Information induction method and system considering dynamic behavior of urban rail passengers under emergency

Abstract

The application provides an information induction method and system considering passenger dynamic behavior under an emergency, belonging to the technical field of rail transit operation management, considering passenger waiting process, dynamically updating personalized heterogeneous induction information, constructing a multi-stage dynamic behavior model to observe Markov decision changes of passengers, defining different affected passenger flows under the emergency and constructing an alternative travel scheme set for the passenger flows, estimating dynamic passenger flow demand under an induction time unit, constructing a dynamic passenger flow induction information release scheme optimization model based on double-layer planning, outputting an optimal recommended scheme of time-sharing passenger flow induction information, realizing line network section passenger flow equilibrium and passenger generalized travel cost minimization. The application discloses a dynamic behavior mechanism of passengers under an urban rail operation emergency, analyzes dynamic evolution characteristics of demand in a line network over time in combination with an alternative travel scheme set, takes into account operation efficiency and safety concepts, realizes accurate optimization of an induction information release scheme effect and develops a corresponding system.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of rail transit operation management, and in particular to an information induction method and system considering dynamic behavior of passengers in a city rail under an emergency. BACKGROUND

[0002] As an important part of urban public transportation, developing city rail is a good strategy to solve urban traffic congestion and traffic pollution. However, due to increasing travel demand, aging infrastructure, and long-term high-load operation, city rail emergencies occur from time to time, seriously affecting the stability of the public transportation system and passenger safety. Passenger flow induction, as a flexible demand management measure, can influence passenger behavior patterns by issuing personalized heterogeneous induction information, adjust the distribution of city rail network passenger flow, and to some extent, alleviate the impact of emergencies. Therefore, under city rail emergencies, how to develop a reasonable and effective induction information release plan is a problem to be solved. It is necessary to analyze the dynamic travel behavior characteristics and preferences of passengers in this scenario in depth and develop an induction information release strategy to provide theoretical support for city rail emergency management.

[0003] On the technical research level, existing technologies often take the design of induction information schemes in the field of road traffic or under the condition of city rail regular operation as the object. However, considering the great difference between city rail network and road network, there are many differences between city rail regular operation and emergency. Therefore, it is necessary to explore how to design information induction strategies under city rail operation emergencies and their impact on passenger travel choice behavior. On the other hand, although existing technologies analyze the static behavior mechanism of passengers under city rail emergencies from different angles. However, passengers under city rail emergencies should be in a continuous evaluation and dynamic decision-making process. Static behavior analysis cannot reflect the dynamic behavior process of passengers and may ignore the Markov behavior dependence relationship of passengers. In addition, passengers under city rail emergencies will be affected to different degrees. The current homogenization induction cannot provide a scientific and reasonable travel plan for passengers. It is necessary to construct different sets of alternative travel plans according to the types of passengers affected and set personalized heterogeneous induction information. SUMMARY

[0004] The present application aims to provide an information induction method and system considering dynamic behavior of passengers in a city rail under an emergency, which analyzes the dynamic evolution characteristics of demand online network over time based on the dynamic behavior of passengers and their alternative travel plan set in the city rail emergency scenario, taking into account the operation efficiency and safety concept, and develops a personalized heterogeneous information induction recommendation scheme to solve at least one of the technical problems existing in the background technology.

[0005] To achieve the above-mentioned purpose, the present application adopts the following technical solutions:

[0006] In a first aspect, the present invention provides an information induction method that considers the dynamic behavior of urban rail passengers during an emergency, comprising:

[0007] Considering the passenger waiting process, personalized heterogeneous induction information is dynamically released, and the dynamic behavior selection model of passengers under operational emergencies is constructed using behavioral sample data;

[0008] Search for efficient routes between all origins and destinations of the urban rail network during emergencies, define different types of affected passenger flows, and construct a set of alternative travel plans for them. Simultaneously, based on dynamic passenger behavior feedback, deduce the evolution of passenger flows at the network level and estimate dynamic passenger flow demand under induced time units.

[0009] A dynamic passenger flow induction information release scheme optimization model based on double-layer planning is constructed to output the optimal time-sharing scheme for induction information. The optimization model for dynamic passenger flow induction information release scheme based on double-layer planning takes the most balanced passenger flow distribution in the line network section and the minimum generalized travel cost of passengers as the optimization objectives. The upper layer is the dual-objective optimization model of the induction information release scheme, and the lower layer is the passenger flow distribution model under urban rail emergencies.

[0010] Furthermore, the waiting process of passengers is considered, personalized heterogeneous induction information is dynamically released, and the behavioral sample data is used to construct a dynamic behavioral selection model of passengers under operational emergencies, including: modeling the dynamic behavioral process of passengers under urban rail operational emergencies through a dynamic discrete choice model, and studying the alternative travel choices made by passengers at any time during continuous waiting; passengers will form behavioral experiential inertia in multi-stage decision-making, and the impact of passenger choice experiential inertia and endogenous preferences on behavior is determined by constructing the Markov dependency relationship of behavior, and the state dependence and serial correlation problems existing in dynamic decision-making are solved by changing the structure of the dynamic behavior model.

[0011] Furthermore, the system searches for efficient routes between all origins and destinations of the urban rail network under emergencies, defines different types of affected passenger flows, and constructs a set of alternative travel plans for them. Furthermore, based on the dynamic behavior feedback of passengers, the system deduces the evolution of passenger flows at the network level and estimates the dynamic passenger flow demand under the induced time unit, including:

[0012] A graph-based depth-first search algorithm is selected to search for effective paths. The search rules with time threshold and transfer number threshold are set. The depth traversal is performed starting from the unvisited adjacent points of a vertex in turn until all points with paths connecting to the vertex are visited. All vertices are marked and traversed to generate effective paths between each OD pair.

[0013] After an urban rail operation emergency occurs, the affected passenger flow is defined based on the different impacts on passengers due to their locations;

[0014] Determine the alternative travel options in the alternative plan set according to the different degrees of impact on the passengers;

[0015] Based on the card swiping data of the urban rail automatic ticket vending and checking system, the passenger travel volume between stations is identified through data mining. At the same time, based on the alternative travel plan sets and dynamic behavior feedback of passengers with different degrees of impact, the evolution law of passenger flow at the line network level is deduced, and the dynamic passenger flow demand under the induced time unit is estimated.

[0016] Furthermore, the affected passenger flows are defined as follows: Category 1: both the starting and ending stations are located in the interruption interval; Category 2: only the starting station is located in the interruption interval; Category 3: only the ending station is located in the interruption interval; Category 4: one or more valid paths of the passenger pass through the interruption interval; Category 5: all valid paths of the passenger do not pass through the interruption interval.

[0017] Furthermore, the alternative travel plans in the alternative plan set are determined: the first category: if the passenger's starting and ending stations are both located in the interruption interval, then waiting for the original route to be restored, bus travel and taxi travel are taken as the alternative travel plan set; the second category: if only the passenger's starting station is located in the interruption interval, then waiting for the original route to be restored, changing the entry station to station A, changing the entry station to station B, bus travel and taxi travel are taken as the alternative travel plan set; the third category: if only the passenger's ending station is located in the interruption interval, then changing the exit station to station A, changing the exit station to station B, bus travel and taxi travel are taken as the alternative travel plan set; the fourth category: if one or more valid paths of the passenger pass through the interruption interval, then waiting for the original route to be restored, urban rail internal detour path 1, urban rail internal detour path 2, bus travel and taxi travel are taken as the alternative travel plan set; the fifth category: if all the passenger's valid paths do not pass through the interruption interval, then the valid path set between OD pairs is taken as the alternative travel plan set.

[0018] Furthermore, the travel plan alternative sets for the affected types of passengers are also different. Therefore, plan recommendation constraints are set: with the goals of minimizing the Theil index of the section full load rate and minimizing the generalized travel cost of passengers, a dual-objective optimization model for the upper-level induction information release plan is constructed. The decision variables of the dual-objective optimization model for the upper-level induction information release plan are the induction information recommendation plans released by each time unit, including: no recommendation, recommending to maintain the original travel route of the urban rail, recommending the first detour route within the urban rail, recommending the second detour route within the urban rail, recommending to change the entry / exit station to station A, recommending to change the entry / exit station to station B, and recommending to switch to ground transportation.

[0019] Further, the constraints including decision state constraints, travel impedance constraints, cross-section full load constraints, expected waiting time constraints and non-negative constraints are considered, the utility function in the dynamic travel selection model is taken as the basis for passenger flow allocation, and a lower-layer passenger flow allocation model under the urban rail emergency is constructed based on the passenger flow between the urban rail sites and the passenger demand evolution characteristics; meanwhile, the allocation result of the lower-layer passenger flow allocation model under the urban rail emergency is fed back to the upper-layer double-target optimization model of the induction information publishing scheme, and the optimal induction scheme is obtained through repeated iteration between the upper-layer double-target optimization model of the induction information publishing scheme and the lower-layer passenger flow allocation model under the urban rail emergency.

[0020] Further, the dynamic passenger flow induction information publishing scheme optimization model of the double-layer planning is solved by a hybrid algorithm of the fast non-dominated sorting genetic algorithm with an elitist strategy and the successive weighted average method, the decision variables are dynamically adjusted within the feasible region, the objective function of the upper-layer double-target optimization model of the induction information publishing scheme is taken as the fitness function in the fast non-dominated sorting genetic algorithm with an elitist strategy, the successive weighted average method is used to solve the stochastic user equilibrium allocation solution, and the information induction recommendation scheme is output.

[0021] Further, the process of the fast non-dominated sorting genetic algorithm with an elitist strategy comprises determining genetic algorithm parameters, initializing a population, generating a first generation of sub-populations, elite selection, generating a new parent population, generating a new sub-population, and finally outputting the passenger flow induction information recommendation scheme to the lower-layer passenger flow allocation model.

[0022] The process of the successive weighted average method algorithm comprises pre-preparation, calculation of the selection probability of each scheme, passenger flow loading, utility function updating, passenger flow iterative calculation and convergence discrimination; the travel selection demand in each induction period is sorted and the passenger flow full load rate of each cross-section is calculated, and the sorted result is input into the upper-layer double-target optimization model, and the upper-layer objective function is solved.

[0023] In the second aspect, the application provides an information induction system considering dynamic behaviors of urban rail passengers under an emergency, comprising:

[0024] A construction module is configured to consider a passenger waiting process, dynamically publish personalized heterogeneous induction information, and construct a dynamic behavior selection model of passengers under an operation emergency by using behavior sample data;

[0025] A definition module is configured to search for effective paths between all origin-destination points of an urban rail network under an emergency, define passenger flows of different affected types and construct a set of alternative travel schemes for the passenger flows, and estimate dynamic passenger flow demands under an induction time unit based on feedback and deduction of evolution rules of passenger flows at the network level.

[0026] The solving module is configured to construct a dynamic passenger flow induction information publishing scheme optimization model based on a bi-level programming, and output a time-division optimal scheme of the induction information, wherein the dynamic passenger flow induction information publishing scheme optimization model based on the bi-level programming takes the most balanced line network section passenger flow distribution and the minimum passenger generalized travel cost as optimization objectives, wherein the upper layer is a double-objective optimization model of the induction information publishing scheme, and the lower layer is a passenger flow distribution model under the urban rail emergency.

[0027] In a third aspect, the present application provides a non-transitory computer readable storage medium for storing computer instructions, which, when executed by a processor, implement the information induction method under the emergency considering the dynamic behavior of urban rail passengers as described in the first aspect.

[0028] In a fourth aspect, the present application provides a computer device comprising a memory and a processor, the processor and the memory being in communication with each other, the memory storing program instructions executable by the processor, and the processor invoking the program instructions to execute the information induction method under the emergency considering the dynamic behavior of urban rail passengers as described in the first aspect.

[0029] In a fifth aspect, the present application provides an electronic device comprising a processor, a memory and a computer program, wherein the processor is connected with the memory, the computer program is stored in the memory, and when the electronic device is running, the processor executes the computer program stored in the memory to enable the electronic device to execute instructions for implementing the information induction method under the emergency considering the dynamic behavior of urban rail passengers as described in the first aspect.

[0030] The present application has the following advantages: by defining the affected passenger flow under the urban rail emergency, exploring the passenger dynamic behavior mechanism under the induction information publishing, and researching the induction information publishing scheme generation method, the present application is helpful to balance the urban rail line network passenger flow distribution, improve the comprehensive operation benefit and transportation resource utilization rate, and relieve the traffic congestion and resource waste caused by the emergency; meanwhile, the present application is helpful to improve the travel benefit and satisfaction of passengers under the urban rail operation emergency.

[0031] The advantages of the additional aspects of the present application will be more apparent from the following description part or be understood through the practice of the present application. BRIEF DESCRIPTION OF DRAWINGS

[0032] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the following will briefly introduce the drawings needed to be used in the embodiment description. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can also be obtained by those skilled in the art without any creative effort on the basis of these drawings.

[0033] Figure 1The flow chart of the information induction scheme generation method considering passenger dynamic behavior under the urban rail operation emergency event according to the embodiment of the present application.

[0034] Figure 2 The schematic diagram of passengers affected by different degrees under the urban rail operation emergency event according to the embodiment of the present application.

[0035] Figure 3 The algorithm flow chart of affected passenger flow definition and alternative travel scheme set construction according to the embodiment of the present application.

[0036] Figure 4 The mixed algorithm flow chart of the successive weighted average algorithm and the genetic algorithm for solving the dynamic passenger flow induction information publishing scheme optimization model based on the double-layer planning according to the embodiment of the present application. DETAILED DESCRIPTION

[0037] The embodiments of the present application are described in detail below, and examples of the embodiments are shown in the drawings, wherein the same or similar notations represent the same or similar elements or elements having the same or similar functions throughout. The embodiments described below through the drawings are exemplary and are only used to explain the present application, and cannot be explained as a limitation of the present application.

[0038] Those skilled in the art can understand that, unless otherwise defined, all terms (including technical terms and scientific terms) used herein have the same meaning as that generally understood by those skilled in the art in the field of the present application.

[0039] It should also be understood that terms such as those defined in general dictionaries should be understood as having meanings consistent with those in the prior art, and will not be interpreted with idealized or overly formal meanings unless defined as such.

[0040] Those skilled in the art can understand that, unless otherwise stated, the singular forms "a", "an" and "the" used herein also include the plural forms. It should be further understood that the phrase "comprising" used in the specification of the present application means that the features, integers, steps, operations, elements and / or groups exist, but does not exclude the presence or addition of one or more other features, integers, steps, operations, elements and / or groups thereof.

[0041] In the description of this specification, reference to the terms "one embodiment," "some embodiments," "examples," "specific examples," or "some examples" means that the specific features, structures, materials, or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present invention. Moreover, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in any one or more embodiments or examples. Those skilled in the art may combine and integrate different embodiments or examples described in this specification, as well as features of different embodiments or examples, unless otherwise contradictory.

[0042] To facilitate understanding of the present invention, the present invention is further explained below with reference to specific embodiments in conjunction with the accompanying drawings, and the specific embodiments do not constitute a limitation on the embodiments of the present invention.

[0043] Those skilled in the art should understand that the drawings are merely schematic diagrams of embodiments, and the components in the drawings are not necessarily necessary for implementing the present invention.

[0044] Example 1

[0045] In this embodiment 1, an information guidance system that considers the dynamic behavior of urban rail passengers during emergencies is first provided. The system includes: a construction module for considering the passenger waiting process, dynamically publishing personalized heterogeneous guidance information, and using behavioral sample data to construct a dynamic behavior selection model for passengers during operational emergencies. A definition module is used to search for valid paths between all starting and ending points of the urban rail network during emergencies, define different types of affected passenger flows, and construct a set of alternative travel plans for them. At the same time, based on passenger dynamic behavior feedback, it deduces the evolution of passenger flow at the network level and estimates the dynamic passenger flow demand within the guidance time unit. A solution module is used to construct an optimization model for the dynamic passenger flow guidance information release scheme based on a two-level planning process and output an optimal time-sharing scheme for the guidance information. The optimization model for the dynamic passenger flow guidance information release scheme based on a two-level planning process aims to achieve the most balanced passenger flow distribution across the network section and minimize the generalized travel cost for passengers. The upper layer is a dual-objective optimization model for the guidance information release scheme, and the lower layer is a passenger flow distribution model for urban rail emergencies.

[0046] In this embodiment 1, the above-mentioned system is used to implement an information induction method that takes into account the dynamic behavior of urban rail passengers under emergency events, including: considering the passenger waiting process, dynamically publishing personalized heterogeneous induction information, and using behavioral sample data to construct a dynamic behavior selection model for passengers under operational emergencies; searching for effective paths between all starting and ending points of the urban rail network under emergency events, defining passenger flows of different affected types and constructing a set of alternative travel plans for them, and at the same time, deducing the evolution law of passenger flow at the network level based on passengers' dynamic behavior feedback, and estimating the dynamic passenger flow demand under the induction time unit; constructing a dynamic passenger flow induction information release scheme optimization model based on double-layer planning, and outputting the optimal time-sharing scheme for induction information; wherein, the dynamic passenger flow induction information release scheme optimization model based on double-layer planning has the most balanced passenger flow distribution in the network section and the minimum generalized travel cost of passengers as optimization objectives, wherein the upper layer is a dual-objective optimization model of the induction information release scheme, and the lower layer is a passenger flow distribution model under urban rail emergencies.

[0047] In this example, a dynamic discrete choice model is used to model the dynamic behavior of passengers during urban rail transit operational emergencies. The study examines the alternative travel choices (internal detours, station changes, exits, or resuming the original route) that passengers make at any time during the waiting period. Furthermore, passengers develop behavioral inertia during multi-stage decision-making. By constructing a Markov dependency relationship for this behavior, the impact of this inertia and endogenous preferences on passenger behavior is explored. The dynamic behavior model structure is modified to address the state dependency and serial correlation issues inherent in dynamic decision-making.

[0048] In this embodiment, valid routes between all starting and ending points of the urban rail network under emergencies are searched, different types of affected passenger flows are defined, and a set of alternative travel plans is constructed for them. Passenger travel between stations at different time periods is counted using the card swiping data of the urban rail automatic ticket vending and checking system. At the same time, the evolution of passenger flow at the line level is deduced based on the dynamic behavior feedback of passengers, and the dynamic passenger flow demand under the induced time unit is estimated; this includes:

[0049] S21: Select a graph-based depth-first search algorithm to search for valid paths. Set search rules based on time thresholds and transfer count thresholds. Start from a vertex's unvisited neighbors and perform a depth traversal until all points connected to the vertex are visited. Mark and traverse all vertices to generate valid paths between each OD pair.

[0050] S22: After an emergency occurs in urban rail operations, passengers are affected differently due to their different locations. The affected passenger flows are defined as follows: Category 1: both the starting and ending stations are located in the interruption interval; Category 2: only the starting station is located in the interruption interval; Category 3: only the ending station is located in the interruption interval; Category 4: one or more valid paths of the passenger pass through the interruption interval; Category 5: all valid paths of the passenger do not pass through the interruption interval.

[0051] S23: According to the different degrees of impact on passengers, the alternative travel plans in their alternative plan set include: Category 1: If the passenger's starting and ending stations are both located in the interruption interval, then waiting for the original route to be restored, bus travel and taxi travel will be taken as their alternative travel plan set; Category 2: If only the passenger's starting station is located in the interruption interval, then waiting for the original route to be restored, changing the entry point to station A, changing the entry point to station B, bus travel and taxi travel will be taken as their alternative travel plan set; Category 3: If only the passenger's ending station is located in the interruption interval, then changing the exit point to station A, changing the exit point to station B, bus travel and taxi travel will be taken as their alternative travel plan set; Category 4: If one or more valid paths of the passenger pass through the interruption interval, then waiting for the original route to be restored, urban rail internal detour path 1, urban rail internal detour path 2, bus travel and taxi travel will be taken as their alternative travel plan set; Category 5: If all of the passenger's valid paths do not pass through the interruption interval, then the valid path set between OD pairs will be taken as their alternative travel plan set.

[0052] S24: Based on the card swipe data from the urban rail automatic ticket vending and checking system, data mining is used to identify passenger travel volume between stations. Simultaneously, based on the alternative travel plans of passengers with different degrees of impact and dynamic behavior feedback, the evolution of passenger flow at the line network level is deduced, and dynamic passenger flow demand under the induced time unit is estimated.

[0053] With the optimization objectives of achieving the most balanced passenger flow distribution across the network and minimizing the generalized travel cost of passengers, a dynamic passenger flow induction information release optimization model based on a two-layer planning approach is constructed to output the optimal time-sharing scheme for induction information. The upper layer is a two-objective optimization model for the induction information release scheme, and the lower layer is a passenger flow allocation model under urban rail emergencies. This includes:

[0054] A two-level planning model is designed to integrate feedback between passenger flow induction schemes and dynamic passenger flow demand. The upper-level planning model is a dual-objective optimization model for the dynamic passenger flow induction information release scheme, while the lower-level planning model is a passenger flow allocation model under emergencies based on the constructed dynamic travel choice model.

[0055] Considering that only one recommended plan can be issued to the same OD pair within the same time unit, and the travel plan alternative sets of different affected types of passengers are also different, the plan recommendation constraints are set. With the goals of minimizing the Theil index of the section load rate and minimizing the generalized travel cost of passengers, a dual-objective optimization model for the upper-level passenger flow induction information release plan is constructed. The decision variables of the upper-level model are the induction information recommendation plans issued by each time unit, including: no recommendation, recommending to maintain the original travel route of the urban rail, recommending the first detour route within the urban rail, recommending the second detour route within the urban rail, recommending to change the entry (exit) station to station A, recommending to change the entry (exit) station to station B, and recommending to switch to ground transportation.

[0056] The constraints include decision state constraints, travel impedance constraints, cross-section full load constraints, expected waiting time constraints, and non-negative constraints. The utility function in the dynamic travel selection model is used as the basis for passenger flow allocation. A lower-level passenger flow allocation model is constructed based on the passenger flow between the urban rail station and the passenger demand evolution characteristics. The allocation results are fed back to the upper model, and the optimal induction scheme is obtained through repeated iteration.

[0057] The double-layer planning model is solved by a hybrid algorithm of the fast non-dominated sorting genetic algorithm with an elitist strategy (NSGA-II) and the method of successive weighted averages (MSWA). The decision variables are dynamically adjusted within the feasible region. The upper-layer objective function is used as the fitness function in the NSGA-II algorithm. The MSWA algorithm is used to solve the stochastic user equilibrium allocation solution. The information induction recommendation scheme is output.

[0058] The NSGA-II algorithm process includes determining genetic algorithm parameters, initializing the population, generating the first generation of sub-population, elite selection, generating a new parent population, and generating a new sub-population. Finally, the passenger flow induction information recommendation scheme is output to the lower-level passenger flow allocation model.

[0059] The MSWA algorithm process includes pre-preparation, calculation of the probability of each scheme being selected, passenger flow loading, utility function updating, passenger flow iteration calculation, and convergence discrimination. The travel selection demand in each induction period is sorted and the passenger flow full load rate of each cross-section is calculated. The sorted results are input into the upper double-objective optimization model, and the upper objective function is solved.

[0060] Embodiment 2

[0061] In this embodiment 2, a method for information induction considering the dynamic behavior of urban rail passengers under emergency is provided. The passenger demand evolution characteristics are considered when inducing passenger flow under urban rail operation emergencies, and the passenger behavior demand is accurately controlled to improve the travel service level under emergencies. By developing a more scientific, efficient, and fair personalized information induction scheme, the coordination and effectiveness of urban rail network operation management under emergencies are ensured.

[0062] As shown in Figure 1 The method for information induction considering the dynamic behavior of urban rail passengers under emergency includes the following processing steps:

[0063] Step S1: Considering the passenger waiting process, we dynamically release personalized, heterogeneous inductive information and use behavioral sample data to construct a dynamic behavioral choice model for passengers during operational emergencies. Driven by dynamic inductive information, we analyze the Markov dynamic decision-making process generated by passengers during multi-stage continuous evaluation, study the impact of choice experience inertia and endogenous preferences on behavior, and address the state dependence and serial correlation issues in dynamic decision-making.

[0064] Step S1 may include: constructing the dynamic behavior process of passengers under urban rail operation emergencies by considering personal socioeconomic characteristics, travel characteristics, and program variables, and studying the alternative travel choices (internal detour, station change, exit, or restoration of the original route) made by passengers at any time during continuous waiting. Based on the utility maximization theory, a dynamic discrete choice model (DDCM) is constructed to quantify passengers' behavioral preferences. In addition, passengers will form behavioral experience inertia in multi-stage decision-making. By constructing the Markov dependency relationship of behavior, the impact of passengers' choice experience inertia and endogenous preferences on behavior is explored. The structure of the dynamic behavior model is changed to solve the problems of state dependence and serial correlation in dynamic decision-making.

[0065] The mathematical representation of the DDCM is as follows:

[0066] According to the utility maximization theory, it is assumed that traveler n has a choice set I, where i∈I. The utility of option i is U in , the utility consists of fixed utility and random utility: U in =V in +ε in ; Among them, V in is the fixed utility; ε in is a random utility.

[0067] The above formula is the general form of the static discrete choice model without considering the time state. Based on the above formula, the time state t is introduced to describe the dynamic Markov decision process of the traveler: U int =V int +ε int Among them, U int V is the utility of traveler n choosing option i at time t; int is the fixed utility at time t; ε int is the random utility at time t.

[0068] In addition, since the current utility level is partially determined by previous choices of the same type, and travelers have certain endogenous preference factors that do not change over time, the Wooldridge correction method, taking into account lagged variables and agency effects, reveals the state dependence and serial correlation in travelers' decisions: in, is a lagged variable, measuring the impact of traveler n's choice of option i at time t-1 on his utility at time t; in is the agency effect, which measures the endogenous long-term preference of traveler n that does not change over time and is composed of the initial decision and exogenous explanatory variables; i ' nt are the remaining unobserved short-term random factors.

[0069] Furthermore, there are:

[0070]

[0071] Among them, y in0 For the first selection; y in(t-1) is the choice at time t-1; y int is the choice under the state at time t; is an exogenous explanatory variable; P(y int |y in0 ,y in(t-1) ,α n ) is the probability of the logit model modified with agent effects considering the first choice and the previous choice; Follows a normal distribution.

[0072] Step S2: Search for valid routes between all origins and destinations in the urban rail network under the emergency, define different types of affected passenger flows, and construct a set of alternative travel plans for them. Passenger travel between stations during different time periods is counted using data from the urban rail automatic ticket vending and checking system. Simultaneously, based on passenger dynamic behavior feedback, the evolution of passenger flow at the network level is deduced, and dynamic passenger flow demand under the induced time unit is estimated.

[0073] The S2 may include:

[0074] S21: Searching for a valid path based on the graph's depth-first search algorithm, including the following steps:

[0075] (1) Construct the urban rail network and select OD pairs. (2) Use the shortest path algorithm to solve the shortest travel time and the minimum number of transfers between OD pairs, and set the search rules for the time threshold and the transfer number threshold. (3) Mark the starting point of a certain OD pair as the current node. Starting from the current node, traverse all unmarked adjacent points and select one to mark. (4) If the travel time from the starting point to the adjacent point along the marked path is less than the time threshold, and the number of transfers is less than the transfer number threshold, then set the adjacent point as the current node and proceed to step (5), otherwise proceed to step (6). (5) If the current node is the end point of the OD pair, proceed to step (6), otherwise proceed to step (3). (6) Traverse the starting points of all OD pairs. If all are marked, the algorithm ends. Otherwise, proceed to step (3).

[0076] S22: Figure 2 As shown in the figure, after an emergency occurs in urban rail operation, passengers are affected differently due to their different locations, and the affected passenger flows are defined as follows: Category 1: both the starting and terminal stations are located in the interruption interval; Category 2: only the starting station is located in the interruption interval; Category 3: only the terminal station is located in the interruption interval; Category 4: one or more valid paths of the passenger pass through the interruption interval; Category 5: all valid paths of the passenger do not pass through the interruption interval.

[0077] S23: Based on the different degrees of impact on passengers, construct alternative travel plans in their alternative plan set:

[0078] according to Figure 3 The algorithm flowchart shown constructs a set of alternative travel plans for passengers of different affected types, including the following steps: (1) First solve the valid path set between all OD pairs according to S21. (2) For a certain OD pair, comprehensively consider the travel time, waiting time, transfer time and on-board time to deduce the time points when the passenger passes through each subway station. (3) Identify the affected type of passengers and construct the alternative travel plan set: ① Determine whether the passenger will pass through the service interruption interval caused by the emergency. If the entire trip does not pass through it, its "affected type" is marked as "5" (the valid path does not pass through the interruption interval), and the valid path set solved by S21 is used as its alternative travel plan set, and proceed to step (4). Otherwise, proceed to steps ② and ③. ② If the passenger's travel starting station is in the interruption interval, determine whether the passenger's travel terminal station is in the interruption interval. If so, the passenger's travel starting and ending stations are both in the interruption interval, and the "affected type" is marked as "1" (the starting and ending stations are both in the interruption interval). At the same time, construct the alternative travel plan set for this type of passenger. Then proceed to step (4). If not, only the starting station of the trip is in the interruption interval, and the "affected type" is marked as "2" (only the starting station is in the interruption interval). At the same time, a set of alternative travel plans for this type of passengers is constructed. Then proceed to step (4). ③ If the passenger's starting station is not in the interruption interval, determine whether the passenger's travel terminal is in the interruption interval. If so, only the terminal is in the interruption interval, and the "affected type" is marked as "3" (only the terminal is in the interruption interval). At the same time, a set of alternative travel plans for this type of passengers is constructed. Then proceed to step (4). If not, there is a valid path between the OD pair passing through the interruption interval, and the "affected type" is marked as "4" (the valid path passes through the interruption interval). At the same time, a set of alternative travel plans for this type of passengers is constructed. Then proceed to step (4). (4) If all OD information is read, the affected type of all passengers is determined, and the set of alternative travel plans is constructed. Otherwise, proceed to step (2).

[0079] S24: Based on the card swipe data from the urban rail automatic ticket vending and checking system, data mining is used to identify passenger travel volume between stations. Simultaneously, based on the alternative travel plans of passengers with different degrees of impact and dynamic behavior feedback, the evolution of passenger flow at the line network level is deduced, and dynamic passenger flow demand under the induced time unit is estimated.

[0080] Step S3: With the optimization objectives of achieving the most balanced passenger flow distribution across the network and minimizing the generalized travel cost for passengers, a dynamic passenger flow guidance information release optimization model based on a two-layer planning framework is constructed to output the optimal time-sharing guidance information scheme. The upper layer is the dual-objective optimization model for the guidance information release scheme, and the lower layer is the passenger flow allocation model under urban rail emergencies.

[0081] The step S3 may include:

[0082] S31: Considering the constraints of the recommended scheme, with the goals of minimizing the Theil index of the section load factor and minimizing the generalized travel cost of passengers, a dual-objective optimization model for the upper-level passenger flow induction information release scheme is constructed.

[0083] The objective function of the upper passenger flow induction information release model is:

[0084]

[0085] Among them, Z1 is the Theil index of the section load rate; T is the control period set, t∈T; L is the urban rail line set, l∈L; E is the urban rail section set, E l is the cross-section set of the l-th urban rail line, e∈E; It is the sum of the full load rates of all sections of the urban rail; is the sum of the full load rates of all sections in the lth urban rail line; n is the total number of sections in the urban rail network; n l is the number of sections in the lth urban rail line; is the full load rate of the section l,e in the lth urban rail line.

[0086]

[0087] Where Z2 is the total generalized travel cost of passengers; W is the set of OD pairs, w∈W, W1, W2, W3, W4, W5 represent the OD pairs of affected types 1-5 respectively; R is the set of alternative options, r∈R; q w,r (t) is the passenger flow of the OD pair w in period t, choosing alternative r; X w,r,g (t) is the g-th attribute value of alternative r in OD pair w during time period t; μ r,g is the generalized travel cost corresponding to the g-th attribute value of alternative r.

[0088] The decision variables of the upper passenger flow induction information release model are the induction information released by each time unit: δ w (t) = [δ w,0 (t),δ w,1 (t),…,δ w,6 (t)],t∈T guidance ,w∈W; where T guidance is the passenger flow induction period; δ w (t) is the recommended travel plan issued to OD pair w during period t; δ w,r (t) is the time period t, whether the travel plan r between OD pair w is recommended, if it is recommended, δ w,r (t) = 1, and the information recommendation dummy variable in the corresponding plan utility function takes the value of 1, otherwise it takes the value of 0. Specifically: r = 0 means that the travel plan is not recommended; r = 1 means that it is recommended to maintain the original urban rail travel route; r = 2 means that it is recommended to take the urban rail internal detour route 1; r = 3 means that it is recommended to take the urban rail internal detour route 2; r = 4 means that it is recommended to change the entry (exit) station to station A; r = 5 means that it is recommended to change the entry (exit) station to station B; r = 6 means that it is recommended to exit the urban rail system and take ground transportation instead.

[0089] The constraints of the upper passenger flow induction information release scheme model are scheme recommendation constraints, where:

[0090] ① In the same time unit t, at most one recommendation plan can be issued to the same OD pair w. The specific representation is as follows:

[0091]

[0092]

[0093] ② When the traveler's starting and ending points are both located in the interruption interval (affected type 1, w∈W1), it is impossible to recommend a detour route within the subway or change the starting and ending points. The specific representation is as follows:

[0094]

[0095] Among them, T disruption The service interruption period caused by an emergency.

[0096] ③ When the traveler's starting station or terminal station is located in the interruption interval (affected type 2 or 3, w∈W2∪W3), the subway internal route cannot be recommended to him / her. The specific representation is as follows:

[0097]

[0098] ④ When a valid path of a traveler passes through the interruption interval (affected type 4, w∈W4), the path cannot be recommended to the traveler, but there is no need to change the starting and ending points. The specific representation is as follows:

[0099]

[0100] ⑤ After the interruption is restored, since the passenger flow distribution has not returned to normal operation in a short period of time, it is necessary to continue passenger flow guidance. Since all paths inside the subway are passable, the operation management only needs to recommend the internal paths of the subway to passengers. The specific form of expression is as follows:

[0101]

[0102] Among them, T recovery This is the interruption recovery period.

[0103] S32: Considering constraints including decision-making state constraints, travel impedance constraints, section load constraints, expected waiting time constraints, and non-negativity constraints, the utility function in the dynamic travel choice model is used as the basis for passenger flow allocation. Based on the passenger flow between urban rail stations and the evolution of passenger demand, a lower-level passenger flow allocation model is constructed. The allocation results are fed back to the upper-level model, and the optimal induction solution is obtained through repeated iterations.

[0104] The constraints of the lower-level passenger flow distribution model are:

[0105] (1) Decision-making state constraints

[0106] Because waiting at a specific location and changing the starting station take time, passengers may not enter the network at that time after making a decision, but only at a future time. Therefore, it is necessary to constrain the passenger flow in each time period.

[0107]

[0108] Among them, q w (t) is the actual travel demand between ODs within time unit t; is the initial travel demand between OD pairs w in time unit t; P w,r (t) is the probability that the passenger between OD pair w chooses alternative r within time unit t; A w,r (t) is the availability of alternative r between OD pair w in time period unit t (1 if available, 0 otherwise); q′ w (t) is the travel demand of passengers waiting at the same location or changing the starting station between OD pair w within time unit t.

[0109] For q′ w(t) Make the following constraints:

[0110] ① Select passengers waiting for the interruption to resume and transfer them to the corresponding time period according to the selected waiting time:

[0111]

[0112] Among them, t wait The waiting time selected for the passenger; P w,h (t) is the probability that a passenger between OD pair w chooses waiting time interval h within time unit t; Δt is the length of the time unit; To round x down.

[0113] ② For passengers who choose to change their starting station, this part of the passenger flow will be transferred to the corresponding time period based on the connection time of their new starting station:

[0114]

[0115] Among them, t transfer The connecting time required for passengers to change their starting station.

[0116] (2) Travel impedance constraints

[0117] The travel impedance constraint is determined based on the utility function of the aforementioned travel choice behavior model. Impedance functions are defined for passengers at different selection stages and under different affected types, which will not be detailed here.

[0118] (3) Section full load constraint

[0119] In order to ensure passenger comfort and safety, as well as the reliability and durability of urban rail equipment, restrictions are imposed on the full load rate of urban rail sections.

[0120]

[0121] in, is the full load rate of section e in line l in time period unit t; q l,e (t) is the passenger flow of section e in line l during time period t; C l is the rated passenger capacity of line l; H l (t) is the departure interval of line l within time unit t.

[0122] Accordingly, the section load factor constraint is:

[0123] in, The maximum load factor of the train is 130%.

[0124] (4) Expected waiting time constraint

[0125] Under normal subway operation, if the following conditions are met: 1) arriving passengers can always take the first arriving train to depart; and 2) the passenger arrival rate at a station is random and independent, unrelated to the train arrival and departure rates. The expected waiting time for passengers at a station can usually be considered to be half of the train departure interval, that is:

[0126] in, is the expected waiting time of passengers on line l; H l is the departure interval of line l.

[0127] However, if an unexpected event causes service interruption, the above equation does not hold. Therefore, the expected waiting time of passengers is re-evaluated based on the real-time situation of passenger flow distribution and transport capacity supply. In this case, the probability of a passenger boarding the train at station s can be expressed as:

[0128]

[0129] in, is the probability that a passenger gets on the bus at station s on line l within time period unit t; is the cumulative number of passengers boarding the bus on line l before station s as of time period t; is the cumulative number of people getting off line l before station s as of time period unit t; is the number of people who get off at station s on line l within time unit t; is the number of people who board the train at station s on line l within time period unit t.

[0130] At the same time, assuming that the process of passengers queuing at the platform to board the bus follows the first-in-first-out principle, the expected waiting time of the passengers can be expressed as:

[0131]

[0132] in, is the expected waiting time of a passenger at station s on line l in time period unit t.

[0133] (5) Non-negative constraints

[0134] All passenger flow-related variables need to satisfy the non-negative constraints:

[0135]

[0136] S33: Solve the bi-level programming model using a hybrid algorithm combining a fast non-dominated sorting genetic algorithm (NSGA-II) with an elitist strategy and the method of successive weighted averages (MSWA). Decision variables are dynamically adjusted within the feasible region. The upper-level objective function is used as the fitness function in the NSGA-II algorithm, and the MSWA algorithm is used to solve the random user equilibrium allocation solution, outputting an information-induced recommendation solution.

[0137] Figure 4 This is a flow chart of a hybrid algorithm combining a sequential weighted average algorithm and a genetic algorithm for solving a passenger flow induction information release optimization model based on bi-level programming. The NSGA-II algorithm process includes determining genetic algorithm parameters, initializing the population, generating the first-generation child population, elite selection, generating a new parent population, and generating a new child population. Ultimately, the recommended passenger flow induction information solution is output to the lower-level passenger flow allocation model. The MSWA algorithm process includes pre-processing, calculating the probability of each solution being selected, passenger flow loading, updating the utility function, iterative passenger flow calculation, and convergence determination. Travel selection demand within each induction period is organized, and the passenger load factor for each section is calculated. The organized results are then input into the upper-level dual-objective optimization model to solve the upper-level objective function.

[0138] The key method parameters shown in Table 1 below can be adopted in relevant case applications.

[0139] Table 1 Method parameter values

[0140]

[0141] In summary, this embodiment reveals the passenger behavior mechanism by constructing an information-induced passenger dynamic behavior model under urban rail operation emergencies. This model enables accurate grasp of the passenger flow demand of each affected type between line ODs, and can effectively solve the problem of generating information-induced solutions under urban rail operation emergencies.

[0142] Example 3

[0143] This embodiment 3 provides a non-transitory computer-readable storage medium for storing computer instructions. When the computer instructions are executed by a processor, the information induction method considering the dynamic behavior of urban rail passengers under an emergency event as described above is implemented. The method includes:

[0144] Considering the passenger waiting process, personalized heterogeneous induction information is dynamically released, and the dynamic behavior selection model of passengers under operational emergencies is constructed using behavioral sample data;

[0145] Search for efficient routes between all origins and destinations of the urban rail network during emergencies, define different types of affected passenger flows, and construct a set of alternative travel plans for them. Simultaneously, based on dynamic passenger behavior feedback, deduce the evolution of passenger flows at the network level and estimate dynamic passenger flow demand under induced time units.

[0146] A dynamic passenger flow induction information release scheme optimization model based on double-layer planning is constructed to output the optimal time-sharing scheme for induction information. The optimization model for dynamic passenger flow induction information release scheme based on double-layer planning takes the most balanced passenger flow distribution in the line network section and the minimum generalized travel cost of passengers as the optimization objectives. The upper layer is the dual-objective optimization model of the induction information release scheme, and the lower layer is the passenger flow distribution model under urban rail emergencies.

[0147] Example 4

[0148] This embodiment 4 provides a computer device, including a memory and a processor, wherein the processor and the memory communicate with each other, the memory stores program instructions executable by the processor, and the processor invokes the program instructions to execute the above-described information induction method considering the dynamic behavior of urban rail passengers under emergency events, the method comprising:

[0149] Considering the passenger waiting process, personalized heterogeneous induction information is dynamically released, and the dynamic behavior selection model of passengers under operational emergencies is constructed using behavioral sample data;

[0150] Search for efficient routes between all origins and destinations of the urban rail network during emergencies, define different types of affected passenger flows, and construct a set of alternative travel plans for them. Simultaneously, based on dynamic passenger behavior feedback, deduce the evolution of passenger flows at the network level and estimate dynamic passenger flow demand under induced time units.

[0151] A dynamic passenger flow induction information release scheme optimization model based on double-layer planning is constructed to output the optimal time-sharing scheme for induction information. The optimization model for dynamic passenger flow induction information release scheme based on double-layer planning takes the most balanced passenger flow distribution in the line network section and the minimum generalized travel cost of passengers as the optimization objectives. The upper layer is the dual-objective optimization model of the induction information release scheme, and the lower layer is the passenger flow distribution model under urban rail emergencies.

[0152] Example 5

[0153] This embodiment 5 provides an electronic device, including: a processor, a memory, and a computer program; wherein the processor is connected to the memory, and the computer program is stored in the memory. When the electronic device is running, the processor executes the computer program stored in the memory to cause the electronic device to execute instructions for implementing the above-mentioned information induction method considering the dynamic behavior of urban rail passengers under emergency events. The method includes:

[0154] Considering the passenger waiting process, personalized heterogeneous induction information is dynamically released, and the dynamic behavior selection model of passengers under operational emergencies is constructed using behavioral sample data;

[0155] Search for efficient routes between all origins and destinations of the urban rail network during emergencies, define different types of affected passenger flows, and construct a set of alternative travel plans for them. Simultaneously, based on dynamic passenger behavior feedback, deduce the evolution of passenger flows at the network level and estimate dynamic passenger flow demand under induced time units.

[0156] A dynamic passenger flow induction information release scheme optimization model based on double-layer planning is constructed to output the optimal time-sharing scheme for induction information. The optimization model for dynamic passenger flow induction information release scheme based on double-layer planning takes the most balanced passenger flow distribution in the line network section and the minimum generalized travel cost of passengers as the optimization objectives. The upper layer is the dual-objective optimization model of the induction information release scheme, and the lower layer is the passenger flow distribution model under urban rail emergencies.

[0157] In summary, the information induction method for considering the dynamic behavior of urban rail passengers under emergencies according to an embodiment of the present invention includes: considering the passenger waiting process, dynamically updating personalized heterogeneous induction information, constructing a multi-stage dynamic behavior selection model to observe changes in passengers' decisions, analyzing the dynamic Markov decision process of passengers driven by induction information, studying the impact of choice experience inertia and endogenous preferences on behavior, and solving the state dependence and sequence correlation problems in dynamic decision-making; searching for effective paths between origin and destination points, defining different types of passenger flows affected under emergencies and constructing a set of alternative travel plans for them, and at the same time, deducing the evolution of passenger flow at the line network level based on the dynamic behavior feedback of passengers, and estimating the dynamic passenger flow demand under the induction time unit; constructing a dynamic passenger flow induction information release scheme optimization model based on two-level planning, outputting the optimal recommendation scheme for time-sharing passenger flow induction information, achieving passenger flow balance in the line network section and minimizing the generalized travel cost of passengers, wherein the upper layer is a two-objective optimization model for the induction information release scheme, and the lower layer is a passenger flow allocation model under urban rail emergencies. The present invention reveals the dynamic behavior mechanism of passengers under urban rail operation emergencies, analyzes the dynamic evolution characteristics of demand online over time in combination with their alternative travel plan set, takes into account both operational efficiency and safety concepts, and achieves precise optimization of the effect of the induction information release plan.

[0158] It will be understood by those skilled in the art that embodiments of the present invention may be provided as methods, systems, or computer program products. Thus, the present invention may take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware. Furthermore, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to magnetic disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0159] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowcharts and / or block diagrams, as well as combinations of processes and / or blocks in the flowcharts and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowcharts and / or block diagrams. Figure 1 a process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0160] These computer program instructions may also be stored in a computer readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 a process or multiple processes and / or boxes Figure 1 The function specified in one or more boxes.

[0161] These computer program instructions can also be loaded onto a computer or other programmable data processing device, and a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, so that the instructions executed on the computer or other programmable device provide the functions for implementing the process. Figure 1 a process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.

[0162] Although the above describes the specific embodiments of the present invention in conjunction with the accompanying drawings, it is not intended to limit the scope of protection of the present invention. Those skilled in the art should understand that various modifications or variations that can be made by those skilled in the art on the basis of the technical solutions disclosed in the present invention without the need for creative work should be included in the scope of protection of the present invention.

Claims

1. An information induction method considering the dynamic behavior of urban rail passengers under emergency events, characterized by: include: Considering the waiting process of passengers, dynamic release of personalized heterogeneous induction information, and using behavioral sample data to build a dynamic behavior selection model of passengers under operational emergencies; including: modeling the dynamic behavior process of passengers under urban rail operational emergencies through a dynamic discrete choice model, and studying the alternative travel choices made by passengers at any time during continuous waiting; passengers will form behavioral experience inertia in multi-stage decision-making, and by constructing the Markov dependency relationship of behavior, the influence of passenger choice experience inertia and endogenous preference on behavior is determined, and by changing the dynamic behavior model structure to solve the state dependence and serial correlation problems existing in dynamic decision-making, the dynamic behavior selection model structure is Among them, U int is the utility of traveler n choosing option i at time t, V int is the fixed utility at time t, is a lagged variable that measures the impact of traveler n’s choice of option i at time t-1 on his utility at time t, α in is the agency effect, which measures the endogenous long-term preference of traveler n that does not change over time and is composed of the initial decision and exogenous explanatory variables, ε′ int are the remaining unobserved short-term random factors; the corresponding probability solution formula is Among them, y in0 For the first selection, y in(t-1) is the choice at time t-1, y int is the choice at time t, is an exogenous explanatory variable, P(y int |y in0 ,y in(t-1) ,α n ) is the probability of the logit model modified by the agent effect under the consideration of the first choice and the previous choice, Follows a normal distribution; Search for efficient routes between all origins and destinations of the urban rail network during emergencies, define different types of affected passenger flows, and construct a set of alternative travel plans for them. Simultaneously, based on dynamic passenger behavior feedback, deduce the evolution of passenger flows at the network level and estimate dynamic passenger flow demand under induced time units. A dynamic passenger flow induction information release optimization model based on double-layer planning is constructed to output the recommended induction information scheme for each time period. The optimization model based on double-layer planning takes the most balanced passenger flow distribution in the line network section and the minimum generalized travel cost for passengers as the optimization objectives: Among them, Z1 is the Theil index of section load rate, T is the control period set, L is the line set, E is the section set, E l is the cross-section set of the lth line, is the sum of all section load factors, is the sum of the full load rates of all sections in the lth line, n is the total number of sections, n l is the number of sections in the lth line, is the full load rate of section l,e on line l, Z2 is the total cost of passenger travel, W is the set of OD pairs, W1, W2, W3, W4, W5 represent the OD pairs of affected types 1-5 respectively, R is the set of alternative options, q w,r (t) is the passenger flow of the OD pair w in the t period, and the alternative r is selected. w,r,g (t) is the g-th attribute value of alternative r in OD pair w during period t, μ r,g is the generalized travel cost corresponding to the g-th attribute value of alternative plan r. The upper layer is the dual-objective optimization model for the induction information release plan, and the lower layer is the passenger flow allocation model under urban rail emergencies. The allocation results are fed back to the upper layer model. The optimal induction plan is obtained through repeated iterations between the upper-layer dual-objective optimization model for the induction information release plan and the lower-layer passenger flow allocation model under urban rail emergencies.

2. The information induction method considering the dynamic behavior of urban rail passengers in an emergency according to claim 1 is characterized in that: Search for efficient routes between all origins and destinations of the urban rail network during emergencies, define different types of affected passenger flows, and construct a set of alternative travel plans for them. Simultaneously, based on dynamic passenger behavior feedback, deduce the evolution of passenger flows at the network level and estimate dynamic passenger flow demand under induced time units, including: A graph-based depth-first search algorithm is used to search for valid paths. A search rule with time thresholds and transfer number thresholds is set. A depth traversal is performed starting from the unvisited adjacent points of a vertex until all points connected to the vertex are visited. All vertices are marked and traversed to generate valid paths between each OD pair. After an urban rail operation emergency occurs, the affected passenger flow is defined based on the different impacts on passengers due to their locations; Determine the alternative travel options in the set of alternative options according to the different degrees of impact on the passengers; Based on the card swiping data of the urban rail automatic ticket vending and checking system, the passenger travel volume between stations is identified through data mining. At the same time, based on the alternative travel plan sets and dynamic behavior feedback of passengers with different degrees of impact, the evolution law of passenger flow at the line network level is deduced, and the dynamic passenger flow demand under the induced time unit is estimated.

3. The information induction method considering the dynamic behavior of urban rail passengers in an emergency according to claim 2 is characterized in that: The affected passenger flows are defined as follows: Category 1: both the starting and ending stations are located in the interruption interval; Category 2: only the starting station is located in the interruption interval; Category 3: only the ending station is located in the interruption interval; Category 4: one or more valid paths of the passenger pass through the interruption interval; Category 5: all valid paths of the passenger do not pass through the interruption interval.

4. The information induction method considering the dynamic behavior of urban rail passengers in an emergency according to claim 2 is characterized in that: Determine the alternative travel plans in the alternative plan set: Category 1: If the passenger's starting and ending stations are both located in the interruption interval, then waiting for the original route to be restored, taking a bus, and taking a taxi are included in the alternative travel plan set; Category 2: If only the passenger's starting station is located in the interruption interval, then waiting for the original route to be restored, changing the entry point to station A, changing the entry point to station B, taking a bus, and taking a taxi are included in the alternative travel plan set; Category 3: If the passenger's only terminal station is located in the interruption interval, then changing the exit station to station A, changing the exit station to station B, bus travel and taxi travel will be taken as the alternative travel plan set; Category 4: If one or more valid paths of the passenger pass through the interruption interval, then waiting for the original path to be restored, urban rail internal detour path 1, urban rail internal detour path 2, bus travel and taxi travel will be taken as the alternative travel plan set; Category 5: If all valid paths of the passenger do not pass through the interruption interval, then the valid path set between OD pairs will be taken as the alternative travel plan set.

5. The information induction method considering the dynamic behavior of urban rail passengers in an emergency according to claim 1 is characterized in that: The travel plan alternative sets for affected types of passengers are also different. Therefore, plan recommendation constraints are set. With the goals of minimizing the Theil index of the section full load rate and minimizing the generalized travel cost of passengers, a dual-objective optimization model for the upper-level induction information release plan is constructed. The decision variables of the dual-objective optimization model for the upper-level induction information release plan are the induction information recommendation plans released by each time unit, including: no recommendation, recommending to maintain the original travel route of the urban rail, recommending the first detour route within the urban rail, recommending the second detour route within the urban rail, recommending to change the entry / exit station to station A, recommending to change the entry / exit station to station B, and recommending to switch to ground transportation.

6. The information induction method considering the dynamic behavior of urban rail passengers in an emergency according to claim 1 is characterized in that: Considering the constraints including decision-making state constraints, travel impedance constraints, section full load constraints, expected waiting time constraints, and non-negativity constraints, the utility function in the dynamic travel choice model is used as the basis for passenger flow allocation. Based on the passenger flow and passenger demand evolution characteristics between urban rail stations, a passenger flow allocation model under lower-level urban rail emergencies is constructed. At the same time, the allocation results of the passenger flow allocation model under lower-level urban rail emergencies are fed back to the upper-level dual-objective optimization model of the induction information release scheme. The optimal induction scheme is obtained through repeated iterations between the upper-level dual-objective optimization model of the induction information release scheme and the lower-level passenger flow allocation model under urban rail emergencies.

7. The information induction method considering the dynamic behavior of urban rail passengers in an emergency according to claim 5 or 6, characterized in that: The dynamic passenger flow induction information release scheme optimization model of the two-level planning is solved by a hybrid algorithm of a fast non-dominated sorting genetic algorithm with an elite strategy and a successive weighted averaging method. The decision variables are dynamically adjusted within the feasible domain. The objective function of the upper-level induction information release scheme dual-objective optimization model is used as the fitness function in the fast non-dominated sorting genetic algorithm with an elite strategy. The successive weighted averaging method is used to solve the random user equilibrium allocation solution, and the information induction recommendation scheme is output.

8. The information induction method considering the dynamic behavior of urban rail passengers in an emergency according to claim 7 is characterized in that: The process of the fast non-dominated sorting genetic algorithm with elite strategy includes determining genetic algorithm parameters, initializing the population, generating the first generation of child populations, elite selection, generating a new parent population, generating a new child population, and finally outputting the passenger flow induction information recommendation plan to the lower-level passenger flow allocation model; The algorithm process of the successive weighted average method includes preliminary preparation, calculation of the probability of selection of each plan, passenger flow loading, utility function update, passenger flow iterative calculation, and convergence judgment; the travel choice demand in each induction period is sorted out and the passenger flow load rate of each section is calculated. The sorted results are input into the upper-level dual-objective optimization model to solve the upper-level objective function.

9. An information guidance system that considers the dynamic behavior of urban rail passengers during emergencies, characterized by: include: The construction module is used to consider the passenger waiting process, dynamically release personalized heterogeneous induction information, and use behavioral sample data to build a dynamic behavior selection model for passengers under operational emergencies; including: modeling the dynamic behavior process of passengers under urban rail operational emergencies through a dynamic discrete choice model, and studying the alternative travel choices made by passengers at any time during continuous waiting; passengers will form behavioral experience inertia in multi-stage decision-making, and determine the impact of passenger choice experience inertia and endogenous preferences on behavior by constructing the Markov dependency relationship of behavior, and solving the state dependence and serial correlation problems in dynamic decision-making by changing the dynamic behavior model structure. The dynamic behavior selection model structure is Among them, U int is the utility of traveler n choosing option i at time t, V int is the fixed utility at time t, is a lagged variable that measures the impact of traveler n’s choice of option i at time t-1 on his utility at time t, α in is the agency effect, which measures the endogenous long-term preference of traveler n that does not change over time and is composed of the initial decision and exogenous explanatory variables, ε′ int are the remaining unobserved short-term random factors; the corresponding probability solution formula is Among them, y in0 For the first selection, y in(t-1) is the choice at time t-1, y int is the choice at time t, is an exogenous explanatory variable, P(y int |y in0 ,y in(t-1) ,α n ) is the probability of the logit model modified by the agent effect under the consideration of the first choice and the previous choice, Follows a normal distribution; The definition module is used to search for effective routes between all origins and destinations of the urban rail network during emergencies, define different types of affected passenger flows and construct a set of alternative travel plans for them. At the same time, based on the dynamic behavior feedback of passengers, it deduces the evolution of passenger flow at the line level and estimates the dynamic passenger flow demand under the induced time unit; The solution module is used to construct a dynamic passenger flow induction information release scheme optimization model based on double-layer planning, and output the recommended scheme of induction information released by each time unit. The optimization model of the dynamic passenger flow induction information release scheme based on double-layer planning aims to achieve the most balanced passenger flow distribution in the line network section and minimize the generalized travel cost of passengers: Among them, Z1 is the Theil index of section load rate, T is the control period set, L is the line set, E is the section set, E l is the cross-section set of the lth line, is the sum of all section load factors, is the sum of the full load rates of all sections in the lth line, n is the total number of sections, n l is the number of sections in the lth line, is the full load rate of section l,e on line l, Z2 is the total cost of passenger travel, W is the set of OD pairs, W1, W2, W3, W4, W5 represent the OD pairs of affected types 1-5 respectively, R is the set of alternative options, q w,r (t) is the passenger flow of the OD pair w in the t period, and the alternative r is selected. w,r,g (t) is the g-th attribute value of alternative r in OD pair w during period t, μ r,g is the generalized travel cost corresponding to the g-th attribute value of alternative plan r. The upper layer is the dual-objective optimization model for the induction information release plan, and the lower layer is the passenger flow allocation model under urban rail emergencies. The allocation results are fed back to the upper layer model. The optimal induction plan is obtained through repeated iterations between the upper-layer dual-objective optimization model for the induction information release plan and the lower-layer passenger flow allocation model under urban rail emergencies.

Images