Method and system for crowd flow simulation analysis based on space-time behavior dynamics

Abstract

The present application relates to the technical field of human flow simulation analysis, and specifically provides a human flow simulation analysis method and system based on space-time behavior dynamics, which comprises the following steps: acquiring city multi-dimensional space-time heterogeneous data and standardizing; constructing a dynamic coupling mechanism of human behavior, space constraint and industrial function supply, and extracting travel decision rules and destination selection logic; taking the mechanism as a constraint, combining with subject modeling to construct a space-time behavior dynamics model, simulating space-time evolution of group flow and outputting human flow characteristic parameters; and predicting target scene human flow distribution and aggregation dissipation trend based on the parameters and visually displaying. The present application combines multi-factor dynamic coupling with subject modeling, improves simulation accuracy and scene adaptability, and the prediction result is intuitive, which can provide scientific decision support for city planning, traffic management, public security guarantee and the like.

Description

Technical Field

[0001] This invention relates to the field of pedestrian flow simulation and analysis technology, specifically to a pedestrian flow simulation and analysis method and system based on spatiotemporal behavioral dynamics. Background Technology

[0002] With the acceleration of urbanization and the diversification of urban functions, the spatiotemporal evolution of population flow has become a core research topic in urban planning, traffic management, and public safety. Accurate population flow simulation and prediction can provide scientific decision-making support for optimal allocation of urban resources, emergency response to public emergencies, and the organization and management of large-scale events, reducing urban operational risks and improving the efficiency of public services. Its technological value and application demand are increasingly prominent.

[0003] Currently, pedestrian flow simulation analysis technology mainly develops in three directions: statistical models (such as gravity models and radiation models), big data-driven models (such as algorithmic models based on neural networks and random forests), and agent modeling (ABM) technology. Among them, statistical models have simple structures and low computational costs, but they rely too much on empirical parameters and are difficult to characterize individual behavioral differences and spatiotemporal dynamic changes. Big data-driven models improve the real-time performance of predictions by using multi-source heterogeneous data such as mobile terminal location data, public transportation card swipe data, and video surveillance data, but they focus on the direct mapping from "data to results" and lack in-depth analysis of travel decision-making mechanisms. Although agent modeling technology can characterize the mapping from micro-behavior to macro-phenomena, its travel decision-making rules and destination selection logic are mostly fixed empirical settings and lack dynamic adaptability.

[0004] These technologies generally suffer from two core problems: First, they lack dynamic coupling of multiple factors and are not adaptable to specific scenarios. They fail to effectively construct the intrinsic relationship between population behavior characteristics, urban spatial constraints, and industrial functional supply, and cannot flexibly respond to the needs of different scenario types (such as commercial areas, transportation hubs, and cultural and tourism venues) or dynamic changes in scenarios (such as temporary traffic control and large-scale event hosting), resulting in limited prediction accuracy and generalization ability. Second, the prediction results are not practical enough. They are mostly presented as single-dimensional data or simple heat maps, lacking hierarchical visualization and decision-oriented analysis for actual application scenarios, making it difficult to directly support actual decision-making needs such as urban management and emergency response.

[0005] In summary, existing pedestrian flow simulation and analysis technologies are still insufficient to meet the application requirements of refinement, scenario-based approach, and high practicality. There is an urgent need for a pedestrian flow simulation and analysis method that can integrate multi-dimensional spatiotemporal data of the city, construct a multi-factor dynamic correlation mechanism, and achieve accurate prediction and efficient visualization. Summary of the Invention

[0006] To overcome the shortcomings of existing technologies, this invention provides a method and system for simulating and analyzing pedestrian flow based on spatiotemporal behavioral dynamics, in order to solve the problems in existing technologies.

[0007] One embodiment of the present invention provides a method for simulating and analyzing pedestrian flow based on spatiotemporal behavioral dynamics, comprising the following steps:

[0008] S10. Obtain multidimensional spatiotemporal heterogeneous data of the city, wherein the multidimensional spatiotemporal heterogeneous data includes at least population distribution data, urban spatial characteristic data, industrial function distribution data and facility layout data.

[0009] S20. Standardize the multidimensional spatiotemporal heterogeneous data to eliminate data format differences and noise interference, and obtain a standardized dataset with a unified spatiotemporal benchmark.

[0010] S30. Based on the standardized dataset, construct a dynamic coupling mechanism among population behavior characteristics, urban spatial constraints, and industrial function supply, and extract population travel decision rules and destination selection logic according to the dynamic coupling mechanism.

[0011] S40. Using dynamic coupling mechanism as the underlying constraint, construct a spatiotemporal behavior dynamics model using subject modeling technology, input the population travel decision rules and destination selection logic into the spatiotemporal behavior dynamics model, simulate the spatiotemporal evolution process of the population flow through a preset path planning algorithm, and output the population flow characteristic parameters based on the evolution results.

[0012] S50. Based on the crowd flow characteristic parameters, predict the crowd distribution data, aggregation and dissipation trends of the preset target spatiotemporal scene, and display the prediction results in a visual manner.

[0013] This application also relates to a pedestrian flow simulation and analysis system based on spatiotemporal behavioral dynamics, including:

[0014] The data acquisition module is used to acquire multidimensional spatiotemporal heterogeneous data of the city, which includes at least population distribution data, urban spatial characteristic data, industrial function distribution data, and facility layout data.

[0015] The processing module is used to standardize the multidimensional spatiotemporal heterogeneous data, eliminate data format differences and noise interference, and obtain a standardized dataset with a unified spatiotemporal benchmark.

[0016] The extraction module is used to construct a dynamic coupling mechanism among population behavior characteristics, urban spatial constraints, and industrial function supply based on the standardized dataset, and to extract population travel decision rules and destination selection logic according to the dynamic coupling mechanism.

[0017] The evolution module is used to construct a spatiotemporal behavior dynamics model with dynamic coupling mechanism as the underlying constraint and subject modeling technology. The population travel decision rules and destination selection logic are input into the spatiotemporal behavior dynamics model. The model simulates the spatiotemporal evolution process of the population flow through a preset path planning algorithm and outputs population flow characteristic parameters based on the evolution results.

[0018] The prediction module is used to predict the distribution data of people flow in a preset target spatiotemporal scene, the trend of gathering and dispersing, based on the crowd flow characteristic parameters, and to display the prediction results in a visual manner.

[0019] This application also relates to a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the above-described method for simulating and analyzing pedestrian flow based on spatiotemporal behavioral dynamics.

[0020] This application also relates to a computer-readable storage medium storing a computer program that, when executed by a processor, implements the steps of the above-described method for simulating and analyzing pedestrian flow based on spatiotemporal behavioral dynamics.

[0021] The above embodiments provide a method and system for simulating and analyzing pedestrian flow based on spatiotemporal behavioral dynamics, which has the following beneficial effects:

[0022] 1. By constructing a dynamic coupling mechanism of population behavior characteristics, urban spatial constraints and industrial function supply, it breaks through the limitations of traditional technologies that rely on multiple factors and fixed empirical parameters, and can accurately capture the intrinsic driving logic of population flow; combined with subject modeling technology to simulate the spatiotemporal evolution of the group, it achieves accurate mapping from micro behavior to macro phenomena, effectively adapts to different scenario types and dynamic change requirements, and significantly reduces prediction errors.

[0023] 2. Based on the simulated output of crowd flow characteristic parameters, it directly predicts the distribution data of crowds and the trends of gathering and dispersing, and displays them intuitively through visualization, solving the problems of the single presentation and poor readability of existing technologies; the prediction results can be directly connected with the actual needs of urban planning, traffic management, public safety, etc., providing a scientific basis for resource optimization, emergency response, and event organization, and improving the efficiency and rationality of decision-making. Attached Figure Description

[0024] Figure 1 A flowchart of a pedestrian flow simulation and analysis method based on spatiotemporal behavioral dynamics provided in this embodiment of the invention;

[0025] Figure 2 This is a schematic block diagram of a computer device provided in an embodiment of the present invention. Detailed Implementation

[0026] The technical solutions in the embodiments of the present invention will now be clearly and completely described in conjunction with the accompanying drawings.

[0027] Reference Figure 1 One embodiment of the present invention provides a method for simulating and analyzing pedestrian flow based on spatiotemporal behavioral dynamics, comprising the following steps:

[0028] S10. Obtain multidimensional spatiotemporal heterogeneous data of the city, wherein the multidimensional spatiotemporal heterogeneous data includes at least population distribution data, urban spatial characteristic data, industrial function distribution data, and facility layout data.

[0029] In this embodiment, the core of step S10 is to collect basic data to support the simulation analysis of pedestrian flow. Among them, multidimensional spatiotemporal heterogeneous data refers to diverse data containing different dimensions, types, and spatiotemporal attributes. Specifically, it includes: population distribution data reflecting the number and spatial distribution of people; urban spatial characteristic data reflecting spatial attributes such as urban road network structure and building layout; industrial function distribution data characterizing the industrial types and functional positioning of various urban areas; and facility layout data recording the location and scale of various service facilities and commercial facilities. This step mainly collects the above-mentioned data through sensor acquisition, public data acquisition, and other methods. The reason for this operation is that pedestrian flow is affected by multiple factors such as the distribution of people themselves, urban spatial constraints, industrial function attraction, and facility accessibility. Existing technologies often lead to simulation prediction deviations due to single or incomplete data. This embodiment provides a comprehensive and accurate basic data source for the subsequent construction of dynamic coupling mechanism, extraction of decision rules, and simulation of group flow through this step, ensuring the smooth implementation of subsequent steps and the reliability of results.

[0030] It should be noted that the acquisition and collection of the above-mentioned types of data are conventional technical means in this field. Those skilled in the art can obtain the data through the Internet from the official websites of urban planning departments, transportation departments, statistics departments, etc., public data channels such as national data open platforms and regional data sharing platforms, or from compliant commercial data service providers. Some real-time data can also be collected through traffic monitoring and urban operation-related sensing equipment deployed in the city. The data acquisition is convenient and accessible, and there is no need to create additional special acquisition methods.

[0031] S20. Standardize the multidimensional spatiotemporal heterogeneous data to eliminate data format differences and noise interference, and obtain a standardized dataset with a unified spatiotemporal benchmark.

[0032] In this embodiment, the core of step S20 is to standardize the collected multidimensional spatiotemporal heterogeneous data. Standardization refers to the process of uniformly standardizing heterogeneous data from different sources and in different formats. Data format differences refer to the inconsistencies in storage formats, field definitions, and units of measurement caused by different data collection channels. Noise interference refers to outliers, redundant information, random errors, and other content in the data that affect data quality. Unifying the spatiotemporal reference refers to calibrating all data to the same time scale and spatial coordinate system. This step primarily utilizes conventional data processing methods such as data format conversion, unified field mapping, outlier identification and removal, and redundant information filtering to achieve standardization. This is necessary because the multidimensional spatiotemporal heterogeneous data acquired in S10 comes from various sources, exhibiting inconsistent formats, noise interference, and inconsistencies in spatiotemporal benchmarks. Directly using such data for subsequent analysis can lead to deviations in the coupling mechanism construction and distortion of simulation results. Existing technologies often suffer from limited analytical accuracy due to neglecting data standardization. This embodiment obtains a standardized dataset that meets consistency, completeness, and reliability standards through this step, providing high-quality data support for subsequent dynamic coupling mechanism construction, decision rule extraction, and population flow simulation. This ensures the analytical accuracy and validity of subsequent steps from the data source.

[0033] S30. Based on the standardized dataset, construct a dynamic coupling mechanism among population behavior characteristics, urban spatial constraints, and industrial function supply, and extract population travel decision rules and destination selection logic according to the dynamic coupling mechanism.

[0034] In this embodiment, the core of step S30 is to establish a multi-factor correlation mechanism and extract the core logic of population flow. The dynamic coupling mechanism refers to the correlation model that describes the interaction and dynamic influence among population behavior characteristics, urban spatial constraints, and industrial function supply. Population behavior characteristics refer to travel-related behavioral attributes such as population travel preferences, activity duration, and flow frequency. Urban spatial constraints refer to spatial restrictions such as road network capacity, building distribution, and regional accessibility. Industrial function supply refers to the factors that attract people, such as the types of industries and service capabilities in different regions. Population travel decision rules are the basis for guiding people to choose travel time and mode. Destination selection logic is the inherent judgment standard for people to determine their travel destination. This step primarily involves data analysis to uncover the intrinsic connections among the three factors, quantify their mutual influence weights, construct a dynamic coupling model, and then extract travel decision-making rules and destination selection logic adapted to real-world scenarios from the model. This operation is necessary because existing technologies often treat factors such as crowd behavior, spatial environment, and functional supply in isolation, lacking a dynamic correlation mechanism. This results in the extracted decision-making rules and selection logic being detached from reality, thus affecting the simulation and prediction effects. This step breaks through the limitations of isolated multi-factor analysis, obtaining a correlation mechanism and core logic that fits the actual crowd flow patterns. This provides underlying theoretical support for the subsequent construction of spatiotemporal behavior dynamics models, improving the rationality and scenario adaptability of the model simulation.

[0035] S40. Using dynamic coupling mechanism as the underlying constraint, a spatiotemporal behavior dynamic model is constructed using subject modeling technology. The population travel decision rules and destination selection logic are input into the spatiotemporal behavior dynamic model. The spatiotemporal evolution process of the population flow is simulated through a preset path planning algorithm. Based on the evolution results, the population flow characteristic parameters are output.

[0036] In this embodiment, the core of step S40 is to construct a model and simulate the evolution of group flow to output core parameters. Here, the subject modeling technique refers to a modeling method that constructs independent subjects simulating individual behavior and drives the emergence of group behavior based on rules. The spatiotemporal behavioral dynamics model is a group flow simulation model that uses dynamic coupling mechanisms as constraints and integrates crowd travel decision rules and destination selection logic. The preset path planning algorithm refers to an algorithm used to calculate the optimal travel route of the subject from the origin to the destination (such as Dijkstra's algorithm, A...). The spatiotemporal evolution of population flow refers to the dynamic changes of a population over time and space. Population flow characteristic parameters are core indicators reflecting the patterns of population flow (such as peak spatiotemporal flow, population movement speed distribution, and regional aggregation intensity). This step primarily uses subject modeling technology to build a model framework, embedding a dynamic coupling mechanism into the model as a bottom-level constraint. Then, extracted travel decision rules and destination selection logic are input, and a preset path planning algorithm is invoked to drive the model's operation, simulating the spatiotemporal evolution of population flow and extracting key parameters. This operation is necessary because existing simulation models lack multi-factor dynamic coupling constraints, and the subject behavior rules are disconnected from reality, leading to distorted simulations of the evolution process and output parameters that fail to reflect real population flow patterns. This step constructs a simulation model that fits the actual scenario, achieving accurate replication of the spatiotemporal evolution of population flow. The output population flow characteristic parameters possess completeness and reliability, providing core data support for subsequent predictive analysis steps and further improving the overall simulation and prediction accuracy of the solution.

[0037] S50. Based on the crowd flow characteristic parameters, predict the crowd distribution data, aggregation and dissipation trends of the preset target spatiotemporal scene, and display the prediction results in a visual manner.

[0038] In this embodiment, the core of step S50 is to realize scenario-based prediction and result presentation based on the core parameters of the simulation output. Here, the preset target spatiotemporal scenario refers to the specific time range and spatial area that needs to be analyzed for pedestrian flow in advance. The pedestrian flow distribution data refers to the number and distribution of people in different spatial locations within the target scenario. The aggregation and dissipation trend refers to the dynamic change pattern of people converging and dispersing within the target scenario. The visualization method refers to the technical means of transforming abstract data and trends into intuitive presentation forms (such as heat maps, dynamic line graphs, spatial distribution maps, etc.). This step primarily involves predictive analysis, such as trend extrapolation and pattern fitting, of crowd flow characteristic parameters to obtain crowd distribution data and aggregation / dispersion trends in the target scenario. The prediction results are then presented in a visual format tailored to actual needs. This step is necessary because existing technologies often stop at simulating parameter output, lacking targeted scenario-based prediction and efficient display. This results in results that are difficult to directly connect with actual decision-making needs, limiting their practicality. This approach transforms abstract simulation parameters into concrete and intuitive prediction results, clearly presenting the crowd flow patterns in the target scenario. This allows relevant personnel to quickly obtain key information, providing direct support for practical decisions such as urban planning, traffic management, and security, thereby improving the practicality of the solution and the efficiency of decision-making.

[0039] In one embodiment, step S30, the construction of the dynamic coupling mechanism includes the following steps:

[0040] S311. Based on the industrial function distribution data and facility layout data in the standardized dataset, construct an attractiveness quantification calculation rule. The attractiveness quantification calculation rule takes POI type priority, facility density and service coverage radius as input parameters and outputs the population attraction intensity value of each spatial area.

[0041] S312. Based on urban spatial feature data in the standardized dataset, construct a spatial constraint intensity quantification calculation rule. The spatial constraint intensity quantification calculation rule takes road network traffic capacity, building barrier coefficient and open space carrying capacity threshold as input parameters, and outputs the population flow constraint intensity value of each spatial area.

[0042] S313. Based on the population distribution data in the standardized dataset, divide the population into at least two types of behavioral profiles, and define a behavioral preference factor for each type of profile. The behavioral preference factor includes destination type preference, route selection priority, and spatial tolerance threshold.

[0043] S314. Construct a scenario-based dynamic weight matrix, wherein the scenario-based dynamic weight matrix dynamically allocates the correlation weights of the crowd attraction intensity value, the correlation weights of the crowd flow constraint intensity value, and the correlation weights of the behavioral preference factor according to the scenario type.

[0044] S315. Based on the crowd attraction intensity value, crowd flow constraint intensity value, behavioral preference factor, and corresponding correlation weights, a quantitative coupling function is constructed. The crowd behavior decision probability is calculated through the quantitative coupling function to form the dynamic coupling mechanism. The quantitative coupling function satisfies the following calculation formula:

[0045] The probability of crowd behavior decision = crowd attraction intensity value × its association weight + crowd mobility constraint intensity value × its association weight + behavior preference factor × its association weight.

[0046] In this embodiment, the core of steps S311-S315 is to stratify and quantify industrial functional attraction, urban spatial constraints, and population behavior preferences, and integrate them with scenario-based dynamic weight allocation and a unified quantitative coupling function. This addresses the pain points of existing technologies, such as fuzzy multi-factor correlations, lack of quantitative support, fixed weights, and poor scenario adaptability. It ensures that the dynamic coupling mechanism possesses computability, dynamic adaptability, and accuracy, providing solid quantitative model support for the subsequent extraction of population travel decision rules and destination selection logic, further improving the scientific rigor and reliability of the overall population flow simulation analysis. Specifically:

[0047] The core of step S311 is to quantify the attractiveness of each spatial area to the population, which is the foundation for building a dynamic coupling mechanism. Among them, the attractiveness quantification calculation rules refer to the rule system that transforms the attractiveness of industrial functions and facility layout to the population into calculable values. POI type priority refers to the attraction priority score of different types of POI to the population (e.g., commercial complexes 8 points, schools 6 points, office parks 5 points, parks 4 points). Facility density refers to the number of facilities within a unit space (e.g., 50 facilities per square kilometer is high density, 20-50 is medium density, and less than 20 is low density). Service coverage radius refers to the spatial range within which facilities can effectively serve the population (e.g., community supermarket service coverage radius 1 kilometer, large business district service coverage radius 3 kilometers). The population attraction intensity value refers to a quantitative indicator that comprehensively reflects the attractiveness of the area (with a value range of 0-10 points). This step primarily involves prioritizing POI types in the standardized dataset, statistically analyzing facility density in each region, defining the service coverage radius of different facilities, and then constructing a weighted calculation model (e.g., population attraction intensity value = POI type priority × 0.4 + facility density standardized value × 0.3 + service coverage radius adaptation coefficient × 0.3, where the service coverage radius adaptation coefficient = distance from the target area to the facility / service coverage radius, with a higher coefficient for closer distances). Substituting the parameters, the attraction intensity value for each region is calculated. This operation is performed because existing technologies do not quantify the attraction effect of industrial functions and facilities, resulting in a lack of accurate data support for the coupling mechanism. This step transforms abstract attraction factors into specific, calculable quantitative indicators, providing standardized attraction capacity data for the dynamic coupling mechanism and improving the quantitative accuracy and operability of the mechanism.

[0048] The core of step S312 is to quantify the degree of constraint of urban space on the flow of people, which complements the attraction quantification in S311. Among them, the spatial constraint intensity quantification calculation rules refer to the rule system that transforms the restrictive effect of urban spatial attributes on the flow of people into calculable values. Road network capacity refers to the maximum number of people that a road can accommodate per unit time (e.g., 1,500 people per hour on a two-way four-lane main road and 300 people per hour on a community side road). Building barrier coefficient refers to the degree of obstruction of the flow of people by the distribution of buildings (e.g., 0.8 for continuous building strips, 0.3 for scattered building distributions, and 0.1 for areas without buildings). Open space carrying capacity threshold refers to the maximum population density that outdoor open areas can accommodate (e.g., 5 people / square meter in urban squares and 3 people / square meter on street pedestrian walkways). The population flow constraint intensity value refers to a quantitative index that comprehensively reflects the spatial constraint capacity of the area (with a value range of 0-10 points, the higher the score, the stronger the constraint). This step mainly involves determining traffic capacity by consulting road network design standards, calculating barrier coefficients based on building density, setting carrying capacity thresholds with reference to industry standards, and then constructing a comprehensive calculation model (for example, the pedestrian flow constraint intensity value = road network traffic capacity reverse standardized value × 0.4 + building barrier coefficient × 0.3 + open space carrying capacity threshold saturation coefficient × 0.3, where saturation coefficient = actual population density in the area / carrying capacity threshold). Substituting the parameters, the constraint intensity value for each area is obtained. This operation is performed because the fuzzy description of urban spatial constraints can lead to the coupling mechanism failing to accurately characterize the impact of space on pedestrian flow. Its beneficial effect is to achieve quantitative modeling of spatial constraint factors, provide standardized constraint capacity data for dynamic coupling mechanisms, and make multi-factor coupling more scientific.

[0049] The core of step S313 is to accurately characterize the individual behavioral differences of the population and inject personalized features into the coupling mechanism. Among them, the behavioral profile is a group of people classified according to their travel habits, purposes and other characteristics (e.g., three core profiles: commuters, leisure travelers and business travelers). The behavioral preference factor refers to the quantitative parameters that reflect the behavioral tendencies of different groups. The destination type preference refers to the weight of the population's preference for different POI types (e.g., commuters prefer transportation hubs with a weight of 0.7, office parks with a weight of 0.2, and others with a weight of 0.1; leisure travelers prefer business districts with a weight of 0.8, parks with a weight of 0.15, and others with a weight of 0.05). The route selection priority refers to the order in which the population prioritizes route attributes when traveling (e.g., commuters prioritize the shortest travel time with a weight of 0.8; leisure travelers prioritize the best scenery with a weight of 0.7). The spatial tolerance threshold refers to the population's acceptance of crowded environments (e.g., the spatial tolerance threshold for commuters is ≤0.6 for crowding and ≤0.4 for leisure travelers; crowding = actual number of people / area carrying capacity). This step mainly uses the K-means clustering algorithm to classify the population distribution data and related travel data, dividing them into different behavioral profiles. Then, by statistically analyzing the travel characteristics of each profile, the corresponding preference factor quantification value is defined. This operation is performed because existing technologies mostly adopt uniform behavioral assumptions, ignoring individual differences and causing the coupling mechanism to deviate from reality. This step achieves a refined characterization of population behavior, provides personalized behavioral data support for the dynamic coupling mechanism, and makes the subsequent decision probability calculation more in line with the actual choice logic of different groups of people.

[0050] The core of step S314 is to realize the scenario-based adaptation of multi-factor weights and improve the flexibility of the coupling mechanism. The scenario-based dynamic weight matrix refers to the matrix model that dynamically adjusts the associated weights of each quantitative factor according to different application scenarios. The scenario type refers to the specific scenario that needs to be simulated for pedestrian flow (e.g., three typical scenarios: weekday commuting scenario, weekend leisure scenario, and large-scale event scenario). The associated weight refers to the contribution ratio of each quantitative factor in the coupling calculation (the total weight is 1). This step primarily involves analyzing the core influencing factors of crowd flow in different scenarios and setting weight allocation rules for each factor (e.g., weekday commuting scenario: crowd attraction intensity value weight 0.3, crowd flow constraint intensity value weight 0.5, behavioral preference factor weight 0.2; weekend leisure scenario: crowd attraction intensity value weight 0.5, crowd flow constraint intensity value weight 0.2, behavioral preference factor weight 0.3; large-scale event scenario: crowd attraction intensity value weight 0.4, crowd flow constraint intensity value weight 0.4, behavioral preference factor weight 0.2). A matrix is ​​then constructed to store the weight combinations for different scenarios. This operation is performed because fixed weights cannot adapt to the differences in the degree of influence of multiple factors in different scenarios, resulting in insufficient generalization ability of the coupling mechanism. This step enables dynamic adjustment of weights, allowing the coupling mechanism to accurately match the core needs of different scenarios and improve the scenario adaptability of the overall solution.

[0051] The core of step S315 is to integrate the preceding quantitative indicators with dynamic weights to form a complete dynamic coupling mechanism. Here, the quantitative coupling function refers to a mathematical model that integrates and calculates multi-factor quantitative indicators with associated weights. The population behavior decision probability refers to the probability value (range 0-1, closer to 1 indicates a higher probability) of a population choosing a specific travel behavior (such as going to a destination or choosing a route). This step mainly involves multiplying the population attraction intensity value obtained in S311, the population flow constraint intensity value obtained in S312, and the behavioral preference factor obtained in S313 with the associated weights of the corresponding scenarios in the weight matrix of S314, and then summing them according to the formula to obtain the decision probability (for example, in a weekend leisure scenario, the attraction intensity value of a certain area is 8 × weight 0.5 = 4, the constraint intensity value is 3 × weight 0.2 = 0.6, the leisure group behavioral preference factor is 0.8 × weight 0.3 = 0.24, and the decision probability = 4 + 0.6 + 0.24 = 4.84, which is 0.48 after standardization). 4) This leads to the formation of a computable and dynamically adjustable coupling mechanism. This operation is performed because the previous steps have completed the quantification and weight allocation of each factor. It is necessary to achieve the organic coupling of multiple factors through a unified function to solve the problem of fuzzy multi-factor correlation in existing technologies. Through this step, the dynamic correlation of population behavior characteristics, urban spatial constraints, and industrial functional supply is transformed into a quantifiable mathematical model, forming a complete dynamic coupling mechanism. This provides accurate quantitative basis for the subsequent extraction of travel decision rules and destination selection logic, and greatly improves the scientificity and fit of the rules and logic.

[0052] It should be noted that the specific calculated values, weight allocation ratios, example scenarios, algorithm selections, and quantization model forms mentioned above are merely illustrative descriptions to facilitate understanding of the technical solutions by those skilled in the art, and do not constitute a limitation on the scope of protection of this invention. Those skilled in the art can reasonably adjust or replace the above values, ratios, models, etc., according to actual application scenarios, data types, and accuracy requirements; such adjustments and replacements all fall within the scope of the technical solutions protected by this invention.

[0053] In one embodiment, step S30, which involves extracting the population travel decision rules and destination selection logic based on the dynamic coupling mechanism, specifically includes the following steps:

[0054] S321. Based on the population behavior decision probability, behavior preference factor, and scenario-based dynamic weight matrix output by the dynamic coupling mechanism, extract population travel decision rules. The extracted population travel decision rules include:

[0055] Based on the time dimension weight changes of the scenario-based dynamic weight matrix, travel time selection rules are generated.

[0056] Based on the correlation between route selection priority and the probability of population behavioral decisions in the behavioral preference factors, travel mode selection rules are generated;

[0057] Based on the matching relationship between the spatial tolerance threshold and the intensity of population flow constraints in the behavioral preference factors, a travel destination priority rule is generated.

[0058] S322. Based on the crowd attraction intensity value, crowd behavior decision probability, and behavior profile output by the dynamic coupling mechanism, extract the destination selection logic. The construction process of the destination selection logic includes:

[0059] Based on the descending order of crowd attraction intensity values, the top-N candidate destinations are selected, where N is a preset positive integer;

[0060] Based on the destination type preferences of the behavioral profile, filter out candidate destinations that do not match;

[0061] Based on the probability of decision-making based on population behavior, the remaining candidate destinations are prioritized to form a complete destination selection logic.

[0062] In this embodiment, the core of steps S321-S322 is to extract travel decision-making rules and destination selection logic hierarchically and multidimensionally by relying on the quantitative output of the dynamic coupling mechanism. This addresses the pain points of existing technologies, such as the lack of scenario adaptability, low personalization matching, and failure to combine multi-factor quantitative correlations in the rules and logic. It ensures that the extracted rules and logic possess accuracy, practicality, and scenario flexibility, providing core behavioral guidance for the spatiotemporal behavioral dynamics model that aligns with actual population flow patterns, further solidifying the rationality and accuracy of the model simulation. Specifically:

[0063] The core of step S321 is to extract scenario-based and personalized travel decision rules based on the quantitative output of the dynamic coupling mechanism. This is a key step in transforming the coupling mechanism into practical logic. Among them, the travel decision rules refer to the clear basis for guiding people to make choices on travel time, mode, and purpose priority. The time dimension weight change of the scenario-based dynamic weight matrix refers to the dynamic adjustment of the weight of each coupling factor in different time periods (for example, in the weekday commuting scenario, the spatial constraint intensity weight is 0.5 in the morning peak from 7-9 am, 0.5 in the evening peak from 5-7 pm, and 0.3 in other time periods). The travel time selection rule refers to the criteria for people to select travel time windows. The travel mode selection rule refers to the basis for people to choose travel modes such as public transportation, driving, and walking. The travel purpose priority rule refers to the criteria for people to rank travel purposes such as commuting, leisure, and shopping. This step primarily involves analyzing the temporal variation patterns of the scenario-based dynamic weight matrix (e.g., the weight difference between peak and off-peak hours) to generate travel time selection rules adapted to the characteristics of each time period (e.g., when the morning peak weight is spatially constrained, priority is given to off-peak travel or high-traffic-efficiency periods); by mining the positive correlation between path selection priority (e.g., for commuters, "shortest time" priority is 0.8) and the probability of behavioral decisions, travel mode selection rules are generated (e.g., when the path selection priority is "shortest time" and the decision probability is ≥0.7, priority is given to efficient travel modes such as subways and BRT); and by matching... The spatial tolerance threshold (e.g., leisure travelers ≤ 0.4) and the crowd flow constraint strength value (e.g., constraint strength value of a certain area 0.3) in the behavioral preference factors are used to generate travel purpose priority rules (e.g., when the spatial tolerance threshold is low and the constraint strength value is below the threshold, leisure travel purposes are given priority). This operation is performed because the travel decision rules of existing technologies are mostly set based on fixed experience and do not take into account the dynamic changes of the scenario and the differences in individual behavior, resulting in the rules being out of touch with reality. This step realizes the scenario-based adaptation and personalized matching of decision rules, making the rules more in line with the actual travel logic of the population and providing accurate behavioral guidance for subsequent model simulation.

[0064] The core of step S322 is to construct a destination selection logic that filters layers and fits the preferences of the target audience. It is a precise replication of the destination selection process of the target audience. The destination selection logic refers to the progressive judgment process by which the target audience determines the final destination from the candidate destinations. The TOP-N candidate destinations refer to the top N highly attractive areas selected after sorting them from high to low attraction intensity values ​​(for example, N=5, filtering the top 5 business districts, parks, transportation hubs, etc. with attraction intensity values). The preset positive integer N can be flexibly set according to the needs of the scenario (for example, N=3 for daily travel and N=8 for long-distance travel). This step primarily involves first sorting the attraction intensity values ​​of each region in descending order to filter out the top-N highly attractive candidate destinations, ensuring that the candidate range focuses on the core attraction areas. Then, based on destination type preferences derived from behavioral profiles (e.g., commuters prefer transportation hubs and office parks), mismatched candidate destinations are filtered out (e.g., parks and amusement parks are removed from the commuter candidate list), achieving personalized adaptation of candidate destinations. Finally, the remaining candidate destinations are prioritized based on the decision probability of the population behavior (e.g., a decision probability of 0.85 for an office park and 0.78 for a transportation hub), forming a complete selection logic of "high attraction + high preference + high probability." This operation is performed because existing destination selection logics often rely on only a single attraction factor without considering population preferences and decision probabilities, leading to distorted selection results. This step constructs a multi-dimensional, progressive destination selection logic that ensures both destination attractiveness and alignment with population behavioral preferences and actual decision probabilities, making the subsequent model simulation of the destination selection process closer to real-world scenarios.

[0065] It should be noted that the specific numerical examples (such as N=5, decision probability threshold 0.7), the associated conditions for rule generation, and the progressive order of the filtering logic involved in this explanation are merely illustrative examples to facilitate understanding of the technical solution by those skilled in the art, and do not constitute a limitation on the scope of protection of this invention. Those skilled in the art can reasonably adjust or optimize the above numerical values, conditions, and order according to the actual scenario type, the complexity of the user profile, and the data accuracy requirements; such adjustments and optimizations all fall within the scope of the technical solution protected by this invention.

[0066] In one embodiment, step S40 specifically includes the following steps:

[0067] S410. Define the subject attributes and set the subject behavior rules. Embed the dynamic coupling mechanism as the underlying constraint into the model to construct a model with a three-layer architecture including subject attributes, subject behavior rules, and dynamic coupling constraints. Perform calibration, optimization, and error analysis on the model to output a spatiotemporal behavior dynamic model. The dynamic coupling constraint is the constraint formed on the subject behavior by the dynamic coupling mechanism through the output quantization results.

[0068] S420. Define a population travel decision rule system based on population travel decision rules, construct a destination selection logic unit based on destination selection logic, and configure a preset path planning algorithm for the model. Input the population travel decision rule system, destination selection logic unit and preset path planning algorithm into the spatiotemporal behavior dynamics model, use the spatiotemporal behavior dynamics model to simulate the spatiotemporal evolution process of the group flow and record the group flow status simultaneously, and extract and output the population flow characteristic parameters based on the evolution results.

[0069] In this embodiment, steps S410-S420 primarily address the pain points of existing technologies—such as lack of dynamic constraints, inefficient core logic calls, and incomplete simulation state recording—through a two-step approach of "model architecture construction + core component integration." This ensures that the spatiotemporal behavioral dynamics model possesses structural integrity, dynamic constraints, and simulation accuracy. Simultaneously, it achieves systematic simulation of group flow evolution and comprehensive extraction of characteristic parameters, further solidifying the overall simulation capability of the solution. Specifically:

[0070] The core of step S410 is to construct a spatiotemporal behavioral dynamic model with dynamic constraints and adaptability. This is the core foundation of the "modeling and simulation" step S40. Here, subject attributes refer to the core characteristic parameters of the simulated individual (e.g., subject identifier ID001, behavioral profile associated label "commuter", initial location "XX community", activity demand threshold "one commute per day"). Subject behavior rules refer to the clear criteria that regulate the subject's actions. Dynamic coupling constraints refer to the restrictive effect of quantitative results such as the probability of crowd behavior decisions and the strength of flow constraints output by the dynamic coupling mechanism on the subject's behavior (e.g., when the constraint strength value is ≥0.6, the subject must not enter the area). The three-layer architecture refers to the model structure based on subject attributes, guided by subject behavior rules, and bounded by dynamic coupling constraints. The preset accuracy threshold refers to the upper limit of the allowable deviation between the model simulation results and the actual data (e.g., the error between the simulated value and the actual value is ≤10%). This step primarily involves defining subject attributes by combining prior behavioral profiles and population distribution data, setting subject behavioral rules based on population behavior decision rules, embedding a dynamic coupling mechanism into the model as the underlying constraint to build a three-layer architecture, and then calibrating the model using historical pedestrian flow monitoring data (such as urban traffic monitoring records and regional pedestrian flow statistics). Error analysis (such as mean square error calculation) is used to adjust behavioral rule parameters and constraint weights until the model accuracy meets a preset threshold. This operation is performed because existing subject modeling techniques lack dynamic coupling constraint support, the model architecture is simplistic, and it has not undergone precise calibration, leading to distorted simulation results. This step constructs a complete, dynamically constrained, and accurate spatiotemporal behavioral dynamics model, providing a high-performance model carrier for subsequent group flow simulations and ensuring the rationality and accuracy of the simulation process.

[0071] The core of step S420 is to integrate core logic and algorithms to drive the model to complete the group flow simulation and output key parameters. It is the core execution link of S40's "simulation output." Here, the group travel decision rule system refers to a systematic set integrating rules for travel time, mode, and destination priority; the destination selection logic unit refers to a functional module that encapsulates destination selection logic and can be directly called by the model; the group flow status refers to real-time information such as the position, number, and direction of movement of the subjects at various spatiotemporal nodes during the simulation (e.g., 200 subjects moving forward on road segment XX at 10:00); and the group flow characteristic parameters refer to quantitative indicators reflecting the core laws of group flow. This step mainly integrates the three types of travel decision rules extracted in S321 to form a rule system, encapsulates the destination selection logic of S322 into an independent functional unit, and configures a preset path planning algorithm adapted to the scenario (e.g., A for urban road scenarios). The algorithm used (Dijkstra's algorithm) for the walking scenario is selected. These three data points are then input into the spatiotemporal behavior dynamics model output by S410. The model generates an initial group based on the subject attributes and drives the group to move spatiotemporally according to the rule system and logical units. The flow state of each spatiotemporal node is recorded synchronously. Finally, feature parameters are extracted from the recorded data. This operation is performed because existing simulations lack a systematic rule system and an efficient logical calling mechanism, and do not fully record the flow state, resulting in incomplete parameter extraction. This step realizes the systematic simulation and full-state recording of the spatiotemporal evolution of the group flow. The output feature parameters are complete and accurate, providing core data support for the subsequent predictive analysis of S50.

[0072] It should be noted that the model architecture details, preset accuracy thresholds, algorithm selections, and examples of main attributes involved in this explanation are merely illustrative descriptions to facilitate understanding of the technical solutions by those skilled in the art, and do not constitute a limitation on the scope of protection of this invention. Those skilled in the art can reasonably adjust or replace the above architecture, thresholds, algorithms, etc., according to the complexity of the actual simulation scenario, data scale, and accuracy requirements; such adjustments and replacements all fall within the scope of the technical solutions protected by this invention.

[0073] In one embodiment, step S410 specifically includes the following steps:

[0074] S411. Based on the behavioral profile and population distribution data, define the main attributes of the model, including the main identifier, behavioral profile association tags, initial location, and activity demand threshold.

[0075] S412. Set subject behavior rules, which include movement rules, aggregation rules and scene adaptation rules. The movement rules dynamically adjust the subject's movement direction and path based on the probability of crowd behavior decisions. The aggregation rules limit the subject's aggregation scale based on the open space carrying capacity threshold. The scene adaptation rules match the corresponding behavior preference factors based on the scene type.

[0076] S413. The dynamic coupling mechanism is used as the underlying constraint embedding model to construct a model with a three-layer architecture including subject attributes, subject behavior rules and dynamic coupling constraints. The underlying constraints dynamically correct the subject behavior decisions by calling the probability of crowd behavior decisions and the strength value of crowd flow constraints output by the dynamic coupling mechanism in real time.

[0077] S414. Obtain historical pedestrian flow monitoring data, use the historical pedestrian flow monitoring data to calibrate and optimize the model, and adjust the parameters of the behavior rules and the weights of the dynamic coupling constraints through error analysis so that the simulation accuracy of the model meets the preset accuracy threshold.

[0078] S415. Based on the calibration optimization and error analysis adjustment results, output a spatiotemporal behavior dynamics model. The spatiotemporal behavior dynamics model can automatically match the corresponding subject behavior rules and dynamic coupling constraint parameters according to the type of the target spatiotemporal scene.

[0079] In this embodiment, steps S411-S415 primarily address the pain points of existing technologies—such as lack of model subject differentiation, rigid behavioral rules, lack of dynamic constraints in the architecture, insufficient accuracy, and poor scene adaptability—through a progressive process of "personalized attribute definition → dynamic rule setting → three-layer architecture construction → historical data calibration → adaptive model output." This ensures that the output spatiotemporal behavioral dynamics model possesses personalization, dynamism, high accuracy, and scene adaptability, providing solid model support for subsequent group flow simulations. Specifically:

[0080] Step S411 is a detailed implementation of "defining subject attributes" in S410. The core is to give the model subject personalized attributes that fit the characteristics of the actual population. Among them, the subject identifier refers to the unique code used to distinguish different simulated individuals (e.g., ID001~ID1000), the behavior profile association tag refers to the identifier that binds the subject to the population type (e.g., "commuters", "leisure group", "business group"), the initial position refers to the spatial coordinates of the subject's simulation start (e.g., "East gate of XX community, longitude 116.4°, latitude 39.9°"), and the activity demand threshold refers to the upper limit of the number of trips and duration of the subject within the simulation period (e.g., commuters travel 2 times a day, with a single trip duration ≤120 minutes). This step mainly uses the behavioral profiles defined in the preceding S313 and the population distribution data obtained in S10 to assign a unique identifier to each simulated subject and bind the corresponding behavioral profile label. The initial location of the subject is determined based on the population distribution data (e.g., more subjects are assigned to high-density residential areas). The activity demand threshold is set in combination with the profile features. This operation is performed because the existing model has a single subject attribute and does not reflect the individual differences of the population, resulting in a lack of realism in the simulation. This step achieves personalized and scenario-based matching of subject attributes, making the simulated subjects closer to the characteristics of real people and providing an accurate individual basis for subsequent behavioral simulation.

[0081] Step S412 is a detailed implementation of "setting subject behavior rules" in S410. The core is to formulate dynamically adaptable action guidelines for the subject. Among these, movement rules refer to the guidelines that regulate the subject's movement behavior (e.g., when the probability of a crowd behavior decision is ≥0.7, the subject prioritizes moving along the shortest path; when the probability is <0.5, the subject adjusts its movement direction to avoid highly constrained areas). Aggregation rules refer to the guidelines that limit the density of subject aggregation (e.g., based on the open space carrying capacity threshold of 5 people / ...). When the number of entities in an area reaches its carrying capacity limit, new entities are not allowed to enter that area. Scene adaptation rules refer to the criteria by which entities switch behavioral preferences based on scene type (for example, when entering a weekend leisure scene, the entity automatically matches the leisure group behavioral preference factor). This step mainly combines the crowd behavior decision probability in S315, the open space carrying capacity threshold in S312, and the scene type in S314 to formulate three types of rules: movement, gathering, and scene adaptation. The triggering conditions and execution standards of these rules are clearly defined. This operation is performed because existing entity behavior rules are fixed and rigid, unable to adapt to dynamic scenes and individual differences, causing simulated behavior to deviate from reality. This constructs a dynamic, multi-dimensional entity behavior rule system, allowing entity behavior to respond to scene changes and individual preferences, thus improving the realism of the simulation process.

[0082] Step S413 is a detailed implementation of S410, "Constructing a three-layer architecture model." Its core is integrating a dynamic coupling mechanism into the model to form constraint boundaries. The three-layer architecture refers to a hierarchical structure with the subject's attributes as the base layer, the subject's behavioral rules as the execution layer, and the dynamic coupling constraints as the constraint layer (the base layer provides individual characteristics, the execution layer regulates actions, and the constraint layer defines boundaries). Real-time invocation of the bottom-level constraints refers to retrieving the latest quantitative results (decision probability, constraint strength value) of the dynamic coupling mechanism at preset intervals (e.g., 5 minutes) during model operation. This step primarily embeds the dynamic coupling mechanism into the constraint layer through model architecture design, sets the real-time invocation frequency, and clarifies the constraint correction logic (e.g., when the constraint strength value ≥ 0.6, corrects the subject's movement rules, prohibiting entry into that area), thus building a complete three-layer architecture. This operation is performed because existing models do not use multi-factor coupling mechanisms as bottom-level constraints, resulting in a lack of dynamic boundary constraints on subject behavior and distorted simulation results. This step achieves deep integration of the dynamic coupling mechanism and the model architecture, ensuring that subject behavior always remains within realistic constraint boundaries, thus improving the model's logical rigor.

[0083] Step S414 is a detailed implementation of S410 "calibration optimization and error analysis adjustment". Here, historical pedestrian flow monitoring data refers to the actual pedestrian flow data of a specific period and area in the past (such as the hourly pedestrian flow of a transportation hub in March 2025, and the gathering density data of a business district on weekends, which are conventionally available data in this field). Error analysis refers to comparing the deviation between the simulation results of the model and the historical actual data (for example, using mean square error (MSE) and mean absolute error (MAE) for calculation). The preset accuracy threshold refers to the minimum simulation accuracy that the model needs to achieve (for example, the error between the simulated value and the true value is ≤10%, that is, the accuracy is ≥90%). This step primarily involves acquiring historical pedestrian flow monitoring data for the target scenario, using it as baseline data to input into the model, running the model to obtain simulation results, calculating the error between the simulation results and the baseline data, and adjusting the parameters of the subject's behavioral rules (such as the moving speed threshold and the upper limit of the gathering density) and the weights of the dynamic coupling constraints (such as the attraction intensity value weight ±0.1) based on the error feedback. This process is iterated repeatedly until the error is ≤ the preset accuracy threshold. This operation is performed because the initial parameters of the model deviate from the actual scenario, and lack of calibration will lead to insufficient simulation accuracy. This step, through historical data calibration and error adjustment, significantly improves the simulation accuracy of the model, ensuring that the model output results are reliable and practical.

[0084] Step S415 is a detailed implementation of S410, "outputting a spatiotemporal behavioral dynamics model." Its core is to output a final model with scene adaptability. Automatic matching refers to the model receiving the target spatiotemporal scene type (e.g., weekday commuting scene, large-scale event scene) and automatically invoking the corresponding main behavioral rules (e.g., invoking the "time-priority" movement rule in a commuting scene) and dynamically coupled constraint parameters (e.g., adjusting the constraint strength weight to 0.4 in a large-scale event scene) without manual intervention. This step mainly integrates the calibration and optimization results from S414, pre-setting a scene-rule-parameter matching mapping relationship in the model (e.g., scene type → set of behavioral rules → combination of constraint parameters), completing model encapsulation and output. This operation is performed because existing models lack scene adaptability, requiring manual parameter adjustment to fit different scenes, which is cumbersome and inefficient. This enables the output model to have automatic scene adaptability, quickly responding to the simulation needs of different target spatiotemporal scenes, improving the model's practicality and ease of operation.

[0085] It should be noted that the specific values ​​such as the main identifier code, initial position coordinates, preset precision threshold, and iterative adjustment range involved in this explanation are merely illustrative examples to facilitate understanding of the technical solution by those skilled in the art, and do not constitute a limitation on the scope of protection of this invention. Those skilled in the art can reasonably adjust the above values, rule triggering conditions, and calling frequency according to the scale of the simulation scenario, data precision, and actual needs; such adjustments all fall within the scope of the technical solution protected by this invention.

[0086] In one embodiment, step S420 specifically includes the following steps:

[0087] S421. Define a population travel decision-making rule system, which includes travel time selection rules, travel mode selection rules, and travel destination priority rules.

[0088] S422. Construct a destination selection logic unit, wherein the destination selection logic unit takes the probability of crowd behavior decision as input and outputs the target destination set and selection probability of the subject based on the behavioral profile and associated tags.

[0089] S423. Configure a preset path planning algorithm, wherein the path planning algorithm includes a basic path algorithm and a dynamic adjustment module;

[0090] S424. Input the population travel decision-making rule system, destination selection logic unit and route planning algorithm into the spatiotemporal behavior dynamics model, which generates an initial population based on subject attributes;

[0091] S425. Based on the population travel decision-making rule system, simulate the spatiotemporal evolution process of the initial population flow. The spatiotemporal evolution process includes destination selection, path planning and movement evolution in sequence, and the population flow status of each spatiotemporal node is recorded synchronously during the spatiotemporal evolution process.

[0092] S426. Based on the simulated spatiotemporal evolution results, extract and output the population flow characteristic parameters, including the spatiotemporal flow peak, the distribution of group movement speed, the destination aggregation density, and the proportion of path selection preference.

[0093] In this embodiment, steps S421-S426 are essentially a complete process of "rule systematization → logical modularization → algorithm dynamization → population generation → ordered simulation → parameter extraction," which addresses the pain points of inefficient core component invocation, chaotic simulation processes, incomplete state recording, and single feature parameter extraction in existing technologies. This ensures the orderliness, realism, and comprehensiveness of the spatiotemporal evolution simulation of population flow, and the output population flow feature parameters possess completeness and accuracy, laying a solid data foundation for subsequent prediction and display steps. Specifically:

[0094] Step S421 is a detailed implementation of S420, "Defining the Crowd Travel Decision-Making Rule System." Its core is to integrate the previously extracted rules into a systematic set. The crowd travel decision-making rule system refers to a set of directly callable rules formed by logically associating the three types of rules (e.g., a progressive rule chain of "scenario type → travel time period → travel mode → travel purpose"). This step mainly involves organizing the travel time period selection rules, travel mode selection rules, and travel purpose priority rules extracted in S321 according to the logical order of "scenario adaptation → time period selection → mode determination → purpose sorting," clarifying the triggering and correlation relationships between rules (e.g., in a weekday commuting scenario, the morning rush hour rule is triggered first, then the efficient travel mode rule is matched, and finally the commuting purpose priority is determined). This operation is performed because scattered rules cannot systematically guide the subject's behavior, leading to a chaotic simulation process. This step forms a clearly structured and logically coherent rule system, facilitating efficient model invocation and ensuring the systematic and rational nature of the subject's travel decisions.

[0095] Step S422 is a detailed implementation of S420's "constructing a destination selection logic unit." Its core is encapsulating the destination selection logic into an independent functional module. Here, the destination selection logic unit refers to encapsulating the three-step selection logic of S322 into a functional component that the model can directly call. The target destination set refers to several highly compatible destinations for the subject to choose from after filtering (e.g., "XX office building, XX subway station" for commuters). The selection probability refers to the probability value of each candidate destination being selected by the subject (e.g., XX office building selection probability 0.85, XX subway station 0.15). This step mainly encapsulates the "TOP-N filtering → preference filtering → probability sorting" logic of S322 through programming. The input parameter is set as the probability of the group's behavioral decision, and the output parameter is the target destination set and its corresponding selection probability. It binds the mapping relationship between behavioral profile-related tags and filtering logic (e.g., the leisure group tag corresponds to shopping districts and park-type destination filtering). This operation is performed because scattered selection logic calls are inefficient and prone to logical conflicts. This step achieves modularization and efficient calling of the destination selection logic, ensuring the accuracy and convenience of the subject's destination selection process.

[0096] Step S423 is a detailed implementation of S420, "Configure a preset path planning algorithm for the model." The core is configuring a path planning tool with dynamic adjustment capabilities. Here, the basic path algorithm refers to a conventional algorithm used to calculate the initial optimal path (e.g., A for urban road scenarios). The algorithm used is Dijkstra's algorithm for pedestrian scenarios and Floyd-Warshall's algorithm for complex road network scenarios. The dynamic adjustment module is a functional module that corrects the path according to real-time constraints (e.g., it calls the crowd flow constraint strength value in real time, and automatically adjusts the path to avoid a road segment when the constraint strength value of a certain road segment is ≥0.7). This step mainly involves selecting an appropriate basic path algorithm according to the target scenario type, developing a dynamic adjustment module and integrating it with the basic algorithm, setting module trigger conditions (such as constraint strength value threshold, crowd congestion threshold), and clarifying the path correction logic. This operation is performed because existing basic path algorithms lack dynamic adjustment capabilities and cannot cope with real-time spatial constraint changes, causing path planning to deviate from reality. Through this step, a composite path planning tool of "basic algorithm + dynamic adjustment" is built, which ensures that the planned path is both optimal and adaptable to real-time constraint changes, improving the realism of the simulation.

[0097] Step S424 is a detailed implementation of S420, "Inputting the Model and Generating the Initial Population." Its core is completing the input of core components and generating the initial simulated population. The initial population refers to the set of simulated individuals created by the model in batches based on subject attributes (e.g., generating 1000 subjects based on population distribution data, including 600 commuters, 300 leisure travelers, and 100 business travelers). This step mainly imports the rule system of S421, the logical units of S422, and the path planning algorithm of S423 into the spatiotemporal behavioral dynamics model through the model's component interface. The model reads the subject attributes defined in S411, generates initial subjects in batches according to the population distribution ratio, and binds corresponding attribute tags and associated components to each subject (e.g., binding commuter subjects to commuter-related rules and logical units). This operation is performed because the simulation cannot be driven without integrating the core components into the input model, and the initial population generation needs to closely match the actual distribution. This step achieves seamless integration of the core components and the model, and the generated initial population conforms to the characteristics of real population distribution, providing high-quality initial objects for subsequent evolutionary simulations.

[0098] Step S425 is a detailed implementation of S420, "simulating spatiotemporal evolution and recording flow status." Its core is to drive the group flow according to the process and record its status throughout the entire cycle. Here, a spatiotemporal node refers to a fixed interval on the simulated time axis and its corresponding spatial coordinates (e.g., every 5 minutes is a time node, and each road segment or area is a spatial node). Group flow status recording refers to the real-time storage of data such as the number of entities, direction of movement, aggregation density, and path selection for each spatiotemporal node (e.g., at 10:05, 150 entities on road segment XX are moving westward, with an aggregation density of 2 people / ...). This step primarily drives the initial population flow through the model in the sequence of "destination selection (calling the S422 logic unit) → path planning (calling the S423 algorithm) → movement evolution (executing the S412 behavior rules)". It sets the spatiotemporal node recording frequency (e.g., recording once every 5 minutes) and establishes a state database to store real-time flow data. This operation is performed because the existing simulation evolution process is unclear and does not fully record the flow state, resulting in a lack of data support for subsequent parameter extraction. This step achieves ordered simulation and full-state recording of population flow evolution, providing a comprehensive and continuous data source for subsequent feature parameter extraction.

[0099] Step S426 is a detailed implementation of S420, "extracting and outputting crowd flow characteristic parameters." The core is to extract key quantitative indicators from the evolutionary record. These include: peak spatiotemporal flow (peak flow) refers to the maximum pedestrian flow at a specific spatiotemporal node within the simulation period (e.g., peak flow of 800 people / hour at XX subway station at 7:30 AM during the morning rush hour); group movement speed distribution refers to the proportion of different groups moving at different speed ranges (e.g., 60% moving at 10-15 km / h, 30% moving at 5-10 km / h); and destination cluster density refers to the average crowd density at each candidate destination (e.g., the average cluster density of XX business district is 3 people / hour). The path selection preference percentage refers to the percentage of entities choosing different paths (e.g., 70% choosing subway lines and 20% choosing bus routes). This step mainly involves statistically analyzing the group flow status data recorded in S425, calculating specific values ​​according to the definitions of the aforementioned feature parameters, forming a parameter set, and outputting it. This operation is performed because existing technologies extract parameters that are too singular to fully reflect the patterns of group flow. The feature parameters extracted in this step are rich in dimensions and precisely quantified, comprehensively depicting the core patterns of group flow and providing high-quality core data support for the predictive analysis of S50.

[0100] It should be noted that the specific numerical values ​​of the feature parameters, recording frequencies, algorithm selections, and rule association logic involved in this explanation are merely illustrative descriptions to facilitate understanding of the technical solutions by those skilled in the art, and do not constitute a limitation on the scope of protection of this invention. Those skilled in the art can reasonably adjust or extend the above parameter types, recording frequencies, algorithms, etc., according to the needs of the simulation scenario, data accuracy, and analysis focus; such adjustments and extensions all fall within the scope of the technical solutions protected by this invention.

[0101] In one embodiment, step S50 specifically includes the following steps:

[0102] S511. Based on the type of the preset target spatiotemporal scene, construct the scene-based prediction adaptation rule, and use the scene adaptation factor of the scene-based prediction adaptation rule to filter and calibrate the characteristic parameters of the crowd flow to obtain the scene-based prediction basic data. The scene adaptation factor includes the spatiotemporal granularity coefficient, the crowd activity duration threshold and the regional functional weight.

[0103] S512. Based on scenario-based prediction data, a hierarchical prediction logic is used to generate prediction results, specifically including:

[0104] Based on the spatiotemporal granularity coefficient in the scene adaptation factor, spatiotemporal interpolation calculations are performed on the scene-based prediction base data according to preset time slices and spatial grid units to output the pedestrian distribution data of the target spatiotemporal scene.

[0105] Based on the scenario-based prediction data and the threshold of crowd activity duration in the scenario adaptation factors, a trend fitting model is used to calculate and predict quantitative indicators of crowd gathering and dispersal trends, including peak gathering time, gathering scale and dispersal rate.

[0106] S513. Quantitatively analyze the quantitative indicators of the crowd gathering and dispersal trends, identify key characteristics and potential risks, and generate a trend analysis report.

[0107] S514. Employ a layered visualization approach to display data on pedestrian flow distribution, quantitative indicators of crowd gathering and dispersal trends, and trend analysis reports.

[0108] In this embodiment, steps S511-S514 primarily address the pain points of existing technologies—such as lack of scenario adaptability, simplistic logic, superficial analysis, and unintuitive presentation—through a complete process of "scenario-based data calibration → hierarchical accurate prediction → in-depth risk analysis → hierarchical visualization." This ensures the accuracy, relevance, and practicality of the prediction results. Simultaneously, visualization lowers the information access barrier, providing comprehensive, intuitive, and valuable support for practical decision-making in urban planning, traffic management, and safety assurance. Specifically:

[0109] The core of step S511 is to construct scenario adaptation rules and calibrate data, providing a scenario-based foundation for accurate prediction. This is a preparatory step for "scenario-based prediction" in S50. Scenario-based prediction adaptation rules refer to a data analysis rule system adapted to preset target spatiotemporal scenario types. Scenario adaptation factors are the core parameters supporting the implementation of these rules (spatiotemporal granularity coefficient refers to the prediction time interval and spatial division accuracy, for example, a 15-minute / 500-meter grid for weekday commuting scenarios and a 5-minute / 200-meter grid for large-scale event scenarios; crowd activity duration threshold refers to the upper limit of the average activity duration of the crowd in this scenario, for example, 2 hours for commuting scenarios and 4 hours for leisure scenarios; regional function weight refers to the proportion of influence of different regional functions on the prediction results, for example, a business district regional function weight of 0.8 and a residential area of ​​0.5). Scenario-based prediction basic data refers to standardized prediction data sources adapted to the target scenarios after screening and calibration. This step primarily involves analyzing the type of the preset target spatiotemporal scenario (such as weekday commuting, large-scale exhibitions, and weekend shopping districts), constructing prediction adaptation rules based on the core features of the scenario, determining the corresponding scenario adaptation factor quantification value, and then filtering the crowd flow characteristic parameters output by S426 according to the rules (retaining scenario-related parameter dimensions, such as retaining the cluster density parameter in large-scale event scenarios) and calibrating the dimensions (adjusting parameter weights according to regional functional weights and unifying data dimensions according to spatiotemporal granularity coefficients). This operation is performed because different scenarios have different prediction requirements and data dimension adaptability. Directly using the original feature parameters will lead to prediction result deviations. Existing technologies lack a scenario-based data adaptation step. This step achieves accurate matching between prediction data and target scenarios, obtaining targeted and dimensionally unified scenario-based prediction base data, providing high-quality data support for subsequent hierarchical predictions, and improving the scenario adaptability of prediction results.

[0110] The core of step S512 is to achieve scenario-based prediction using hierarchical logic. It is the core execution link of "generating prediction results" in S50. Hierarchical prediction logic refers to the logical system of predicting and integrating results according to the dimensions of "people flow distribution data + gathering and dissipation trends." Preset time slices refer to prediction time intervals divided according to spatiotemporal granularity coefficients (e.g., 5 minutes / slice, 15 minutes / slice). Spatial grid units refer to the smallest spatial analysis units divided according to spatiotemporal granularity coefficients (e.g., 200m×200m, 500m×500m). Spatiotemporal interpolation calculation refers to the calculation method for supplementing missing values ​​in the spatiotemporal dimensions of the data (e.g., using Kriging interpolation or inverse distance weighted interpolation). Trend fitting models refer to mathematical models used to deduce the laws of data change (e.g., linear regression models, LSTM time series prediction models). Among the quantitative indicators of gathering and dissipation trends, the peak gathering time refers to the specific time when the crowd gathering scale is the largest (e.g., 14:30 for a large event), and the gathering scale refers to the number or density of people at the peak time (e.g., 5000 people, 4 people / ...). The dissipation rate refers to the rate at which the population decreases after the peak (e.g., 100 people / minute, 0.2 people / minute, etc.). • Minutes). This step mainly involves dividing time slices and spatial grid cells according to spatiotemporal granularity coefficients, performing spatiotemporal interpolation calculations on the basic data for scenario-based prediction, filling in data gaps, and outputting the pedestrian distribution data for each grid cell in each time slice (e.g., 14:00-14:05, pedestrian density of grid XX is 3.2 people / minute). Then, the scenario-based prediction base data and the threshold of crowd activity duration are input into the trend fitting model. By analyzing the changing patterns of historical data, quantitative indicators such as peak gathering time, gathering scale, and dissipation rate are deduced. This operation is performed because the existing technology has a single prediction logic and does not predict in layers according to data types, resulting in a lack of specificity and accuracy in the prediction results. Through this step, accurate prediction of crowd distribution and gathering and dissipation trends is achieved. The output prediction data has clear dimensions and precise quantification, which can fully meet the analysis needs of the target scenario.

[0111] Step S513 is the core of in-depth analysis of the prediction results, mining data value and highlighting risks. It is a key step in improving the practicality of the prediction. Among them, quantitative analysis refers to the numerical interpretation and correlation analysis of indicators such as peak gathering time, gathering scale, and dissipation rate (e.g., comparing with the regional carrying capacity threshold to determine whether the gathering exceeds the limit, analyzing the correlation between dissipation rate and spatial constraints). Key features refer to the prominent information reflecting the core laws of population flow (e.g., "the peak gathering period is from 7:30 to 8:00 am, and the gathering scale in the core area exceeds the carrying capacity threshold by 1.2 times"). Potential risks refer to the safety hazards or operational pressures that may occur in the prediction results (e.g., excessive gathering scale in a certain area may lead to the risk of stampede, and excessively slow dissipation rate may cause traffic congestion). The trend analysis report refers to the structured report that integrates the quantitative analysis results, key features, and potential risks. This step primarily involves setting quantitative analysis standards (such as regional carrying capacity thresholds and safe dissipation rate ranges) to compare and analyze quantitative indicators of aggregation and dissipation trends, identify key characteristics, and then assess potential risks by combining factors such as urban spatial constraints and facility carrying capacity. Finally, a trend analysis report is generated according to the structure of "data interpretation - key characteristics - risk warning - optimization suggestions" (e.g., suggesting increasing the number of people to manage crowds and optimizing transportation capacity during peak aggregation times). This operation is performed because existing technologies often stop at predictive data output, lacking in-depth analysis and risk warnings, making it difficult for prediction results to directly support decision-making. This step transforms abstract predictive indicators into valuable decision-making references, clarifies key patterns and potential risks of population movement, and provides accurate basis for subsequent safety assurance, resource allocation, and other decisions.

[0112] Step S514's core is to present the prediction results intuitively through layered visualization. It is the final step in S50's "Result Presentation." Layered visualization refers to classifying and displaying data according to data type and importance using different visualization formats (e.g., dynamic heatmaps for pedestrian distribution data, dynamic line charts + bar charts for aggregation and dissipation trends, structured dashboards + text interpretations for trend analysis reports, and red highlighting for key risk points). This step primarily uses visualization tools (such as ECharts and Tableau) to match appropriate visualization formats for different types of prediction results: dynamic heatmaps intuitively present the distribution of pedestrian density at various spatiotemporal nodes; dynamic line charts show the trend of aggregation scale changes over time; bar charts compare aggregation peaks in different areas; structured dashboards summarize core quantitative indicators and risk levels; and text interpretations clearly present the core content of the trend analysis report. This operation is performed because existing technologies use a single visualization format, failing to clearly distinguish different types of prediction results, leading to inefficient information transmission. This step achieves a clear, layered presentation of prediction results, allowing relevant personnel to quickly capture core data, key features, and potential risks, significantly improving information acquisition efficiency and decision-making response speed.

[0113] It should be noted that the specific values ​​of scene adaptation factors, spatiotemporal granularity, prediction model selection, visualization tools, and analysis standards involved in this explanation are merely illustrative descriptions to facilitate understanding of the technical solutions by those skilled in the art, and do not constitute a limitation on the scope of protection of this invention. Those skilled in the art can reasonably adjust or replace the above values, granularity, models, tools, etc., according to the complexity of the target scene, data accuracy, and actual decision-making needs; such adjustments and replacements all fall within the scope of the technical solutions protected by this invention.

[0114] In one feasible embodiment, it is assumed that the target spatiotemporal scenario is a "weekend leisure scenario in the core urban business district" (the preset time range is Saturday 9:00-21:00, the spatial range is the core area of ​​the business district of 3 square kilometers, and the core requirements are to predict peak traffic flow, identify the risk of congestion, and provide support for traffic management decisions):

[0115] Step S10: Targeted acquisition of POI data for shopping malls / restaurants / entertainment, main and secondary road network structure data, weekend crowd distribution statistics for the past 3 months, parking lot / subway entrance / bus stop layout data, and historical pedestrian flow monitoring data for 30 monitoring points within the business district, ensuring that the data covers all dimensions of "people-space-industry-facilities".

[0116] Step S20: Standardize the format of the above data (convert Excel spreadsheets, GPS coordinates, and surveillance video statistics into JSON format), map fields (unify the definitions of fields such as "people flow" and "density"), remove outliers in the monitoring data (such as zero values ​​or abnormal peak values ​​caused by equipment failure), and calibrate all data to the same latitude and longitude coordinate system (such as WGS-84) and Beijing time standard to obtain a standardized dataset.

[0117] Steps S311~S315: ① Construct attraction quantification rules (shopping mall POI priority 8 points, catering 7 points, entertainment 6 points, facility density is set at 30 per square kilometer as medium density standard, service coverage radius is set at 2 kilometers, and the attraction intensity value of each area is calculated); ② Construct spatial constraint rules (main road traffic capacity 1200 people / hour, branch road 300 people / hour, building-dense area barrier coefficient 0.6, open square 0.1); ③ Divide into two categories: "young leisure group" and "family group" (young leisure group route selection priority "convenience" weight 0). 7. Spatial tolerance threshold ≤ 0.4; Family groups prioritize route selection with a "safety" weight of 0.8 and a spatial tolerance threshold ≤ 0.3); ④ Construct a dynamic weight matrix for weekend scenarios (crowd attraction intensity value weight 0.5, spatial constraint intensity value weight 0.2, behavioral preference factor weight 0.3); ⑤ Calculate the decision probability of young leisure groups going to XX shopping center (0.78) and the decision probability of family groups going to XX cultural and creative block (0.82) using a quantitative coupling function (decision probability = attraction intensity value × 0.5 + constraint intensity value × 0.2 + preference factor × 0.3);

[0118] Steps S321~S322: ① Extract travel decision rules (10:00-18:00 is the peak travel period; young leisure travelers prioritize subway / walking, while families with children prioritize driving + short walking distances); ② Extract destination selection logic (filter the top 5 candidate destinations based on attraction intensity: XX Shopping Center, XX Cultural and Creative Street, XX Food Court, XX Park, XX Supermarket, then filter out XX Supermarket based on family preferences, and finally sort by decision probability to obtain the selection order of "XX Cultural and Creative Street → XX Food Court → XX Shopping Center");

[0119] Steps S411~S415: ① Define 2000 simulated subjects, tagged with "Young Leisure Group" (1200 subjects) and "Parent-Child Family Group" (800 subjects), initially distributed in 3 large residential areas and 2 subway entrances around the business district; ② Set subject behavior rules (movement rule: when the decision probability is ≥0.6, priority is given to option A). Algorithm-planned shortest path; Clustering rule: Open square capacity threshold 4 people / When the threshold is exceeded, the subject is not allowed to enter; Scene adaptation rule: switch to "dining scene" preference factor from 12:00 to 14:00); ③ Construct a three-layer architecture model of "subject attributes + behavior rules + dynamic coupling constraints", and call the constraint strength value in real time (e.g., when the constraint strength value of XX road section is 0.7, the subject is prohibited from entering); ④ Use the weekend pedestrian flow monitoring data of the same period last year to calibrate the model, and adjust the behavior rule parameters through mean square error calculation (e.g., adjust the walking speed of young leisure people from 5km / h to 4.5km / h) so that the simulation error is ≤8%; ⑤ Output a spatiotemporal behavior dynamic model adapted to the scene of this business district;

[0120] Steps S421~S426: ① Input the progressive rule system of "time period-mode-destination"; ② Invoke the destination selection logic unit; ③ Configure A Path planning algorithm (including dynamic adjustment module: when the pedestrian density of a certain road segment exceeds 3 people / (Automatically adjust path at the same time); ④ The model generates an initial group of 2000 people and drives its evolution, synchronously recording the number of people and their direction of movement every 5 minutes and every 200-meter grid; ⑤ The simulation yields the core characteristic parameters: 15:00 is the peak period for human traffic, and the gathering density in the central square of the business district is 3.8 people / The subway station to shopping mall route accounted for 65% of the choices;

[0121] Steps S511~S514: ① Set scene adaptation factors (spatiotemporal granularity 5 minutes / 200 meters grid, crowd activity duration threshold 6 hours, core business district functional weight 0.8), filter and retain relevant parameters such as "gathering density" and "path proportion" and calibrate them; ② Output the pedestrian distribution data of each grid using the inverse distance weighted interpolation method, and predict using the LSTM time series model: peak gathering scale of 5000 people at 15:00, dissipation rate of 80 people / minute, and pedestrian density drops back to 1.2 people / minute after 18:00. ③ Quantitative analysis and identification: Central square gathering density (3.8 people / Approaching the carrying capacity threshold (4 people / ) ), 14:30-15:30 is a high-risk period with potential for localized overcrowding, generating suggestions such as "adding crowd control signs in the central square and optimizing subway entrance access channels"; ④ Through dynamic heat maps, the distribution of people in each grid is displayed in real time (high-risk areas are marked in red), line graphs show the gathering and dissipation trends from 9:00 to 21:00, and structured dashboards summarize peak times, risk levels, and crowd control suggestions, allowing managers to intuitively obtain core information and quickly deploy crowd control measures.

[0122] In one embodiment, a pedestrian flow simulation and analysis system based on spatiotemporal behavioral dynamics is provided. This system corresponds to a pedestrian flow simulation and analysis method based on spatiotemporal behavioral dynamics described in the above embodiments. The spatiotemporal behavioral dynamics-based pedestrian flow simulation and analysis system includes:

[0123] The data acquisition module is used to acquire multidimensional spatiotemporal heterogeneous data of the city, which includes at least population distribution data, urban spatial characteristic data, industrial function distribution data, and facility layout data.

[0124] The processing module is used to standardize the multidimensional spatiotemporal heterogeneous data, eliminate data format differences and noise interference, and obtain a standardized dataset with a unified spatiotemporal benchmark.

[0125] The extraction module is used to construct a dynamic coupling mechanism among population behavior characteristics, urban spatial constraints, and industrial function supply based on the standardized dataset, and to extract population travel decision rules and destination selection logic according to the dynamic coupling mechanism.

[0126] The evolution module is used to construct a spatiotemporal behavior dynamics model with dynamic coupling mechanism as the underlying constraint and subject modeling technology. The population travel decision rules and destination selection logic are input into the spatiotemporal behavior dynamics model. The model simulates the spatiotemporal evolution process of the population flow through a preset path planning algorithm and outputs population flow characteristic parameters based on the evolution results.

[0127] The prediction module is used to predict the distribution data of people flow in a preset target spatiotemporal scene, the trend of gathering and dispersing, based on the crowd flow characteristic parameters, and to display the prediction results in a visual manner.

[0128] For specific limitations regarding a pedestrian flow simulation and analysis system based on spatiotemporal behavioral dynamics, please refer to the limitations of a pedestrian flow simulation and analysis method based on spatiotemporal behavioral dynamics mentioned above, which will not be repeated here. Each module in the aforementioned pedestrian flow simulation and analysis system based on spatiotemporal behavioral dynamics can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in or independent of the processor in a computer device, or stored in the memory of a computer device as software, so that the processor can call and execute the corresponding operations of each module.

[0129] In one embodiment, a computer device is provided, which may be a server, and its internal structure diagram may be as follows. Figure 2As shown, the computer device includes a processor, memory, network interface, and database connected via a system bus. The processor provides computational and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system, computer programs, and database. The internal memory provides the environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The database is used for data storage, data processing, and data analysis. The network interface is used for communication with external terminals via a network connection. When the computer program is executed by the processor, it implements a human flow simulation analysis method based on spatiotemporal behavioral dynamics.

[0130] In one embodiment, a computer device is provided, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements a human flow simulation analysis method based on spatiotemporal behavioral dynamics.

[0131] In one embodiment, a computer-readable storage medium is provided having a computer program stored thereon, which, when executed by a processor, implements a method for simulating and analyzing pedestrian flow based on spatiotemporal behavioral dynamics.

[0132] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium. When executed, the computer program can include the processes of the embodiments of the above methods. Any references to memory, storage, databases, or other media used in the embodiments provided in this application can include non-volatile and / or volatile memory. Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in a variety of forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), dual data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), RAMbus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), etc.

[0133] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the above-described division of functional units and modules is used as an example. In practical applications, the above functions can be assigned to different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above.

[0134] The above-described embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application, and should all be included within the protection scope of this application.

Claims

1. A method for simulating and analyzing pedestrian flow based on spatiotemporal behavioral dynamics, characterized in that, Includes the following steps: S10. Obtain multidimensional spatiotemporal heterogeneous data of the city, wherein the multidimensional spatiotemporal heterogeneous data includes at least population distribution data, urban spatial characteristic data, industrial function distribution data and facility layout data. S20. Standardize the multidimensional spatiotemporal heterogeneous data to eliminate data format differences and noise interference, and obtain a standardized dataset with a unified spatiotemporal benchmark. S30. Based on the standardized dataset, construct a dynamic coupling mechanism among population behavior characteristics, urban spatial constraints, and industrial function supply; extract population travel decision rules and destination selection logic according to the dynamic coupling mechanism; the construction of the dynamic coupling mechanism includes the following steps: S311. Based on the industrial function distribution data and facility layout data in the standardized dataset, construct an attractiveness quantification calculation rule. The attractiveness quantification calculation rule takes POI type priority, facility density and service coverage radius as input parameters and outputs the population attraction intensity value of each spatial area. S312. Based on urban spatial feature data in the standardized dataset, construct a spatial constraint intensity quantification calculation rule. The spatial constraint intensity quantification calculation rule takes road network traffic capacity, building barrier coefficient and open space carrying capacity threshold as input parameters, and outputs the population flow constraint intensity value of each spatial area. S313. Based on the population distribution data in the standardized dataset, divide the population into at least two types of behavioral profiles, and define a behavioral preference factor for each type of profile. The behavioral preference factor includes destination type preference, route selection priority, and spatial tolerance threshold. S314. Construct a scenario-based dynamic weight matrix, wherein the scenario-based dynamic weight matrix dynamically allocates the correlation weights of the crowd attraction intensity value, the correlation weights of the crowd flow constraint intensity value, and the correlation weights of the behavioral preference factor according to the scenario type. S315. Based on the crowd attraction intensity value, crowd flow constraint intensity value, behavioral preference factor, and corresponding correlation weights, a quantitative coupling function is constructed. The crowd behavior decision probability is calculated through the quantitative coupling function to form the dynamic coupling mechanism. The quantitative coupling function satisfies the following calculation formula: Probability of crowd behavior decisions = crowd attraction intensity value × its correlation weight + crowd mobility constraint intensity value × its correlation weight + behavior preference factor × its correlation weight; S40. Using a dynamic coupling mechanism as the underlying constraint, a spatiotemporal behavioral dynamics model is constructed using subject modeling technology. The population travel decision rules and destination selection logic are input into the spatiotemporal behavioral dynamics model. A preset path planning algorithm is used to simulate the spatiotemporal evolution process of population flow, and population flow characteristic parameters are output based on the evolution results. Specifically, the following steps are included: S410. Define the subject attributes and set the subject behavior rules. Embed the dynamic coupling mechanism as the underlying constraint into the model to construct a model with a three-layer architecture including subject attributes, subject behavior rules, and dynamic coupling constraints. Perform calibration, optimization, and error analysis on the model to output a spatiotemporal behavior dynamic model. The dynamic coupling constraint is the constraint formed on the subject behavior by the dynamic coupling mechanism through the output quantization results. S420. Define a population travel decision rule system based on population travel decision rules, construct a destination selection logic unit based on destination selection logic, and configure a preset path planning algorithm for the model. Input the population travel decision rule system, destination selection logic unit and preset path planning algorithm into the spatiotemporal behavior dynamics model, use the spatiotemporal behavior dynamics model to simulate the spatiotemporal evolution process of the group flow and record the group flow status simultaneously, and extract and output the population flow characteristic parameters based on the evolution results. S50. Based on the crowd flow characteristic parameters, predict the crowd distribution data, aggregation and dissipation trends of the preset target spatiotemporal scene, and display the prediction results in a visual manner.

2. The pedestrian flow simulation and analysis method based on spatiotemporal behavioral dynamics as described in claim 1, characterized in that, In step S30, the extraction of population travel decision rules and destination selection logic based on the dynamic coupling mechanism specifically includes the following steps: S321. Based on the population behavior decision probability, behavior preference factor, and scenario-based dynamic weight matrix output by the dynamic coupling mechanism, extract population travel decision rules. The extracted population travel decision rules include: Based on the time dimension weight changes of the scenario-based dynamic weight matrix, travel time selection rules are generated. Based on the correlation between route selection priority and the probability of population behavioral decisions in the behavioral preference factors, travel mode selection rules are generated; Based on the matching relationship between the spatial tolerance threshold and the intensity of population flow constraints in the behavioral preference factors, a travel destination priority rule is generated. S322. Based on the crowd attraction intensity value, crowd behavior decision probability, and behavior profile output by the dynamic coupling mechanism, extract the destination selection logic. The construction process of the destination selection logic includes: Based on the descending order of crowd attraction intensity values, the top-N candidate destinations are selected, where N is a preset positive integer; Based on the destination type preferences of the behavioral profile, filter out candidate destinations that do not match; Based on the probability of decision-making based on population behavior, the remaining candidate destinations are prioritized to form a complete destination selection logic.

3. The pedestrian flow simulation and analysis method based on spatiotemporal behavioral dynamics as described in claim 1, characterized in that, Step S410 specifically includes the following steps: S411. Based on the behavioral profile and population distribution data, define the main attributes of the model, including the main identifier, behavioral profile association tags, initial location, and activity demand threshold. S412. Set subject behavior rules, which include movement rules, aggregation rules and scene adaptation rules. The movement rules dynamically adjust the subject's movement direction and path based on the probability of crowd behavior decisions. The aggregation rules limit the subject's aggregation scale based on the open space carrying capacity threshold. The scene adaptation rules match the corresponding behavior preference factors based on the scene type. S413. The dynamic coupling mechanism is used as the underlying constraint embedding model to construct a model with a three-layer architecture including subject attributes, subject behavior rules and dynamic coupling constraints. The underlying constraints dynamically correct the subject behavior decisions by calling the probability of crowd behavior decisions and the strength value of crowd flow constraints output by the dynamic coupling mechanism in real time. S414. Obtain historical pedestrian flow monitoring data, use the historical pedestrian flow monitoring data to calibrate and optimize the model, and adjust the parameters of the behavior rules and the weights of the dynamic coupling constraints through error analysis so that the simulation accuracy of the model meets the preset accuracy threshold. S415. Based on the calibration optimization and error analysis adjustment results, output a spatiotemporal behavior dynamics model. The spatiotemporal behavior dynamics model can automatically match the corresponding subject behavior rules and dynamic coupling constraint parameters according to the type of the target spatiotemporal scene.

4. The pedestrian flow simulation and analysis method based on spatiotemporal behavioral dynamics as described in claim 3, characterized in that, Step S420 specifically includes the following steps: S421. Define a population travel decision-making rule system, which includes travel time selection rules, travel mode selection rules, and travel destination priority rules. S422. Construct a destination selection logic unit, wherein the destination selection logic unit takes the probability of crowd behavior decision as input and outputs the target destination set and selection probability of the subject based on the behavioral profile and associated tags. S423. Configure a preset path planning algorithm, wherein the path planning algorithm includes a basic path algorithm and a dynamic adjustment module; S424. Input the population travel decision-making rule system, destination selection logic unit and route planning algorithm into the spatiotemporal behavior dynamics model, which generates an initial population based on subject attributes; S425. Based on the population travel decision-making rule system, simulate the spatiotemporal evolution process of the initial population flow. The spatiotemporal evolution process includes destination selection, path planning and movement evolution in sequence, and the population flow status of each spatiotemporal node is recorded synchronously during the spatiotemporal evolution process. S426. Based on the simulated spatiotemporal evolution results, extract and output the population flow characteristic parameters, including the spatiotemporal flow peak, the distribution of group movement speed, the destination aggregation density, and the proportion of path selection preference.

5. The pedestrian flow simulation and analysis method based on spatiotemporal behavioral dynamics as described in claim 1, characterized in that, Step S50 specifically includes the following steps: S511. Based on the type of the preset target spatiotemporal scene, construct the scene-based prediction adaptation rule, and use the scene adaptation factor of the scene-based prediction adaptation rule to filter and calibrate the characteristic parameters of the crowd flow to obtain the scene-based prediction basic data. The scene adaptation factor includes the spatiotemporal granularity coefficient, the crowd activity duration threshold and the regional functional weight. S512. Based on scenario-based prediction data, a hierarchical prediction logic is used to generate prediction results, specifically including: Based on the spatiotemporal granularity coefficient in the scene adaptation factor, spatiotemporal interpolation calculations are performed on the scene-based prediction base data according to preset time slices and spatial grid units to output the pedestrian distribution data of the target spatiotemporal scene. Based on the scenario-based prediction data and the threshold of crowd activity duration in the scenario adaptation factors, a trend fitting model is used to calculate and predict quantitative indicators of crowd gathering and dispersal trends, including peak gathering time, gathering scale and dispersal rate. S513. Quantitatively analyze the quantitative indicators of the crowd gathering and dispersal trends, identify key characteristics and potential risks, and generate a trend analysis report. S514. Employ a layered visualization approach to display data on pedestrian flow distribution, quantitative indicators of crowd gathering and dispersal trends, and trend analysis reports.

6. A pedestrian flow simulation and analysis system based on spatiotemporal behavioral dynamics, used to implement the steps of the pedestrian flow simulation and analysis method based on spatiotemporal behavioral dynamics as described in any one of claims 1-5, characterized in that, include: The data acquisition module is used to acquire multidimensional spatiotemporal heterogeneous data of the city, which includes at least population distribution data, urban spatial characteristic data, industrial function distribution data, and facility layout data. The processing module is used to standardize the multidimensional spatiotemporal heterogeneous data, eliminate data format differences and noise interference, and obtain a standardized dataset with a unified spatiotemporal benchmark. The extraction module is used to construct a dynamic coupling mechanism among population behavior characteristics, urban spatial constraints, and industrial function supply based on the standardized dataset, and to extract population travel decision rules and destination selection logic according to the dynamic coupling mechanism. The evolution module is used to construct a spatiotemporal behavior dynamics model with dynamic coupling mechanism as the underlying constraint and subject modeling technology. The population travel decision rules and destination selection logic are input into the spatiotemporal behavior dynamics model. The model simulates the spatiotemporal evolution process of the population flow through a preset path planning algorithm and outputs population flow characteristic parameters based on the evolution results. The prediction module is used to predict the distribution data of people flow in a preset target spatiotemporal scene, the trend of gathering and dispersing, based on the crowd flow characteristic parameters, and to display the prediction results in a visual manner.

7. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the human flow simulation analysis method based on spatiotemporal behavioral dynamics as described in any one of claims 1-5.

8. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by the processor, it implements the steps of the human flow simulation analysis method based on spatiotemporal behavioral dynamics as described in any one of claims 1-5.

Images