Urban flood disaster dynamic risk simulation method and system based on intelligent agent interactive modeling

Abstract

The embodiment of the invention discloses an urban flood disaster dynamic risk simulation method and system based on intelligent agent interactive modeling, and relates to the technical field of natural disaster prevention and control. According to the technical scheme provided by the embodiment of the invention, active response behaviors (for example, returning home in advance, reducing going out and the like) of residents in disasters are truly described by constructing the intelligent agent interaction model and utilizing crowd position service data and interest point data, and the behavior mechanism is embedded into the dynamic risk simulation process based on the intelligent agent, so that the dynamic risk simulation of the residents is realized. And thus, a more realistic individual exposure degree estimation result is obtained. Compared with a traditional method of considering a population as a static state and passively exposing a disaster-bearing body, the method can effectively improve the precision and reliability of a risk simulation result.

Description

TECHNICAL FIELD

[0001] The embodiment of the application relates to the technical field of natural disaster prevention and control, and particularly relates to a city flood disaster dynamic risk simulation method and system based on agent interaction modeling. BACKGROUND

[0002] Extreme rainstorm events present the characteristics of high frequency, short duration and high intensity, and waterlogging disasters have the characteristics of suddenness and high locality. Traditional static or semi-static city flood disaster population exposure evaluation cannot reflect the linkage changes of the "man-land-road" system in a short time, resulting in that risk research and judgment lags behind disaster evolution. Most of the existing risk evaluation methods regard the population as a passive disaster-bearing body following the conventional movement mode, and cannot fully consider the interaction between individual residents and disaster environment, and lack consideration of the response behavior of the residents in actively adjusting the travel plan to avoid risks during disasters, resulting in deviation of the risk evaluation results from the actual situation, especially in the densely populated urban environment.

[0003] Therefore, there is an urgent need for a new city rainstorm flood dynamic risk evaluation method. SUMMARY

[0004] The embodiment of the application provides a city flood disaster dynamic risk simulation method and system based on agent interaction modeling, so as to at least partially solve the above problems.

[0005] The first aspect of the embodiment of the application provides a city flood disaster dynamic risk simulation method based on agent interaction modeling, and the method comprises the following steps: Collecting multi-source data of a target research area, wherein the multi-source data at least comprises population location service data, interest point data, road network data and rainfall data in a rainstorm period and a normal period; Comparing the differences in the spatio-temporal distribution of the population in the rainstorm period and the normal period based on the population location service data and the interest point data in the rainstorm period and the normal period, and identifying the spatio-temporal response behavior characteristics of the population under the influence of the rainstorm disaster event; Generating an individual behavior rule library in a response mode based on the spatio-temporal response behavior characteristics of the population; Constructing a city flood inundation simulation environment based on the road network data and the rainfall data, and dynamically outputting disaster factor data; Driving an agent-based model to perform dynamic simulation in the city flood inundation simulation environment, wherein the agents in the agent-based model make movement path decisions according to the individual behavior rule library in the response mode, and output dynamic population exposure degree data; Calculating and outputting the dynamic risk results of the city rainstorm flood disaster based on the disaster factor data and the population exposure degree data.

[0006] Optionally, the method further comprises: determining the spatial distribution density and weight of each type of urban functional area based on the point of interest data and the road network data, to obtain an urban functional area layer; identifying the spatiotemporal response behavior characteristics of the population under the influence of the rainstorm disaster event, including: calculating the population flow index and the functional area attraction index of each urban functional area based on the population location service data and the point of interest data during the rainstorm period and the normal period; comparing the population flow index during the rainstorm period and the normal period to quantify the spatiotemporal response behavior characteristics of the population in the time dimension for different functional areas, including the travel peak time shift and the activity duration change; comparing the functional area attraction index during the rainstorm period and the normal period to quantify the spatiotemporal response behavior characteristics of the population in the spatial dimension for different functional areas, including the relative change of the functional area attraction.

[0007] Optionally, based on the spatiotemporal response behavior characteristics of the population, a response mode individual behavior rule library is generated, including: establishing a normal mode individual behavior rule based on a probabilistic finite state machine; using the spatiotemporal response behavior characteristics of the population in the time dimension for different functional areas to correct the activity time parameter in the finite state machine; using the spatiotemporal response behavior characteristics of the population in the spatial dimension for different functional areas to correct the state transition probability in the finite state machine.

[0008] Optionally, a city flood inundation simulation environment is constructed based on the road network data and the rainfall data, including: based on the road network data and the rainfall data, using an underground drainage pipe network model to simulate the hydraulic process of the underground drainage pipe network, and using a two-dimensional surface flow model to simulate the surface flow process; coupling the underground drainage pipe network model and the two-dimensional surface flow model to dynamically simulate the generation of inundation water depth and inundation range as the disaster-causing factor data.

[0009] Optionally, in the city flood inundation simulation environment, an agent-based model is driven for dynamic simulation, including: setting city residents as agents with decision-making ability; driving each agent to determine a travel destination according to the response mode individual behavior rule library at each time step; based on the travel destination, path planning is performed to determine a path with the minimum threat weighted value considering the road length, traffic capacity, and inundation water depth; Determine the spatio-temporal evolution of the population distribution of each functional area, the dynamic crowd flow of the road segment and the crowd exposure data.

[0010] Optionally, based on the disaster-causing factor data and the crowd exposure data, calculate and output the dynamic risk result of urban rainstorm flood disaster, including: Based on the disaster-causing factor data and the crowd exposure data, calculate the risk value of each functional block and each road segment; Based on the risk value, determine the risk level; Based on the risk level of each functional block and each road segment, output the dynamic risk map.

[0011] Optionally, the method further comprises: Based on the risk value of each road segment, output the road risk list; Based on the risk level of each functional block, output the high-risk area migration trajectory; Based on the risk value of each functional block, output the risk change curve of each functional block.

[0012] The second aspect of the embodiment of the application provides a city flood disaster dynamic risk simulation system based on agent interaction modeling, the system comprises: A data acquisition module is configured to acquire multi-source data of a target research area, wherein the multi-source data at least includes crowd location service data, interest point data, road network data and rainfall data in a rainstorm period and a normal period; A feature recognition module is configured to compare the spatio-temporal distribution difference of the crowd in the rainstorm period and the normal period based on the crowd location service data and the interest point data in the rainstorm period and the normal period, and recognize the spatio-temporal response behavior characteristics of the crowd under the influence of the rainstorm disaster event; A rule generation module is configured to generate an individual behavior rule library in a response mode based on the spatio-temporal response behavior characteristics of the crowd; A disaster-causing factor output module is configured to construct a city flood inundation simulation environment based on the road network data and the rainfall data, and dynamically output disaster-causing factor data; A simulation module is configured to drive an agent-based model to perform dynamic simulation in the city flood inundation simulation environment, wherein the agent in the agent-based model moves according to the individual behavior rule library in the response mode to make path decision and output dynamic crowd exposure data; A risk output module is configured to calculate and output the dynamic risk result of the urban rainstorm flood disaster based on the disaster-causing factor data and the crowd exposure data.

[0013] The third aspect of the embodiment of the present application provides an electronic device, comprising a memory, a processor and a computer program stored in the memory and executable on the processor, and the processor implements the urban flood disaster dynamic risk simulation method based on agent interaction modeling as described in the first aspect of the present application when executed.

[0014] The fourth aspect of the embodiment of the present application provides a computer readable storage medium having a computer program stored thereon, and the program is executed by a processor to implement the urban flood disaster dynamic risk simulation method based on agent interaction modeling as described in the first aspect of the present application.

[0015] The fifth aspect of the embodiment of the present application provides a computer program product comprising computer programs / instructions, and the computer programs / instructions are used to implement the steps in the urban flood disaster dynamic risk simulation method based on agent interaction modeling as described in the first aspect of the present application.

[0016] The technical solution provided by the embodiment of the present application uses the population location service data and the point of interest data to realistically depict the active response behavior of residents in disasters (for example, returning home in advance, reducing going out, etc.), and integrates these behaviors into the risk simulation simulation based on the agent, so that more realistic individual exposure estimation results are obtained. Compared with the traditional method of regarding the population as a static and passive exposure disaster body, the present application can significantly improve the accuracy and reliability of the risk simulation results.

[0017] In the embodiment of the present application, the dynamic flood evolution process is combined with the dynamic population response behavior, which can output the minute-level or hour-level spatiotemporal continuous risk map, and reveal the evolution law of the risk in different time periods (such as morning and evening peak) and different regions (such as commercial and residential areas) in a day, which provides the possibility for fine and time-period-specific emergency response.

[0018] The final urban storm flood dynamic risk assessment result of the embodiment of the present application can reveal the risk heterogeneity change of different functional areas under disaster response, for example, the risk of commercial and educational areas decreases significantly, while the risk of residential areas may increase in some time periods due to the concentration of population. It can provide more targeted resource allocation and risk control strategies for city managers, thereby improving the efficiency and effectiveness of emergency management. BRIEF DESCRIPTION OF DRAWINGS

[0019] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the following will briefly introduce the drawings needed to be used in the description of the embodiments of the present application. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can also be obtained by those skilled in the art without creative labor on the basis of these drawings.

[0020] Figure 1 is a step flow chart of a city flood disaster dynamic risk simulation method based on agent interaction modeling provided by an embodiment of the present application; Figure 2 is a structural block diagram of a city flood disaster dynamic risk simulation system based on agent interaction modeling provided by an embodiment of the present application. DETAILED DESCRIPTION

[0021] In order to make the above-mentioned purposes, features and advantages of the present application more obvious and easy to understand, the present application will be further described in detail below with reference to the drawings and specific embodiments.

[0022] Existing researches are mostly based on normal commuting mode or empirical O-D, lacking quantitative characterization and parameterization mechanism of human behavior "time displacement" "activity duration change" "space redistribution" under disaster conditions, and it is difficult to close-loop inject the observed abnormal travel / stay rules into the risk assessment process. Existing models are generally based on normal sample learning fixed probability or fixed utility parameters, lacking linkage update with disaster observation (such as rain intensity, water accumulation, congestion change); this leads to the difficulty of the model to reproduce the real behavior transfer and congestion migration under extreme scenarios.

[0023] An evaluation method capable of dynamically coupling disaster-time human response behavior and flood environment is proposed in the embodiments of the present application, which can not only identify and quantify the two-dimensional behavior changes in time and space using high spatio-temporal resolution data, but also parameterize them into the model rules based on agent interaction modeling, drive exposure evolution together with dynamic water depth field, and form a dynamic risk assessment process.

[0024] Specifically, as shown in Figure 1 the step flow chart of the city flood disaster dynamic risk simulation method based on agent interaction modeling provided by the embodiments of the present application is shown, and the method comprises the following steps: S101, collecting multi-source data of a target research area, wherein the multi-source data at least includes human location service data, interest point data, road network data and rainfall data in a rainstorm period and a normal period.

[0025] S102, comparing the differences in human spatio-temporal distribution in the rainstorm period and the normal period based on the human location service data and the interest point data in the rainstorm period and the normal period, and identifying human spatio-temporal response behavior characteristics under the influence of a rainstorm disaster event.

[0026] S103, generating an individual behavior rule library under a response mode based on the human spatio-temporal response behavior characteristics.

[0027] S104, constructing a city flood inundation simulation environment based on the road network data and the rainfall data, and dynamically outputting disaster-causing factor data.

[0028] S105, in the urban flood inundation simulation environment, driving the agent-based model to simulate, wherein the agent in the agent-based model follows the individual behavior rule base in the response mode to make a moving path decision, and outputs dynamic crowd exposure data.

[0029] S106, based on the disaster-causing factor data and the crowd exposure data, calculating and outputting the dynamic risk result of urban rainstorm flood disaster.

[0030] In the embodiment of the application, in step S101, multi-source data of city, such as geography, hydrology, social economy and crowd movement, can be collected and integrated, including road network, digital elevation model (DEM), drainage system, rainfall process, point of interest (POI) and location-based service (LBS) data, coordinate unification and data cleaning are performed, and “normal period” and “rainstorm period” are divided, and spatial units for analysis are constructed, to establish a unified and standardized data basis for subsequent analysis.

[0031] In the embodiment of the application, the administrative boundary of the research area, road network data based on OpenStreetMap (OSM), digital elevation model (DEM) data, and underground drainage pipe network data provided by the local water department (including rainwater pipelines, inspection wells, water outlets, etc.) can be collected. At the same time, the flood-prone point monitoring data published by the relevant water department is obtained, which is used to check the accuracy of the model.

[0032] In the embodiment of the application, for rainfall data, the hourly or high-resolution observation data of the target rainstorm process can be selected, and in actual application, the time resolution can be set to be less than or equal to 5 minutes, to ensure that the rainfall intensity variation characteristics can be described in detail, and reliable input can be provided for subsequent water power model simulation.

[0033] In the embodiment of the application, the target rainstorm process is determined based on the selected representative rainstorm event in the research period in the research area. The rainstorm time can usually be divided into two sources: historical typical rainstorm flood disaster events; scenario design rainstorm events (rainstorm events occurring once every 50 years, rainstorm events occurring once every 100 years). Correspondingly, the target rainfall process includes the complete rainfall process of the entire rainstorm event, including the rainfall start, peak and end period.

[0034] In this embodiment of the invention, Point of Interest (POI) data is collected for the identification and division of urban functional areas. Location-based services (LBS) data is also collected, which includes the geographic coordinates and timestamps of anonymous users to characterize the spatiotemporal distribution dynamics of the population. To calibrate the validity of the LBS data, authoritative third-party population spatial distribution data, such as LandScan global population raster data, can also be introduced.

[0035] In this embodiment of the invention, all collected spatial data can be unified under the same geographic coordinate system (such as WGS84). The raw data is cleaned, including removing outliers and filling in missing values. Based on timestamps, LBS data and rainfall data are divided into "normal periods" (as a baseline) and "heavy rain periods" (as the analysis objects) to ensure the effectiveness of subsequent comparative analysis.

[0036] In this embodiment of the invention, the study area can be divided into several plot units based on the road network, or a spatial unit set can be generated according to a preset rule grid method. Each spatial unit serves as the smallest analytical unit for subsequent calculations and modeling. Within each unit, data is recorded at each time step. The LBS request count is used, and combined with POI data, its functional attribute category is determined. In this way, we can not only achieve spatiotemporal characterization of crowd activities, but also dynamically couple functional area characteristics with crowd behavior, laying the foundation for subsequent modeling and risk assessment.

[0037] In this embodiment of the invention, the method further includes: determining the spatial distribution density and weight of various urban functional areas based on the point of interest data and the road network data, thereby obtaining an urban functional area layer.

[0038] In this embodiment of the invention, the urban functional area layer will be used as the basic spatial framework in subsequent steps. On the one hand, it is used to aggregate LBS location request data according to spatial units and functional categories, thereby calculating the population flow index; on the other hand, it is used to define the functional categories of the attractiveness index, so that the attractiveness characteristics of different functional areas (such as residential, commercial, educational, etc.) can be quantified and compared.

[0039] In this embodiment of the invention, based on POI data, methods such as kernel density estimation can be used to calculate the distribution density and weight of various functions within a spatial unit, identify and classify the dominant functional category of each spatial unit, and generate an urban functional area layer.

[0040] In this embodiment of the invention, based on POI and OSM data, reclassification, weighted kernel density calculation and normalization can be performed to determine the dominant functional category of each spatial unit, and finally generate an urban functional area layer.

[0041] Among them, POI data mainly reflects the type and density of points of interest, and can characterize the attribute features of functional areas (such as residential, commercial, and educational). OSM data provides road and land use boundary information, which can assist in spatial unit division and boundary correction, making the functional area division more consistent with geographical boundaries.

[0042] During the zoning process, the functional attributes of POIs can be used as the primary basis for determining functional categories. The functional zoning results of POIs can be spatially clipped and corrected using the road and plot boundaries of OSM to ensure that the functional zoning results are consistent with the actual geographical boundaries. When the distribution of POIs in a certain spatial unit is sparse or there is ambiguity in classification, the land use tags of OSM can be used for auxiliary determination.

[0043] In this embodiment of the invention, after obtaining the urban functional area layer, step S102 includes the following sub-steps: S1021, Based on the population location service data and point of interest data during the rainstorm period and normal period, calculate the population flow index and functional area attractiveness index of each urban functional area.

[0044] S1022, compare the population flow index during the rainstorm period with that during the normal period to quantify the spatiotemporal response behavior characteristics of different functional areas in the time dimension, including the time shift of travel peaks and changes in the duration of activities.

[0045] S1023, compare the functional area attractiveness index during the rainstorm period and the normal period to quantify the spatiotemporal response behavior characteristics of different functional areas in the spatial dimension, including the relative change in the attractiveness of functional areas.

[0046] In this embodiment of the invention, LBS data and POI data are used to construct the Population Flow Index (PFI) and Attraction Index (ATI) for each urban functional area. By comparing the PFI during rainstorm periods and normal periods, the temporal response characteristics of residents' travel are quantified, such as peak shift and changes in activity duration. By comparing the ATI, the changes in the attractiveness of different functional areas to the population are quantified, revealing the spatial response characteristics of residents, such as gravitating towards residential areas and reducing unnecessary travel.

[0047] In this embodiment of the invention, the Population Flow Index (PFI) is used to describe the intensity of net inflow or outflow of population in a functional area, while the Attractiveness Index (ATI) comprehensively measures its attractiveness by integrating the population density of the functional area itself and the distance decay effect of surrounding hotspot areas.

[0048] Specifically, in this embodiment of the invention, the following calculation method can be used to calculate the population mobility index to quantify the population dynamics of different functional areas: Step 1: Estimate the total population of the study area: ,in, Indicates the functional area determined based on LBS data. In time Number of location requests; This refers to the number of sampling time points, specifically the time slice corresponding to LBS data acquisition. In particular, the time granularity of LBS sampling is 1 hour. This represents the total number of functional zones. This represents the estimated total population of the study area.

[0049] Step 2: Population estimation for functional areas: ,in, Indicates functional area The number of request grids within, Indicates functional area Internal Time of the first The number of grid location requests is allocated proportionally to ensure that the sum of the populations of each functional area equals the total population of each functional area. .

[0050] Step 3: Calculate the population mobility index for each functional category: ,in, Functional category The number of functional blocks below; For category At any moment The PFI is a liquidity index, where PFI>0 indicates net inflow and PFI<0 indicates net outflow, and the absolute value indicates intensity.

[0051] Step 4: Based on the PFI time series curve, calculate the "Peak Time Shift" by comparing the peak times during the rainstorm period and the normal period. This is the amount by which the peak time of population inflow or outflow for a certain functional category is advanced or delayed during the rainstorm period compared to the normal period, quantifying the advance or delay of travel time. The calculation formula is as follows: ,in, Number of days for sampling; , These represent the peak times of functional category c during the rainy / flood season and the normal season on day k, respectively.

[0052] Step 5: Calculate the duration change ratio by comparing the duration of the PFI peak: ,in, , These represent the start and end times of activities in functional category c under different scenarios. This indicator reflects the change in the duration of travel activities in functional areas under disaster scenarios.

[0053] By using the ratio of peak time shift to activity duration change, the time response characteristics of different functional categories under rainstorm events can be quantified, such as earlier travel, delayed travel, or peak compression.

[0054] Specifically, in this embodiment of the invention, the following calculation method can be used to calculate the functional area attractiveness index to measure the change in the attractiveness of the functional area to the population and to characterize the spatial response pattern of the population during a disaster: Step 1: Calculate the functional area attractiveness index: ,in, Let be the population density of functional zone j at time t; Let be the hotspot influence factor of functional region j at time t. The attractiveness of functional region j at time t; Step 2: Construct the hotspot impact index using the distance decay function. ,in for Hotspot Areas The degree of population density concentration for Hotspot Areas population density Functional area hotspot areas European distance, This is the distance attenuation coefficient, typically ranging from 1.0 to 2.0. A lower value (close to 1.0) reflects a more developed urban transportation system, where the effect attenuates more slowly with distance. Step 3: Calculate the average ATI for each function category: ,in, Indicates functional area category In time The average attractiveness index; Step 4: Calculate the average ATI of various functional zones during the rainstorm period and the normal period, and calculate their ATI change rate: ,in, Functional area categories In time The ATI change rate during heavy rain and normal periods is used to reflect the relative change in the attractiveness of functional zones under disaster conditions.

[0055] In this embodiment of the invention, if the ATI (Active Time Index) of a certain functional area increases significantly during heavy rain, it indicates that residents' tendency to stay or seek refuge in that area is enhanced; conversely, it indicates that travel demand in that area is reduced. Through time series comparison and spatial distribution display, the spatial response behavior characteristics of people under disaster conditions, such as concentrated refuge and reduction of unnecessary travel, can be intuitively revealed.

[0056] This invention uses urban functional zone division as the spatial basis, identifies and parameterizes crowd response behavior based on location service (LBS) data, generates dynamic behavior rules using probabilistic finite state machine (PFSM), and embeds them into an agent-based simulation model (ABM) coupled with a hydrodynamic model. Finally, it achieves dynamic and refined assessment of urban rainstorm and flood risk under the framework of "hazard factor-exposure-vulnerability (HEV)".

[0057] In this embodiment of the invention, step S103 includes the following sub-steps: S1031, Establish rules for individual behavior in normal modes based on probabilistic finite state machines; S1032, using the spatiotemporal response behavior characteristics of crowds in different functional areas in the time dimension, the activity time parameters in the finite state machine are corrected; S1033, using the spatiotemporal response behavior characteristics of crowds in different functional areas in the spatial dimension to correct the state transition probability in the finite state machine.

[0058] In this embodiment of the invention, a crowd travel activity chain model based on a probabilistic finite state machine (PFSM) can be established. Using the spatiotemporal response features (changes in PFI and ATI) identified in step S102, the state transition time parameters and spatial transition probabilities in the PFSM are corrected, generating a set of response pattern rule bases that can reflect crowd behavior under disaster scenarios.

[0059] Specifically, a probabilistic finite state machine (PFSM) can be established to represent the response behavior of different population groups (such as students, employed individuals, retirees, or unemployed individuals). The PFSM is formalized as a quintuple: ,in: It is a finite set of activity states, including {residence, work, study, dining and shopping, leisure and entertainment, and others}; This is the initial state; Let be the state transition probability function, satisfying ; Mapping between status and functional area categories; This is a rule update function used to adjust state transition probabilities under disaster scenarios. This PFSM can comprehensively describe the spatiotemporal movement chains of people between urban functional areas. Furthermore, in this embodiment of the invention, residents' travel behavior can be abstracted as a state transition process of a finite state machine. The illustrated state transition relationship is as follows: Initial State Corresponding to "residence status"; under normal circumstances, individuals can proceed from [location] according to the normal timeline. Shift to a work, study, or other activity state (such as...) Its transition probability is determined by Given: In response mode, the individual transfer rules are consistent with the normal mode, but the activity time parameters and spatial transfer probabilities are modified by subsequent steps, thus forming a new... Therefore, the same state machine corresponds to two sets of transition probability rules in normal mode and response mode: and ; Under the influence of disasters, the peak time shift calculated by the aforementioned steps in this embodiment of the invention is utilized. and duration change ratio The time parameter in the state transition probability is corrected; specifically, the activity time is corrected using the ratio of the peak time shift to the duration change of the PFI. ,in, For activities in response mode Time parameters; Activities under normal conditions The time parameter. The average peak time shift for the functional categories involved in activity n, This represents the ratio of the average duration variation of the functional categories involved in activity n.

[0060] Under the influence of disasters, this embodiment utilizes the calculations made in step S32. Specifically, the spatial parameters in the state transition probability are modified by using... Corrected transition probability: ,in, This represents the transition probability under normal conditions. for Time Function Category The rate of change in attractiveness, For the event The number of functional categories involved needs to be normalized after correction.

[0061] In this embodiment of the invention, after dual correction in both time and spatial dimensions, the PFSM transition rules are updated to form a probability library under the "response mode". This probability library contains detailed transition probabilities for different groups (students, employed, retired, etc.), time periods (e.g., peak, off-peak, night), and activities (residence, work, education, leisure, etc.). Through this library, corresponding crowd response behavior rules can be invoked during agent-based simulations, enabling the simulation results to reflect real-world crowd travel patterns under rainstorm and flood disaster conditions.

[0062] In this embodiment of the invention, step S104 includes the following sub-steps: S1041, based on road network data and rainfall data, uses an underground drainage network model to simulate the hydraulic process of the underground drainage network; and uses a two-dimensional surface runoff model to simulate the surface runoff process. S1042, the underground drainage network model and the two-dimensional surface runoff model are coupled to dynamically simulate and generate the inundation depth and inundation range, which are used as the disaster-causing factor data.

[0063] In this embodiment of the invention, a combination of underground drainage models (such as SWMM) and surface runoff models (such as LISFLOOD-FP) can be used to construct an underground-surface coupled urban flooding simulation environment, dynamically generating the total flood depth and range.

[0064] Specifically, a SWMM (Storm Flood Management Model) can be used to simulate the hydraulic processes of underground drainage networks, while a LISFLOOD-FP (Two-Dimensional Hydrodynamic Model) can be used to simulate surface runoff. By setting a unified rainfall input and time step, the two models are coupled to achieve bidirectional feedback between overflow from the drainage network to the surface and surface water infiltration into the drainage network, thereby dynamically generating high-precision maps of urban inundation extent and water depth distribution. This provides dynamic input for subsequent risk simulations.

[0065] Specifically, in the environmental intelligent agent construction phase, the urban road network is digitized and represented as a graph model. ,in For intersection nodes, The roadside is represented by various attributes, each assigned to traffic and disaster-related properties, including road length, traffic capacity, and water depth. A "threat-weighted cost" model is constructed based on these attributes, and Dijkstra's algorithm is used to find the path with the minimum cost during path planning; that is, a path that comprehensively considers the shortest distance, good traffic capacity, and low risk of water accumulation under the current environment. This path result guides the movement decisions of individual residents in an agent-based model (ABM), enabling simulated individuals to avoid roads with severe water accumulation or restricted access, thus more realistically reflecting travel choices under heavy rain and flooding conditions.

[0066] In this embodiment of the invention, in addition to the road network, functional areas are also embedded as surface agents in the environment model. Each functional area agent includes attributes such as area range, function type, population density, and water depth, and serves as the starting point or end point for crowd movement. During the simulation, the water depth information of the functional area agents can be connected in real time with the results of SWMM and LISFLOOD-FP, thereby dynamically adjusting their accessibility and attractiveness, and providing an environmental basis for crowd behavior simulation.

[0067] After completing the underground-surface coupling and the construction of road and functional area proxies, a dynamically updated urban flood environment is finally obtained. This environment can output a global water depth raster, a road accessibility matrix, and functional area water accumulation characteristics at each time step.

[0068] Therefore, in this embodiment of the invention, the collected rainfall data (e.g., monitoring data from rain gauges) can be input, and the output of the total flood depth and range can be used as the core environmental input for subsequent road accessibility determination and functional area exposure calculation.

[0069] In this embodiment of the invention, step S105 includes the following sub-steps: S1051, city residents are defined as intelligent agents with decision-making capabilities; S1052, drive each agent to determine the travel destination at each time step according to the individual behavior rule base of the response mode; S1053, Based on the travel destination, perform route planning to determine the route with the minimum threat-weighted cost that takes into account road length, traffic capacity and flood depth. S1054, determine the spatiotemporal evolution map of population distribution in each functional area, and the dynamic population flow and population exposure data of road sections.

[0070] In this embodiment of the invention, agent-based dynamic simulation (ABM) is adopted. In a coupled flood environment, individual residents are treated as agents. Based on the individual behavior rule base generated in step S103 and the decision model based on utility maximization, travel decisions and route planning are performed, and the simulation outputs the dynamic spatiotemporal distribution of the population.

[0071] In this embodiment of the invention, an agent-based interactive modeling (ABM) is run in an urban flooding simulation coupled environment. Individuals select destinations and plan routes according to utility functions, and dynamically adjust their travel behavior in combination with PFSM rules to output the spatiotemporal distribution of the population.

[0072] In this embodiment of the invention, the destination selection utility function of the agent comprehensively considers the attractiveness of the target functional area, the distance from the current location, and the average flooding depth of the target functional area; the path planning uses algorithms such as Dijkstra to find the path with the minimum threat-weighted cost that comprehensively considers road length, traffic capacity, and average water depth in the road network.

[0073] Specifically, urban residents can be set as intelligent agents with autonomous decision-making capabilities. Each intelligent agent has attributes such as the group they belong to (students, employees, etc.), place of residence, place of work, etc., and follows the individual behavior rule library generated in step S103.

[0074] Specifically, at each time step, an individual resident agent needs to select a travel destination according to behavioral rules. In this embodiment of the invention, a destination selection model based on utility maximization is constructed, and the utility function is defined as follows: ,in, For the group exist Always check the functional areas The attraction For the group exist Time to reach the functional area The shortest path distance, The average water depth of the functional zone. The greater the utility, the more likely the functional zone is to be selected as the destination; if... If the individual postpones or cancels their trip, then it is determined that the maximum utility value maxU is lower than a preset threshold among all candidate functional areas. This indicates that the combined attractiveness of all destinations under the current circumstances is insufficient to support an individual's travel. In this case, it is determined that the individual should postpone or cancel their trip, maintain their current status, or remain in their current safe location. This preset threshold... It can be calibrated using historical behavioral data or validation experiments to simulate residents' behavior patterns of "reducing non-essential travel" in disaster scenarios.

[0075] Once an individual has determined their destination, they can use a road network map. The simulation process employs pathfinding, where individuals utilize Dijkstra's algorithm within the road network to select paths based on minimizing threat-weighted costs. The simulation proceeds in discrete time steps, with each time step... The time interval can be set from 5 to 60 minutes. At each time step, individuals update their status (residence, work, leisure, etc.) according to the PFSM rules, and move or stay based on the utility function and path planning results. Through ABM simulation, the spatiotemporal evolution map of population distribution in functional areas, dynamic crowd flow and risk exposure of road segments, and individual travel trajectories and path change records can be output.

[0076] In this embodiment of the invention, the simulation results of the model can not only be used for subsequent risk simulation based on HEV, but also serve as independent supporting information for urban disaster prevention decision-making, such as hotspot area monitoring, congestion risk warning, and evacuation route recommendation.

[0077] In this embodiment of the invention, step S106 includes the following sub-steps: S1061, Calculate the risk value of each functional block and each road segment based on the disaster-causing factor data and the population exposure data; S1062, Based on the risk value, determine the risk level; S1063 outputs a dynamic risk map based on the risk levels of each functional block and each road segment.

[0078] In this embodiment of the invention, within the HEV framework, risk values ​​for functional zones and roads can be calculated by combining disaster-causing factor data (H), population exposure data (E), and vulnerability coefficients (V), and risk zones and high-risk zones can be divided using empirical thresholds. Specifically, the flooding depth output in step S104 can be used as the disaster-causing factor (H), and the population exposure data (E) output from the simulation in step S105 can be combined with preset functional zone and road vulnerability coefficients (V) to calculate and output the risk values ​​of each spatial unit at different times.

[0079] Specifically, for any functional block, the formula for calculating its risk value is as follows: ,in, for Time of the first Average water depth in each functional area The vulnerability coefficient of the functional area. for Time of the first The vulnerability coefficient for the exposed population in each functional area can be preset according to the functional category. For example, it can be 0.3 for residential areas, 0.4 for educational areas, 0.6 for commercial areas, and 0.1 for green spaces. Through this calculation formula, the disaster risk level of different functional areas under different time conditions can be quantitatively described.

[0080] For any road segment, the formula for calculating its risk value is: ,in, for Time of the first The average water depth of each section of the road Road vulnerability (usually set to 1.0). For Time of the first This calculation method can identify the degree of threat posed to public safety by flooded roads by the exposed population in each road segment.

[0081] To transform continuous risk values ​​into actionable grading results, this embodiment of the invention introduces a threshold determination method based on empirical distribution. First, a first threshold is selected based on the risk value distribution of all functional zones and roads throughout the entire simulation period. This is used to define "risk zones," such as when R > 0.01, a zone is considered risky; then a second threshold is selected. Used to delineate "high-risk areas", such as R> This area is designated as a high-risk zone. The risk level classification not only helps identify key areas of concern but also provides tiered early warnings for emergency response.

[0082] In this embodiment of the invention, after risk simulation and result classification are completed, the results are output in the form of GIS raster layers and road vector maps to form a dynamic risk map. This map can intuitively display the risk distribution of different time periods, functional areas, and roads, supporting decision-making departments in resource allocation and intervention measure design for key areas.

[0083] In this embodiment of the invention, the simulation results from the above steps can be compared with LBS data to verify the accuracy of the model in terms of spatial distribution, hotspot identification, and behavioral patterns. Parameter sensitivity analysis and cross-validation calibration can also be performed to ensure the robustness and rationality of the results.

[0084] Specifically, to verify the reliability of the model in simulating population distribution, a key period during a rainstorm (e.g., 16:00–23:00) can be selected as a comparison window. The spatial distribution of the population output by the ABM simulation is compared with the population distribution represented by actual LBS data to calculate the overall spatial consistency index. If the spatial distributions of the two remain highly consistent for most of the time period, it indicates that the model has good spatial stability and can effectively depict population migration patterns during disasters.

[0085] In this embodiment of the invention, accuracy verification can also be performed on hotspot areas where crowds congregate. Specifically, the top 10% of hotspot areas in the measured population distribution are selected, and their spatial overlap with the predicted hotspot areas in the simulation results is calculated. If the hotspot identification accuracy remains at a high level (e.g., greater than 80%), it indicates that the model can accurately capture the trend of crowd concentration under disaster conditions. Simultaneously, the matching accuracy of zero-population areas can also be calculated to verify the model's effectiveness in identifying "no-man's land."

[0086] In this embodiment of the invention, during the behavioral level verification of the model, it can be checked whether the crowd behavior output by ABM is consistent with the response patterns revealed by the PFI and ATI indicators. Typical verification content includes: whether behavioral patterns such as "returning home early, postponing travel, and reducing non-essential activities" appear; whether there is a time shift or compression in the duration of travel peaks in different functional areas; and whether the changes in the attractiveness of residential areas, educational areas, and commercial areas conform to actual trends. If the simulated behavior matches the observed characteristics, it indicates that the PFSM rule update and ATI correction mechanism is reasonable and effective.

[0087] In one optional implementation, the method further includes: S107, based on the risk values ​​of each road segment, outputs a road risk list; S108, based on the risk level of each functional block, outputs the migration trajectory of high-risk areas; S109, based on the risk value of each functional block, outputs the risk change curve of each functional block.

[0088] In this embodiment of the invention, risk heat maps, road risk lists, risk zone proportions and high-risk area migration trajectories, risk change curves of each functional block, dynamic population distribution animations and key indicator statistics tables can also be output based on the assessment results obtained in step S106, for emergency management and decision support.

[0089] Specifically, risk heatmaps for each functional area at different time steps can be generated first based on the risk assessment results. These heatmaps, based on the risk values ​​obtained in step S106, are rasterized using GIS to visually display the risk intensity and spatiotemporal evolution characteristics of each functional area. The results can be exported as standard map format files (such as Shapefiles), facilitating the overlay of other geographic information on the emergency command platform for comprehensive analysis.

[0090] For the road network, the system can statistically analyze the risk values ​​and trends of each roadside at each time step, generating a road risk list. The list includes fields such as road number, average water depth, number of travelers, and risk level. This report provides real-time reference for urban traffic management departments, supporting measures such as traffic control, temporary closures, or emergency evacuations.

[0091] In this embodiment of the invention, the dynamic changes in the proportion of risk areas and high-risk areas throughout the entire simulation period can also be calculated, and their migration trajectories can be output. Time-series line graphs can be used to observe the trend of changes in the number of risk areas; spatial trajectory visualization can identify the spatial transfer patterns of risk areas at different time periods, thereby assisting decision-makers in understanding the disaster evolution path.

[0092] In this embodiment of the invention, risk change curves statistically categorized by function can also be output. For example, the average and maximum risks for categories such as residential areas, commercial areas, educational areas, and industrial areas at different time periods. This result can intuitively reveal the vulnerability differences and response patterns of different functional areas, providing a basis for zoned governance and differentiated resource allocation.

[0093] In this embodiment of the invention, to enhance the visualization of the results, a dynamic crowd distribution animation can be generated based on ABM simulation data, demonstrating the spatiotemporal migration process of crowds between functional areas and roads. Simultaneously, snapshots of crowd distribution at key moments can be exported for post-disaster review or plan comparison. This function helps emergency managers intuitively understand crowd response patterns under disaster conditions.

[0094] In this embodiment of the invention, during the results output stage, a key indicator statistical table is automatically generated, including the average risk reduction, the road risk reduction ratio, and the functional area risk reduction rate. The indicators can be compared by time period (morning peak, evening peak, night), spatial unit (functional area or road segment), and scenario (normal and response modes) to form structured statistical data, which is convenient for archiving, comparison, and subsequent research.

[0095] To verify the effectiveness of the method provided in this embodiment, an exemplary embodiment is also provided, which selects Futian District, Shenzhen City, Guangdong Province as the research object. This area covers approximately 78.66 square kilometers and has a resident population of approximately 1.55 million, making it one of Shenzhen's core urban areas. From September 7th to 8th, 2023, affected by the outer circulation of Typhoon Haikui, Shenzhen experienced extreme heavy rainfall, with Futian District becoming one of the most severely affected areas. The maximum water depth reached 2.1 meters, and the flooded area accounted for more than 12% of the built-up area. This embodiment uses the "9.7" catastrophic rainstorm event of 2023 as an application scenario to verify and demonstrate the method.

[0096] 1. Data was collected on the administrative boundaries and road network of Futian District (from OpenStreetMap), digital elevation model (DEM), and underground drainage network and flood-prone area monitoring data provided by the Shenzhen Water Authority. For rainfall data, 5-minute resolution rainfall data from Shenzhen meteorological monitoring stations were used as the model-driven input. For socioeconomic and population data, POI data from Gaode Maps (a total of 65,023 valid points) was obtained and deduplicated and categorized; LBS location request data from Baidu Maps' Insight platform was also obtained, covering the normal period (September 4-5) and the rainy season (September 7-8), with a time resolution of 1 hour. Simultaneously, LandScan raster population data was used to validate the LBS data.

[0097] 2. Through kernel density estimation (200-meter bandwidth) and weighted processing, Futian District was divided into 10 functional zones, including 7 single-function zones (residential, commercial, educational, industrial, public service, transportation, and green space), 2 mixed-function zones (residential-commercial and commercial-industrial), and 1 comprehensive functional zone. The results show that residential and mixed-function zones cover a large area, commercial and educational zones are concentrated in the central area, and industrial zones are relatively scattered in the peripheral areas.

[0098] 3. Population Flow Index (PFI) and Functional Area Attractiveness Index (ATI) were calculated based on LBS data. The results show that during the rainy season, the peak inflow to residential areas occurred approximately 2 hours earlier, and the peak intensity increased by about 26.7%. The peak inflow to commercial and educational areas decreased significantly, with the inflow to educational areas decreasing by more than 60%. Spatially, the ATI values ​​of residential and mixed-use areas increased substantially, while the ATI values ​​of educational and commercial areas decreased by more than half, indicating that residents generally chose to shorten their time away from home, return home earlier, or reduce non-essential travel.

[0099] 4. Based on the changes in PFI and ATI, the finite state machine (PFSM) rules for the three groups (students, employed individuals, and retirees / unemployed individuals) were revised. For example, during the rainy season, only about 20% of students leave home for school between 06:00 and 08:00 in the morning, while the remaining 80% stay at home; employed individuals' probability of going out in the evening (dining, leisure) decreases by more than half, and their return home time is 1-2 hours earlier; retirees' daytime travel decreases by more than 30%. The revised rules are used to drive the ABM simulation.

[0100] 5. Input this rule base into the ABM model and simulate a flood environment coupled with SWMM and LISFLOOD-FP. Simulation results show that, compared with the normal mode, the population in commercial and educational areas decreases significantly under the response mode, while the peak population duration in residential areas is prolonged; in terms of road traffic, the evening peak period (17:20-22:30) is shortened to (20:20-21:40), and the average number of people on the road decreases by 65.38%.

[0101] 6. Risk simulation results show that, after considering population response behavior, the overall average risk of the city decreased by 69.05%. Among them, road risk decreased by more than 70% at all times. The risk of functional areas also decreased significantly, but showed heterogeneity: the risk of educational areas decreased by more than 90% in the morning, and the risk of commercial areas decreased by 73.9% in the evening; however, due to the concentration of population in residential areas, the risk of residential and mixed residential and commercial areas actually increased by 1.7 to 2.5 times during certain periods (such as 8:00-11:00 am).

[0102] 7. In terms of spatial distribution, under normal conditions, 26.32% of Futian District was identified as a risk area; while under response conditions, this proportion dropped to 12.58%, indicating that risk areas showed a stronger clustering effect.

[0103] 8. By comparing the simulated population distribution with real LBS data, the overall spatial accuracy of the model exceeds 60% during the period from 16:00 to 23:00, and the identification accuracy of the population hotspot area (Top 10%) is about 83%, indicating that the present invention can effectively and reliably simulate the spatiotemporal dynamics of the population under disaster scenarios.

[0104] Beyond the specific embodiments provided in this invention, various variations can be designed based on data conditions and application scenarios. For example, in hydrodynamic simulation, when underground drainage network data is lacking, only shallow surface water models (such as LISFLOOD-FP or HEC-RAS2D) can be used for simulation, or empirical flood depth-rainfall intensity relationships can be used to replace physical models. Regarding data sources, location service (LBS) data can be replaced with cellular signaling data, public transportation card swipe data, or social media check-in data; POI data can also be integrated with building foundation or land use data. In behavioral modeling, PFSM can be extended to a semi-Markov chain or multi-agent game model, and the utility function can be enhanced with facility carrying capacity or emergency refuge point guidance terms. In rapid response applications, path replanning for the road layer can be omitted, and only functional area risks can be estimated, thereby improving computational efficiency. Any of the above variations or combinations fall within the scope of protection of this invention.

[0105] Traditional flood risk assessment methods often treat populations as "passive disaster bearers," employing static population data or routine commuting patterns, leading to significant discrepancies between assessment results and actual disaster situations. This invention, through quantitative modeling and dynamic coupling of population response behavior, fundamentally improves assessment accuracy, specifically in the following ways: Based on LBS big data, the “Population Flow Index (PFI)” and “Agency Index for Functional Areas (ATI)” are constructed to accurately depict the dynamic response patterns of the population and quantify the differences in population behavior between the rainstorm period and the normal period. For example, it can identify real response characteristics such as “the peak inflow to the residential area is 2 hours earlier” and “the travel volume in the education area is reduced by 60%”, thus avoiding the bias of “inferring disaster risk from normal behavior”.

[0106] In this embodiment of the invention, behavioral characteristics (PFI / ATI changes) are transformed into behavioral rules for agents by using a "probabilistic finite state machine (PFSM)," which corrects activity time parameters (such as peak shift) and spatial transfer probabilities (such as increased attractiveness of residential areas). This results in a matching accuracy of over 60% between the simulated spatiotemporal distribution of the population and real LBS data, and an accuracy of over 83% in identifying hotspot areas, significantly reducing the risk misjudgment rate.

[0107] In this embodiment of the invention, a coupled model of SWMM (Subsurface Drainage Model) + LISFLOOD-FP (Surface Flooding Model) is used to simulate real flood environments and population decisions, and dynamically output the inundation depth / range. At the same time, the agent selects the destination and path based on the "utility maximization decision model" (combining attractiveness, distance, and water depth), avoiding the disconnect between "fixed environment + static population" and making the risk simulation more in line with the actual disaster situation.

[0108] Traditional flood risk assessment methods are mostly "static summaries after a disaster," which are insufficient to meet the needs of real-time response during a disaster. This invention upgrades risk assessment from a "static snapshot" to a "dynamic video" through dynamic simulation and real-time output.

[0109] In this embodiment of the invention, the risk evolution pattern is captured with a time step of 5 to 60 minutes, and the risk values ​​of different time periods (such as morning peak, evening peak, and night) are output to reveal the changes in risk peaks and valleys, providing a basis for time-segmented emergency management (such as staggered evacuation).

[0110] In this embodiment of the invention, dynamic risk heat maps of functional areas / roads can be output to identify risk heterogeneity and migration trajectories, clearly presenting the risk "aggregation-diffusion" process, and guiding emergency resources to be tilted towards "high-risk aggregation areas".

[0111] In this embodiment of the invention, risk comparison and prediction in multiple scenarios are used to compare the risks of "normal mode" and "response mode" and predict the impact of population behavior on risks in advance (e.g., simulating the risk difference between "no population response" and "with response"), providing data support for "guiding people to avoid disasters in advance".

[0112] In this embodiment of the invention, not only is the focus on "risk simulation" emphasized, but also on the transformation and practical application of results. Its output covers the entire process of emergency decision-making: "monitoring-early warning-response-review". Output dynamic risk heat map (GIS format): Overlay geographic information to intuitively display the current high-risk areas for real-time analysis by the command platform; Output a road risk list: including road number, water depth, exposed population, and risk level, supporting traffic control (such as closing flooded road sections) or evacuation route planning; Output crowd migration animation: Visualize the flow trajectory of crowds between functional areas / roads, helping to identify congestion points or blind spots.

[0113] Output functional area risk curves: Output risk change trends according to categories such as residential, commercial, and educational areas, revealing the vulnerability differences of different areas (e.g., the risk in educational areas decreases by 90% in the morning and in commercial areas by 73.9% in the evening), providing a basis for urban flood control planning (e.g., upgrading drainage facilities in commercial areas); Output key indicator statistics table: including average risk reduction, road risk reduction ratio, etc., to support post-disaster review and plan optimization (such as comparing the effects of different disaster avoidance strategies).

[0114] In summary, in this embodiment of the invention, "crowd response behavior" is introduced as a core mediating variable, forming a complete technology chain of "data collection → behavior recognition → rule modeling → environmental coupling → dynamic simulation → risk simulation", thereby realizing closed-loop analysis of "natural disaster - human behavior - risk consequences".

[0115] Specifically, the deep integration of big data and physical models has for the first time combined LBS big data (socially perceived data) with SWMM / LISFLOOD-FP (physical model) and ABM (agent simulation), solving the industry pain points of "macro big data is difficult to implement in micro modeling" and "physical models lack human factors input", providing a model for interdisciplinary (hydrology, geography, sociology) risk assessment.

[0116] Based on the same inventive concept, embodiments of the present invention also provide a dynamic risk simulation system for urban flooding disasters based on intelligent agent interaction modeling, such as... Figure 2 The diagram illustrates the structural block diagram of the urban flood disaster dynamic risk simulation system based on intelligent agent interaction modeling provided in an embodiment of the present invention. Specifically, the system includes: Data acquisition module 201 is used to collect multi-source data of the target study area. The multi-source data includes at least population location service data, point of interest data, road network data and rainfall data during the rainstorm period and normal period. The feature recognition module 202 is used to identify the spatiotemporal response behavior characteristics of the population under the influence of rainstorm disaster events by comparing the spatiotemporal distribution differences of the population during the rainstorm period and the normal period based on the population location service data and point of interest data during the rainstorm period and the normal period. Rule generation module 203 is used to generate an individual behavior rule library under the response mode based on the spatiotemporal response behavior characteristics of the population. The disaster-causing factor output module 204 is used to construct an urban flood inundation simulation environment based on the road network data and the rainfall data, and dynamically output disaster-causing factor data. The simulation module 205 is used to drive an agent-based model to perform dynamic simulation in the urban flooding simulation environment, wherein the agents in the agent-based model follow the individual behavior rule base of the response mode to make movement path decisions and output dynamic population exposure data. The risk output module 206 is used to calculate and output the dynamic risk results of urban rainstorm and flood disasters based on the disaster-causing factor data and the population exposure data.

[0117] Optionally, the system further includes: The urban functional area determination module is used to determine the spatial distribution density and weight of various urban functional areas based on the point of interest data and the road network data, and to obtain an urban functional area layer. The feature recognition module 202 is used for: Based on the population location service data and point of interest data during the rainstorm period and normal period, the population flow index and functional area attractiveness index of each urban functional area are calculated. By comparing the population flow index during the rainstorm period with that during the normal period, the spatiotemporal response behavior characteristics of different functional areas are quantified in the time dimension, including the time shift of peak travel and changes in the duration of activities; By comparing the attractiveness index of functional areas during the rainstorm period and the normal period, the spatiotemporal response behavior characteristics of people in different functional areas are quantified in the spatial dimension, including the relative changes in the attractiveness of functional areas.

[0118] Optionally, the rule generation module 203 is used for Establish rules for individual behavior in normal patterns based on probabilistic finite state machines; By utilizing the spatiotemporal response behavior characteristics of crowds in different functional areas over time, the activity time parameters in the finite state machine are corrected. By utilizing the spatiotemporal response behavior characteristics of crowds in different functional areas in the spatial dimension, the state transition probabilities in the finite state machine are modified.

[0119] Optionally, the disaster-causing factor output module 204 is used for: Based on road network data and rainfall data, an underground drainage pipe network model is used to simulate the hydraulic process of the underground drainage pipe network; a two-dimensional surface runoff model is used to simulate the surface runoff process. The underground drainage network model and the two-dimensional surface runoff model are coupled to dynamically simulate and generate the inundation depth and inundation range, which are used as the disaster-causing factor data.

[0120] Optionally, the simulation module 205 is used for: Urban residents are defined as intelligent agents with decision-making capabilities; Each agent is driven to determine its travel destination at each time step based on an individual behavior rule base under the aforementioned response pattern; Based on the travel destination, route planning is performed to determine the route with the minimum threat-weighted cost that takes into account road length, traffic capacity, and flood depth. Determine the spatiotemporal evolution map of population distribution in each functional area, and the dynamic population flow and population exposure data for road sections.

[0121] Optionally, the risk output module 206 is used for: Based on the disaster-causing factor data and the population exposure data, calculate the risk value of each functional block and each road segment; Based on the risk value, the risk level is determined; Based on the risk level of each functional block and each road segment, a dynamic risk map is output.

[0122] Optionally, the system further includes: The list output module is used to output a road risk list based on the risk value of each road segment; The trajectory output module is used to output the migration trajectory of high-risk areas based on the risk level of each functional block; The curve output module is used to output the risk change curve of each functional block based on the risk value of each functional block.

[0123] Based on the same inventive concept, embodiments of the present invention also provide an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the steps of the dynamic risk simulation method for urban flood disasters based on intelligent agent interaction modeling as described in any of the above embodiments.

[0124] Based on the same inventive concept, embodiments of the present invention also provide a computer-readable storage medium storing a computer program thereon, which, when executed by a processor, implements the steps in the urban flood disaster dynamic risk simulation method based on intelligent agent interaction modeling described in any of the above embodiments.

[0125] Based on the same inventive concept, embodiments of the present invention provide a computer program product, including a computer program / instruction, which, when executed by a processor, implements the steps in the urban flood disaster dynamic risk simulation method based on intelligent agent interaction modeling described in any of the above embodiments.

[0126] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on the differences from other embodiments. The same or similar parts between the various embodiments can be referred to each other.

[0127] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, embodiments of the present invention can take the form of entirely hardware embodiments, entirely software embodiments, or embodiments combining software and hardware aspects. Furthermore, embodiments of the present invention can take the form of computer program products implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0128] Embodiments of the present invention are described with reference to flowchart illustrations and / or block diagrams of methods, terminal devices (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable terminal device to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable terminal device, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0129] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable terminal device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0130] These computer program instructions can also be loaded onto a computer or other programmable terminal device to cause a series of operational steps to be performed on the computer or other programmable terminal device to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable terminal device for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0131] Although preferred embodiments of the present invention have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments as well as all changes and modifications falling within the scope of the embodiments of the present invention.

[0132] Finally, it should be noted that in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or terminal device that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or terminal device. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or terminal device that includes said element.

[0133] The present invention provides a detailed description of a dynamic risk simulation method and system for urban flood disasters based on intelligent agent interaction modeling. Specific examples have been used to illustrate the principles and implementation methods of the present invention. The descriptions of the above embodiments are only for the purpose of helping to understand the method and core ideas of the present invention. At the same time, for those skilled in the art, there will be changes in the specific implementation methods and application scope based on the ideas of the present invention. Therefore, the content of this specification should not be construed as a limitation of the present invention.

Claims

1. A method for dynamic risk simulation of urban flooding disasters based on intelligent agent interaction modeling, characterized in that, The method includes: Collect multi-source data of the target study area, including at least population location service data, point of interest data, road network data, and rainfall data during rainstorm and normal periods; Based on the population location service data and point of interest data during the rainstorm period and the normal period, the spatiotemporal distribution differences of the population during the rainstorm period and the normal period are compared to identify the spatiotemporal response behavior characteristics of the population under the influence of rainstorm disaster events. Based on the spatiotemporal response behavior characteristics of the population, a base of individual behavior rules under the response pattern is generated; Based on the road network data and the rainfall data, an urban flooding simulation environment is constructed, and disaster-causing factor data is dynamically output. In the urban flooding simulation environment, a model based on intelligent agents is driven to perform dynamic simulation. The agents in the model follow the individual behavior rule base of the response mode to make movement path decisions and output dynamic population exposure data. Based on the disaster-causing factor data and the population exposure data, the dynamic risk results of urban rainstorm and flood disasters are calculated and output.

2. The method for dynamic risk simulation of urban flood disasters based on intelligent agent interaction modeling according to claim 1, characterized in that, The method further includes: Based on the point of interest data and the road network data, the spatial distribution density and weight of various urban functional zones are determined to obtain an urban functional zone layer. Identify the spatiotemporal response behaviors of populations under the influence of rainstorm disasters, including: Based on the population location service data and point of interest data during the rainstorm period and normal period, the population flow index and functional area attractiveness index of each urban functional area are calculated. By comparing the population flow index during the rainstorm period with that during the normal period, the spatiotemporal response behavior characteristics of different functional areas are quantified in the time dimension, including the time shift of peak travel and changes in the duration of activities; By comparing the attractiveness index of functional areas during the rainstorm period and the normal period, the spatiotemporal response behavior characteristics of people in different functional areas are quantified in the spatial dimension, including the relative changes in the attractiveness of functional areas.

3. The method for dynamic risk simulation of urban flood disasters based on intelligent agent interaction modeling according to claim 2, characterized in that, Based on the spatiotemporal response behavior characteristics of the population, a rule base for individual behavior under the response pattern is generated, including: Establish rules for individual behavior in normal patterns based on probabilistic finite state machines; By utilizing the spatiotemporal response behavior characteristics of crowds in different functional areas over time, the activity time parameters in the finite state machine are corrected. By utilizing the spatiotemporal response behavior characteristics of crowds in different functional areas in the spatial dimension, the state transition probabilities in the finite state machine are modified.

4. The method for dynamic risk simulation of urban flood disasters based on intelligent agent interaction modeling according to claim 1, characterized in that, Based on the road network data and the rainfall data, an urban flooding simulation environment is constructed, including: Based on road network data and rainfall data, an underground drainage pipe network model is used to simulate the hydraulic process of the underground drainage pipe network; a two-dimensional surface runoff model is used to simulate the surface runoff process. The underground drainage network model and the two-dimensional surface runoff model are coupled to dynamically simulate and generate the inundation depth and inundation range, which are used as the disaster-causing factor data.

5. The method for dynamic risk simulation of urban flood disasters based on intelligent agent interaction modeling according to claim 1, characterized in that, In the urban flooding simulation environment, driving the agent-based model to perform dynamic simulation includes: Urban residents are defined as intelligent agents with decision-making capabilities; Each agent is driven to determine its travel destination at each time step based on an individual behavior rule base under the aforementioned response pattern; Based on the travel destination, route planning is performed to determine the route with the minimum threat-weighted cost that takes into account road length, traffic capacity, and flood depth. Determine the spatiotemporal evolution map of population distribution in each functional area, and the dynamic population flow and population exposure data for road sections.

6. The method for dynamic risk simulation of urban flooding disasters based on intelligent agent interaction modeling according to claim 1, characterized in that, Based on the disaster-causing factor data and the population exposure data, the dynamic risk results of urban rainstorm and flood disasters are calculated and output, including: Based on the disaster-causing factor data and the population exposure data, calculate the risk value of each functional block and each road segment; Based on the risk value, the risk level is determined; Based on the risk level of each functional block and each road segment, a dynamic risk map is output.

7. The method for dynamic risk simulation of urban flooding disasters based on intelligent agent interaction modeling according to claim 6, characterized in that, The method further includes: Based on the risk values ​​of each road segment, output a road risk list; Based on the risk level of each functional block, output the migration trajectory of high-risk areas; Based on the risk value of each functional block, the risk change curve of each functional block is output.

8. A dynamic risk simulation system for urban flooding disasters based on intelligent agent interaction modeling, characterized in that, The system includes: The data acquisition module is used to collect multi-source data of the target study area. The multi-source data includes at least population location service data, point of interest data, road network data, and rainfall data during the rainstorm period and the normal period. The feature recognition module is used to compare the spatiotemporal distribution differences of the population during the rainstorm period and the normal period based on the population location service data and point of interest data during the rainstorm period and the normal period, and to identify the spatiotemporal response behavior characteristics of the population under the influence of rainstorm disaster events. The rule generation module is used to generate a rule library of individual behaviors under the response pattern based on the spatiotemporal response behavior characteristics of the population. The disaster-causing factor output module is used to construct an urban flood inundation simulation environment based on the road network data and the rainfall data, and dynamically output disaster-causing factor data. The simulation module is used to drive an agent-based model to perform dynamic simulation in the urban flooding simulation environment. The agent in the agent-based model makes movement path decisions according to the individual behavior rule base of the response mode and outputs dynamic population exposure data. The risk output module is used to calculate and output the dynamic risk results of urban rainstorm and flood disasters based on the disaster-causing factor data and the population exposure data.

9. An electronic 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 urban flood disaster dynamic risk simulation method based on intelligent agent interaction modeling as described in any one of claims 1-7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When executed by a processor, the computer program implements the urban flood disaster dynamic risk simulation method based on intelligent agent interaction modeling as described in any one of claims 1-7.

Images