of transfers of the path k, represents the penalty coefficient of the number of transfers,
Figure FDA0003252541710000069
indicating the i-th transfer of the slave line1To l2The running time of (2);
the congestion perception coefficient of the passenger is as follows:
Figure FDA00032525417100000610
wherein, CijRepresenting the perceptual coefficients of the intervals i-j,
Figure FDA00032525417100000611
is the congestion coefficient, mijIs the full load of the interval i-j, m0Grading the threshold for the first full load, m2Grading a threshold for a second fullness; o isIRepresenting the set of affected regions, KYRepresenting a set of affected passengers;
the generalized travel cost of the passengers is as follows:
Figure FDA0003252541710000071
wherein,
Figure FDA0003252541710000072
represents the travel cost, ξ, of the OD for path k from a to bj,kIs a subordinate mark, indicates whether the section/station is on the route k,has xij,k＝{0,1}；
In the discrete selection model, the total utility consists of two parts, wherein one part is a fixed utility and is a generalized cost for passengers to go out; the other part is random utility;
Figure FDA0003252541710000073
in the formula:
Figure FDA0003252541710000074
represents the total utility of path k from a to b,
Figure FDA0003252541710000075
representing a random utility portion;
constructing a discrete selection model:
Figure FDA0003252541710000076
in the formula:
Figure FDA0003252541710000077
Figure FDA0003252541710000078
representing the probability of the passenger selecting a route;
Figure FDA0003252541710000079
representing the minimum generalized cost of a through b.
8. The system for accurately inducing passenger flow under multiple scenes of urban rail transit according to claim 7, wherein the feasible path sequencing module comprises a passenger path selection calculating unit, a congestion degree calculating unit and a feasible path sequencing unit;
the passenger path selection computing unit is used for establishing a passenger path selection behavior model with the effectiveness maximized as the target under the multi-scene inducing information;
the congestion degree calculating unit is used for calculating the time when the passenger passes through each section and the transfer station on the effective path, and quantitatively evaluating the overall congestion level of the path based on the congestion condition of each section and the corresponding time of the transfer station to reflect the congestion degree;
the feasible path sorting unit is used for selecting a behavior model based on the passenger path and sorting the feasible paths according to the generalized utility.
9. The system for accurately inducing passenger flow under multiple scenes in urban rail transit according to claim 8, further comprising: the system comprises a query module, an information push module, a last bus query module, a trip correction module and a data management module;
the query module is configured to: inquiring feasible paths and states thereof according to the appointed starting and stopping point and starting time input by the passenger, and displaying the feasible paths according to the principle of small crowding degree and short running time and less total station number for transfer to order and recommend the feasible paths to the passenger;
the information pushing module is used for: under a train delay scene, according to a common OD, a path and travel time set by a passenger, and in combination with a time-space influence range of delay, pushing guidance information for changing the path to the passenger affected by the delay for reminding;
the last bus query module is used for: displaying the reachable condition of the specified station going to other stations at the specified time and the reachable condition and the latest time of the travel path of the specified OD for the passenger;
the trip correcting module is used for: providing travel correction service according to the real-time positioning information of the passenger and combining the travel path of the passenger;
the data management module is used for: basic data support is provided for the system, and the basic data support comprises road network prediction data, passenger ticket card information, road network basic data, feasible path set data, a train schedule, an information template, train delay information, last-class train information and passenger actual travel path data.
CN201910753308.6A 2019-08-15 2019-08-15 Passenger flow accurate induction method and system under urban rail transit multi-scene Active CN110428117B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201910753308.6A CN110428117B (en) 2019-08-15 2019-08-15 Passenger flow accurate induction method and system under urban rail transit multi-scene

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201910753308.6A CN110428117B (en) 2019-08-15 2019-08-15 Passenger flow accurate induction method and system under urban rail transit multi-scene

Publications (2)

Publication Number Publication Date
CN110428117A CN110428117A (en) 2019-11-08
CN110428117B true CN110428117B (en) 2021-10-22

Family

ID=68416388

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201910753308.6A Active CN110428117B (en) 2019-08-15 2019-08-15 Passenger flow accurate induction method and system under urban rail transit multi-scene

Country Status (1)

Country Link
CN (1) CN110428117B (en)

Families Citing this family (17)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN111539565B (en) * 2020-04-20 2023-04-07 交通运输部科学研究院 Bus fare preferential method based on vehicle and station congestion degree
CN111861841B (en) * 2020-06-30 2024-07-23 南昌轨道交通集团有限公司 Method, device, equipment and storage medium for determining traffic network passenger flow distribution
CN111737826B (en) * 2020-07-17 2020-11-24 北京全路通信信号研究设计院集团有限公司 Rail transit automatic simulation modeling method and device based on reinforcement learning
CN112016008B (en) * 2020-08-27 2024-03-01 广州地铁集团有限公司 Urban rail transit passenger flow accurate induction system under multiple scenes
CN112231870B (en) * 2020-09-23 2022-08-02 西南交通大学 Intelligent generation method for railway line in complex mountain area
CN112183889B (en) * 2020-10-26 2023-06-06 中国联合网络通信集团有限公司 Riding route recommendation method and device
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
CN112949078B (en) * 2021-03-17 2023-12-05 北京交通大学 Matching degree calculation method for urban rail transit passenger flow and traffic flow
CN113537555B (en) * 2021-06-03 2023-04-11 太原理工大学 Traffic sub-region model prediction sliding mode boundary control method considering disturbance
CN113723659B (en) * 2021-06-22 2023-11-21 北京交通大学 Urban rail transit full-scene passenger flow prediction method and system
CN113935595B (en) * 2021-09-28 2023-07-28 北京交通大学 Urban rail transit road network peak large passenger flow dredging system
CN113807026A (en) * 2021-10-08 2021-12-17 青岛理工大学 Passenger flow line optimization and dynamic guide signboard system in subway station and design method
CN114566041B (en) * 2022-01-27 2022-12-06 苏州大学 Multi-mode travel induction method and device for multi-source data fusion urban congested road section
CN114611807B (en) * 2022-03-16 2023-04-21 武汉大学 Construction method of transportation ticket buying transfer recommendation index
CN114923497B (en) * 2022-04-21 2023-07-21 西南交通大学 Method, device, equipment and storage medium for planning path of railway travel
CN114896507B (en) * 2022-05-27 2024-03-19 桂林电子科技大学 Subway path recommendation method based on space-time structure
CN117690301B (en) * 2024-02-04 2024-04-23 福建省高速公路科技创新研究院有限公司 Expressway diversion induction method considering induction compliance rate

Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US9146127B2 (en) * 2012-06-27 2015-09-29 International Business Machines Corporation Navigation system providing lane guidance to driver based on driver's driving habits and preferences
CN107194497A (en) * 2017-04-27 2017-09-22 北京交通大学 Urban track traffic passenger trip route planing method under a kind of accident

Patent Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US9146127B2 (en) * 2012-06-27 2015-09-29 International Business Machines Corporation Navigation system providing lane guidance to driver based on driver's driving habits and preferences
CN107194497A (en) * 2017-04-27 2017-09-22 北京交通大学 Urban track traffic passenger trip route planing method under a kind of accident

Non-Patent Citations (2)

* Cited by examiner, † Cited by third party
Title
基于D2D信息发布及满载率饱和有效路径集模型的新型客流诱导***;颜开;《中国优秀硕士学位论文全文数据库 工程科技Ⅱ辑》;20150315(第03期);全文 *
轨道交通乘客个性化出行路径规划算法;刘莎莎等;《交通运输***工程与信息》;20141031(第5期);全文 *

Also Published As

Publication number Publication date
CN110428117A (en) 2019-11-08

Similar Documents

Publication Publication Date Title
CN110428117B (en) Passenger flow accurate induction method and system under urban rail transit multi-scene
CN110428096B (en) Ticket information-based urban rail transit multi-traffic-road transportation organization optimization method
Gutiérrez et al. Transit ridership forecasting at station level: an approach based on distance-decay weighted regression
Bhatta et al. Errors in variables in multinomial choice modeling: A simulation study applied to a multinomial logit model of travel mode choice
CN106127357A (en) A kind of customization public transport automatic routing system based on reservation data and method
Zhu et al. The potential of ride-pooling in VKT reduction and its environmental implications
CN103366224B (en) Passenger demand prediction system and method based on public transport network
JP5273106B2 (en) Traffic flow calculation device and program
CN116720997A (en) Bus route evaluation system and optimization method based on big data analysis
Kim et al. Effect of taxi information system on efficiency and quality of taxi services
JP2020135231A (en) Traffic demand prediction device and traffic demand prediction system
CN116070033A (en) Novel shared public transportation transfer demand estimation method based on mobile phone signaling data
Saha et al. Network model for rural roadway tolling with pavement deterioration and repair
Wu et al. Time-dependent customized bus routing problem of large transport terminals considering the impact of late passengers
Ning et al. Robust and resilient equilibrium routing mechanism for traffic congestion mitigation built upon correlated equilibrium and distributed optimization
Jung et al. Assessment of the transit ridership prediction errors using AVL/APC data
Liu et al. Modeling the effects of population density on prospect theory-based travel mode-choice equilibrium
Cui et al. Dynamic pricing for fast charging stations with deep reinforcement learning
Feng et al. Choices of intercity multimodal passenger travel modes
Wang et al. A simulation-based metro train scheduling optimization incorporating multimodal coordination and flexible routing plans
JP2020160960A (en) Movement support system and method
Zhen et al. Vehicle routing for customized on-demand bus services
CN115495701A (en) Method for predicting time-space distribution of charging load of consistent electric automobile
CN105023063A (en) Establishing method of public transport network new energy bus operation energy consumption index system
Dui et al. Simulations for urban taxi sharing system on routes and passengers with numerical experiments

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
GR01 Patent grant
GR01 Patent grant