Dam break flood emergency evacuation path planning method and system based on improved NSGA-III

Abstract

The proposed method and system for dam-break flood emergency evacuation route planning based on an improved NSGA-III integrates dam-break flood simulation, evacuation point and refuge point selection, multi-objective optimization, and heuristic strategies. This method can achieve efficient evacuation route optimization for groups of people with different priorities under extreme dam-break scenarios, thereby improving overall evacuation efficiency and priority. By conducting two-dimensional unsteady flow dam-break simulation in HEC-RAS and combining the scientific selection and prioritization of evacuation points and refuge points using the ArcGIS Pro platform, an evacuation optimization model that adapts to the spatial distribution characteristics of flood risk is constructed.

Description

Technical Field

[0001] The present invention relates to the field of dam break flood disaster emergency management and path optimization, and specifically to a dam break flood emergency evacuation path planning method and system based on an improved NSGA-III, which is suitable for multi-objective optimization and auxiliary decision-making of personnel evacuation paths in a rapid and orderly manner under flood disaster scenarios. Background Art

[0002] With the intensification of climate change and the frequent occurrence of extreme rainfall events, the risk of flood disasters caused by reservoir dam failures is increasing, posing a serious threat to densely populated areas downstream. Dam failure floods are characterized by sudden onset, high peaks, and rapid propagation, easily leading to large-scale casualties and property losses. Therefore, after the issuance of a flood warning, how to quickly plan efficient and reasonable evacuation routes based on the severity of the flood has become a key issue in disaster emergency management.

[0003] Traditional evacuation route planning methods are mostly based on single-objective shortest path algorithms (such as the Dijkstra algorithm). While these algorithms can quickly solve the shortest path, they struggle to comprehensively consider multiple practical requirements, such as personnel risk priorities, shelter capacity constraints, and regional distribution balance. Consequently, they lack adaptability and practicality in extreme flood scenarios. NSGA-III, a novel multi-objective evolutionary algorithm, possesses excellent global search capabilities and multi-objective trade-off capabilities, making it suitable for complex optimization problems. However, directly applying NSGA-III to large-scale evacuation route planning still faces challenges such as slow convergence, poor solution feasibility, and high computational resource consumption.

[0004] Therefore, an improved NSGA-III evacuation path optimization method that integrates dam break flood simulation results, priority classification, path caching mechanism and heuristic strategy is urgently needed to improve the timeliness, scientificity and life protection capability of evacuation scheduling and adapt to the rapid emergency response needs in extreme scenarios of sudden dam break floods. Summary of the Invention

[0005] The technical problem to be solved by the present invention is to address the above shortcomings and provide a priority planning method for emergency evacuation paths for dam break floods based on the improved NSGA-III. The method solves the problem that the existing evacuation path planning method is difficult to take into account multi-objective optimization, evacuation priority classification and refuge point capacity constraints in the face of sudden and wide-ranging dam break flood scenarios. The method realizes differentiated evacuation path optimization for people with high, medium and low priorities, improves the overall evacuation efficiency and emergency response capabilities, and maximizes the protection of personnel life safety.

[0006] In order to solve the above technical problems, the present invention adopts the following technical solutions:

[0007] The priority planning method for dam-break flood emergency evacuation paths based on the improved NSGA-III includes the following steps:

[0008] Step S1, collecting relevant data and parameter information of the study area; determining the digital elevation model map, satellite image map, river distribution map, road network map and dam structure parameter information of the study area;

[0009] Step S2, setting parameters of the HEC-RAS model to carry out dam break simulation;

[0010] Based on the acquired digital elevation model, satellite imagery, river distribution map, road network map, and dam structural parameters, extreme flood scenarios and breach parameters were set in the HEC-RAS model. A two-dimensional unsteady flow dam breach simulation was conducted in the HEC-RAS module. The maximum flood inundation depth map was output as a result file, and the simulation results were imported into the ArcGIS Pro platform.

[0011] Step S3, selecting evacuation points and refuge points based on flood inundation and grading them;

[0012] Using ArcGIS Pro, evacuation and refuge points were selected based on a comprehensive analysis of satellite imagery, road network maps, flood inundation depth maps, high-resolution population density maps, and building block vector maps. Evacuation points were also prioritized into high, medium, and low levels based on flood inundation depth and impact.

[0013] Step S4: Build the topology, calculate all shortest paths and cache them;

[0014] Use Python NetworkX tools to build the road network topology, calculate the shortest path distance from each evacuation point to all refuge points based on the Dijkstra algorithm, and cache the calculation results;

[0015] Step S5, based on the spatial location of the evacuation points and their priority weights, perform weighted K-means clustering initialization to construct the initial population solution of the NSGA-III algorithm;

[0016] Step S6, NSGA-III multi-objective optimization iteratively updates the initial population solution of the NSGA-III algorithm;

[0017] In the framework of the NSGA-III multi-objective optimization algorithm, the goal is to minimize the multi-objective function, combined with the capacity limit of the refuge point, and iteratively update the initial population solution to obtain the initial Pareto solution set;

[0018] Step S7, fine-tuning the Pareto solution set using a heuristic strategy;

[0019] A heuristic fine-tuning strategy is introduced for the initially obtained Pareto solution set: based on the path length comparison results between evacuation points of different priorities, the allocation relationship between evacuation points and refuge points is optimized through local exchange. For evacuation points whose path length exceeds the preset improvement threshold, load exchange optimization is performed.

[0020] Step S8, obtaining the optimal evacuation path structure and visualizing it;

[0021] After optimization using the heuristic fine-tuning strategy, the optimal Pareto solution set is obtained, and the optimal refuge point and path allocation plan corresponding to each evacuation point are output; the final evacuation path results are imported into the ArcGIS Pro platform for visualization, and results such as the distribution map of evacuation paths of various priorities and the evacuation pressure heat map are generated for reference in emergency dispatch.

[0022] Furthermore, in step S2, performing the two-dimensional unsteady flow dam break simulation in the HEC-RAS module includes the following steps:

[0023] Use the Storage Area tool to map the reservoir storage area, the 2D Flow Area tool to define the flood simulation area boundary, and the SA / 2D Connection tool to set the dam axis and breach parameters. Then, set the simulation conditions, including the initial water level, dam breach level, simulation duration, and time step. After constructing the 2D computational grid, initiate the unsteady flow simulation.

[0024] Furthermore, the process of selecting evacuation points and refuge points in the ArcGIS Pro platform in step S3 includes the following steps:

[0025] Based on satellite images, building block vector maps and high-resolution population density maps, intersections and open places in high-population areas in the flood-inundated area are selected as evacuation points; shelters are set up in residential areas and large public facilities areas outside the flood-affected area.

[0026] Furthermore, when constructing the road network topology structure using the Python NetworkX tool in step S4, the following steps are included:

[0027] First, the pre-processed road data is converted into a graph structure, where each road segment corresponds to a weighted edge in the graph, where the edge weight is the actual road distance; and nodes represent road intersections or key geographical locations. Subsequently, a spatial mapping method is used to project the locations of evacuation areas and shelters onto the nearest nodes in the network graph, thereby establishing a correspondence between evacuation points and shelters in the road network.

[0028] Ensure that the network topology conforms to the actual road traffic logic and avoid unreasonable cross-connections; check the network connectivity and make connectivity corrections if there are non-connected segments; confirm that all evacuation points and refuge points are located on the nodes, and make adjustments if they are not.

[0029] Furthermore, the process of calculating the shortest path distance from each evacuation point to all refuge points using the Dijkstra algorithm in step S4 includes the following steps:

[0030] The Dijkstra algorithm is used to sequentially calculate the path lengths between all evacuation points and each refuge point. The tuples consisting of the evacuation point and the refuge point and their path lengths are stored in a cache structure such as a hash table to support subsequent fast query and reuse, avoiding repeated path calculations in each model iteration.

[0031] The tuple consisting of an evacuation point and a refuge point refers to an ordered pair consisting of an evacuation point and a refuge point, which is used to uniquely identify the correspondence between the two locations. In the path cache, each such tuple represents a specific evacuation path, and the corresponding shortest path length is stored as a value in the hash table.

[0032] Furthermore, the weighted K-means clustering initialization method in step S5 includes the following steps:

[0033] The weights are set according to the priorities of the evacuation points, and weighted clustering is performed to generate the initial groups. The nearest refuge point is matched to each cluster center as its default refuge target, and the initial evacuation point allocation is completed to construct the initial population solution of the NSGA-III algorithm.

[0034] Furthermore, the objective function used in the NSGA-III multi-objective optimization in step S6 includes:

[0035] The total length of all evacuation routes;

[0036] Total path length weighted by priority;

[0037] The total length of high-priority evacuation routes.

[0038] Furthermore, the objective function calculation process used in the NSGA-III multi-objective optimization is as follows:

[0039] ;

[0040] in, , decision variables, N is the total number of evacuation points 192; Overview represents the index of the refuge point assigned to the i-th evacuation point, and S is the total number of refuge points, 32; , evacuation point Go to the evacuation point The shortest path length; represents the flood priority weight ( , corresponding weights [10, 5, 1]); : High priority evacuation point collection; is the capacity constraint penalty term.

[0041] Furthermore, the heuristic fine-tuning strategy is divided into three stages:

[0042] In the first stage, after NSGA-III calculates the preliminary Pareto solution set in step S7, if it is found that the path length of a high-priority evacuation point exceeds that of a low- or medium-priority point, the evacuation point allocation and exchange operation is performed. When the path length of the medium-priority evacuation point is significantly longer than that of the low-priority evacuation point, the allocation and exchange operation is also performed to improve the high-priority evacuation efficiency;

[0043] Phase 2: After Phase 1 is completed, if it is found that the path length of a high-priority evacuation point exceeds that of a low-priority point, the refuge point exchange operation will continue until there are no more exchange pairs that can be significantly improved or the preset number of iterations is reached, and the optimized Pareto solution set is obtained;

[0044] In the third stage, based on the optimized Pareto solution set obtained in the second stage, extreme evacuation points whose path length exceeds the preset threshold are screened out. All refuge points are traversed to screen out candidate refuge points that do not exceed the capacity limit and can shorten the path to the greatest extent. The original assigned refuge points are replaced with the candidate refuge points, and the corresponding path and refuge point allocation information are updated.

[0045] The dam break flood emergency evacuation path priority planning system based on the improved NSGA-III includes:

[0046] Parameter determination unit, used to obtain digital elevation model map, satellite image map, river distribution map, road network map and dam structure parameter information of the study area;

[0047] Flood simulation unit, used to perform two-dimensional unsteady flow dam break simulation in HEC-RAS and export simulation results;

[0048] The spatial point selection unit is used to comprehensively analyze the flood impact range, building distribution, and population density in the ArcGIS Pro platform, scientifically select evacuation points and refuge points, and classify evacuation point priority categories;

[0049] Path pre-calculation unit, used to build the road network topology and cache the shortest paths between all evacuation points and refuge points through the Dijkstra algorithm;

[0050] Cluster initialization unit, used to generate initial groups based on the weighted K-means method and construct the initial population solution of NSGA-III;

[0051] The multi-objective optimization unit is used to perform path optimization under the NSGA-III multi-objective optimization framework and obtain the initial Pareto solution set;

[0052] A heuristic fine-tuning unit is used to optimize the high-priority paths and the extreme path load exchange of the initial solution set;

[0053] The result visualization unit is used to export the optimal Pareto solution set allocation results between evacuation points and refuge points to a file, and display the distribution map of evacuation paths of various priorities and the evacuation pressure heat map on the ArcGIS Pro platform.

[0054] The present invention adopts the above technical solution, which has the following technical effects compared with the prior art:

[0055] The proposed method and system for dam-break flood emergency evacuation route planning based on the improved NSGA-III integrates dam-break flood simulation, evacuation point and refuge point selection, multi-objective optimization, and heuristic strategies. This method can achieve efficient evacuation route optimization for groups of people with different priorities under extreme dam-break scenarios, thereby improving overall evacuation efficiency and priority. By conducting two-dimensional unsteady flow dam-break simulation in HEC-RAS and combining the scientific selection and prioritization of evacuation points and refuge points using the ArcGIS Pro platform, an evacuation optimization model that adapts to the spatial distribution characteristics of flood risk is constructed.

[0056] The method described in this paper introduces a path caching mechanism and weighted K-means initialization within the NSGA-III multi-objective optimization framework, significantly improving the convergence speed and feasibility of the solution. Furthermore, the initial solution set is locally fine-tuned through heuristic priority exchange and load regulation strategies to further optimize the evacuation efficiency of high-priority groups. Compared to traditional evacuation methods based on shortest paths or single-objective optimization, this method more rationally balances path shortestness, priority, and refuge capacity constraints, providing timely and operational decision-making support for emergency response to dam-break flood disasters. BRIEF DESCRIPTION OF THE DRAWINGS

[0057] To more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the following briefly describes the drawings required for the specific embodiments or the description of the prior art. Similar elements or parts are generally identified by similar reference numerals throughout the drawings. Elements or parts in the drawings are not necessarily drawn to scale.

[0058] Figure 1This is a flow chart of a dam-break flood emergency evacuation path planning method and system based on the improved NSGA-III according to an embodiment of the present invention;

[0059] Figure 2 The location map of the three priority evacuation points and refuge points of high, medium and low in the specific example;

[0060] Figure 3 To improve the 3D distribution diagram of evacuation paths for each priority refuge point of the NSGA-III model in a specific example;

[0061] Figure 4 This is the evacuation path distribution map of high-priority evacuation points in a specific example;

[0062] Figure 5 This is a structural diagram of a dam-break flood emergency evacuation path planning system based on the improved NSGA-III according to an embodiment of the present invention. DETAILED DESCRIPTION

[0063] The present invention will be further described below with reference to the following examples, but it should not be understood that the scope of the present invention is limited to the following examples. Without departing from the above technical ideas of the present invention, various substitutions and modifications can be made according to common technical knowledge and customary means in the art, and all should be included in the scope of protection of the present invention.

[0064] Examples, such as Figure 1 As shown in FIG, the priority planning method for emergency evacuation paths for dam break floods based on the improved NSGA-III includes the following steps:

[0065] Step S1, collecting relevant data and parameter information of the study area;

[0066] Determine the digital elevation model (DEM) map, satellite image map, river distribution map, road network map and dam structure parameter information of the study area.

[0067] The dam structural parameter information includes: dam elevation, dam length, dam width, reservoir capacity-water level curve, design flood level and water storage capacity.

[0068] Step S2, setting parameters of the HEC-RAS model to carry out dam break simulation;

[0069] Based on the acquired digital elevation model, satellite imagery, river distribution map, road network map, and dam structural parameters, extreme flood scenarios and breach parameters were set in the HEC-RAS model. A two-dimensional unsteady flow dam breach simulation was carried out in the HEC-RAS module, and the result file of the maximum flood inundation depth map was output. The simulation results were then imported into the ArcGIS Pro platform.

[0070] The breach parameters include: breach mode, breach location, breach shape, breach weir flow coefficient, and breach formation time.

[0071] The steps for conducting a two-dimensional unsteady flow dam-break simulation in the HEC-RAS module include: using the Storage Area tool to map the reservoir storage area, using the 2D Flow Area tool to define the flood simulation area boundary, and using the SA / 2DConnection tool to set the dam axis and breach parameters. The simulation conditions are then set, including the initial water level, dam-break water level, simulation duration, and time step. The unsteady flow simulation is then initiated after constructing the two-dimensional computational grid.

[0072] Step S3, selecting evacuation points and refuge points based on flood inundation and grading them;

[0073] In ArcGIS Pro, evacuation and refuge points were selected after comprehensive analysis of satellite imagery, road network maps, flood inundation depth maps, high-resolution population density maps, and building block vector maps. Evacuation points were then divided into three priority levels: high, medium, and low, based on the depth of flood inundation and the extent of impact.

[0074] Specifically, 80 high-priority evacuation points with a water depth of more than 10 meters, 56 medium-priority evacuation points with a water depth of 5-10 meters, and 56 low-priority evacuation points with a water depth of less than 5 meters were selected, totaling 192 evacuation points and 32 refuge points.

[0075] After HEC-RAS completes the dam-break flood evolution simulation, the maximum water depth (Max Depth) and other raster results can be exported through RAS Mapper and then imported into the ArcGIS Pro platform to produce and display the flood inundation depth map. The high-resolution population density map comes from the publicly released China Seventh National Population Census grid dataset (resolution 100m), which has been publicly shared on the Figshare platform. The building block vector map is based on Google Earth remote sensing imagery (spatial resolution 0.5m) acquired from 2020 to 2022 and is generated using a building boundary extraction algorithm. It can effectively represent the spatial distribution and morphological characteristics of buildings in the study area. This dataset is a publicly available resource and can be downloaded from the Zenodo database to support the spatial reference of evacuation and refuge point selection.

[0076] The process of selecting evacuation points and refuge points in the ArcGIS Pro platform includes the following steps:

[0077] Based on satellite images, building block vector maps and high-resolution population density maps, intersections and open places in high-population areas in the flood-inundated area are selected as evacuation points; shelters are set up in residential areas and large public facilities areas outside the flood-affected area.

[0078] To scientifically and rationally select evacuation and refuge points, the study combined road network maps with satellite imagery. Within high-density built-up areas within the flood impact zone, evacuation points were prioritized at road intersections and key nodes through visual interpretation of satellite imagery. At the same time, refuge points were established in densely populated areas outside the flood impact zone. To further verify the rationality of evacuation and refuge point selection, the study also introduced 100-meter-resolution population density maps and building block vector data. Although these data have certain limitations in spatial resolution and accuracy, they can still serve as auxiliary references. Based on the overlay analysis of this data, the locations of the initially established evacuation and refuge points were appropriately adjusted to enhance the scientificity and rationality of their spatial distribution.

[0079] Step S4: Build the topology, calculate all shortest paths and cache them;

[0080] The Python NetworkX tool is used to construct the road network topology structure, and the shortest path distance from each evacuation point to all refuge points is calculated based on the Dijkstra algorithm, and the calculation results are cached.

[0081] The use of Python's NetworkX tool to construct a road network topology structure includes:

[0082] First, preprocessed road data (including information such as road start and end points and distances) is converted into a graph structure, where each road segment corresponds to a weighted edge in the graph, with the edge weight representing the actual road distance; nodes represent road intersections or key geographical locations. Subsequently, a spatial mapping method is used to project the locations of evacuation areas and shelters onto the nearest nodes in the network graph, thereby establishing a correspondence between evacuation points and shelters in the road network.

[0083] Ensure that the network topology conforms to the actual road traffic logic and avoid unreasonable cross-connections; check the network connectivity and make connectivity corrections if there are non-connected segments; confirm that all evacuation points and refuge points are located on network nodes, and make adjustments if they are not.

[0084] The Dijkstra algorithm calculates the shortest path distance from each evacuation point to all refuge points, including:

[0085] The Dijkstra algorithm is used to calculate the path lengths between all evacuation points and each refuge point in sequence, and the (evacuation point, refuge point) tuple and its path length are stored in a cache structure such as a hash table to support subsequent fast query and reuse, avoiding repeated path calculation in each model iteration.

[0086] A (evacuation point, refuge point) tuple is an ordered pairing of an evacuation point and a refuge point, uniquely identifying the correspondence between the two locations. In the path cache, each such tuple represents a specific evacuation path, and the corresponding shortest path length is stored as a value in a hash table. This allows for quick querying of the path length from any evacuation point to a refuge point in subsequent model iterations, avoiding duplicate calculations.

[0087] In step S5, based on the spatial locations of the evacuation points and their priority weights, weighted K-means clustering initialization is performed to construct the initial population solution of the NSGA-III algorithm.

[0088] The weighted K-means clustering initialization method includes: setting weights according to the priorities of evacuation points, performing weighted clustering to generate initial groups; matching the nearest refuge point to each cluster center as its default refuge target, completing the initial evacuation point allocation, and constructing the initial population solution of the NSGA-III algorithm.

[0089] Step S6, NSGA-III multi-objective optimization iteratively updates the initial population solution of the NSGA-III algorithm;

[0090] In the framework of the NSGA-III multi-objective optimization algorithm, the goal is to minimize the multi-objective function, combined with the capacity limit of the refuge point, and iteratively update the initial population solution to obtain the initial Pareto solution set;

[0091] The objective functions used in the NSGA-III multi-objective optimization include:

[0092] ;

[0093] in, , decision variables, N is the total number of evacuation points 192; Overview represents the index of the refuge point assigned to the i-th evacuation point, and S is the total number of refuge points, 32; , evacuation point Go to the evacuation point The shortest path length; represents the flood priority weight ( , corresponding weights [10, 5, 1]); : High priority evacuation point collection; is the capacity constraint penalty term.

[0094] Step S7, fine-tuning the Pareto solution set using a heuristic strategy;

[0095] A heuristic fine-tuning strategy is introduced for the initially obtained Pareto solution set: based on the comparison results of path lengths between evacuation points with different priorities, the allocation relationship between evacuation points and refuge points is locally exchanged and optimized. For evacuation points whose path length exceeds the preset improvement threshold, load exchange optimization is performed.

[0096] The heuristic fine-tuning strategy is divided into three stages:

[0097] In the first stage, after NSGA-III calculates the preliminary Pareto solution set, if it finds that the path length of a high-priority evacuation point exceeds that of a low- or medium-priority point, it performs a shelter point allocation exchange operation. When the path length of a medium-priority evacuation point is significantly longer than that of a low-priority evacuation point, it also performs an allocation exchange to improve the efficiency of high-priority evacuation.

[0098] In the second stage, after the first stage is completed, if it is found that the path length of a high-priority evacuation point exceeds that of a low-priority point, the refuge point exchange operation will continue to be performed until there are no more exchange pairs that can be significantly improved or the preset number of iterations is reached, and the optimized Pareto solution set is obtained.

[0099] In the third stage, based on the optimized Pareto solution set obtained in the second stage, extreme evacuation points whose path length exceeds the preset threshold are screened out. All refuge points are traversed to screen out candidate refuge points that do not exceed the capacity limit and can shorten the path to the greatest extent. The original assigned refuge points are replaced with the candidate refuge points, and the corresponding path and refuge point allocation information are updated.

[0100] Based on the original NSGA-III algorithm, the Pareto solution set is optimized after three stages of local path adjustment and refuge point redistribution, resulting in a better solution set with higher evacuation efficiency and more reasonable priority path allocation.

[0101] Step S8, obtaining the optimal evacuation path structure and visualizing it;

[0102] After optimization using the heuristic fine-tuning strategy, the optimal Pareto solution set is obtained, and the optimal refuge point and path allocation plan corresponding to each evacuation point are output; the final evacuation path results are imported into the ArcGIS Pro platform for visualization, and results such as the distribution map of evacuation paths of various priorities and the evacuation pressure heat map are generated for reference in emergency dispatch.

[0103] To further understand the technical solution of the present invention, the method of the present invention is described below through a specific example. This example selects a dam and reservoir outlet area as the research area for implementation, and selects the locations of refuge points and evacuation points of high, medium and low priority levels, such as Figure 2 The specific implementation process is as follows:

[0104] 1) Obtain digital elevation model (DEM) maps, satellite images, river distribution maps, road network maps, and dam structural parameters of the study area;

[0105] 2) Set extreme flood scenarios and breach parameters in the HEC-RAS 2D hydrodynamic model, conduct 2D unsteady flow dam breach simulations, and output result files such as maximum flood depth maps;

[0106] 3) The simulation results were imported into the ArcGIS Pro platform. After comprehensive analysis, satellite imagery, road network maps, flood inundation depth maps, population density maps, and building block vector maps were combined to scientifically select evacuation and refuge points. Evacuation points were also prioritized into high, medium, and low levels based on the depth of flooding and the extent of its impact.

[0107] Figure 3 The improved NSGA-III model shows the distribution of evacuation paths for each priority level of the evacuation point. The points of different shapes in the figure represent evacuation points of different flood priorities: black squares represent high priority (≥10m), gray dots represent medium priority (5-10m), and gray triangles represent low priority (<5m). After heuristic fine-tuning, the improved NSGA-III model exhibits a significant hierarchical effect in the path lengths of evacuation points for the three flood priority levels. As can be seen from the figure, the path lengths of high-priority evacuation points are mostly concentrated in a relatively short range; medium priority points are second; and the path length distribution of low-priority points is relatively wide. At the same time, the extreme maximum value of the evacuation path has been reduced to approximately 10,000 meters, effectively avoiding the occurrence of excessively long paths and improving the rationality of evacuation path planning.

[0108] Figure 4 This demonstration demonstrates evacuation route planning based on evacuation points affected by flood depths ≥10m, specifically the distribution of evacuation routes for high-priority evacuation points. The evacuation routes connect high-priority evacuation points and refuges at different locations, visually demonstrating the planned layout of high-priority evacuation points. This demonstrates the evacuation route arrangements for high-priority areas in response to dam-break flood risks, helping to understand the evacuation directions of high-priority evacuation points and the distribution of accessible refuges.

[0109] Based on the same inventive concept, according to the above embodiments of the present application, the method, system, and device for planning emergency evacuation paths for dam break floods based on the improved NSGA-III are provided. Accordingly, another embodiment of the present invention also provides priority planning for emergency evacuation paths for dam break floods based on the improved NSGA-III, such as Figure 5 FIG. 1 is a schematic diagram of the system structure, which includes:

[0110] The parameter determination unit 501 is used to obtain information such as a digital elevation model map, satellite image map, river distribution map, road network map, and dam structural parameters of the study area;

[0111] A flood simulation unit 502 is used to perform two-dimensional unsteady flow dam break simulation in HEC-RAS and export simulation results;

[0112] The spatial point selection unit 503 is used to comprehensively analyze the flood impact range, building distribution and population density in the ArcGIS Pro platform, scientifically select evacuation points and refuge points, and classify the evacuation points into priority categories;

[0113] The path pre-calculation unit 504 is used to construct a road network topology and cache the shortest paths between all evacuation points and refuge points using the Dijkstra algorithm;

[0114] A cluster initialization unit 505 is used to generate initial groups based on a weighted K-means method and construct an initial population solution for NSGA-III;

[0115] The multi-objective optimization unit 506 is used to perform path optimization under the NSGA-III multi-objective optimization framework to obtain an initial Pareto solution set;

[0116] a heuristic fine-tuning unit 507 for performing high-priority path tuning and extreme path load exchange optimization on the initial solution set;

[0117] The result visualization unit 508 is used to export the optimal Pareto solution set allocation result between evacuation points and refuge points to a file, and to display the evacuation path distribution map of each priority level and the evacuation pressure heat map in the ArcGIS Pro platform.

[0118] The description of the present invention has been presented for purposes of illustration and description and is not intended to be exhaustive or to limit the invention to the form disclosed. Many modifications and variations will be apparent to those skilled in the art. The embodiments were chosen and described in order to better illustrate the principles of the invention and its practical application and to enable those skilled in the art to understand the invention and design various embodiments with various modifications as are suited for specific applications.

Claims

1. A priority planning method for emergency evacuation routes for dam-break floods based on an improved NSGA-III, characterized by: The following steps are involved: Step S1, collecting relevant data and parameter information of the study area; determining the digital elevation model map, satellite image map, river distribution map, road network map and dam structure parameter information of the study area; Step S2, setting parameters of the HEC-RAS model to carry out dam break simulation, and importing the simulation results into the ArcGIS Pro platform; Step S3, selecting evacuation points and refuge points based on flood inundation and grading them; In ArcGIS Pro, select evacuation points and refuge points; classify evacuation points into three priority levels: high, medium, and low, based on the depth of flooding and the degree of impact; Step S4: Build the topology, calculate all shortest paths and cache them; Use Python NetworkX tools to build the road network topology, calculate the shortest path distance from each evacuation point to all refuge points based on the Dijkstra algorithm, and cache the calculation results; Step S5, based on the spatial location of the evacuation points and their priority weights, perform weighted K-means clustering initialization to construct the initial population solution of the NSGA-III algorithm; Step S6, NSGA-III multi-objective optimization iteratively updates the initial population solution of the NSGA-III algorithm to obtain the initial Pareto solution set; Step S7, introducing a heuristic fine-tuning strategy for the initial Pareto solution set to obtain the optimal Pareto solution set; Based on the comparison results of the path lengths between evacuation points of different priorities, the allocation relationship between evacuation points and refuge points is optimized locally. For evacuation points whose path length exceeds the preset improvement threshold, load exchange optimization is performed; Step S8, obtaining the optimal evacuation path structure and visualizing it; Based on the optimal Pareto solution set, the optimal refuge point and its path allocation plan corresponding to each evacuation point are output, and the results are imported into the ArcGIS Pro platform for visualization, generating a distribution map of evacuation paths of different priorities and an evacuation pressure heat map; The heuristic fine-tuning strategy is divided into three stages: In the first stage, after NSGA-III calculates the preliminary Pareto solution set in step S7, if it is found that the path length of a high-priority evacuation point exceeds that of a low- or medium-priority point, the evacuation point allocation and exchange operation is performed. When the path length of the medium-priority evacuation point is significantly longer than that of the low-priority evacuation point, the allocation and exchange operation is also performed to improve the high-priority evacuation efficiency; Phase 2: After Phase 1 is completed, if it is found that the path length of a high-priority evacuation point exceeds that of a low-priority point, the refuge point exchange operation will continue until there are no more exchange pairs that can be significantly improved or the preset number of iterations is reached, and the optimized Pareto solution set is obtained; In the third stage, based on the optimized Pareto solution set obtained in the second stage, extreme evacuation points whose path length exceeds the preset threshold are screened out. All refuge points are traversed to screen out candidate refuge points that do not exceed the capacity limit and can shorten the path to the greatest extent. The original assigned refuge points are replaced with the candidate refuge points, and the corresponding path and refuge point allocation information are updated.

2. The dam-break flood emergency evacuation path priority planning method based on the improved NSGA-III according to claim 1, characterized in that: In step S2, the HEC-RAS model sets parameters to carry out dam break simulation. The specific steps are as follows: Based on the acquired digital elevation model, satellite imagery, river distribution map, road network map, and dam structural parameters, extreme flood scenarios and breach parameters were set in the HEC-RAS model. A two-dimensional unsteady flow dam breach simulation was conducted in the HEC-RAS module. The maximum flood inundation depth map was output as a result file, and the simulation results were imported into the ArcGIS Pro platform. Carrying out a 2D unsteady flow dam-break simulation in the HEC-RAS module involves the following steps: Use the Storage Area tool to map the reservoir storage area, the 2D Flow Area tool to define the flood simulation area boundary, and the SA / 2D Connection tool to set the dam axis and breach parameters. Then, set the simulation conditions, including the initial water level, dam breach level, simulation duration, and time step. After constructing the 2D computational grid, initiate the unsteady flow simulation.

3. The dam-break flood emergency evacuation path priority planning method based on the improved NSGA-III according to claim 1, characterized in that: In step S3, the process of selecting evacuation points and refuge points in the ArcGIS Pro platform includes the following steps: Based on satellite images, building block vector maps and high-resolution population density maps, intersections and open places in high-population areas in the flood-inundated area are selected as evacuation points; shelters are set up in residential areas and large public facilities areas outside the flood-affected area.

4. The dam-break flood emergency evacuation path priority planning method based on the improved NSGA-III according to claim 1, characterized in that: In step S4, when constructing the road network topology structure using the Python NetworkX tool, the following steps are included: First, the preprocessed road data is converted into a graph structure, where each road segment corresponds to a weighted edge in the graph, with the weight of the edge representing the actual road distance. Nodes represent road intersections or key geographical locations. Then, using a spatial mapping method, the locations of evacuation areas and shelters are projected to the nearest nodes in the network graph, thereby establishing a correspondence between evacuation points and shelters in the road network. Ensure that the network topology conforms to the actual road traffic logic and avoid unreasonable cross-connections; check the network connectivity and make connectivity corrections if there are non-connected segments; confirm that all evacuation points and refuge points are located on the nodes, and make adjustments if they are not.

5. The dam-break flood emergency evacuation path priority planning method based on the improved NSGA-III according to claim 1, characterized in that: In step S4, the Dijkstra algorithm calculates the shortest path distance from each evacuation point to all refuge points, including the following steps: The Dijkstra algorithm is used to sequentially calculate the path lengths between all evacuation points and each refuge point. The tuples consisting of the evacuation point and the refuge point and their path lengths are stored in a cache structure such as a hash table to support subsequent fast query and reuse, avoiding repeated path calculations in each model iteration. The tuple consisting of an evacuation point and a refuge point refers to an ordered pair consisting of an evacuation point and a refuge point, which is used to uniquely identify the correspondence between the two locations. In the path cache, each such tuple represents a specific evacuation path, and the corresponding shortest path length is stored as a value in the hash table.

6. The dam-break flood emergency evacuation path priority planning method based on the improved NSGA-III according to claim 1, characterized in that: In step S5, the weighted K-means clustering initialization method includes the following steps: The weights are set according to the priorities of the evacuation points, and weighted clustering is performed to generate the initial groups. The nearest refuge point is matched to each cluster center as its default refuge target, and the initial evacuation point allocation is completed to construct the initial population solution of the NSGA-III algorithm.

7. The dam-break flood emergency evacuation path priority planning method based on the improved NSGA-III according to claim 1, characterized in that: In step S6, the objective function used in the NSGA-III multi-objective optimization includes: The total length of all evacuation routes; Total path length weighted by priority; The total length of high-priority evacuation routes.

8. The dam-break flood emergency evacuation path priority planning method based on the improved NSGA-III according to claim 7, characterized in that: The objective function calculation process used in the NSGA-III multi-objective optimization is as follows: ； in, , decision variables, N is the total number of evacuation points; Overview represents the index of the refuge point assigned to the i-th evacuation point, and S is the total number of refuge points; , evacuation point Go to the evacuation point The shortest path length; represents the flood priority weight; , corresponding weights [10, 5, 1]; : High priority evacuation point collection; is the capacity constraint penalty term.

9. A dam-break flood emergency evacuation path priority planning system based on the improved NSGA-III is characterized by: The planning system is applied to the planning method according to any one of claims 1 to 8, comprising: Parameter determination unit, used to obtain digital elevation model map, satellite image map, river distribution map, road network map and dam structure parameter information of the study area; Flood simulation unit, used to perform two-dimensional unsteady flow dam break simulation in HEC-RAS and export simulation results; The spatial point selection unit is used to comprehensively analyze the flood impact range, building distribution, and population density in the ArcGIS Pro platform, scientifically select evacuation points and refuge points, and classify evacuation point priority categories; Path pre-calculation unit, used to build the road network topology and cache the shortest paths between all evacuation points and refuge points through the Dijkstra algorithm; Cluster initialization unit, used to generate initial groups based on the weighted K-means method and construct the initial population solution of NSGA-III; The multi-objective optimization unit is used to perform path optimization under the NSGA-III multi-objective optimization framework and obtain the initial Pareto solution set; A heuristic fine-tuning unit is used to optimize the high-priority paths and the extreme path load exchange of the initial solution set; The result visualization unit is used to export the optimal Pareto solution set allocation results between evacuation points and refuge points to a file, and display the distribution map of evacuation paths of various priorities and the evacuation pressure heat map on the ArcGIS Pro platform.

Images