An urban flood disaster emergency evacuation optimization method

Abstract

The application belongs to the technical field of disaster emergency management, and discloses a kind of urban flood disaster emergency evacuation optimization method, comprising: S1, the basic data of urban area is collected;S2, the basic data is input into the flood inundation model based on InfoWorks ICM, simulates the flood inundation scene of urban area under a rainstorm and outputs flood inundation simulation result;S3, the attribute of evacuation agent in the multi-agent emergency evacuation model based on NetLogo is set, and the evacuation agent includes autonomous decision-making population and dependent guide type population, and the guide is set in autonomous decision-making population;S4, the flood inundation model is integrated into the multi-agent emergency evacuation model, a plurality of different evacuation strategies are set, the evacuation of evacuation agent is simulated, and different evacuation results are obtained;S5, the implementation effect of a plurality of different evacuation strategies is evaluated according to the evacuation result, and the optimal strategy is determined. The urban waterlogging water dynamic and the evacuation behavior of different populations are accurately simulated, and the evacuation efficiency is improved.

Description

Technical Field

[0001] This invention belongs to the field of disaster emergency management technology, and in particular relates to an optimized method for emergency evacuation in urban flood disasters. Background Technology

[0002] Global climate change and accelerated urbanization have led to frequent extreme rainfall events, making urban flooding a major urban risk. Urban public spaces, due to their dense populations, diverse functions, expansion of impermeable surfaces, and low drainage standards, face particularly significant evacuation pressures during floods.

[0003] Existing technologies, such as the "Urban Flood Pedestrian Evacuation Method for Solving Complex Problems of Emergency Evacuation in High-Density Urban Areas" with patent number CN119005676A, can take into account both the evacuation efficiency and the safety of flood evacuation in high-density urban areas. It identifies refuge demand points based on flood risk, selects refuge sites, considers the safety of pedestrian evacuation, and combines them with path optimization algorithms to rationally plan emergency evacuation routes. However, it has the following shortcomings: (1) The stormwater simulation model (InfoWorks ICM) is a one-way data transmission, which makes it difficult to reflect the feedback relationship between water accumulation dynamics and population movement; (2) It does not classify the evacuation subjects in detail, and ignores the differences in risk perception and decision-making ability among different groups. For example, the elderly, children, or tourists lack the ability to find evacuation routes independently compared with local young adults, which can easily lead to being stranded; (3) The emergency response strategy lacks a hierarchical design, and the optimization mechanism for the configuration of guides and the layout of refuge sites is unclear; (4) It lacks a numerical efficiency evaluation system, making it difficult to accurately identify the key bottlenecks in the evacuation method.

[0004] Therefore, the technical problem to be solved by this invention is how to design a technology that can accurately simulate the dynamics of urban flooding and the evacuation behavior of different groups of people, thereby improving evacuation efficiency. Summary of the Invention

[0005] This invention provides an optimized method for emergency evacuation in urban flood disasters, which can accurately simulate the dynamics of urban waterlogging and the evacuation behavior of different groups of people, thereby improving evacuation efficiency.

[0006] To achieve the above-mentioned technical objectives, the present invention adopts the following technical solution:

[0007] This invention provides an optimized method for emergency evacuation during urban flooding, characterized by the following steps:

[0008] S1, collect basic data for urban areas;

[0009] S2, Input the basic data into the flood inundation model built based on InfoWorks ICM, simulate the flood inundation scenario of the urban area under a rainstorm, and output the flood inundation simulation results;

[0010] S3, set the attributes of the evacuation subjects in the multi-subject emergency evacuation model built based on NetLogo. The evacuation subjects include autonomous decision-making groups and guide-dependent groups. The autonomous decision-making groups are those who can plan evacuation routes independently and evacuate on their own. A guide is set in the autonomous decision-making groups. The guide-dependent groups are those who need to follow the guide to evacuate.

[0011] S4, integrate the flood inundation model into the multi-subject emergency evacuation model, generate several evacuation subjects based on the regional population information and the attributes of the evacuation subjects in the basic data, set several different evacuation strategies, simulate the situation where the evacuation subjects evacuate to the refuge places in the basic data in the flood inundation scenario according to the evacuation strategies under different evacuation strategies, and obtain the evacuation results of the evacuation subjects corresponding to different evacuation strategies;

[0012] S5. Evaluate the implementation effect of the several different evacuation strategies based on the evacuation results of the evacuation subjects, and determine the optimal strategy among the several different evacuation strategies.

[0013] The basic data in the urban flood disaster emergency evacuation optimization method described above includes: topographic data, water system distribution data, pipeline layout data, regional population data, refuge site data, and POI data of the urban area.

[0014] The urban flood disaster emergency evacuation optimization method described above includes flood inundation simulation results such as water depth and water area determined based on topographic data, water system distribution data, and pipeline data, and flood risk zoning determined based on the water depth and water area.

[0015] In the urban flood disaster emergency evacuation optimization method described above, the attributes of the evacuation subject set in step S3 include:

[0016] The average response time, movement speed, and movement speed reduction rules due to flooding are set for the autonomous decision-making group and the guided group, respectively. The autonomous decision-making group is set to use the Dijkstra algorithm to plan the optimal passable path and move along the optimal passable path. The guided group is set to follow the guide according to the Flocking rule.

[0017] In the urban flood disaster emergency evacuation optimization method described above, step S3 includes setting the attributes of the evacuation subject, which includes setting the initial location of the evacuation subject to be distributed according to the population heat map.

[0018] As described above, in the urban flood disaster emergency evacuation optimization method, in step S3, setting the attributes of the evacuation subject further includes: setting the initial movement direction of the evacuation subject, wherein the initial movement direction of the evacuation subject is set to move from the initial position of the evacuation subject to a certain type of POI in the urban area, and the POI is determined according to the function of the evacuation subject in the urban area.

[0019] The urban flood disaster emergency evacuation optimization method described above includes the following evacuation strategies:

[0020] The facilitator configuration is as follows: the facilitators are selected proportionally from the self-decision-making population.

[0021] The early warning response is tiered, comprising: Level 4 response: broadcasting the early warning information, with evacuation voluntary and no guides assigned; Level 3 response: broadcasting and text simultaneously, updating flood information at frequency 'a', with evacuation voluntary and no guides assigned, and a recommended evacuation route pushed; Level 2 response: broadcasting and text / image simultaneously, updating flood information at frequency 'b' (where 'b' is greater than 'a'), with evacuation semi-forced, a recommended evacuation route pushed, and a guide assigned to guide those dependent on guidance; Level 1 response: broadcasting, text / image, and voice simultaneously, updating flood information at frequency 'c' (where 'c' is greater than 'b'), with evacuation forced, a recommended evacuation route pushed, and a guide assigned to guide those dependent on guidance.

[0022] The urban flood disaster emergency evacuation optimization method described above further includes refuge site optimization, which includes:

[0023] Add more shelters in remote suburban areas;

[0024] Reserve redundant capacity for refuge sites around high-risk areas in flood risk zones.

[0025] As described above, the optimization method for emergency evacuation in urban flood disasters includes evacuation success rate, average evacuation time per person, and retention rate. The higher the evacuation success rate, the shorter the average evacuation time per person, and the lower the retention rate of the evacuation entity, the better the implementation effect of the evacuation strategy.

[0026] The urban flood disaster emergency evacuation optimization method described above includes the following steps for adjusting the evacuation strategy of the evacuation entity:

[0027] The proportion of the guides was adjusted and different levels of early warning response were used to simulate the evacuation results.

[0028] By comparing the evacuation results, the optimal evacuation strategy is determined.

[0029] Compared with the prior art, the advantages and positive effects of the present invention are:

[0030] By importing the flood inundation model into a multi-agent emergency evacuation model, the co-simulation framework achieves two-way feedback between water accumulation dynamics and evacuation behavior, effectively overcoming the technical shortcomings of the existing model's one-way data transmission and significantly improving the quantitative accuracy of the impact of water accumulation on evacuation. By classifying evacuation subjects and setting up guides to lead guided groups, the evacuation success rate of this group is significantly improved, while reducing disorderly congestion and route confusion, fully leveraging the positive regulatory role of guides in evacuation order. The method in this invention can be directly applied to flood emergency management for various urban public safety situations, providing quantitative basis for evacuation plan formulation and offering strong practical guidance. Attached Figure Description

[0031] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0032] Figure 1 This is a flowchart of an embodiment of the urban flood disaster emergency evacuation optimization method of the present invention;

[0033] Figure 2 This is a schematic diagram illustrating the principle of urban area population heat map extraction in one embodiment of the urban flood disaster emergency evacuation optimization method of the present invention;

[0034] Figure 3 This is a land use overview map of area a in one embodiment of the urban flood disaster emergency evacuation optimization method of the present invention;

[0035] Figure 4 This is a map showing the water depth in region a of the study area in one embodiment of the urban flood disaster emergency evacuation optimization method of the present invention. Detailed Implementation

[0036] like Figure 1 As shown, this application provides an optimization method for emergency evacuation in urban flood disasters, which solves problems such as insufficient quantification of the impact of water accumulation, lack of dynamic feedback in the model, and neglect of population differences in urban flood disaster evacuation. It can significantly improve the efficiency of guided population evacuation and provide technical support for urban flood emergency management. The optimization method for emergency evacuation in urban flood disasters includes the following specific steps:

[0037] S1 collects basic data for urban areas.

[0038] The basic data includes regional topographic data, water system distribution data, pipeline network layout data, regional population data, and refuge site data.

[0039] Among them, topographic data, water system distribution and pipeline layout data are used for subsequent flood inundation model simulation of flood disasters; regional population data includes population structure, population distribution data and population heat map; refuge site data includes refuge site location and refuge site capacity. These two types of information are used for subsequent multi-subject emergency evacuation model simulation of emergency evacuation of people.

[0040] The population structure includes local young and middle-aged adults, elderly, children, and tourists. The local young and middle-aged adults, elderly, and children consist of local permanent residents and working populations, the numbers and proportions of which are obtained based on data from the Seventh National Population Census; the number and proportion of tourists are obtained based on surveys and information collection in the study area.

[0041] Population distribution data is obtained by integrating data on local permanent residents, working population, and tourists. This application illustrates the distribution using the proportions of "60% local young adults, 30% elderly and children, and 10% tourists." Shelters can be located in schools, shopping malls, or plazas, and the initial capacity of these shelters can be allocated according to building area.

[0042] S2 inputs basic data into a flood inundation model built on InfoWorks ICM to simulate a flood inundation scenario in an urban area during a rainstorm and outputs the flood inundation simulation results. Specifically:

[0043] The flood inundation model is a one- or two-dimensional coupled flood inundation model built on InfoWorks ICM (Integrated Catchment Modeling) software. By using a seamless 1D+2D integrated surface runoff model and pipe network hydrodynamic model in InfoWorks ICM, it can simultaneously simulate underground pipe network hydrodynamics (1D), surface runoff (2D), and rivers and floodplains, achieving unified simulation of rainwater, sewage, combined sewer systems, and natural water systems.

[0044] The flood inundation simulation results include water depth, waterlogging extent, and flood risk zones. The solution process is as follows:

[0045] First, the flood inundation model uses the Saint-Venant equations to describe the unsteady flow evolution of the underground pipe network, and employs the Preissmann slot method to simulate the overload pressure flow and backflow phenomenon of the pipe network under extreme rainfall in real time. When the flow rate at the pipe network node exceeds the interception or water passage capacity, the excess water is injected into the surface runoff model as a constant flow source term through the overflow outlet of the inspection well, and is superimposed with the net surface runoff to jointly drive the two-dimensional runoff simulation based on the Shallow Water Equation (SWE).

[0046] The pipeline hydrodynamic model uses the Saint-Venant equations as the governing equations to describe backflow, counterflow and pressure flow phenomena under unsteady flow, and uses the Preissmann slot method (also known as the Preissmann slot method) to handle overload pressure flow.

[0047] The Saint-Venant equations are a pair of equations concerning the conservation of mass and momentum. The Preissmann Slot method is used for simulating overload, enabling the InfoWorks ICM model to perform simulations of various complex hydraulic systems. Furthermore, storage compensation methods can mitigate the inadequacy of pipeline storage space caused by simplified models, thus avoiding erroneous predictions of pipeline overload and flooding.

[0048] The Saint-Venant equations, also known as the dynamic wave equations, consider the flow conditions at both upstream and downstream boundary conditions. They can simulate the impact of backflow on upstream flow, reverse flow, pressure flow, infiltration and outflow losses (relative to the pipe), and the attenuation of flood peaks during pipe propagation. This process is quite complex and difficult to solve analytically. In most cases, the Saint-Venant equations can simplify the calculations and be solved using computers. This method offers high accuracy and wide applicability for dynamic simulations.

[0049] The formula for calculating channel flow using the Saint-Venant equations in the pipeline hydrodynamic model is shown below:

[0050] ;

[0051] ;

[0052] Where A represents the cross-sectional area of ​​the water passage, in m². 2 Q represents flow rate, measured in m³ / s. 3 / s; t represents time, in seconds; The length along the pipe is expressed in meters (m); g is the acceleration due to gravity. Indicates the horizontal tilt angle; S0 represents the distance from the bottom of the pipe to the free water surface; K represents the water conveyance rate; and S0 represents the bottom slope of the pipe.

[0053] When the Saint-Venant equations calculate that the flow capacity of the pipe network reaches its limit, the excess water will be simulated as overflowing from the inspection well through the Preissmann slot method. This "water spit out by the pipe network" becomes the input source of the surface runoff model. The overflowing water accumulates on the surface, and the water depth of each grid is calculated through the Shallow Water Equation (SWE).

[0054] Then, at the numerical computation level, the finite volume method (FVM) combined with the Godunov scheme is used to perform unstructured grid discretization of the simulation region. The time step is automatically adjusted based on the Courant-Friedrichs-Lewy (CFL) condition. Active computational grids are dynamically identified by setting a wet-drythreshold of 0.001m, and grid-level water depth matrix data with spatiotemporal continuity is generated.

[0055] Since turbulence effects cannot be directly simulated in the InfoWorks ICM model, they are included in the energy loss from bottom friction and simulated using Manning's values. The essence of the Shallow Water Equation (SWE) lies in maintaining the conservation of mass and momentum. This form of equation can describe the discontinuities in flow and the changes between gradually varied and rapidly varied flows.

[0056] The shallow water equations are solved using the finite volume method. The finite volume method uses a control volume to describe the relevant region, dividing the simulation region into small spatial grids. The shallow water equations are then integrated between these geometrically divided regions based on the flow rates through the control volume boundaries to obtain relevant equations. The flow rates at the control volume boundaries are calculated using the Riemann solver, and the integrated shallow water equations are solved using the Godunov numerical model. The advantages of the finite volume method lie in its good computational stability, high geometric flexibility, and relatively simple concept.

[0057] Furthermore, the finite volume method is an explicit solution that does not require iterative stabilization. For each grid cell, the necessary time step is calculated using the Courant-Friedrichs-Lewy condition, as shown below:

[0058] ;

[0059] Where C represents a dimensionless Courant number, and to control the stability of the time step, the default value of C in the 2D model is 0.9; This indicates the side length of the spatial grid. In this embodiment, the spatial grid is set to a size of 100m × 100m, i.e. =100; This represents the time step. This formula can calculate the time step using real-time constraints. This ensures that the transmission of flood waves within the 100m spatial grid is neither "instantaneous" nor "missing," thereby guaranteeing numerical stability and physical accuracy during the hydrodynamic simulation process.

[0060] InfoWorks ICM uses unstructured meshes to mesh the simulation area. To reflect the wet / dry state of the mesh, the flood inundation model uses a threshold depth as a standard to determine whether a mesh is wet, and when the water depth in the mesh is less than the threshold depth, the flow velocity is set to 0. In the InfoWorks ICM 2D model, this threshold depth is typically set to 0.001m by default. This avoids artificially creating excessively high flow velocities in wet / dry areas.

[0061] Import the DEM-formatted elevation model into InfoWorks ICM to create a TIN model (irregular triangular face model). After the TIN model is created, determine the 2D calculation area as needed. Use 2D intervals in the TIN model to delineate the area requiring 2D calculations. The 2D intervals in the model are used for mesh generation, with each mesh reading an elevation data point from the TIN model. For buildings, the water level generally does not reach the roof, so the roof is usually considered an impermeable area and can be excluded from the meshed area as a blank area.

[0062] InfoWorks ICM's 2D computational engine is based on solving the Shallow Water Equations (SWE) and uses the TVD shock capture format. The flood inundation model uses the Shallow Water Equations, specifically the Navier-Stokes equations in the mean depth form, to mathematically describe the two-dimensional flow. The Shallow Water Equations used in the InfoWorks ICM model assume that the water flow primarily extends horizontally, neglecting variations in velocity in the vertical direction.

[0063] Therefore, the formula for calculating the depth of accumulated water is as follows:

[0064] ;

[0065] ;

[0066] ;

[0067] Where h represents the water depth; u represents the velocity component of the water flow in the x-direction; and v represents the velocity component of the water flow in the y-direction. This represents the slope component of the water flow in the x-direction; This represents the bottom slope component of the water flow in the y-direction; This represents the frictional component of the water flow in the x-direction; This represents the frictional component of the water flow in the y-direction; This represents the outflow rate per unit area. express Velocity component in the x-direction; express The velocity component in the y-direction.

[0068] The extent of water accumulation was derived based on the following geospatial logic:

[0069] 1. Topology discrimination: Using the 8-neighborhood connectivity rule, the depth matrix is ​​scanned globally to automatically identify and extract spatially adjacent connected wet grid clusters with water level height;

[0070] 2. Scale Aggregation: An adaptive spatial aggregation algorithm is executed. According to the spatial granularity of the distribution density of manholes in the pipeline network of the study area, the discrete grid is resampled and scaled to ensure that the water accumulation pattern and the hydraulic feedback of the pipeline network are logically consistent.

[0071] 3. Rule-based filtering: By applying a rule-based mapping strategy and setting a 'minimum water accumulation area threshold' and a 'typical feature overflow correction coefficient', isolated false water accumulation grids generated by digital elevation model (DEM) errors or numerical oscillations are automatically removed;

[0072] 4. Vector Smoothing: The cleaned mesh boundary is converted from raster to vector, and then fitted with a cubic quasi-uniform B-spline curve or Gaussian smoothing algorithm to generate a vector water accumulation range surface that is dynamically mapped to the water depth, spatially topologically continuous, and closely matches the actual physical characteristics of flood spread.

[0073] Through the above-mentioned end-to-end collaborative simulation, a quantitative characterization of "rainfall, pipeline overload overflow, and path topology blockage" can be achieved.

[0074] Flood risk zoning is based on water depth and extent. Different water depths are divided into different levels of flood risk zones, each corresponding to different traffic conditions, as shown in Table 1 below:

[0075] Table 1. Traffic Status Corresponding to Flood Risk Zones

[0076]

[0077] In the table, h1 < h2 < h3 < h4. The specific values ​​of h1, h2, h3, and h4 can be set according to actual needs. For example, when h ≤ 0.15, the road traffic status is free passage, and evacuees can freely cross the flooded area; when 0.15 < h ≤ 0.3, the road traffic status is slow-moving and restricted, and evacuees need to move slowly when crossing the flooded area, with speed restrictions; when 0.3 < h ≤ 0.5, the road traffic status is dangerous passage, and it is relatively dangerous for evacuees to cross the flooded area; when 0.5 < h ≤ 0.8, the road traffic status is critically prohibited, and it is even more dangerous for evacuees to cross the flooded area, and crossing the water is not recommended; when h > 0.8, the road traffic status is completely prohibited.

[0078] The original output of water depth, water area, and flood risk area are all grid data displayed by ArcGIS, mainly in Shapefile (vector format) and GeoTIFF (raster format).

[0079] During the simulation, a once-in-a-century rainstorm scenario was simulated in InfoWorks ICM. The flood inundation model outputs the water depth, water area, and flood risk zones. The road traffic status is divided according to the water depth, and evacuation routes are selected based on the road traffic status.

[0080] S3 sets the attributes of the evacuation subjects in the multi-subject emergency evacuation model built on NetLogo. The evacuation subjects include autonomous decision-making groups and guide-dependent groups. Autonomous decision-making groups are those who can plan their own evacuation routes and evacuate independently. Guides are set in the autonomous decision-making groups, while guide-dependent groups are those who need to follow the guides to evacuate.

[0081] The multi-agent emergency evacuation model is an ABM (Agent-Based Model) built on NetLogo software. The evacuation agents include autonomous decision-makers and guide-dependent individuals. Autonomous decision-makers include local young adults who are physically fit and have rich local experience, enabling them to independently identify water distribution and road conditions, and plan their own routes. In the event of floods, they can quickly find evacuation routes. A portion of the autonomous decision-makers are selected as guides. Guide-dependent individuals include the elderly, children, and tourists. These individuals are physically poor or lack local experience, and therefore do not have the ability to plan their own routes. In the event of floods, they need to rely on guides for evacuation.

[0082] In NetLogo software, attributes can be set separately for self-decision-making users and guidance-dependent users, including:

[0083] The average response time, movement speed, and movement speed reduction rules due to flooding were set for autonomous decision-making groups. Autonomous decision-making groups were set to use Dijkstra's algorithm to plan the optimal passable path and move along the optimal passable path. Guided groups were set to follow the guide and move according to the Flocking rule.

[0084] This application provides a set of reference walking speeds for evacuation subjects in an ideal environment without water accumulation:

[0085] The average response time for local young adults was 10 minutes. When there was no water accumulation, the movement speed of young males was 1.43 m / s, and the movement speed of young females was 1.39 m / s. As autonomous decision-making groups, young males and young females used Dijkstra's algorithm to plan the optimal passable path and moved and evacuated along the passable path.

[0086] The average response time for the elderly, children, and tourists was 25 minutes. When there was no water accumulation, the movement speed of elderly men was 1.07 m / s, that of elderly women was 1.10 m / s, and that of children was 0.66 m / s. The movement speed of each group of tourists was the same as that of the local population of the same age. As guide-dependent groups, the elderly, children, and tourists followed the Flopking rule (a rule of group behavior) and automatically followed the guide within 100 meters of the guide for movement and evacuation.

[0087] The movement speed of both autonomous decision-making and guidance-dependent individuals is affected by the depth of floodwater, and this effect is manifested through the rules governing the reduction of movement speed caused by floodwater accumulation:

[0088] ;

[0089] in, This represents the actual moving speed of the evacuation subject i at time t after being constrained by the accumulated water; This represents the baseline walking speed of the evacuation team in an ideal environment with no standing water. This represents the velocity reduction factor, a weight determined by the current water depth in the grid.

[0090] The formula for calculating the speed reduction factor is as follows:

[0091] ;

[0092] The specific values ​​of h1, h2, h3, and h4 can be set according to actual needs.

[0093] The formula for calculating the optimal passable path planned by Dijkstra's algorithm for autonomous decision-making populations is as follows:

[0094] ;

[0095] in, Indicates the planned accessible path; This represents the original physical length of road segment e; This represents the difficulty coefficient of passage caused by the water depth h at time t; This represents the delay coefficient caused by crowd congestion on the road segment at time t.

[0096] The calculation formula is as follows:

[0097] ;

[0098] The calculation formula is as follows:

[0099] ;

[0100] in, represents the real-time traffic volume, which is represented in NetLogo as the total number of evacuation agents currently falling on the road segment edge e; c represents the design capacity, which is the maximum number of agents allowed to pass through the corresponding road segment per unit time, c = road width × ideal maximum traffic flow, and the ideal maximum traffic flow can be set to 1.20p / (m·s); The model parameters can be set in accordance with traffic engineering standards. , It determines the initial intensity of congestion; The model parameters can be set in accordance with traffic engineering standards. , This ensures that when the number of people exceeds the road's capacity, the weights will increase exponentially, forcing Dijkstra's algorithm to recognize that "this road is overcrowded," as shown below. =0.

[0101] Before Dijkstra's algorithm begins planning routes, areas with water depths exceeding h4 are designated as restricted zones. Therefore, road segments with water depths exceeding h4 within these zones need to be removed. This is achieved using a dynamic road network topology reduction model, with the formula shown below:

[0102] ;

[0103] in, This represents the effective road network topology available for evacuation at time t. This represents the original, complete road connectivity graph topology within the region; Indicates a specific road segment within a road network; This indicates the depth of water accumulation at point e on the road, as transmitted in real time by InfoWorks ICM. This represents the passage limit threshold, i.e. .

[0104] This formula is used to simulate infrastructure failure and physical water outages caused by extreme disasters, when the water depth in a road section exceeds [a certain value]. At that time, the road section was directly removed from the available map data to completely avoid extremely high-risk areas.

[0105] When planning the optimal path, Dijkstra's algorithm needs to ensure both traversability (light congestion) and safety (low water depth). Therefore, it requires a non-linear path penalty function based on environmental thresholds to score traversable paths. The calculation formula is shown below:

[0106] ;

[0107] in, The score indicates the passable path; Indicates the actual length of the road; This represents the actual moving speed of the evacuation subject i at time t after being constrained by the accumulated water; The score representing the risk-averse mindset of the evacuees is calculated using the following formula:

[0108] ;

[0109] in, , , The value can be set according to the road traffic conditions, for example... This indicates that slow-moving traffic is restricted; This indicates a dangerous passage. =20 represents a critical forbidden condition; The route has been determined to be strictly prohibited and has been removed from the list.

[0110] The elderly, children, and tourists move according to the Flocking rule (group behavior rule). The movement formula for a dependent, guided group i is as follows:

[0111] + + ;

[0112] in, This indicates the direction and speed of the following dependent-guided group i; This indicates the direction and speed at which i is moving; The separation force is represented by the proximity of i to other dependent-guided groups of people (hereinafter referred to as neighbors). The greater the separation force, the more it ensures that the simulated evacuation subjects will not overlap, meaning that real people will not experience stampedes. This indicates the alignment force; whichever direction the neighbor moves in, i will move in the same direction. This represents cohesion, indicating that when i discovers it is far from its neighbors, a cohesive force will arise that pulls it toward the center; It represents the guiding force, which is the attraction vector emitted by the guide, and can correct the movement direction of the guide-dependent group i; , , , These are all weighting coefficients. By adjusting these coefficients, different group behaviors can be created; for example, the higher the water depth, the more likely the group will behave differently. The higher the weight, the deeper the water is trapped, and the more the people who rely on guidance listen to the guidance of the guide.

[0113] The formula for calculating the separation force is as follows:

[0114] ;

[0115] in, Indicates the untreated separation force; This represents the direction vector from the neighbor to the dependent guided group i; This represents the distance from the neighbor to the dependent guided population i. The negative sign at the beginning of the formula indicates the opposite direction.

[0116] To prevent the separation force from being too great and causing the guided group i to be instantly "bounced away," this force will be normalized, transforming it into a standard guide with a length of 1 representing only direction; this is the final separation force. .from arrive The unitization process is existing technology and will not be explained in detail here.

[0117] The formula for calculating alignment force is as follows:

[0118] ;

[0119] in, Indicates alignment force; This represents the average speed of all neighbors within the range; This represents the movement speed of the dependent guide group i.

[0120] The formula for calculating cohesion is as follows:

[0121] ;

[0122] in, It represents cohesion; This represents the average location of the neighbors, i.e., the center point; This indicates the position of the dependent guide group i.

[0123] The formula for calculating guiding force is as follows:

[0124] ;

[0125] in, Indicates guiding force; Indicates the real-time location of the guide; Indicates the position of the dependent-guided group i; The attraction coefficient, which is affected by the depth of the water accumulation, is set as shown in Table 2 below:

[0126] Table 2. Attraction coefficients affected by water depth for passage conditions

[0127]

[0128] The attributes for setting the evacuation subject include: setting the initial location of the evacuation subject to be distributed according to the population heat map of the urban area.

[0129] In a flood-inundated scenario where the floodwaters have not yet arrived, the initial locations of the evacuation subjects are distributed based on the population heat map of the urban area. Compared to random distribution, this reduces randomness and better reflects the actual situation of the urban area, allowing for the design of the optimal emergency evacuation plan when a flood disaster actually occurs.

[0130] This application uses the population heatmap dataset from the Baidu Insight platform to analyze the initial spatiotemporal distribution characteristics of the population within an urban area. This dataset is generated based on terminal location information collected by the Baidu Maps location service SDK, and the urban area is gridded using the BD09MC coordinate system (i.e., the Baidu 2009 Mercator projection coordinate system). As shown in Figure 2, the urban area is divided into multiple rectangular grid cells, each enclosed by blue meridians and horizontal dashed lines. For a single grid cell, the population heatmap value (value) for the corresponding area is calculated by statistically analyzing the total amount of terminal location data within the cell hourly. This application uses the centroid of each grid cell as a reference to extract data from the corresponding locations and conduct subsequent analysis.

[0131] Setting the attributes of the evacuation subject also includes setting the initial movement direction of the evacuation subject. Compared to random movement, the purposeful movement of the evacuation subject is closer to the current situation and is more valuable for subsequent simulations of people moving during floods.

[0132] The initial movement direction of the evacuation subject is: from the initial position of the evacuation subject to a certain type of POI in the area. The type of POI is determined according to the function of the urban area where the evacuation subject is located.

[0133] POI data can abstract real-world geographic entities into point data containing spatial coordinates and attribute categories, intuitively reflecting the land use functions of urban areas. However, traditional POI data generally suffers from redundancy issues such as "one object, multiple categories" and "one object, multiple points." To avoid statistical bias caused by duplicate functional attributes, this application refers to the national standard "Classification of Urban Land Use and Standards for Planning and Construction Land Use" (GB 50137-2011) issued by the Ministry of Housing and Urban-Rural Development in 2011. Combining the actual development characteristics of urban areas, and for the sake of simplifying analysis, this application simplifies the eight categories of land use classification into six categories: residential land, commercial office, living services, leisure and entertainment, green space and square, and public services.

[0134] Meanwhile, in view of the limitations of traditional POI data, which can only represent spatial location and cannot reflect land use scale and public usage frequency, this application introduces type influence coefficients to quantify the importance of different categories of POIs. The original number of POIs is multiplied by the influence coefficient on the application land category to obtain the corrected effective number of POIs. The values ​​of the influence coefficients for each category are detailed in Table 3 below.

[0135] Table 3 Impact coefficients of various POI data

[0136]

[0137] The above functional zone division adopts two core indicators: urban area frequency density (BFD) and urban area type proportion (BCR). The specific calculation formula is as follows:

[0138] ;

[0139] ;

[0140] in This indicates the degree of clustering of the x-th type of POI within the corresponding city area; x is the type of POI. This represents the number of POIs of type x within the urban area; This represents the total number of POIs of type x. This indicates the proportion of different types of POIs within a city area, reflecting the relative proportion of different types of POIs within the city area; Indicates the type of POI.

[0141] When the BFD value of a certain type of POI is ≥0.5, the area is determined to be a single functional area of ​​the corresponding type; when the BFD value of all types of POIs in the urban area is less than 0.5, and there are multiple types of non-zero POI data, the mixed functional area attribute of the area is determined based on the top three POI types with the highest BCR values.

[0142] If the area is a single-function zone, the initial movement direction of the evacuees is towards the POI (Point of Interest) belonging to that function within that area. For example, if the current functional zone is determined to be a business office area, the evacuees within the current functional zone will move towards the business office building.

[0143] If the area is a mixed-use zone, the initial movement direction of the evacuees will be towards the POIs belonging to the three functions within the area. For example, if the current functional zone is determined to be a residential, office, and leisure and entertainment area, the evacuees within the current functional zone will move towards the nearest residential community, commercial office building, and shopping mall, respectively.

[0144] S4 integrates the flood inundation model into the multi-subject emergency evacuation model. Based on the regional population information and the attributes of the evacuation subjects in the basic data, several evacuation subjects are generated, and several different evacuation strategies are set. The simulation shows that under different evacuation strategies, the evacuation subjects evacuate to the refuge places in the basic data in the flood inundation scenario, and the evacuation results of the evacuation subjects corresponding to different evacuation strategies are obtained.

[0145] The flood inundation model is imported into a multi-entity emergency evacuation model. GIS conversion enables dynamic interaction between the flood inundation model's water accumulation data and the multi-entity emergency evacuation model, with periodic synchronous updates of water accumulation information. The two models are coordinated and advanced according to a "pre-start-synchronous operation-finalization" time-sequence control. Specifically, the flood inundation model is pre-started, running for a period (e.g., 1 hour) to generate an initial water accumulation field. Then, the flood inundation model is imported into the multi-entity emergency evacuation model to complete initialization. Both models then run synchronously with a unified step size. In the finalization phase, simulation stops when the evacuation success rate of the evacuation entities is ≥95%, the stranded entities do not move for three consecutive steps, or the water accumulation in over 90% of the urban area is ≤0.1m. "Three consecutive steps without movement" refers to the NetLogo software, which uses a tick-based movement method for the set population. This means that if the entity does not move for three consecutive ticks, the simulation can stop. This is existing technology and will not be described in detail.

[0146] Evacuation strategies include:

[0147] The facilitator configuration is as follows: the facilitators are selected proportionally from the self-decision-making population.

[0148] The early warning response is tiered, including: Level 4 response: early warning information is broadcast, and the evacuation is voluntary without designated guides; Level 3 response: early warning information is simultaneously broadcast and text-based, water accumulation information is updated at frequency 'a', the evacuation is voluntary without designated guides, and recommended evacuation routes are pushed; Level 2 response: early warning information is simultaneously broadcast and text-based, water accumulation information is updated at frequency 'b' (where 'b' is greater than 'a'), the evacuation is semi-forced, recommended evacuation routes are pushed, and guides are assigned to guide those who rely on guidance; Level 1 response: early warning is simultaneously broadcast, text-based, and voice-based, water accumulation information is updated at frequency 'c' (where 'c' is greater than 'b'), the evacuation is forced, recommended evacuation routes are pushed, and guides are assigned to guide those who rely on guidance.

[0149] Refuge site optimization includes the following methods:

[0150] Add more shelters in remote suburban areas;

[0151] Reserve redundant capacity for refuge sites around high-risk areas in flood risk zones.

[0152] The evacuation results of the evacuation entities include the evacuation success rate, the average evacuation time per person, and the delay rate.

[0153] S5. Evaluate the effectiveness of several different evacuation strategies based on the evacuation results of the evacuation subjects, and determine the optimal strategy among them. Evacuation results include evacuation success rate, average evacuation time per person, and retention rate. Evacuation strategies that achieve higher evacuation success rates, shorter average evacuation times per person, and lower retention rates are more effective. The threshold ranges and effectiveness levels for evacuation success rate, average evacuation time per person, and retention rate are shown in Table 4 below.

[0154] Table 4 Threshold ranges and effectiveness levels for evacuation success rate, average evacuation time per person, and retention rate.

[0155]

[0156] The steps for adjusting the evacuation strategy of the evacuation entity include:

[0157] Adjust the proportion of facilitators to match different levels of early warning response to obtain evacuation results. Facilitators can be set to comprise 5%, 10%, 15%, and 20% of the self-decision-making population. Then, different numbers of facilitators can be paired with different levels of early warning response (one to four), and the output results can be observed.

[0158] Compare the evacuation results and determine the optimal evacuation strategy.

[0159] Taking region a as an example, the optimized emergency evacuation method for urban flood disasters in this application is implemented.

[0160] Figure 3 This is a land use overview map of region a; Figure 4 This is a map showing the depth of waterlogging in region a, output by a flood inundation model.

[0161] Based on the combination of four percentage gradients of guides (5%, 10%, 15%, 20%) and four-level early warning response, 10 typical combination scenarios were selected for simulation evaluation. The "four-level early warning + 0% guide" scenario was used as the basic control scenario. All scenarios were based on the unified settings of the study area of ​​region a, 2 million evacuation subjects (60% local young adults, 30% elderly and children, and 10% tourists), and a once-in-a-century rainstorm (deepest water accumulation of 1.23m).

[0162] 1. Evacuation success rate: increased from 62% in the control scenario to 96% in the optimal scenario. The core driving factors are "Level 1 warning triggered 6 hours in advance (allowing sufficient evacuation time) + 20% full coverage of guides (solving the path confusion of people who rely on guides)". The evacuation rate of people in high-risk areas increased from 35% to 80%.

[0163] 2. Average evacuation time per person: reduced from 185 minutes to 65 minutes. This is due to two factors: firstly, early warnings allow people to avoid peak flooding periods; secondly, guides use Dijkstra's algorithm to plan the optimal passable route, reducing detours. The average evacuation time for guided groups has been reduced from 210 minutes to 78 minutes.

[0164] 3. Detention rate: Reduced from 38% to 4%. Detained people were mainly concentrated in local low-lying areas where the initial water level rose rapidly. After optimization, only a small number of elderly people with mobility difficulties were detained due to temporary route blockages, and they could be quickly transferred through subsequent rescue efforts.

[0165] Considering the synergistic optimization effect of the three major indicators, the optimal combination is a Level 1 early warning response (triggered at T=-6h) + a 20% facilitator ratio. The core basis for this is as follows:

[0166] 1. The evacuation success rate reached 96%, meeting the target threshold of ≥95%, and covering the evacuation of personnel in all risk level areas;

[0167] 2. The average evacuation time per person is 65 minutes, which is 65% shorter than the control scenario, avoiding secondary risks caused by prolonged stays;

[0168] 3. The retention rate is only 4%, and all those who remain are in low-risk areas (≤0.3m), making rescue relatively easy;

[0169] 4. This combination can maximize the synergistic effect of "early warning and diversion + precise navigation by guides", adapt to the extreme water accumulation scenario of 1.23m in the study area, and at the same time take into account the evacuation balance between the core urban area and the suburbs.

[0170] The implementation logic of the optimal solution is as follows: initiate a full-channel early warning (broadcast + text + voice) 6 hours in advance, evacuate in batches according to "elderly and children → tourists → young adults", 20% of the guides (about 240,000 people) cover the core road network and the area around the refuge site, guide the evacuation of the guide-dependent population through the Flocking rule, and allow the autonomous decision-making population to avoid obstacles autonomously through the Dijkstra algorithm, so as to achieve efficient and orderly evacuation.

[0171] In the description of this specification, the references to terms such as "one embodiment," "some embodiments," "illustrative embodiment," "example," "specific example," or "some examples," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of the invention. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0172] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit them. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments or make equivalent substitutions for some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions claimed by the present invention.

[0173] Whenever possible, the various aspects and features described and shown in the specification can be applied individually, and these individual aspects can serve as the subject of a divisional application.

Claims

1. An optimized method for emergency evacuation in urban flood disasters, characterized in that, Includes the following steps: S1, collect basic data for urban areas; S2, Input the basic data into the flood inundation model built based on InfoWorks ICM, simulate the flood inundation scenario of the urban area under a rainstorm, and output the flood inundation simulation results; S3, set the attributes of the evacuation subjects in the multi-subject emergency evacuation model built based on NetLogo. The evacuation subjects include autonomous decision-making groups and guide-dependent groups. The autonomous decision-making groups are those who can plan evacuation routes independently and evacuate on their own. A guide is set in the autonomous decision-making groups. The guide-dependent groups are those who need to follow the guide to evacuate. S4, integrate the flood inundation model into the multi-subject emergency evacuation model, generate several evacuation subjects based on the regional population information and the attributes of the evacuation subjects in the basic data, set several different evacuation strategies, simulate the situation where the evacuation subjects evacuate to the refuge places in the basic data in the flood inundation scenario according to the evacuation strategies under different evacuation strategies, and obtain the evacuation results of the evacuation subjects corresponding to different evacuation strategies; S5. Evaluate the implementation effect of the several different evacuation strategies based on the evacuation results of the evacuation subjects, and determine the optimal strategy among the several different evacuation strategies.

2. The optimized method for emergency evacuation in urban flood disasters according to claim 1, characterized in that, The basic data includes: topographic data, water system distribution data, pipeline layout data, regional population data, refuge site data, and POI data of the urban area.

3. The optimized method for emergency evacuation in urban flood disasters according to claim 2, characterized in that, The flood inundation simulation results include the water depth and water area determined based on the topographic data, water system distribution data, and pipeline data, and the flood risk zones determined based on the water depth and water area.

4. The optimized method for emergency evacuation in urban flood disasters according to claim 1, characterized in that, The attributes of the evacuation subject set in step S3 include: The average response time, movement speed, and movement speed reduction rules due to flooding are set for the autonomous decision-making group and the guided group, respectively. The autonomous decision-making group is set to use the Dijkstra algorithm to plan the optimal passable path and move along the optimal passable path. The guided group is set to follow the guide according to the Flocking rule.

5. The optimized method for emergency evacuation in urban flood disasters according to claim 1, characterized in that, In step S3, setting the attributes of the evacuation subject includes setting the initial location of the evacuation subject to be distributed according to the population heat map.

6. The optimized emergency evacuation method for urban flood disasters according to claim 5, characterized in that, In step S3, setting the attributes of the evacuation subject further includes: setting the initial movement direction of the evacuation subject, wherein the initial movement direction of the evacuation subject is set to move from the initial position of the evacuation subject to a certain type of POI in the city area, and the POI is determined according to the function of the evacuation subject in the city area.

7. The optimized method for emergency evacuation in urban flood disasters according to claim 1, characterized in that, The evacuation strategy includes: The facilitator configuration is as follows: the facilitators are selected proportionally from the self-decision-making population. The early warning response is tiered, comprising: Level 4 response: broadcasting the early warning information, with evacuation voluntary and no guides assigned; Level 3 response: broadcasting and text simultaneously, updating flood information at frequency 'a', with evacuation voluntary and no guides assigned, and a recommended evacuation route pushed; Level 2 response: broadcasting and text / image simultaneously, updating flood information at frequency 'b' (where 'b' is greater than 'a'), with evacuation semi-forced, a recommended evacuation route pushed, and a guide assigned to guide those dependent on guidance; Level 1 response: broadcasting, text / image, and voice simultaneously, updating flood information at frequency 'c' (where 'c' is greater than 'b'), with evacuation forced, a recommended evacuation route pushed, and a guide assigned to guide those dependent on guidance.

8. The optimized method for emergency evacuation in urban flood disasters according to claim 7, characterized in that, The evacuation strategy also includes refuge site optimization, and the refuge site optimization method includes: Add more shelters in remote suburban areas; Reserve redundant capacity for refuge sites around high-risk areas in flood risk zones.

9. The optimized method for emergency evacuation in urban flood disasters according to claim 1, characterized in that, The evacuation results of the evacuation entity include evacuation success rate, average evacuation time per person, and retention rate. The higher the evacuation success rate, the shorter the average evacuation time per person, and the lower the retention rate of the evacuation entity, the better the implementation effect of the evacuation strategy.

10. The optimized method for emergency evacuation in urban flood disasters according to claim 1, characterized in that, The steps for adjusting the evacuation strategy of the evacuation entity include: The proportion of the guides was adjusted and different levels of early warning response were used to simulate the evacuation results. By comparing the evacuation results, the optimal evacuation strategy is determined.

Images