A subway passenger flow simulation and deduction method considering individual trip chain and activity mode

Abstract

The application discloses a subway passenger flow simulation and deduction method considering individual trip chains and activity modes, and relates to the technical field of subway operation management. The specific steps of the method are as follows: constructing a topological framework, calculating external factor attention weights, constructing an individual trip behavior library, dynamically adjusting a topological structure, and performing passenger flow simulation and deduction. The application improves the precision and real-time performance of subway passenger flow simulation by constructing a space-time heterogeneous topological model and dynamically adjusting the weight values of nodes and edges. Not only physical parameters and historical operation data of urban subway lines are collected, but also an external factor input layer and a calibration layer are added, so that the space-time heterogeneous topological model can reflect the influence of external factors on subway passenger flow in real time. Through calculating external factor attention weights and combining data in the individual trip behavior library, the station retention weight and the line passing weight are dynamically corrected, the prediction of subway passenger flow changes is realized, and a scientific basis is provided for subway operation managers.

Description

Technical Field

[0001] This invention relates to the field of subway operation management technology, specifically a subway passenger flow simulation and extrapolation method that considers individual travel chains and activity patterns. Background Technology

[0002] With the acceleration of global urbanization and the continuous growth of urban population, the subway, as a high-capacity and high-efficiency urban public transportation mode, directly affects the overall operational efficiency of the urban transportation system. Accurate prediction and management of subway passenger flow is not only key to ensuring passenger safety and improving service experience, but also an important basis for optimizing subway operation strategies and improving resource utilization efficiency. However, the subway system has significant spatiotemporal heterogeneity, and passenger flow distribution is affected by multiple factors such as time, space and external factors, exhibiting dynamic changes.

[0003] However, traditional subway passenger flow forecasting technologies are mostly based on static models or historical data averages, which make it difficult to accurately reflect the spatiotemporal heterogeneity of the subway system and capture real-time changes in passenger flow over time and space, resulting in a significant decrease in forecast accuracy. Secondly, the lack of effective integration of external factors leads to a large deviation between simulation results and actual passenger flow when facing sudden weather changes, holidays, and other situations. Furthermore, traditional technologies often ignore individual passenger behavior differences, such as the distribution of travel time and transfer preferences under different activity patterns, further reducing the accuracy of simulations. Finally, in response to emergencies, the lack of flexible adjustment mechanisms makes it difficult to quickly respond to and re-analyze changes in passenger flow, thus limiting decision-making capabilities in emergency situations.

[0004] Therefore, we need to develop a subway passenger flow simulation method that takes into account individual travel chains and activity patterns. Summary of the Invention

[0005] The purpose of this invention is to overcome the shortcomings of existing technologies and provide a subway passenger flow simulation and extrapolation method that considers individual travel chains and activity patterns. This invention improves the accuracy and real-time performance of subway passenger flow simulation by constructing a spatiotemporal heterogeneous topology model and dynamically adjusting the weights of nodes and edges. It not only collects physical parameters and historical operational data of urban subway lines but also adds an external factor input layer and a calibration layer, enabling the spatiotemporal heterogeneous topology model to reflect the impact of external factors on subway passenger flow in real time. By calculating the attention weights of external factors and combining them with data from the individual travel behavior database, the station dwell weights and line passage weights are dynamically adjusted, achieving the prediction of subway passenger flow changes. This dynamic adjustment mechanism is closer to the actual situation, providing a scientific basis for subway operation managers, helping to optimize operational strategies and improve the passenger travel experience.

[0006] To solve the above-mentioned technical problems, this invention provides the following technical solution: a subway passenger flow simulation and extrapolation method considering individual travel chains and activity patterns, the specific steps of which are as follows:

[0007] Constructing a topology framework: Collect physical parameters and historical operation data of urban subway lines, construct a spatiotemporal heterogeneous topology model and define topology nodes and edges, set weight values ​​for topology nodes and edges, and add an external factor input layer and a calibration layer to form a topology framework;

[0008] External factor attention weight calculation: classify and quantify external factors to determine correlation coefficients, collect data on the intensity and spatial correlation of external factors, and calculate the weight of external factors on the target in each time period;

[0009] Construct an individual travel behavior database: Obtain subway IC card data for a certain period of time and conduct user travel surveys. Clean and analyze the IC card data to extract individual travel chain characteristics. Combine the survey data to label each travel chain with activity mode tags, construct an individual travel behavior database, and set differentiated travel time distribution and transfer preferences for different activity modes.

[0010] Dynamic topology adjustment: Based on the topology framework, external factor attention weights and individual travel behavior database data are introduced, station dwell weights and route passage weights are then calibrated, and the route passage weights are corrected in combination with the individual travel behavior database to complete the dynamic topology adjustment.

[0011] Perform passenger flow simulation and extrapolation: Based on the adjusted topology framework, construct intelligent agents through multi-agent simulation technology to simulate the entire travel process, calculate and record passenger flow data at each time period and organize and output it to complete the extrapolation of passenger flow changes. When encountering sudden scenarios, update the external factor data, recalculate the attention weight of external factors and set the calibration coefficient, and then extrapolate again.

[0012] Furthermore, in constructing the topology framework, physical parameters of the target city's subway lines are collected, including the maximum capacity of each station. Maximum design speed of the line Maximum passenger capacity of the line Collect historical operational data for different periods and time periods, including station passenger flow density at each time period. Historical average passenger flow density of the station A spatiotemporal heterogeneous topology model is constructed based on physical parameters and historical operational data. Subway stations are defined as topological nodes, and line segments between adjacent stations are defined as topological edges. A weight value is set for each topological node using a dynamic topological node weight calculation formula, and a weight value is set for each topological edge using a dynamic topological edge weight calculation formula. The weight values ​​of nodes and edges evolve dynamically over time. An external factor input layer and a calibration layer are added to the spatiotemporal heterogeneous topology model. The external factor input layer receives collected external factor data, including weather, holidays, and large-scale events. The calibration layer stores the adjustment parameters of the external factor data on the weights of topological nodes and edges, forming the topological framework of a dynamically calibrated subway network.

[0013] Furthermore, in the constructed topology framework, a weight value is set for each topology node through a dynamic topology node weight calculation formula, which is as follows: ,in, For the first Time Period Site The node weights are used to characterize the real-time capacity of the site. For the site Real-time passenger count For the site Maximum capacity For the first Time Period Site Passenger density, For the first Time Period Site Historical average passenger flow density These are the weighting coefficients.

[0014] Furthermore, in the constructed topology framework, a weight value is set for each topology edge using a dynamic topology edge weight calculation formula, which is as follows: ,in, For the first Time-of-day routes The edge weights are used to characterize the real-time throughput efficiency of the line. For the first Time-of-day routes Real-time speed of Chinese trains For the line The maximum speed set by the train For the first Time-of-day routes Real-time passenger volume, For the line The maximum passenger capacity is set. These are the weighting coefficients.

[0015] Furthermore, in the calculation of attention weights for external factors, three categories of external factors—weather, holidays, and large-scale events—are considered. The external factors are categorized, and their coefficients are quantified for each category. And collect the intensity of external factors at each time period; Target acquisition through GIS map matching External factors Spatial correlation ,in, At that time, it was a station. The time is the route; then, the external factors for each time period are calculated using the external factor attention weighting calculation formula. For the target The weights are determined, and calibration coefficients are set based on the intensity and duration of the influence of external factors. .

[0016] Furthermore, in the calculation of external factor attention weights, the external factors for each time period are calculated using the external factor attention weight calculation formula. For the target The weight of external factor attention is calculated using the following formula: ,in, For the first External factors during the period For the target The weight, External factors coefficient, For the first External factors during the period strength, For the goal External factors Spatial correlation.

[0017] Furthermore, in the aforementioned dynamic topology adjustment, based on the topology framework of a dynamically calibrated metro network, attention weights for external factors are introduced. By combining individual travel behavior database data with the entry / exit percentages of users with different activity patterns in the database, the entry / exit passenger flow weights of stations are adjusted, and simultaneously... and The target is calculated using the calibrated formula for the weighting of station dwell time and line traffic. The weights are calibrated, and the route selection preferences of users with different activity patterns in the individual travel behavior database are combined to correct the route passage weights and complete the dynamic topology adjustment.

[0018] Furthermore, in the dynamic topology adjustment, the target is adjusted using the calibrated formula for calculating the weights of station stagnation and line traffic. The weights are calibrated, and the formula for calculating the weights of station lingering and line passage after calibration is as follows: ,in, For the first Target after time period calibration The weight, when hour, The node weight of the site is calibrated to obtain the site retention weight. hour, The edge weights of the line are used, and the line passage weights are obtained after calibration. For the first External factors during the period For the target The weight, This is the calibration coefficient.

[0019] Furthermore, in the passenger flow simulation and extrapolation, based on the adjusted topology framework, multi-agent simulation technology is used to set the total number of simulation agents according to the average daily passenger volume of the target city's subway. For each agent, travel chain features and activity pattern labels from the individual travel behavior database are matched; the entire travel process of the agent from the departure station to the destination station is simulated, and the station at each time period is calculated in real time using the multi-agent simulation passenger flow calculation formula. The system records and organizes passenger flow data for each time period, including inbound, outbound, and lingering passenger flow, to predict changes in passenger flow. In case of unforeseen circumstances, it updates data on external factors. Recalculate and set Then, we re-analyze the scenarios, including but not limited to sudden line failures, temporary station flow restrictions, and temporary train delays.

[0020] Furthermore, in the passenger flow simulation and extrapolation process, the passenger flow at each station in real time is calculated using a multi-agent simulation passenger flow calculation formula. The multi-agent simulation passenger flow calculation formula is as follows: ,in, For the first Time Period Site Passenger flow The total number of simulated intelligent agents, For intelligent agents No. Is the time period within the site? Yes = 1, No = 0 For intelligent agents No. The probability of travel during a given time period is set according to the activity mode.

[0021] Compared with existing technologies, this subway passenger flow simulation and extrapolation method that considers individual travel chains and activity patterns has the following advantages:

[0022] I. This invention improves the accuracy and real-time performance of subway passenger flow simulation by constructing a spatiotemporal heterogeneous topology model and dynamically adjusting the weights of nodes and edges. It not only collects physical parameters and historical operational data of urban subway lines but also adds an external factor input layer and a calibration layer. This allows the spatiotemporal heterogeneous topology model to reflect the impact of external factors on subway passenger flow. By calculating the attention weights of external factors and combining them with data from an individual travel behavior database, the invention dynamically corrects station dwell weights and line passage weights, enabling the prediction of subway passenger flow changes. This dynamic adjustment mechanism is closer to the actual situation, providing a scientific basis for subway operation managers, helping to optimize operational strategies and improve the passenger travel experience.

[0023] Second, this invention utilizes multi-agent simulation technology, combined with travel chain features and activity pattern tags from an individual travel behavior database, to simulate passenger flow changes under sudden scenarios. When encountering a sudden scenario, it updates external factor data, recalculates the attention weights of external factors, and sets calibration coefficients, thereby re-inferring passenger flow changes. This flexible adjustment mechanism enables subway operation managers to promptly understand the impact of sudden scenarios on passenger flow and quickly take countermeasures, such as adjusting train operation plans and increasing staff, thereby effectively ensuring the safe and efficient operation of the subway.

[0024] Other advantages, objectives and features of the invention will be set forth in part in the description which follows, and in part will be apparent to those skilled in the art from the following examination or study, or may be learned from the practice of the invention. Attached Figure Description

[0025] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the accompanying drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are merely some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without any creative effort.

[0026] Figure 1 A flowchart for a subway passenger flow simulation and extrapolation method that considers individual travel chains and activity patterns;

[0027] Figure 2 This is a framework diagram of a subway passenger flow simulation and extrapolation method that considers individual travel chains and activity patterns. Detailed Implementation

[0028] To further illustrate the technical means and effects of the present invention in achieving its intended purpose, the following detailed description of the specific implementation methods, structure, features and effects of the present invention, in conjunction with the accompanying drawings and preferred embodiments, is provided below.

[0029] Example 1:

[0030] Constructing a topology framework: In a simulation scenario of passenger flow on a city's core metro commuter line during weekday morning rush hour (7:00-9:00), physical parameters and historical operational data of the metro line are collected. Physical parameters need to cover the maximum capacity of all stations, the highest design speed of the entire line, and the maximum passenger capacity of each train. This data forms the basis for subsequent evaluation of station capacity and line capacity. Historical operational data focuses on station passenger flow density and the historical average passenger flow density during corresponding weekday morning rush hours over the past 3-6 months, capturing the periodic variation patterns of morning rush hour passenger flow. Based on physical parameters and historical operational data, a spatiotemporal heterogeneous topology model is constructed. Each metro station on the line is defined as a topology node, and the actual line segment between two adjacent stations is defined as a topology edge. Subsequently, dynamic topology node weight calculation formulas and dynamic topology edge weight calculation formulas are used to assign dynamically changing weight values ​​over time to each topology node and topology edge. The dynamic topology node weight calculation formula is as follows: ,in, For the first Time Period Site The node weights are used to characterize the real-time capacity of the site. For the site Real-time passenger numbers are collected through station turnstiles. For the site Maximum capacity For the first Time Period Site Passenger flow density is obtained through historical operational data. For the first Time Period Site Historical average passenger flow density is obtained through historical operational data. The weighting coefficients are determined using historical operational data; the formula for calculating the dynamic topology edge weights is as follows: ,in, For the first Time-of-day routes The edge weights are used to characterize the real-time throughput efficiency of the line. For the first Time-of-day routes Real-time speed of Chinese trains For the line The maximum speed set by the train For the first Time-of-day routes Real-time passenger volume, For the line The maximum passenger capacity is set. The weighting coefficients are determined through historical operational data; node weights reflect the real-time capacity status of stations at different times, and edge weights reflect the real-time traffic efficiency of the line. At the same time, an external factor input layer and a calibration layer are added to the spatiotemporal heterogeneous topology model. The external factor input layer is used to receive external factor data that may affect the morning peak passenger flow, such as weather, holidays, and large-scale events. The calibration layer stores the adjustment parameters of external factors on node and edge weights, and finally forms a dynamic calibration type metro network topology framework that can be adapted to the morning peak scenario.

[0031] External Factor Attention Weight Calculation: External factors that may affect passenger flow are categorized into three types: weather, holidays, and large-scale events. Considering the actual scenario of weekday morning rush hour, which typically lacks holidays and large-scale events to avoid overlapping with commuter traffic, only the impact of weather factors needs to be considered. Weather factors such as sunny days, rainy days, and smog are quantified. Based on historical data showing the changing patterns of morning rush hour passenger flow under different weather conditions, corresponding external factor coefficients are determined. Specific weather intensity data for each time period of the morning rush hour, such as rainfall intensity and smog concentration, are collected. Spatial correlation between each station and each line segment and the weather factor is obtained through GIS map matching. Since the impact of morning rush hour weather on subway lines within the same urban area is relatively uniform, the spatial correlation between each station and line is relatively small. The external factor attention weight calculation formula is used to calculate the weight of weather factors for each station and each line at each time period. The external factor attention weight calculation formula is as follows: ,in, For the first External factors during the period For the target The weight, At that time, it was a station. The time is the line, External factors coefficient, For the first External factors during the period strength, For the goal External factors Spatial correlation; and combined with the stability and duration of the weather impact during the morning rush hour, since the weather conditions usually change little during the entire morning rush hour, a calibration coefficient is set to provide a basis for the influence of external factors for subsequent dynamic topology adjustments.

[0032] Constructing an individual travel behavior database: This involves acquiring subway IC card data for the core commuter line over the past month. The IC card data comprehensively records key information such as users' entry and exit times, stations, and travel routes. Simultaneously, a sample travel survey was conducted targeting major commuter groups within the line's coverage area, including commuters and students, to understand details such as users' travel purposes, daily fixed travel times, and route preferences during transfers. The collected IC card data was then cleaned and processed, removing invalid data such as duplicate card swipes and abnormal data caused by turnstile malfunctions. Finally, individual travel chain characteristics for each user were extracted, such as entering station A in the suburbs at 7:10 AM and exiting station B in the city center at 8:05 AM. Typical commuting routes, such as entering station C in the residential area at 7:30 AM and exiting station D in the business district at 8:20 AM, are analyzed. Based on travel survey results, each extracted route is labeled with a commuting activity mode tag. This tag has a high proportion in routes during the morning rush hour. Differentiated travel time distributions and transfer preferences are set for commuting activity modes. The travel time distribution reflects the peak entry times (e.g., 7:00-7:30 AM) and exit times (e.g., 8:00-8:30 AM). Transfer preferences align with commuters' habits of prioritizing direct routes and minimizing transfers to save time. Ultimately, an individual travel behavior database tailored to the morning rush hour commuting scenario is constructed. Figure 1 As shown.

[0033] Dynamic Topology Adjustment: Based on the constructed dynamic calibration metro network topology framework, external factor attention weights and individual travel behavior database data are introduced. This is combined with the entry and exit percentages of commuting users in the individual travel behavior database. For example, during the early morning rush hour (7:00-7:20), the entry percentage of commuters at suburban stations is significantly higher than that at city center stations; during the later morning rush hour (8:00-8:30), the exit percentage of commuters at city center stations increases significantly. Based on this, the entry and exit passenger flow weights of the corresponding stations are adjusted. Simultaneously, combining external factor attention weights and a set calibration coefficient, the station dwell time and line traffic weights are calibrated using the calibrated station dwell time and line traffic weight calculation formulas. The calibrated station dwell time and line traffic weight calculation formulas are as follows: ,in, For the first Target after time period calibration The weight, when hour, The node weight of the site is calibrated to obtain the site retention weight. hour, The edge weights of the line are used, and the line passage weights are obtained after calibration. For the first External factors during the period For the target The weight, The calibration coefficients are used; the dwell weight reflects the congestion level of passengers staying at the station, and the passage weight reflects the transportation efficiency of the line. Finally, based on the route selection preferences of commuters in the individual travel behavior database, the line passage weights are further adjusted. For example, the passage weights of direct routes that commuters prefer need to be adjusted appropriately to match the actual passenger flow distribution, ensuring that the adjusted dynamic topology can accurately match the travel characteristics of commuters during the morning peak, thus completing the entire dynamic topology adjustment process.

[0034] Passenger flow simulation and extrapolation: Based on the adjusted topology framework, multi-agent simulation technology is used to construct simulated agents. The total number of agents is set according to the average daily passenger volume during the weekday morning peak on the core commuter route to ensure that the simulation scale matches the actual passenger flow scale. Each simulated agent is matched with travel chain characteristics and activity pattern labels from the individual travel behavior database for commuter categories, allowing the agents to simulate the entire travel process of real commuters, including the journey from the departure point to the starting station, entering the station, taking the subway (including possible transfers), and exiting the destination station. The entire simulation process strictly follows the passenger flow change patterns during the morning peak. During the simulation, the passenger flow data for each station at each time period is calculated in real time using the multi-agent simulation passenger flow calculation formula, which is: ,in, For the first Time Period Site Passenger flow The total number of simulated intelligent agents, For intelligent agents No. Is the time period within the site? Yes = 1, No = 0 For intelligent agents No. The probability of travel during a specific time period is set according to the activity mode. Then, the inbound passenger flow, outbound passenger flow, and stranded passenger flow of each station during each time period are recorded and organized. Among them, the inbound passenger flow reflects the passenger flow inflow of different stations, the outbound passenger flow reflects the passenger flow dispersal of different stations, and the stranded passenger flow assesses the congestion level of the station. Finally, these data are organized and output to form a complete prediction result of the morning peak passenger flow change. This result can directly provide data support for the subway operator to formulate operational strategies such as adjusting the frequency of train departures, station staff allocation, and passenger flow guidance during the morning peak period.

[0035] In summary, this study simulates passenger flow on a core commuter subway line in a city during weekday morning rush hour. A spatiotemporal heterogeneous topology model was constructed by collecting physical parameters and historical operational data of the line, and an external factor input layer and calibration layer were added to form the topology framework. Then, attention weights for weather-related external factors were calculated for the morning rush hour scenario. Subsequently, an individual travel behavior database labeled with commuting tags was constructed based on IC card data and travel surveys. Next, dynamic topology adjustments were completed by combining the external factor weights and the behavior database data. Finally, passenger flow simulation was performed using multi-agent simulation technology to recreate the characteristics of morning rush hour passenger flow, providing scientific support for line operation scheduling and passenger flow management.

[0036] Example 2:

[0037] Constructing a Topology Framework: In a simulation scenario of passenger flow on a subway shuttle line connecting a sports venue in a city from 21:00 to 22:00 after a large-scale weekend sporting event, physical parameters and historical operational data of the subway shuttle line were collected. Among the physical parameters, the maximum capacity of the direct shuttle station to the sports venue is crucial, as this station is the main entry point for passenger flow after the event, making capacity assessment essential. The maximum design speed of the entire line and the maximum passenger capacity per train are also required. Historical operational data includes station passenger flow density and the historical average passenger flow density for the same time period on weekends over the past 3-6 months. By comparing the differences in passenger flow data between regular weekends and the event's closing time, a reference is provided for subsequent calibration. Based on the physical parameters and historical operational data, a spatiotemporal heterogeneous topology model is constructed. Each subway station in the line is defined as a topology node, with the sports venue shuttle station as a core topology node and the line segments between adjacent stations defined as topology edges. Dynamic topology node weight calculation formulas and dynamic topology edge weight calculation formulas are used to set dynamically evolving weight values ​​for each topology node and topology edge over time. The dynamic topology node weight calculation formula is as follows: The formula for calculating the dynamic topological edge weight is as follows: The node weights primarily reflect the real-time capacity changes of each station during the event's closing period, while the edge weights focus on the fluctuations in line efficiency under surges in passenger flow. Simultaneously, an external factor input layer and a calibration layer are added to the spatiotemporal heterogeneous topology model. The external factor input layer receives data on three types of external factors: weather, holidays, and large-scale events. Weather data includes the weather type and intensity during the event's closing period; holiday data includes whether the weekend is a statutory holiday and the typical passenger flow characteristics of the line during holidays; and large-scale event data is the core influencing factor for this scenario, focusing on information related to the large-scale sporting event, specifically including the expected number of spectators, the planned end time (i.e., closing time), and spectator departure routes. This provides crucial information for quantifying the impact of large-scale events on passenger flow. The calibration layer stores the adjustment parameters of these three types of external factors on the topology nodes and edge weights, ultimately forming a dynamically calibrated metro network topology framework adapted to the event closing scenario. Figure 2 As shown.

[0038] External Factor Attention Weight Calculation: External factors are categorized into three types: weather, holidays, and large-scale events. In this scenario, the core external factor is the large-scale sporting event, while also considering the impact of weather factors, such as the possibility that rain at the end of the event might lead more spectators to choose the subway. These two types of factors are quantified separately. For large-scale sporting events, the corresponding external factor coefficients are determined based on the expected number of spectators and the ease of connection between the venue and the subway station. For weather factors, corresponding external factor coefficients are set based on spectator travel preferences under different weather types, such as sunny days, rainy days, and strong winds. Weather intensity data is also collected for each time period during the event's end. For example, data on rainfall intensity and the intensity of large-scale sporting events can be used. The intensity of the event can be determined based on the rate at which spectators leave, with the peak departure time typically occurring 30-40 minutes after the event ends. Spatial correlations between each station, each route, and large-scale sporting events and weather factors are obtained through GIS map matching. The spatial correlation between event venue shuttle stations and large-scale sporting events is the highest, gradually decreasing with increasing distance between the station and the venue. Weather factors have a relatively uniform correlation with each station along the route. Furthermore, the weights of the two types of external factors for each station and each route at different times are calculated using the external factor attention weight calculation formula. The external factor attention weight calculation formula is as follows: Considering the concentrated and short-lived impact of large-scale sporting events, the fact that peak passenger flow typically lasts 1-1.5 hours, and the relatively stable nature of weather influences, a calibration coefficient suitable for this scenario is set to provide external factor impact data for subsequent dynamic topology adjustments.

[0039] Constructing an individual travel behavior database: Obtaining subway IC card data for the event venue shuttle routes within the past period, with a focus on IC card data from past event days. Simultaneously, conducting a sample survey of event attendees and regular weekend travelers within the route coverage area. The survey included event attendees' expected departure time, destination area, acceptance of the routes during transfers (i.e., willingness to accept 1-2 transfers for faster travel), and regular users' travel purposes and time arrangements. The collected IC card data was cleaned and analyzed, and invalid data was removed to extract individual travel chain characteristics, with particular attention to travel chains originating from the event venue shuttle stations, such as the 21:10-21:20 arrival at the event shuttle station and arrival in the city. The central transfer station sees passengers entering between 21:40 and 21:50, while the residential area station sees passengers exiting between 22:10 and 22:20. This type of travel chain accounts for a high proportion during the event's departure period. Based on travel survey results, each travel chain is labeled with an activity mode tag for leaving the event, and the proportion of this tag in all travel chains during the departure period is calculated. At the same time, differentiated travel time distributions and transfer preferences are set for the event departure activity modes. The travel time distribution should highlight the characteristics of the peak time for spectators to enter the station after the event, such as 21:00-21:30 being the peak time for entering the station. The transfer preferences should meet the needs of spectators who want to reach their destination quickly, are willing to accept 1-2 reasonable transfers, and prioritize routes with greater capacity. Ultimately, an individual travel behavior database that closely matches the event departure scenario is constructed.

[0040] Dynamic Topology Adjustment: Based on the constructed dynamic calibration metro network topology framework, external factor attention weights and individual travel behavior database data are introduced. Combining the entry and exit percentages of users exhibiting the "spectator departure" activity pattern in the individual travel behavior database, the entry percentage at venue connection stations surges to its peak during the initial post-event period (21:00-21:30), while the exit percentage at city center transfer stations and stations near residential areas gradually increases during the later period (21:30-22:00). Based on this, the entry passenger flow weights at venue connection stations and the exit passenger flow weights at transfer stations and stations near residential areas are adjusted. Simultaneously, combining external factor attention weights and set calibration coefficients, the station dwell time and line traffic weights are calibrated using the calibrated station dwell time and line traffic weight calculation formulas. The calibrated station dwell time and line traffic weight calculation formulas are as follows: Special attention is paid to the risk of congestion at event shuttle stations and the transportation capacity of routes under surges in passenger flow. Finally, based on the route selection preferences of spectators leaving the event in the individual travel behavior database, the route traffic weights are further adjusted. For example, the traffic weights of high-capacity routes or direct transfer routes that spectators prefer need to be adjusted according to actual demand to ensure that the adjusted dynamic topology can accurately reflect the passenger flow characteristics during the event's closing time and complete the dynamic topology adjustment.

[0041] Passenger flow simulation and extrapolation: Based on the adjusted topology framework, multi-agent simulation technology is used to construct simulated agents. The total number of agents is set according to the expected passenger volume of the shuttle route after the event ends, and must include both event spectators and regular travelers, with a focus on the additional passenger flow brought by event spectators. Each agent is matched with the spectator departure travel chain characteristics and activity mode tags from the individual travel behavior database. The simulation simulates the entire process of spectators departing from the event venue, walking to the shuttle station, entering the station, taking the subway including transfers, and exiting the station after arriving at their destination. The simulation process needs to reproduce the characteristics of concentrated passenger flow and gradual dispersal after the event ends. During the simulation, the passenger flow at each station at each time period is calculated in real time using the multi-agent simulation passenger flow calculation formula, which is as follows: Then, record and organize the passenger flow entering, exiting, and stranded at each station in detail, paying special attention to the peak entry volume at the event shuttle stations, tracking the evacuation volume at transfer stations and destination stations, assessing whether there is a risk of congestion at the stations, and forming preliminary passenger flow projection results. If an emergency occurs, such as the failure of some turnstiles at the event shuttle stations or the temporary speed limit on a certain section of the line, immediately update the external factor data, such as adjusting the external factor coefficients of the event shuttle stations, correcting the relevant parameters of the faulty lines, recalculating the attention weight of external factors and setting new calibration coefficients, and then conduct passenger flow projection again. The final projection results can provide a scientific basis for decision-making on emergency line scheduling during the event's end period, such as adding temporary trains, and guiding passenger flow at stations, such as adding temporary ticket gates and guiding diversion, to ensure the safe and orderly evacuation of passengers.

[0042] In summary, this study simulates passenger flow on subway shuttle lines after large-scale weekend sporting events. It collects physical parameters and historical operational data, focusing on parameters of shuttle stations and constructs a spatiotemporally heterogeneous topology model with core nodes and added relevant layers. It then calculates the attention weights for two external factors: weather and large-scale sporting events. Next, it builds a database of individual travel behaviors labeled with spectator departure tags based on IC card data and surveys. Finally, it adjusts the topology by combining the weights and the behavior database. Finally, it simulates passenger flow through multi-agent simulations, considering contingency plans, providing a basis for decision-making regarding emergency scheduling and safe passenger evacuation during event dispersal periods.

[0043] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention in any way. Although the present invention has been disclosed above with reference to preferred embodiments, it is not intended to limit the present invention. Any person skilled in the art can make some modifications or alterations to the above-disclosed technical content to create equivalent embodiments without departing from the scope of the present invention. Any simple modifications, equivalent changes and alterations made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the scope of the present invention.

Claims

1. A method for simulating and extrapolating subway passenger flow considering individual travel chains and activity patterns, characterized in that, The specific steps of this method are as follows: Constructing a topology framework: Collect physical parameters and historical operation data of urban subway lines, construct a spatiotemporal heterogeneous topology model and define topology nodes and edges, set weight values ​​for topology nodes and edges, and add an external factor input layer and a calibration layer to form a topology framework; External factor attention weight calculation: classify and quantify external factors to determine correlation coefficients, collect data on the intensity and spatial correlation of external factors, and calculate the weight of external factors on the target in each time period; Construct an individual travel behavior database: Obtain subway IC card data for a certain period of time and conduct user travel surveys. Clean and analyze the IC card data to extract individual travel chain characteristics. Combine the survey data to label each travel chain with activity mode tags, construct an individual travel behavior database, and set differentiated travel time distribution and transfer preferences for different activity modes. Dynamic topology adjustment: Based on the topology framework, external factor attention weights and individual travel behavior database data are introduced. By using calibrated formulas for calculating station dwell time and route passage weights, and combining this with the individual travel behavior database to correct route passage weights, dynamic topology adjustment is completed. The calibrated formulas for calculating station dwell time and route passage weights are as follows: ,in, For the first Target after time period calibration The weight, when hour, The node weight of the site is calibrated to obtain the site retention weight. hour, The edge weights of the line are used, and the line passage weights are obtained after calibration. For the first External factors during the period For the target The weight, For calibration coefficients; Perform passenger flow simulation and extrapolation: Based on the adjusted topology framework, construct intelligent agents through multi-agent simulation technology to simulate the entire travel process, calculate and record passenger flow data at each time period and organize and output it to complete the extrapolation of passenger flow changes. When encountering sudden scenarios, update the external factor data, recalculate the attention weight of external factors and set the calibration coefficient, and then extrapolate again.

2. The subway passenger flow simulation and extrapolation method considering individual travel chains and activity patterns according to claim 1, characterized in that, In constructing the topology framework, physical parameters of the target city's subway lines are collected, including the maximum capacity of each station. Maximum design speed of the line Maximum passenger capacity of the line Collect historical operational data for different periods and time periods, including station passenger flow density at each time period. Historical average passenger flow density of the station A spatiotemporal heterogeneous topology model is constructed based on physical parameters and historical operational data. Subway stations are defined as topological nodes, and line segments between adjacent stations are defined as topological edges. A weight value is set for each topological node using a dynamic topological node weight calculation formula, and a weight value is set for each topological edge using a dynamic topological edge weight calculation formula. The weight values ​​of nodes and edges evolve dynamically over time. An external factor input layer and a calibration layer are added to the spatiotemporal heterogeneous topology model. The external factor input layer receives collected external factor data, including weather, holidays, and large-scale events. The calibration layer stores the adjustment parameters of the external factor data on the weights of topological nodes and edges, forming the topological framework of a dynamically calibrated subway network.

3. The subway passenger flow simulation and extrapolation method considering individual travel chains and activity patterns according to claim 2, characterized in that, In the constructed topology framework, a weight value is set for each topology node using a dynamic topology node weight calculation formula. The dynamic topology node weight calculation formula is as follows: ,in, For the first Time Period Site The node weights, For the site Real-time passenger count For the site Maximum capacity For the first Time Period Site Passenger density, For the first Time Period Site Historical average passenger flow density These are the weighting coefficients.

4. The subway passenger flow simulation and extrapolation method considering individual travel chains and activity patterns according to claim 2, characterized in that, In the constructed topology framework, a weight value is set for each topology edge using a dynamic topology edge weight calculation formula, which is as follows: ,in, For the first Time-of-day routes edge weights, For the first Time-of-day routes Real-time speed of Chinese trains For the line The maximum speed set by the train For the first Time-of-day routes Real-time passenger volume, For the line The maximum passenger capacity is set. These are the weighting coefficients.

5. The subway passenger flow simulation and extrapolation method considering individual travel chains and activity patterns according to claim 1, characterized in that, In the calculation of attention weights for external factors, three categories of external factors are considered: weather, holidays, and large-scale events. The external factors are categorized, and their coefficients are quantified for each category. ; And collect the intensity of external factors at different time periods. Target acquisition through GIS map matching External factors Spatial correlation ,in, At that time, it was a station. The time is the route; then, the external factors for each time period are calculated using the external factor attention weighting calculation formula. For the target The weights are determined, and calibration coefficients are set based on the intensity and duration of the influence of external factors. .

6. The subway passenger flow simulation and extrapolation method considering individual travel chains and activity patterns according to claim 5, characterized in that, In the calculation of external factor attention weights, the external factors for each time period are calculated using the external factor attention weight calculation formula. For the target The weight of external factor attention is calculated using the following formula: ,in, For the first External factors during the period For the target The weight, External factors coefficient, For the first External factors during the period strength, For the goal External factors Spatial correlation.

7. The subway passenger flow simulation and extrapolation method considering individual travel chains and activity patterns according to claim 1, characterized in that, In the aforementioned dynamic topology adjustment, based on the topology framework of a dynamically calibrated metro network, attention weights for external factors are introduced. By combining individual travel behavior database data with the entry / exit percentages of users with different activity patterns in the database, the entry / exit passenger flow weights of stations are adjusted, and simultaneously... and The target is calculated using the calibrated formula for the weighting of station dwell time and line traffic. The weights are calibrated, and the route selection preferences of users with different activity patterns in the individual travel behavior database are combined to correct the route passage weights and complete the dynamic topology adjustment.

8. The subway passenger flow simulation and extrapolation method considering individual travel chains and activity patterns according to claim 1, characterized in that, In the passenger flow simulation and extrapolation, based on the adjusted topology framework, multi-agent simulation technology is used to set the total number of simulation agents according to the average daily passenger volume of the target city's subway. For each intelligent agent, travel chain features and activity pattern labels from the individual travel behavior database are matched; The simulation project covers the entire travel process of an intelligent agent from its departure station to its destination station, and uses a multi-agent simulation passenger flow calculation formula to calculate station flow in real time for each time period. The system records and organizes passenger flow data for each time period, including inbound, outbound, and lingering passenger flow, to predict changes in passenger flow. In case of unforeseen circumstances, it updates data on external factors. Recalculate and set Then we re-analyze the scenario.

9. A subway passenger flow simulation and extrapolation method considering individual travel chains and activity patterns according to claim 8, characterized in that, In the passenger flow simulation and extrapolation, the passenger flow at each station in real time is calculated using a multi-agent simulation passenger flow calculation formula. The multi-agent simulation passenger flow calculation formula is as follows: ,in, For the first Time Period Site Passenger flow The total number of simulated intelligent agents, For intelligent agents No. Is the time period within the site? Yes = 1, No = 0 For intelligent agents No. Probability of travel during a given time period.

Images