Urban rainstorm waterlogging disaster dynamic risk agent model and method and application
By constructing a dynamic risk intelligent body model for urban heavy rain and waterlogging disasters, combining the intelligent body and hydrological and hydrodynamic coupling model, the problems of static analysis, multi-factor comprehensive consideration and insufficient emergency management of flooding simulation in the existing technology are solved, and urban waterlogging is achieved Dynamic simulation and risk assessment of the process have improved the ability to deal with waterlogging disasters.
Patent Information
- Application Number
- CN202411032401.5
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-07-30
- Publication Date
- 2024-12-06
AI Technical Summary
The existing urban flooding simulation method is mainly static analysis, which is difficult to accurately reflect the dynamic changes of the flooding process. It lacks the role of comprehensively considering multiple factors, ignores the dynamic behavior of people, vehicles and other subjects and their mutual influences. Insufficient emergency management can not be provided. Scientific decision-making support.
Build a dynamic risk intelligent body model for urban heavy rain and flooding disasters. Through the intelligent body model, we comprehensively consider factors such as terrain, buildings, roads, rainfall, etc., and use the NetLogo platform to simulate the dynamic behavior and interactions of different subjects in the waterlogging process, combining with hydrological and hydrodynamic dynamics Coupled model to realize dynamic simulation of urban flooding processes and risks.
It improves the accuracy and practicality of the simulation results, can more comprehensively simulate the urban flooding process, provide scientific emergency management decision support, and enhances the city's ability to respond to flooding disasters.
Smart Images

Figure CN119089766A_ABST
Abstract
Description
Technical Field
[0001] This embodiment belongs to the field of urban waterlogging process and its risk dynamic intelligent response simulation, and specifically involves building a dynamic risk intelligent agent model, method and application of urban rainstorm waterlogging disasters. Background Art
[0002] With the acceleration of urbanization, urban waterlogging has become increasingly prominent. Frequent waterlogging incidents have not only caused serious damage to urban infrastructure, but also had a profound impact on residents' lives and the urban economy. Waterlogging is mainly caused by heavy rainfall, imperfect drainage systems, low-lying terrain and other reasons that lead to the inability to drain rainwater in time, resulting in waterlogging and even flooding. Waterlogging not only affects transportation, but may also cause secondary disasters such as power outages and damage to water supply systems, seriously threatening urban safety and residents' lives and property.
[0003] In the existing technology, urban waterlogging simulation methods mainly include simulations based on hydrological models and hydrodynamic models. These models have achieved certain results in describing rainfall-runoff processes and surface water flow, but they still have the following shortcomings in practical applications:
[0004] (1) Static analysis: Traditional simulation methods are mostly static analysis, which is difficult to accurately reflect the dynamic changes of the waterlogging process and cannot respond to the rapid development and changes of waterlogging events in a timely manner.
[0005] (2) Lack of comprehensive consideration: Existing models usually only consider factors such as rainfall, topography or drainage system, but lack the simulation of waterlogging process under the combined effect of multiple factors, making it difficult to comprehensively predict waterlogging risks.
[0006] (3) Ignoring the behavior of the main body: Most models ignore the dynamic behavior of the main body such as people, vehicles, and buildings during the flooding process and their mutual influence, and cannot truly reflect the actual impact of flooding disasters on urban operations and residents' lives.
[0007] (4) Inadequate emergency management: Existing simulation methods have limited application in emergency management and cannot provide scientific decision-making support for urban waterlogging emergency response and disaster prevention and mitigation.
[0008] In recent years, agent-based models have gradually attracted attention. Agent-based models can more realistically reflect the dynamic changes of complex systems by simulating the behaviors and interactions of multiple subjects. In urban waterlogging simulation, the introduction of agent-based models can effectively make up for the shortcomings of traditional models, and realize the dynamic simulation of waterlogging processes and risks by comprehensively considering multiple factors and their interactions. However, the application of agent-based models in urban waterlogging simulation is still in the exploratory stage, and there are challenges such as complex model construction, difficult data acquisition, and limited simulation accuracy. Therefore, it is urgent to construct an agent-based model, method, and application of the dynamic risk of urban rainstorm waterlogging disasters, which can comprehensively and accurately simulate the waterlogging process, provide a scientific basis for urban planning and emergency management, and improve the ability of cities to cope with waterlogging disasters. Summary of the invention
[0009] In view of the problems existing in the above-mentioned prior art, this embodiment constructs an intelligent agent model, method and application for the dynamic risk of urban rainstorm waterlogging disasters. Its purpose is to simulate the urban waterlogging process more comprehensively by comprehensively considering multiple factors such as terrain, buildings, roads, rainfall, etc. through the intelligent agent model. The NetLogo platform is used to simulate the dynamic behaviors and interactions of different subjects in the waterlogging process in real time to improve the accuracy and practicality of the simulation results. By constructing a hydrological and hydrodynamic coupling model and combining multiple data sources and model coupling technology, a dynamic simulation of urban waterlogging processes and risks is achieved, which provides an effective way to solve the deficiencies in the prior art.
[0010] In order to achieve the above purpose, the specific scheme of this embodiment is as follows:
[0011] The method for constructing a dynamic risk agent model for urban rainstorm waterlogging disasters includes the following steps:
[0012] S1, obtain the modeling data of the hydrological and hydrodynamic coupling model and the intelligent agent model respectively;
[0013] S2, based on the modeling data of the hydrological and hydrodynamic coupling model obtained in step S1, constructing a hydrological and hydrodynamic coupling model to simulate and output flooding data including flooding depth, flooding area and flooding time;
[0014] S3, based on the modeling data of the intelligent agent model obtained in step S1, combined with the inundation data described in step S2, set the disaster-prone environment intelligent agent, the disaster-causing factor intelligent agent and the disaster-affected body intelligent agent, and set the interaction rules between the disaster-prone environment intelligent agent, the disaster-causing factor intelligent agent and the disaster-affected body intelligent agent to construct the intelligent agent model.
[0015] Furthermore, the hydrological and hydrodynamic coupling model in step S1 is a coupling model of a storm runoff management model and a flood inundation plain model, and the modeling data of the hydrological and hydrodynamic coupling model are pipe networks, nodes, drains and rainfall data; the modeling data required for the intelligent model are terrain elevation raster data, study area range vector data, building vector data, road vector data, population grid, population characteristic information, and traffic data.
[0016] Furthermore, the steps of constructing the hydrological and hydrodynamic coupling model in step S2 to simulate and output flooding data including flooding depth, flooding area and flooding time are as follows:
[0017] S21, obtaining pipe network, node, drainage outlet, and rainfall data;
[0018] S22, constructing a storm runoff management model: generalizing the rainwater pipe network, generalizing the rainwater pipe network of the study area according to the pipe network, node, and drainage outlet data obtained in step S21, and dividing the study area into several sub-catchment areas by manual division or Thiessen polygon method;
[0019] S23, setting basic modeling parameters: basic modeling parameters include geometric parameters and calibration parameters. Geometric parameters include catchment area, average slope, characteristic width, and imperviousness. Calibration parameters include surface depression storage capacity in impervious areas, surface depression storage capacity in permeable areas, initial infiltration rate, stable infiltration rate, attenuation coefficient, and drought time.
[0020] S24, using measured rainfall data to simulate the stormwater runoff management model, and using the coefficient of variation to determine the accuracy of the simulated value and the actual value of the stormwater runoff management model;
[0021] S25, based on the topographic elevation data of the floodplain, the continuity equation and momentum equation were used to construct a flood plain model;
[0022] S26, the flow, node water level and pipeline flow output by the storm runoff management model simulation are used as input data of the flood inundation plain model to couple to form a hydrological and hydrodynamic coupling model, and after simulation, the inundation data of the inundation area, inundation depth and inundation time are output.
[0023] Furthermore, the coefficient of variation calculation formula described in step S24 is as follows:
[0024]
[0025] In formula (2): C V is the coefficient of variation; is the difference between the actual comprehensive runoff coefficient and the simulated comprehensive runoff coefficient; is the average value of the actual comprehensive runoff coefficient and the simulated runoff coefficient;
[0026] The formula of the continuity equation in step S25 is as follows:
[0027]
[0028] In formula (3), v is the total flow of the unit grid; t is the time; Q up , Q down , Q left , Q right are the flow rates of the grid cells adjacent to it, upstream, downstream, left, and right respectively;
[0029] The momentum equation is given by:
[0030]
[0031] In formula (4), Q ij is the flow rate between the i-th grid and the j-th grid; A ij With R ij are the hydraulic radius of the interface area at the junction of adjacent grids i and j; S ij The hydraulic slope between adjacent grids i and j;
[0032] The calculation formula of the simulation error in step S26 is as follows:
[0033]
[0034] Where: h is the relative error; H0 is the measured water depth; H s To simulate water depth, the water depth forecast uses 20% of the measured value as the allowable error.
[0035] Furthermore, in step S3, the disaster-prone environment agent, the disaster-causing factor agent, and the disaster-bearing agent agent are respectively set as different cell points or turtles in NetLogo according to their attributes and meanings;
[0036] The steps of constructing the agent model are as follows:
[0037] S31, using the gis:load-dataset command, load the terrain elevation raster data, the study area vector data, the building vector data, and the road vector data in NetLogo to set up a disaster-prone environment agent, which includes the terrain environment, geographic boundaries, building distribution and attributes, and transportation network;
[0038] S32, the disaster factor agent includes flooding data of flooding depth, flooding area and flooding time. The flooding data is obtained by constructing a hydrological and hydrodynamic coupling model in step S2, and the output result is obtained under a 30-year rainstorm scenario;
[0039] S33, set up the disaster-sustaining intelligent agent in NetLogo through the to set up command. The disaster-sustaining intelligent agent includes three types of subjects: people, houses, and vehicles. It exists in the form of Turtle in NetLogo.
[0040] S34, setting interaction rules among the disaster-prone environment agent, the disaster-causing factor agent and the disaster-bearing agent agent;
[0041] S35, on the NetLogo platform, the agent model is constructed by completing the set disaster-prone environment agent, disaster-causing factor agent, disaster-bearing agent agent and the interaction rules between agents.
[0042] Furthermore, the interaction rules between the disaster-prone environment agent, the hazard-causing factor agent and the disaster-bearing agent in step S34 include the interaction rules between people and the disaster-prone environment agent and the hazard-causing factor agent, the interaction rules between houses and people and the hazard-causing factor agent, and the interaction rules between vehicles and the disaster-prone environment agent and the hazard-causing factor agent;
[0043] The interaction rules between people and the disaster-prone environment agent and the disaster-causing factor agent, and the interaction rules between houses and people and the disaster-causing factor agent are respectively divided into five levels of risk status of people, namely safe, low risk, medium risk, high risk and extremely high risk, through the NetLogo platform using color coding, and the same water depth parameter threshold is set as follows:
[0044] Safe status: water depth <0.15m, green;
[0045] Low risk state: 0.15m≤water depth<0.27m, blue;
[0046] Medium risk status: 0.27m≤water depth<0.40m, yellow;
[0047] High-risk status: 0.40m≤water depth<0.60m, orange;
[0048] Extremely high risk status: water depth ≥ 0.60m, red;
[0049] When people are at different water depth parameter thresholds, NetLogo uses the move person command to make people choose safe paths and shelters, giving priority to areas with high terrain and buildings. Based on real-time water depth data and road conditions, it simulates people's behavior in choosing the optimal escape path and shelter location.
[0050] The interaction rules between the vehicle and the disaster-prone environment agent and the disaster-causing factor agent are divided into five levels of risk status: safe, low risk, medium risk, high risk and extremely high risk using color coding on the NetLogo platform, and the vehicle submergence depth threshold is set as follows:
[0051] Safe status: vehicle submergence depth <0.15m, green;
[0052] Low risk state: 0.15m≤vehicle submergence depth<0.30m, blue;
[0053] Medium risk status: 0.30m≤vehicle submergence depth<0.50m, yellow;
[0054] High-risk status: 0.50m≤vehicle submergence depth<0.70m, orange;
[0055] Extremely high risk status: vehicle submergence depth ≥ 0.70m, red;
[0056] When the submergence depth of a vehicle exceeds a certain threshold, the vehicle needs to slow down or change routes. In NetLogo, the movecar command is used to simulate the vehicle's selection of a moving route based on real-time water depth data and road conditions.
[0057] Furthermore, the interaction rules between the human and the disaster-pregnant environment agent and the disaster-causing factor agent are set as follows:
[0058] In NetLogo, we use the if and set commands to first determine whether the individual is in a safe state; if so, the individual remains motionless; if the individual is in an unsafe state, we determine whether there are buildings around; if there are no buildings, we continue to determine whether there is an open area without water accumulation; if so, the individual moves to that area; if not, we further determine whether it is in a high or extremely high danger state; if so, the individual waits for rescue; if not, there are individuals who have made decisions around, and the individual obeys their decisions; if there are no individuals who have made decisions, the individual will move to a higher elevation area to avoid danger; if there are buildings, we determine whether the building is safe; if it is safe, the individual remains motionless; if the building is in a severely dangerous state, we wait for rescue; if the building is in a moderately dangerous state, the individual will move to a safe building or a slightly dangerous building; if the building is in a slightly dangerous state, the individual moves to a safe building;
[0059] The interaction rules between the house, people and the hazard agent are set as follows:
[0060] In NetLogo, we use the if and set instructions to first determine whether the building is safe. If it is safe, no further processing is required. If it is not safe, we continue to determine whether the building is in an extremely high or high risk state. If so, the people in the building need to wait for rescue. For buildings in a medium risk state, the people in the building will be transferred to low-risk or safe buildings. If the building is in a low-risk state, the people in the building will be transferred to a safe building.
[0061] The interaction rules between the vehicle and the disaster-prone environment agent and the disaster-causing factor agent are set as follows:
[0062] In NetLogo, the if and set instructions are used to first determine whether the vehicle is safe. If it is safe, no further processing is required. If it is not safe, the system continues to determine whether the vehicle is in an extremely high or high risk state. If so, the vehicle waits for rescue. For vehicles in a medium risk state, the system drives at the lowest speed or takes a detour. If the vehicle is in a low risk state, the system reduces speed or takes a detour.
[0063] A dynamic risk intelligent agent model of urban rainstorm waterlogging disasters constructed by the method.
[0064] An application of the model in dynamic risk simulation and assessment of urban rainstorm waterlogging disasters.
[0065] Furthermore, the dynamic risk simulation assessment of urban rainstorm waterlogging disasters includes dynamic vulnerability assessment and dynamic hazard assessment. The expression of the dynamic vulnerability assessment is as follows:
[0066] V t =W E E t +W S S t -W Re R et (6)
[0067] Where: V t is dynamic vulnerability; E t S is the dynamic exposure of the hazard-bearing body; t Dynamic sensitivity of the hazard-bearing body; R et Dynamic restoring force; W E is the exposure weight coefficient; W S is the sensitivity weight coefficient; W Re Restoration force weight coefficient; the weight coefficient range is [0, 1];
[0068] The expression of the dynamic risk assessment is as follows:
[0069] R t =f(H t ,V t )=W H H t ×W V V t =W H H t ×W V (W E E t +W S S t -W Re R et)(7)
[0070] Where: H t W is the dynamic risk; H W is the risk factor; V is the vulnerability coefficient.
[0071] Advantages of this embodiment
[0072] 1. The dynamic risk agent intelligent simulation method of urban rainstorm waterlogging disasters in this embodiment realizes the dynamic risk agent intelligent simulation of urban rainstorm waterlogging disasters by constructing a dynamic risk agent model of urban rainstorm waterlogging disasters. The agent model comprehensively considers multiple factors such as terrain, buildings, roads, rainfall, etc., and can simulate the urban waterlogging process more comprehensively. The NetLogo platform is used to simulate the dynamic behaviors and interactions of different subjects in the waterlogging process in real time, thereby improving the accuracy and practicality of the simulation results. By constructing a hydrological and hydrodynamic coupling model, accurate flooding depth and flooding depth data are obtained, and efficient assessment of urban waterlogging risks is achieved, providing a scientific basis for urban planning and emergency management.
[0073] 2. The simulation method of this embodiment combines intelligent body modeling technology and hydrological and hydrodynamic coupling technology, can be used to formulate and optimize urban waterlogging emergency response strategies, can improve the city's ability to cope with waterlogging disasters, reduce economic losses and threats to life safety, has high technological advancement and innovation, and fills the gap in existing technologies in dynamic simulation and comprehensive risk assessment. BRIEF DESCRIPTION OF THE DRAWINGS
[0074] Figure 1 A framework flow chart for constructing a dynamic risk agent model for urban rainstorm waterlogging disasters in this embodiment.
[0075] Figure 2 for Figure 1 Framework flow chart of the interaction rules between humans and disaster-prone environment agents and disaster-causing factor agents in the model.
[0076] Figure 3 for Figure 1 Framework flow chart of the interaction rules among houses, people and disaster-causing agents in the model.
[0077] Figure 4 for Figure 1 Framework flow chart of the interaction rules between vehicles in the model and the intelligent agents of disaster-prone environments and disaster-causing factors.
[0078] Figure 5 This is the NetLogo platform interface and the effect diagram of running the intelligent agent model.
[0079] Figure 6The figure is a diagram showing the dynamic risk changes of people, houses and vehicles during urban flooding. (1) to (3) represent the effects when the time step ticks are 120, 240 and 360 respectively.
[0080] Figure 7 This is a quantitative result map of the comprehensive dynamic risk of the population in the study area suffering from heavy rain and waterlogging disasters.
[0081] Figure 8 This is a quantitative result map of the comprehensive dynamic risk of houses in the study area suffering from heavy rain and waterlogging disasters.
[0082] Fig. 9 This is a quantitative result diagram of the comprehensive dynamic risk of vehicles suffering from heavy rain and waterlogging disasters in the study area. DETAILED DESCRIPTION
[0083] The present embodiment is further explained below in conjunction with the accompanying drawings and specific embodiments. It should be noted that this specific embodiment is not intended to limit the scope of rights of the present embodiment.
[0084] like Figures 1 to 9 As shown, this specific embodiment provides a method and application for constructing a dynamic risk intelligent agent model of urban rainstorm waterlogging disasters. Taking the Chaoyangxi Research Area of Nanning City (hereinafter referred to as the Research Area) as an example, the method includes the following steps:
[0085] S1, obtain the modeling data of the hydrological and hydrodynamic coupling model and the intelligent agent model respectively;
[0086] The modeling data of the hydrological and hydrodynamic coupling model are the pipe network, node, drainage outlet and rainfall data obtained through on-site investigation and measured data.
[0087] The modeling data required for the intelligent model are terrain elevation grid data, study area range vector data, building vector data, road vector data, population grid, population characteristic information, and traffic flow data. Specifically,
[0088] Terrain elevation raster data: Digital elevation model (DEM) is a digital surface representation method used to represent terrain height data models. DEM usually exists in two formats: raster (grid) or triangulated network (TIN). Raster DEM is the most common format. Its function is to be divided into regular grids, and each grid cell stores a height value. The DEM used in this embodiment is generated by extracting the GF-7 satellite stereo image pair and stored in a raster format.
[0089] Vector data of the study area: In ArcGIS (Geographic Information System) software, the study area is divided based on terrain elevation data and the distribution of roads and buildings in high-definition satellite images, and saved in shapefile (vector) format;
[0090] Building vector data: Digital surface model (DSM) data and digital elevation model (DEM) data were extracted based on the Gaofen-7 satellite stereo image pair. The difference between DSM and DEM is that it includes not only the elevation of the terrain surface, but also all objects on the ground, such as buildings and vegetation. Overlay analysis was performed on DSM and DEM to obtain the normalized digital surface model (nDSM). nDSM represents the height of the surface object relative to the ground. The outlines and heights of all buildings in the study area were extracted through nDSM and saved in vector format.
[0091] Road vector data: interpret and analyze high-definition satellite images to obtain road line vector data. The specific process is to first extract road edges, then extract linear features, connect disconnected road segments and extract road skeleton lines through morphological processing and thinning algorithms, convert the extracted road lines into vector data format, add attributes to the vector data, perform topology checks and repairs, and ensure the connectivity and correctness of the road network;
[0092] Population grid, population characteristics information, traffic data: through the population data website ( http: / / landscan.ornl.gov / ) Download the population grid data of the study area with a resolution of 1km×1km, obtain the characteristic information data of the population through the Nanning Statistical Yearbook, and obtain the traffic data through on-site investigation and measured data;
[0093] The hydrological-dynamic coupling model is a coupling model of a storm runoff management model and a flood inundation plain model.
[0094] The Stormwater Runoff Management Model (hereinafter referred to as the SWMM hydrological model) is a dynamic rainfall-runoff simulation model developed by the U.S. Environmental Protection Agency (EPA). It is widely used in urban hydrology research, especially in drainage systems, stormwater management, and non-point source pollution control. The workflow steps of the SWMM hydrological model are as follows:
[0095] S11, building the model structure, including setting nodes and catchment points, defining pipes and channels in the drainage network, and dividing sub-basins;
[0096] S12, collect input data, such as rainfall data and time series data, and determine watershed boundaries and drainage divisions;
[0097] S13, setting the geographical attributes and hydraulic attributes of each sub-basin, node and pipeline, and inputting the time series data of rainfall events to set the corresponding parameters;
[0098] S14, after the model is calibrated, the simulation is run to generate rainfall-runoff results.
[0099] The flood plain model (hereinafter referred to as Lisflood-FP hydrodynamic model) is a two-dimensional hydrodynamic model developed by the University of Bristol and widely used in flood risk assessment, flood prediction and water resource management. The Lisflood-FP hydrodynamic model can simulate the propagation of water flow in the floodplain. The workflow steps of the Lisflood-FP hydrodynamic model are as follows:
[0100] S15, collect and prepare geographic information data (such as terrain data) and hydrological data (such as rainfall data);
[0101] S16, define the computational grid of the simulation area, set the boundary condition (.bdy) and initial condition (.bcy) files, and input the terrain data (.asc) and model parameter files (.par);
[0102] S17, use Python scripts or other tools to run the model and calculate the spread and flow process of the flood based on the continuity equation and momentum equation;
[0103] S18, after the model runs, the output results are analyzed and visualized.
[0104] This embodiment constructs an agent model through the NetLogo platform. The NetLogo platform is a programming platform for building and running agent models. NetLogo enables users to easily build, run and analyze agent models through its concise programming language, interactive user interface and powerful visualization function. This embodiment uses NetLogo to define the behavior rules of the agent, set up the simulation environment, and run the simulation to observe the dynamic evolution of the system; the GIS extension plug-in of the platform can facilitate the import of vector and raster data; the terrain elevation data in the model data is in asc raster format, the study area, buildings, and roads are in shapefile vector format, and the data projection coordinates used are all WGS1984. The agent model (Agent-Based Modeling, hereinafter referred to as ABM model) is a computer simulation technology used to study a system composed of multiple interacting autonomous individuals (i.e., agents). The agent model explores and understands the dynamic characteristics and collective behavior of complex systems by simulating the behavior of these agents and their interactions.
[0105] S2, constructing a hydrological and hydrodynamic coupling model based on the modeling data of the hydrological and hydrodynamic coupling model obtained in step S1, and obtaining flooding data including flooding depth, flooding area and flooding time, the specific steps are as follows:
[0106] S21, through field survey and measured data, obtain the pipe network, nodes, drainage outlets, and rainfall data used to construct the hydrological and hydrodynamic coupling model;
[0107] S22, constructing a SWMM hydrological model: generalizing the rainwater pipe network, generalizing the rainwater pipe network of the study area according to the pipe network, node, and drainage outlet data obtained in step S21, and dividing the study area into several sub-catchment areas by manual division or Thiessen polygon method; the specific steps are as follows:
[0108] S221, first generalize the rainwater pipe network, generalize the rainwater pipe network in the study area according to the pipe network, node, and drain outlet data obtained in step S2; obtain the road surface elevation and average slope through the terrain elevation data to arrange the rainwater outlet position; determine the pipe diameter, length and shape of the drainage pipe through hydraulic calculation; rainwater flows from high to low terrain through the drainage pipe to the drain outlet in the form of gravity flow and then flows into the water system; the result of the generalization of the pipe network in the study area is to set 545 rainwater pipes with a roughness coefficient of 0.013; set 539 rainwater outlets (nodes); 44 drain outlets;
[0109] S222, then divide the sub-catchment areas. There are two main methods for dividing sub-catchment areas. The first is manual division of sub-catchment areas. The sub-catchment areas are directly divided according to the direction of the pipeline network, the distribution of buildings and streets. When the data of the study area is complete and the scope of the study area is small, the manual division method divides the sub-catchment areas according to the actual drainage situation and has a good effect; the second is the Thiessen polygon method to divide the sub-catchment areas. The sub-catchment areas are first divided according to the distribution of nodes, and then manual local adjustments are made. This method is suitable for the situation where the scope of the study area is large and has the characteristics of saving time and effort; this embodiment divides the sub-catchment areas by combining automatic and manual methods. The watershed area is first extracted through the hydrological analysis function of ArcGIS, and then the sub-catchment areas are divided by comprehensively considering the distribution information of buildings and roads in the pipeline network and high-definition remote sensing images. Finally, the study area is divided into 88 sub-catchments.
[0110] S23, setting basic modeling parameters: basic modeling parameters include geometric parameters and calibration parameters. Geometric parameters include catchment area, average slope, characteristic width, and imperviousness. Calibration parameters include surface depression storage capacity in impervious areas, surface depression storage capacity in permeable areas, initial infiltration rate, stable infiltration rate, attenuation coefficient, and drought time.
[0111] Specifically, after completing the generalization of the rainwater pipe network and the division of the sub-catchment area in step S22, the basic modeling parameters are set; the basic parameters for constructing the SWMM hydrological model are divided into two categories: geometric parameters and calibration parameters. Generally speaking, the geometric parameters include the sub-catchment area, average slope, characteristic width, imperviousness, etc., and the calibration parameters include the surface depression storage capacity of the impervious area, the surface depression storage capacity of the permeable area, the initial infiltration rate, the stable infiltration rate, the attenuation coefficient, and the drought time; the detailed method and steps for setting the basic parameters are as follows:
[0112] (1) Sub - catchment area: Use the area geometric calculation tool in ArcGIS to batch obtain the areas of all sub - catchments;
[0113] (2) Average slope: Utilize the slope analysis function of ArcGIS to calculate the average slope of each sub - catchment based on DEM data;
[0114] (3) Characteristic width: The SWMM manual defines the characteristic width as the ratio of the sub - catchment area to the longest overland flow path length. In this embodiment, the following formula is used to calculate the characteristic width:
[0115]
[0116] Where: Width is the flow length, m; K is the adjustment coefficient, with 0.2 < K < 5; Area is the sub - catchment area, m 2 ;
[0117] (4) Impervious rate: The impervious rate refers to the proportion of the impervious part in the entire area within the region. By using the area tabulation tool in ArcGIS to count the types and area sizes of land uses in each sub - catchment, and combining with the runoff coefficient values of each land use type in the "Outdoor Drainage Design Standard" (GB50014 - 2021), the impervious rate of each sub - catchment is calculated through weighted calculation;
[0118] (5) The value ranges of the calibrated parameters are shown in Table 1 below:
[0119] Table 1 Value ranges of calibrated parameters
[0120]
[0121]
[0122] S24, use the measured rainfall data to simulate the stormwater runoff management model, and use the coefficient of variation to determine the accuracy of the simulated values and actual values of the stormwater runoff management model;
[0123] Specifically, use the measured rainfall data of two or more 2 - hour rainfall events to simulate the SWMM hydrological model constructed in the above step S22, and use the coefficient of variation to determine the accuracy of the model simulated values and actual values. The coefficient of variation (Coefficient of Variation, abbreviated as CV) is a statistical index used to measure the dispersion or variation degree of a set of data; when the coefficient of variation is within ±5%, it indicates that the simulation accuracy is good. The coefficient of variation calculation formula is as follows:
[0124]
[0125] Where: C V is the coefficient of variation; is the difference between the actual comprehensive runoff coefficient and the simulated comprehensive runoff coefficient; It is the average value of the actual comprehensive runoff coefficient and the simulated runoff coefficient.
[0126] S25, based on the topographic elevation data of the floodplain, the Lisflood-FP hydrodynamic model was constructed using the continuity equation and momentum equation;
[0127] Specifically, after the SWMM hydrological model is built, the Lisflood-FP hydrodynamic model is constructed next. Based on the terrain elevation data of the floodplain, the continuity equation and momentum equation are used to consider the water balance between adjacent grids. The continuity equation and momentum equation are expressed as follows:
[0128] Continuity equation:
[0129] Momentum equation:
[0130] In equations (3) and (4), V is the total flow rate of the unit grid; t is the time; Q up , Q down , Q left , Q right are the flow rates of the grid cells adjacent to it, upstream, downstream, left, and right respectively; Q ij is the flow rate between the i-th grid and the j-th grid; A ij With R ij are the hydraulic radius of the interface area at the junction of adjacent grids i and j; S ij The hydraulic gradient between adjacent grids i and j.
[0131] The coupling type between the SWMM one-dimensional hydrological model and the Lisflood-FP two-dimensional hydrodynamic model is one-way coupling. The data between the two models are not interoperable. The data needs to be converted into the corresponding file format first, such as DEM in asc format, node coordinates in bci format, overflow time series in bdy format, and main parameters such as simulation time and simulation step solution mode in par format;
[0132] The flow, node water level, pipe flow and other data output by the SWMM model simulation are used as the input data of the Lisflood-FP model. After coupled simulation, the final output is the flooded area, flooded depth and flooded time data.
[0133] S26, using the flow, node water level, and pipeline flow output from the stormwater runoff management model simulation as input data for the flood plain model to couple to form a hydrological and hydrodynamic coupling model, and outputting flooding data of flooded area, flooding depth, and flooding time after simulation;
[0134] According to the requirements of my country's Hydrological Information Forecasting Specification (GB / T 22482-2008), the simulation error is used as an indicator to evaluate the effect of the rain and flood waterlogging model on simulating urban waterlogging. In this embodiment, the simulation error of the flooding depth is expressed as the relative error δ h The calculation formula is as follows:
[0135]
[0136] Where: h is the relative error; H0 is the measured water depth; H s To simulate water depth, the water depth forecast uses 20% of the measured value as the allowable error.
[0137] When the relative error is less than 20%, it indicates that the accuracy of the model simulation meets the requirements. It has been verified that the flooding depth data output by the hydrological and hydrodynamic model constructed in this embodiment meets the requirements.
[0138] S3, based on the modeling data of the intelligent model obtained in step S1, combined with the inundation data described in step S2, set the disaster-prone environment intelligent agent including terrain, boundaries, and roads, set the inundation depth, inundation area, and inundation time as the disaster-causing factor intelligent agent, set the disaster-bearing body intelligent agent with people, houses, and vehicles as the main body, and set the interaction rules between the disaster-prone environment intelligent agent, the disaster-causing factor intelligent agent, and the disaster-bearing body intelligent agent to construct an intelligent agent model and realize the dynamic simulation of urban waterlogging risk.
[0139] Specifically, the disaster-pregnant environment agent, disaster-causing factor agent and disaster-bearing agent agent are set as different Patches (cell points) or Turtles (turtles) in NetLogo according to their attributes and meanings; a Patch is a square grid unit in the NetLogo world, and all Patches together constitute the entire NetLogo world. Each Patch has a unique coordinate (pxcor, pycor), where pxcor and pycor represent the position of the patch on the x-axis and y-axis respectively. The center coordinate of the grid is (0,0), and the coordinate values increase or decrease from the center to the outside. The Patch has multiple attributes, such as color, mark and custom variables; NetLogo commands are used to set and obtain the attributes of the Patch, for example, the color of the Patch can be set to red by set pcolor red, or the color of all Patches can be set to green by ask patches [set pcolor green]. Conditional checks and interactions can be performed based on the attributes of the Patch. Based on the attributes of the Patch, complex environments and scenes can be created to simulate the real scene of urban waterlogging.
[0140] A turtle is a mobile agent in the NetLogo world that can move on a two-dimensional grid, rotate, change properties, and interact with other agents. Each turtle has multiple properties, such as position (xcor and ycor), direction (heading), color (color), and shape (shape). Turtles can perform various behaviors through NetLogo commands, such as moving, rotating, and changing color.
[0141] The specific steps to build the agent model are as follows:
[0142] S31, load terrain elevation raster data, study area vector data, building vector data and road vector data in NetLogo through the gis:load-dataset command to set up the disaster-prone environment intelligent body. The disaster-prone environment intelligent body exists in the form of Patch in NetLogo, which serves as a static condition to constrain the movement of the disaster-prone body.
[0143] The disaster-pregnant environment agent Patch includes terrain environment, geographical boundaries, building distribution and attributes, and transportation network; the details are as follows:
[0144] (1) Terrain environment: Enter the command set dem-data gis:load-dataset in NetLogo to load the terrain elevation grid data as the terrain environment patch, which is used to simulate the terrain characteristics of the study area. The terrain environment affects the movement path and speed of the subject (such as people and vehicles). According to the terrain elevation grid data, the highest elevation of the study area is 114m and the lowest elevation is 50m.
[0145] (2) Geographic boundary: Enter the command set yanjiuqu-data gis:load-dataset in NetLogo to load the vector data of the study area as the geographic boundary patch, define the geographic boundary of the simulation area, and ensure that the model runs within the set geographic scope. The setting of the geographic boundary can prevent the subject from moving out of the simulation scope, ensuring the accuracy and reality of the simulation results.
[0146] (3) Houses: Enter the command set house-data gis:load-dataset in NetLogo to load building vector data as house patches, which are used to represent the location, shape, and size of buildings in the simulation area and reflect the actual urban layout. Use the command set pcolor green to set the initial color of the 10,233 houses to green. Each building can have different attributes, such as residential or office buildings. These attributes affect the behavior and interaction of the subject. For example, different types of buildings are affected differently by waterlogging disasters.
[0147] (4) Traffic network patch: Enter the command road-data gis:load-dataset in NetLogo to load the road vector data as a traffic network patch, which is used to construct the traffic network in the area and define the main paths for vehicles and pedestrians to move. Different types of roads have different effects on vehicle speed and path selection. According to the road vector data, 95 roads in the study area are set.
[0148] S32, the disaster factor agent includes flooding depth, flooding area and flooding time. The data is obtained by constructing a hydrological and hydrodynamic coupling model through step S3, and the output result is obtained under the 30-year rainstorm scenario; the dynamic results of waterlogging indicators (flooding depth, flooding area and flooding time) simulated and output by the hydrological and hydrodynamic coupling model are input into the agent model to construct the disaster factor agent; in NetLogo, the flooding depth and flooding area data with a step length of 30 minutes are loaded and read in the same way as the disaster-prone environment to form the disaster factor agent; the set time command is used to set the change time step of the disaster factor agent to 1 minute to facilitate more detailed observation of the dynamic changes of the flooding depth and flooding area over time;
[0149] S33, in NetLogo, set up the disaster-prone intelligent agent through the to set up command. The disaster-prone intelligent agent includes three types of subjects: people, houses, and vehicles. They exist in the form of turtles in NetLogo. Houses are disaster-prone environments in the construction of intelligent agent models, but they are evaluated as disaster-prone bodies in urban waterlogging risk analysis. Their settings have been explained in step S31. Here, only the settings of human and vehicle intelligent agents are explained in detail. The specific contents are as follows:
[0150] (1) People: People represent individual behavior subjects in the intelligent agent model. This embodiment only considers pedestrian traffic. The characteristics of people are set according to gender and movement speed. When waterlogging occurs, people need to move quickly to a safe area. The movement speed and path selection of individuals under different waterlogging severity levels are simulated, including avoiding areas with deep water. According to the population grid data of the study area, enter the command to setup-person in NetLogo to set 4 types of population, totaling 306,645 population turtles; according to the population characteristic information data, enter the set speed command to set the movement speed parameter, setting the movement speed of male population aged 15-59 to 1, the movement speed of female population aged 15-59 to 0.9, the movement speed of male and female aged 60 and above to 0.75, and the movement speed of male and female aged 0-14 to 0.7.
[0151] (2) Vehicles: Vehicles represent transportation tools in the intelligent agent model. According to the traffic flow data, input the command to setup-car in NetLogo to set the vehicle Turtle, input the command set speed to set the initial moving speed of the vehicle, and adjust the traffic flow parameters according to different time periods by setting the slider number-car to achieve a realistic simulation effect. The driving path and speed of the vehicle are affected by the flooding, simulating the driving ability and path selection of the vehicle under different water depths;
[0152] S34, setting the interaction rules among the disaster-pregnant environment agent, the disaster-causing factor agent and the disaster-bearing body agent. In NetLogo, the disaster-pregnant environment agent and the disaster-causing factor agent respectively provide constraints and response conditions for the disaster-bearing body agent. The house agent in the disaster-pregnant environment agent is also affected by the disaster-causing factor agent and changes. The disaster-pregnant environment agent, the disaster-causing factor agent and the disaster-bearing body agent recognize and respond to each other, thus realizing the coordinated operation among the disaster-pregnant environment agent, the disaster-causing factor agent and the disaster-bearing body agent.
[0153] The interaction rules among the disaster-prone environment agent, the disaster-causing factor agent and the disaster-bearing agent agent include the interaction rules between people and the disaster-prone environment agent and the disaster-causing factor agent, the interaction rules between houses and people and the disaster-causing factor agent, and the interaction rules between vehicles and the disaster-prone environment agent and the disaster-causing factor agent.
[0154] In NetLogo, the if depth, set color command is used to set the color of the person according to the water depth, so as to clearly distinguish the risk status of different human agents. For example, if depth<0.15, set color green means that when the water depth of the person is less than 0.15m, it is set to green. The specific water depth parameter thresholds, risk status and color of the interaction rules between the person and the disaster-pregnant environment agent and the disaster-causing factor agent, and the interaction rules between the house and the person and the disaster-causing factor agent are set as follows:
[0155] Safe status: water depth <0.15m, green;
[0156] Low risk state: 0.15m≤water depth<0.27m, blue;
[0157] Medium risk status: 0.27m≤water depth<0.40m, yellow;
[0158] High-risk status: 0.40m≤water depth<0.60m, orange;
[0159] Extremely high risk status: water depth ≥ 0.60m, red;
[0160] When people are at different water depth thresholds, the move person command in NetLogo enables them to choose safe paths and shelters, giving priority to areas with high terrain and buildings. Based on real-time water depth data and road conditions, the behavior of people choosing the optimal escape path and shelter is simulated.
[0161] The specific flooding depth parameter threshold, risk status and color settings of the interaction rules between the vehicle and the disaster-prone environment agent and the disaster-causing factor agent are as follows:
[0162] Safe status: vehicle submergence depth <0.15m, green;
[0163] Low risk state: 0.15m≤vehicle submergence depth<0.30m, blue;
[0164] Medium risk status: 0.30m≤vehicle submergence depth<0.50m, yellow;
[0165] High-risk status: 0.50m≤vehicle submergence depth<0.70m, orange;
[0166] Extremely high risk status: vehicle submergence depth ≥ 0.70m, red;
[0167] The speed and ability of vehicles are affected in flooding. When the depth of water on the road (or the depth of vehicle submergence) exceeds a certain threshold, the vehicle needs to slow down or change routes. The move car command simulates the vehicle to choose a moving route based on real-time water depth data and road conditions.
[0168] (1) Figure 2 As shown, the interaction rules between the human and the disaster-pregnant environment agent and the disaster-causing factor agent are set as follows:
[0169] First, determine whether the individual is in a safe state. If so, the individual remains motionless. If not, further determine whether there are buildings around; if there are no buildings, continue to determine whether there is an open area without water accumulation. If so, the individual moves to that area. If not, further determine whether it is in a severe (high or extremely high) dangerous state. If so, the individual waits for rescue; if not, there are individuals who have made decisions around, and the individual obeys their decisions. If there are no individuals who have made decisions, the individual will move to a higher elevation area to avoid danger; if there are buildings, determine whether the building is safe. If safe, the individual remains motionless; if the building is in a severe dangerous state, wait for rescue; if the building is in a moderately dangerous state, the individual will move to a safe building or a slightly dangerous building; if the building is in a slightly dangerous state, the individual moves to a safe building.
[0170] (2) Figure 3 As shown, the interaction rules between the house, people and disaster-causing factor agents are set as follows:
[0171] First, determine whether the building is safe. If it is safe, no further processing is required. If it is not safe, continue to determine whether the building is in an extremely high or high risk state. If so, the people in the building need to wait for rescue. For buildings in a medium risk state, the people in the building will be transferred to a low risk or safe building. If the building is in a low risk state, the people in the building will be transferred to a safe building. The specific flooding depth parameter threshold, risk state and color settings are the same as the above human agent.
[0172] (3) Figure 4 As shown, the interaction rules between the vehicle and the disaster-prone environment agent and the disaster-causing factor agent are set as follows:
[0173] First, determine whether the vehicle is safe. If it is safe, no further action is required. If it is not safe, continue to determine whether the vehicle is in an extremely high or high-risk state. If so, the vehicle waits for rescue. For vehicles in a medium-risk state, drive at the lowest speed or take a detour. If the vehicle is in a low-risk state, reduce speed or take a detour.
[0174] After completing the setting of the disaster-prone environment agent, the disaster-causing factor agent and the disaster-affected body agent, and setting the interaction rules among the disaster-prone environment agent, the disaster-causing factor agent and the disaster-affected body agent, an agent model is constructed on the NetLogo platform, and the simulation of the above three types of agents is realized through code and parameter setting. The model is run, the simulation results are observed, and the parameters and rules are adjusted as needed to optimize the model effect.
[0175] Based on the above-mentioned intelligent agent model, dynamic risk simulation of urban rainstorm waterlogging disasters is realized. The specific steps are as follows:
[0176] Step 1: Load the constructed model in NetLogo and click the Set Up button. The platform will read the code and execute the instructions to build the agent. The NetLogo platform interface and the model effect after the Set Up is completed are as follows: Figure 5 As shown in the figure; click the go button, the platform will run the model according to the established interaction rules, and the dynamic risk change process of people-houses-vehicles during urban waterlogging can be clearly observed in the main interface, as shown in the figure below: Figure 6 As shown, Figure 6 (1) to (3) in the figure represent the effect diagrams when the time step ticks are 120, 240 and 360 respectively. You can also observe the change of the number of people, houses and vehicles with different risk levels over time steps through the line graph in the main interface of the platform. By running the intelligent agent model, the dynamic simulation results of urban waterlogging risk are obtained.
[0177] Step 2: Combine the simulation results to evaluate the dynamic risk of urban waterlogging. The dynamic risk is reflected in the dynamic process of the sponge city system responding to the danger of rainstorm waterlogging disasters. The dynamic response capacity can be comprehensively defined as dynamic vulnerability, which is specifically refined into the coupled comprehensive response of the exposure and sensitivity of the disaster-bearing body in the study area and the disaster prevention and mitigation capacity of the urban system. The quantitative expression of dynamic vulnerability assessment is as follows:
[0178] V t =W E E t +W S S t -W Re R et (6)
[0179] Where: V t is dynamic vulnerability; E t S is the dynamic exposure of the hazard-bearing body; t Dynamic sensitivity of the hazard-bearing body; R et Dynamic restoring force; W E is the exposure weight coefficient; W S is the sensitivity weight coefficient; W Re Restoration force weight coefficient; the weight coefficient range is [0,1];
[0180] In addition, the comprehensive dynamic risk of sponge city system in response to rainstorm waterlogging is the dynamic hazard H t With dynamic vulnerability V t As a result of the coupling effect, the quantitative expression of dynamic hazard assessment is as follows:
[0181] R t =f(H t ,V t )=W H H t ×W V V t =W H H t ×W V (W E E t +W S S t -W Re R et )(7)
[0182] Where: H t W is the dynamic risk; H W is the risk factor; V is the vulnerability coefficient;
[0183] In particular, the above variables are all time-varying quantities, but their weight coefficients do not consider temporal and spatial variability. The weight coefficients of each factor are determined by the analytic hierarchy process. The results are shown in Table 2. The analytic hierarchy process (AHP) is a systematic method for multi-criteria decision-making. The process is as follows:
[0184] (1) Establish a hierarchical structure: There are three levels: overall goal, criteria, and alternatives;
[0185] (2) Construct a judgment matrix: Compare two factors and construct a judgment matrix
[0186]
[0187] Among them, a ij Indicates the importance of the i-th factor to the j-th factor;
[0188] (3) Consistency test: when CR < 0.1, the consistency of the judgment matrix is reasonable; otherwise, a new consistency test is required;
[0189] (4) Calculate weights: Calculate the feature vector v and normalize it to obtain the weight vector w i , the calculation formula is as follows:
[0190]
[0191] Among them, w i is the weight of the ith factor;
[0192] Table 2 Weights of indicators for dynamic risk assessment of urban waterlogging disaster-bearing bodies
[0193]
[0194] At the same time, combined with the simulation results, the correlation coefficients of the primary and secondary indicators of the hazard and vulnerability factors are gradually determined, and the consistency test of the indicators is carried out to finally obtain the reasonable dynamic risk distribution results of the study area;
[0195] Step 3, then conduct a specific assessment of the dynamic hazard and dynamic vulnerability of urban waterlogging;
[0196] Taking population as an example, the dynamic danger of urban flooding is t It refers to the degree of loss that the disaster-bearing body (pedestrian flow) may suffer from waterlogging under the dual effects of the risk of the disaster-prone environment and the danger of the disaster-causing factors. It has the characteristics of changing continuously with the process of waterlogging. The definition of waterlogging disaster degree is a quantitative definition. t The key points may include inundation depth and inundation area;
[0197] Based on the simulation results of the waterlogging process of the 30-year return period design rainstorm in the study area by SWMM and LISFLLOD-FP, the results of the change of waterlogging inundation depth with time are obtained; the waterlogging inundation depth (ID) in the study area is divided into 4 levels, and the thresholds of each level are: safe (ID < 0.15m), low risk (0.15m < ID < 0.27m), medium risk (0.27m ≤ ID < 0.40m), high risk (0.40m ≤ ID < 0.60m), extremely high risk (0.60m ≤ ID). Specifically, based on the sub-catchment area division, the dynamic risk level of the sub-catchment inundation depth is defined according to the maximum inundation depth of each sub-catchment area;
[0198] From the perspective of the heterogeneity of time, according to the divided levels of the inundation depth and the differences of each level, the law of the change of the affected area with time in the study area from 1 to 8 hours is statistically analyzed; secondly, from the perspective of spatial heterogeneity, the cumulative affected areas of each sub-catchment area at each level within the time period from 1 to 8 hours are statistically analyzed, and according to the natural break method of ArcGIS, the range of the affected area is divided; based on the simulation results of the waterlogging process, the time-varying process of the inundation area (IA) of 88 sub-catchment areas in the Chaoyangxi River Basin is obtained, and further the inundation area level of each sub-catchment area is determined according to the IA area;
[0199] Combined with the above dynamic risk results of the inundation depth and the inundation area (the distribution weights are shown in Table 2), the evolution process of the risk of the rainstorm waterlogging disaster degree in the Chaoyangxi River Basin of Nanning City can be comprehensively determined, that is, the dynamic risk H of the pedestrian flow in the study area that may be affected by the waterlogging disaster is determined t ;
[0200] Step 4, after completing the dynamic risk assessment, specifically evaluate the dynamic vulnerability of urban waterlogging;
[0201] In this embodiment, the comprehensive vulnerability (V t ) is defined as the superposition of the exposure (E t ) of the pedestrian flow affected by the waterlogging disaster, the sensitivity (S t ) and the disaster prevention and mitigation ability (R et ) of the study area system, where the disaster prevention and mitigation ability includes resistance, adaptability and resilience; under the action of the dynamic risk of the waterlogging disaster, the comprehensive vulnerability dynamics of the disaster-bearing bodies in the study area to the waterlogging disaster can be quantitatively determined through agent modeling and simulation;
[0202] Step 5, the dynamic exposure risk (E t) refers to the exposed population elements and their spatiotemporal distribution within a certain range of waterlogging disasters under the scenario of waterlogging disasters. Combined with the waterlogging process and population distribution data information, the human flow exposure risk in the study area is divided into five levels according to the number or density of exposed disaster-affected population. The thresholds of the number of exposed population at each level are: safe (E t = 0 people / area), low risk (1≦E t <15 people / area), medium risk (16≦E) t <30 people / area), high risk (30≦E) t <50 people / area), extremely high risk (50 people / area ≦ E) t ), finally, ArcGIS spatial analysis was used to quantify the time-varying risk level distribution of hourly pedestrian flow disaster exposure in the study area;
[0203] Step 6, Dynamic Sensitivity Risk (S t ) refers to the possible loss degree of people exposed to the flood disaster in the flood disaster area due to their own sensitivity to the disaster. In this embodiment, S t The main reflection of this phenomenon is the difference in the exposure of pedestrians of different genders and ages to waterlogging disasters. First, based on the waterlogging process and the exposure of the affected pedestrians, the S of pedestrians affected by waterlogging in each period of the study area were statistically analyzed according to the differences in the sensitivity of pedestrians to waterlogging of different genders and ages. t Population; secondly, the spatial sensitivity of pedestrian flow to disasters of different genders S t Difference analysis, due to the different sub-catchment areas of population distribution, the cumulative number of people of different genders affected in the sub-catchment area during the 1-8h period was counted; then, the spatial sensitivity St of different age groups to disasters was analyzed. Consistent with the different genders, this embodiment counted the number of different sub-catchments of people of different age groups affected by disasters during the 1-8h period under the 30-year rainfall scenario; finally, based on the above-mentioned St population number and spatial distribution characteristics of the affected population in each period, the intelligent agent response simulation was obtained to obtain the St risk level distribution of pedestrian flow affected by disasters in the study area;
[0204] Step 7: Disaster resilience and mitigation capacity of the study area (R et) refers to the ability of the urban system to resist and adapt to flood disasters and recover from disasters within a certain period of time during the occurrence of torrential rain and waterlogging disasters, including engineering and non-engineering disaster prevention and reduction measures, as well as the disaster-bearing and disaster-resistance capacity of social and economic elements, that is, the comprehensive ability to protect the disaster-bearing body (pedestrian flow) in the study area from the impact of waterlogging disasters; first, the disaster resistance capacity of the study area is manifested as the comprehensive ability of various underlying engineering or infrastructure to resist, regulate, adapt and recover from torrential rain and waterlogging disasters; secondly, in view of the waterlogging disaster process under the comprehensive response of the underlying engineering or infrastructure in the study area, it is necessary to further adopt non-engineering measures to carry out emergency rescue; based on the interpretation and recognition of high-definition remote sensing images, the distribution density map of government agencies and hospitals in the study area was obtained, and further combined with the time-varying characteristics of waterlogging disasters, the disaster resistance and reduction capacity was generalized to each sub-catchment area with the help of ArcGIS spatial analysis, and finally the risk level distribution of the disaster resistance and reduction capacity of the emergency rescue agencies in the study area was obtained;
[0205] Step 8: Steps 3 to 7 obtain the dynamic risk of rainstorm waterlogging disaster in the study area (Ht) and the dynamic vulnerability of the disaster-bearing intelligent agent (i.e., pedestrian flow) (V t ) risk results. Based on the above results, according to the dynamic risk assessment formula of urban rainstorm waterlogging disasters (Formula 7), the analytic hierarchy process is used to determine the weights of each factor (Table 2), and finally the quantitative results of the comprehensive dynamic risk of pedestrian flow in the study area suffering from rainstorm waterlogging disasters are obtained as follows: Figure 7 As shown, the calculation formula is as follows:
[0206] R t =0.333H t ×0.667V t =0.333H t ×0.667(0.4187E t +0.3575S t -0.2238R et )(10)
[0207] The specific calculation process is as follows: First, H t 、E t , S t , R et The results of the factors are normalized, and the risk level of each factor is uniformly defined as 0, 1, 2, 3, and 4, corresponding to safety, low risk, medium risk, high risk, and extremely high risk, respectively. Secondly, formula (10) is used for calculation. This process is realized through spatial overlay analysis and raster calculation of ArcGIS.
[0208] Taking the risk of pedestrian flow in flooding as an example, Figure 7It can be seen from the figure that during the 1-hour period, the risk of human flow disasters is mainly low and medium risk; during the 2-hour period, the risk of human flow disasters reaches the maximum; during the 3-hour period, the overall risk of human flow disasters is slightly lower than that of the 2-hour period, but its risk level and spatial distribution are similar to those of the 2-hour period; during the 4-5-hour period, the risk of human flow disasters shows a rapid downward trend; during the 6-8-hour period, the risk of human flow disasters quickly recovers from a low risk in local areas to a safe state.
[0209] Step 9. For houses and vehicles, the methods and processes of dynamic risk assessment are similar to those for population. Specifically, in terms of the basis of division, the difference between the three is that the population uses sub-catchment areas as the basis for division, while houses and vehicles use functional areas and roads as the basis for division of risk levels respectively; in terms of dynamic hazard assessment, the inundation area and inundation depth are used as the key to define dynamic hazard; in terms of dynamic exposure assessment, the three are based on different data: the population is based on the waterlogging process and population distribution data information, the houses are based on the waterlogging process and the distribution data of the affected houses, and the vehicles are based on the waterlogging process and the measured traffic data information; in terms of dynamic sensitivity assessment, the population reflects its sensitivity through the difference in the waterlogging disasters suffered by pedestrians of different genders and ages, the houses reflect its sensitivity through the difference in the types and levels of the houses affected, and the vehicles reflect their sensitivity through the difference in the waterlogging disasters suffered by different types of traffic;
[0210] The dynamic risk of rainstorm waterlogging in the study area (Ht) and the dynamic vulnerability of disaster-bearing objects (houses and vehicles) (Vt) were obtained by the same method as for population. t ) risk results, and then according to the dynamic risk assessment formula of urban rainstorm waterlogging disaster formula (7) and the weights of each factor in Table 2, the comprehensive dynamic risk quantitative results of houses and vehicles in the study area suffering from rainstorm waterlogging disasters are finally obtained, as shown in Figure 8 and 9 shown.
Claims
1. A method for constructing a dynamic risk agent model for urban rainstorm waterlogging disasters, characterized in that: The steps include: S1, obtain the modeling data of the hydrological and hydrodynamic coupling model and the intelligent agent model respectively; S2, based on the modeling data of the hydrological and hydrodynamic coupling model obtained in step S1, constructing a hydrological and hydrodynamic coupling model to simulate and output flooding data including flooding depth, flooding area and flooding time; S3, based on the modeling data of the intelligent agent model obtained in step S1, combined with the inundation data described in step S2, set the disaster-prone environment intelligent agent, the disaster-causing factor intelligent agent and the disaster-affected body intelligent agent, and set the interaction rules between the disaster-prone environment intelligent agent, the disaster-causing factor intelligent agent and the disaster-affected body intelligent agent to construct the intelligent agent model.
2. The method according to claim 1, characterized in that: The hydrological and hydrodynamic coupling model in step S1 is a coupling model of a storm runoff management model and a flood inundation plain model. The modeling data of the hydrological and hydrodynamic coupling model are pipe networks, nodes, drainage outlets and rainfall data; the modeling data required for the intelligent model are terrain elevation raster data, study area range vector data, building vector data, road vector data, population grid, population characteristic information, and traffic flow data.
3. The method according to claim 1, characterized in that: The steps of constructing the hydrological and hydrodynamic coupling model to simulate and output flooding data including flooding depth, flooding area and flooding time in step S2 are as follows: S21, obtaining pipe network, node, drainage outlet, and rainfall data; S22, constructing a storm runoff management model: generalizing the rainwater pipe network, generalizing the rainwater pipe network of the study area according to the pipe network, node, and drainage outlet data obtained in step S21, and dividing the study area into several sub-catchment areas by manual division or Thiessen polygon method; S23, setting basic modeling parameters: basic modeling parameters include geometric parameters and calibration parameters. Geometric parameters include catchment area, average slope, characteristic width, and imperviousness. Calibration parameters include surface depression storage capacity in impervious areas, surface depression storage capacity in permeable areas, initial infiltration rate, stable infiltration rate, attenuation coefficient, and drought time. S24, using measured rainfall data to simulate the stormwater runoff management model, and using the coefficient of variation to determine the accuracy of the simulated value and the actual value of the stormwater runoff management model; S25, based on the topographic elevation data of the floodplain, the continuity equation and momentum equation were used to construct a flood plain model; S26, the flow, node water level and pipeline flow output by the storm runoff management model simulation are used as input data of the flood inundation plain model to couple to form a hydrological and hydrodynamic coupling model, and after simulation, the inundation data of the inundation area, inundation depth and inundation time are output.
4. The method according to claim 3, characterized in that: The coefficient of variation calculation formula described in step S24 is as follows: In formula (2): C V is the coefficient of variation; is the difference between the actual comprehensive runoff coefficient and the simulated comprehensive runoff coefficient; is the average value of the actual comprehensive runoff coefficient and the simulated runoff coefficient; The formula of the continuity equation in step S25 is as follows: In formula (3), v is the total flow of the unit grid; t is the time; Q up , Q down , Q left , Q right are the flow rates of the grid cells adjacent to it, upstream, downstream, left, and right respectively; The momentum equation is given by: In formula (4), Q ij is the flow rate between the i-th grid and the j-th grid; A ij With R ij are the hydraulic radius of the interface area at the junction of adjacent grids i and j; S ij The hydraulic slope between adjacent grids i and j; The calculation formula of the simulation error in step S26 is as follows: Where: h is the relative error; H0 is the measured water depth; H s To simulate water depth, the water depth forecast uses 20% of the measured value as the allowable error.
5. The method according to claim 1, characterized in that: In step S3, the disaster-prone environment agent, the disaster-causing factor agent, and the disaster-bearing agent agent are respectively set as different cell points or turtles in NetLogo according to their attributes and meanings; The steps of constructing the agent model are as follows: S31, using the gis:load-dataset command, load the terrain elevation raster data, the study area vector data, the building vector data, and the road vector data in NetLogo to set up a disaster-prone environment agent, which includes the terrain environment, geographic boundaries, building distribution and attributes, and transportation network; S32, the disaster factor agent includes flooding data of flooding depth, flooding area and flooding time. The flooding data is obtained by constructing a hydrological and hydrodynamic coupling model in step S2, and the output result is obtained under a 30-year rainstorm scenario; S33, set up the disaster-sustaining intelligent body through the to setup command in NetLogo. The disaster-sustaining intelligent body includes three types of entities: people, houses, and vehicles. It exists in the form of Turtle in NetLogo. S34, setting interaction rules among the disaster-prone environment agent, the disaster-causing factor agent and the disaster-bearing agent agent; S35, on the NetLogo platform, the agent model is constructed by completing the set disaster-prone environment agent, disaster-causing factor agent, disaster-bearing agent agent and the interaction rules between agents.
6. The method according to claim 5, characterized in that: The interaction rules between the disaster-prone environment agent, the hazard-causing factor agent and the disaster-bearing agent in step S34 include the interaction rules between people and the disaster-prone environment agent and the hazard-causing factor agent, the interaction rules between houses and people and the hazard-causing factor agent, and the interaction rules between vehicles and the disaster-prone environment agent and the hazard-causing factor agent; The interaction rules between people and the disaster-prone environment agent and the disaster-causing factor agent, and the interaction rules between houses and people and the disaster-causing factor agent are respectively divided into five levels of risk status of people, namely safe, low risk, medium risk, high risk and extremely high risk, through the NetLogo platform using color coding, and the same water depth parameter threshold is set as follows: Safe status: water depth <0.15m, green; Low risk state: 0.15m≤water depth<0.27m, blue; Medium risk status: 0.27m≤water depth<0.40m, yellow; High-risk status: 0.40m≤water depth<0.60m, orange; Extremely high risk status: water depth ≥ 0.60m, red; When people are at different water depth parameter thresholds, NetLogo uses the move person command to make people choose safe paths and shelters, giving priority to areas with high terrain and buildings. Based on real-time water depth data and road conditions, it simulates people's behavior in choosing the optimal escape path and shelter location. The interaction rules between the vehicle and the disaster-prone environment agent and the disaster-causing factor agent are divided into five levels of risk status: safe, low risk, medium risk, high risk and extremely high risk using color coding on the NetLogo platform, and the vehicle submergence depth threshold is set as follows: Safe status: vehicle submergence depth <0.15m, green; Low risk state: 0.15m≤vehicle submergence depth<0.30m, blue; Medium risk status: 0.30m≤vehicle submergence depth<0.50m, yellow; High-risk status: 0.50m≤vehicle submergence depth<0.70m, orange; Extremely high risk status: vehicle submergence depth ≥ 0.70m, red; When the submergence depth of a vehicle exceeds a certain threshold, the vehicle needs to slow down or change routes. In NetLogo, the move car command is used to simulate the vehicle selecting a moving route based on real-time water depth data and road conditions.
7. The method according to claim 5, characterized in that: The interaction rules between the human, the disaster-prone environment agent, and the disaster-causing factor agent are set as follows: In NetLogo, we use the if and set commands to first determine whether the individual is in a safe state; if so, the individual remains motionless; if the individual is in an unsafe state, we determine whether there are any buildings around; If there are no buildings, continue to determine whether there is an open area without water accumulation. If so, the individual moves to that area. If not, further determine whether it is in a high or extremely high danger state. If so, the individual waits for rescue. If not, there are individuals who have made decisions around, and the individual obeys their decisions. If there are no individuals who have made decisions, the individual will move to a higher elevation area to avoid danger. If there are buildings, determine whether the building is safe. If it is safe, the individual remains motionless. If the building is in a serious danger state, wait for rescue. If the building is in a moderately dangerous state, the individual will move to a safe building or a slightly dangerous building; if the building is in a slightly dangerous state, the individual will move to a safe building; The interaction rules between the house, people and the hazard agent are set as follows: In NetLogo, we use the if and set instructions to first determine whether the building is safe. If it is safe, no further processing is required. If it is not safe, we continue to determine whether the building is in an extremely high or high risk state. If so, the people in the building need to wait for rescue. For buildings in a medium risk state, the people in the building will be transferred to low-risk or safe buildings. If the building is at low risk, the people in the building are moved to a safe building; The interaction rules between the vehicle and the disaster-prone environment agent and the disaster-causing factor agent are set as follows: In NetLogo, we first determine whether the vehicle is safe by using the if and set instructions. If it is safe, no further processing is required. If it is unsafe, continue to determine whether the vehicle is in an extremely high or high risk state. If so, the vehicle waits for rescue; for vehicles in a medium risk state, drive at the lowest speed or take a detour; If the vehicle is in a low-risk state, reduce speed or take a detour.
8. A dynamic risk intelligent agent model of urban rainstorm waterlogging disasters constructed by the method described in any one of claims 1 to 7.
9. An application of the model described in claim 8 in dynamic risk simulation and assessment of urban rainstorm waterlogging disasters.
10. The use according to claim 9, characterized in that: The dynamic risk simulation assessment of urban rainstorm waterlogging disasters includes dynamic vulnerability assessment and dynamic hazard assessment. The expression of the dynamic vulnerability assessment is as follows: V t =W E E t +W S S t -W Re R et (6) Where: V t is dynamic vulnerability; E t S is the dynamic exposure of the hazard-bearing body; t Dynamic sensitivity of the hazard-bearing body; R et Dynamic restoring force; W E is the exposure weight coefficient; W S is the sensitivity weight coefficient; W Re Restoration force weight coefficient; the weight coefficient range is [0, 1]; The expression of the dynamic risk assessment is as follows: R t =f(H t ,V t )=W H H t ×W V V t =W H H t ×W V (W E E t +W S S t -W Re R et )(7) Where: H t W is the dynamic risk; H W is the risk factor; V is the vulnerability coefficient.
Citation Information
Cited By
Urban inland inundation situation real-time prediction method integrating particle filtering and physical model
CN120145788A
Urban system simulation model coupling method, device and product based on disaster chain
CN120745178A
Urban inland inundation prediction method and device, computer equipment and storage medium
CN121071399A
Flood disaster population exposure calculation method and system based on spatial constraint
CN121882484A
A method and system for calculating population exposure to flood disasters based on spatial constraints
CN121882484B
