Drainage network reconstruction and waterlogging simulation method in area without network data and related device

Abstract

This invention discloses a method and related apparatus for reconstructing drainage networks and simulating urban flooding in areas lacking pipe network data. Relating to the field of urban hydrology, the method involves acquiring discrete coordinates of surface drainage facilities and road vector data of the target area. The discrete coordinates are projected onto the road centerlines in the road vector data. Based on the projection position and lateral attributes, the discrete coordinates are grouped. The grouped discrete coordinates are then merged into nodes according to their distribution spacing to generate connecting pipe segments. At road intersections, the nodes are stitched together based on proximity relationships to reconstruct an approximate drainage network for hydraulic simulation. A one-dimensional hydrodynamic model is constructed based on the approximate drainage network. Water exchange data between the one-dimensional hydrodynamic model and the surface is extracted. This water exchange data drives a two-dimensional surface hydrodynamic model, outputting the surface inundation results for the target area. This application can achieve refined simulation of urban flooding and accurately characterize local water accumulation points even without measured pipe network data.

Description

Technical Field

[0001] This invention relates to the field of urban hydrology technology, and more specifically, to a method and related apparatus for reconstructing drainage networks and simulating urban flooding in areas without pipe network data. Background Technology

[0002] Numerical simulation of urban flooding is a key technical means to assess urban flood control and drainage capabilities, optimize drainage system planning, and formulate emergency disaster reduction strategies. In conventional refined urban flooding simulation work, building a high-precision hydraulic model usually relies on complete and accurate basic data of the underground drainage pipe network. This data covers the geometric dimensions, burial depth and elevation, and complex topological connections of the pipe network. Currently, the commonly used technical approach in the industry is to directly build a one-dimensional pipe network hydrodynamic model based on measured pipe network drawings or geographic information system databases, and then perform coupled calculations with a two-dimensional surface runoff model to simulate the surface water accumulation process under rainfall events. This modeling method based on real pipe network data constitutes the mainstream paradigm for current urban flooding risk assessment and is widely used in scenarios such as evaluating the effectiveness of sponge city construction and identifying flood-prone areas.

[0003] However, in many urban areas, parts of historic old towns, and rapidly expanding new urban areas, there are often realities such as a lack of underground drainage network data, incomplete records, data confidentiality restrictions, or excessively high costs for on-site surveying. In these areas lacking real network data, existing equivalent processing methods typically assume spatial homogenization of drainage capacity or use simplified empirical formulas for generalization, ignoring the heterogeneous characteristics of the actual spatial distribution of drainage facilities. This approach makes it difficult for simulation results to accurately reflect the local runoff collection and discharge processes determined by the layout of drainage facilities, and to precisely characterize the dynamics of water accumulation, diffusion, and receding at specific locations. Therefore, conducting high-precision and reliable urban flooding simulations in areas without network data has been a long-standing technical challenge in this field. Summary of the Invention

[0004] In view of this, the present invention discloses a method and related apparatus for reconstructing drainage pipe networks and simulating urban flooding in areas without pipe network data, in order to solve the technical problem that has long been a challenge in this field to carry out high-precision and reliable urban flooding simulation in areas without pipe network data.

[0005] A method for reconstructing drainage pipe networks and simulating urban flooding in areas lacking pipe network data includes:

[0006] Obtain the discrete coordinates of the surface drainage facilities in the target area, as well as the road vector data of the target area;

[0007] The discrete coordinates are projected onto the road centerline in the road vector data, and the discrete coordinates are grouped based on the projection position and lateral attributes.

[0008] The grouped discrete coordinates are merged into nodes according to the distribution interval and connected pipe segments are generated. At road intersections, they are stitched together based on proximity to reconstruct an approximate drainage network for hydraulic simulation.

[0009] A one-dimensional hydrodynamic model is constructed based on the approximate drainage network, and the water exchange data between the one-dimensional hydrodynamic model and the surface is extracted.

[0010] Based on the water exchange data, a two-dimensional surface hydrodynamic model is driven to output the surface inundation results of the target area.

[0011] Optionally, the step of projecting the discrete coordinates onto the road centerline in the road vector data, and grouping the discrete coordinates based on the projection position and lateral attributes, includes:

[0012] Project each discrete coordinate onto the centerline of the road to which it belongs, and calculate the mileage position of each projection point along the centerline of the road.

[0013] The lateral attribute is determined based on the deviation of each discrete coordinate relative to the road centerline;

[0014] Discrete coordinates belonging to the same road and having the same lateral attributes are sorted according to their mileage positions.

[0015] Optionally, the step of merging the grouped discrete coordinates into nodes according to the distribution spacing and generating connecting pipe segments, and stitching them together at road intersections based on proximity relationships to reconstruct an approximate drainage network for hydraulic simulation includes:

[0016] Among discrete coordinates belonging to the same road and having the same lateral attributes, those with a projection distance between adjacent discrete coordinates that is less than the merging threshold are grouped into one cluster group.

[0017] Based on the centroid position of the discrete coordinates within the cluster group, a manhole node is generated by offsetting a preset distance towards the road centerline along a direction perpendicular to the road centerline.

[0018] Adjacent manhole nodes are connected sequentially along the centerline of the road to form a main pipeline;

[0019] Connect each discrete coordinate with the corresponding inspection well node to form a branch pipe;

[0020] At road intersections, a disjoint-set data structure algorithm is used to stitch together spatially adjacent but unconnected branches of the pipe network, so that the approximate drainage pipe network forms a connected network.

[0021] Optionally, before grouping discrete coordinates belonging to the same road and having the same lateral attributes into a single cluster group, if the projection distance between adjacent discrete coordinates is less than the merging threshold, the method further includes:

[0022] Spatial clustering and fusion are performed on the discrete coordinates of the surface drainage facilities that belong to the category of rainwater inlets, and multiple discrete coordinates located around the same actual facility location are merged into one rainwater inlet node.

[0023] Statistical analysis of the spacing distribution characteristics between adjacent rainwater inlet nodes after fusion;

[0024] The merging threshold is determined based on the spacing distribution characteristics.

[0025] Optionally, generating manhole nodes by offsetting a preset distance towards the road centerline along a direction perpendicular to the road centerline based on the centroid position of the discrete coordinates within the cluster group includes:

[0026] Obtain the centroid position of the discrete coordinates within the cluster group;

[0027] The corresponding offset distance is determined based on the road classification of the road in the road vector data;

[0028] The centroid position is shifted by the offset distance along a direction perpendicular to the road centerline towards the road centerline to obtain the planar coordinates of the manhole node.

[0029] Optionally, the step of constructing a one-dimensional hydrodynamic model based on the approximate drainage network and extracting water exchange data between the one-dimensional hydrodynamic model and the surface includes:

[0030] The designed rainfall is applied to the one-dimensional hydrodynamic model to simulate the generation, confluence, and unsteady flow transport process within the approximate drainage network.

[0031] When the water level at a node in the one-dimensional hydrodynamic model exceeds the surface elevation of that node, an overflow is determined to have occurred and the overflow flow rate of that node is extracted.

[0032] The overflow flow rate is used as the water exchange data between the one-dimensional hydrodynamic model and the surface.

[0033] Optionally, the step of driving a two-dimensional surface hydrodynamic model based on the water exchange data and outputting the surface inundation results for the target area includes:

[0034] The overflow flow rate is used as the sole water source of the two-dimensional surface hydrodynamic model, and mapped to two-dimensional surface grid cells according to geographic coordinates, serving as the point source inflow boundary condition of the two-dimensional surface hydrodynamic model.

[0035] The two-dimensional surface hydrodynamic model is driven by the point source inflow boundary conditions to solve the two-dimensional shallow water equation to simulate the surface runoff process and obtain the solution results of the surface runoff.

[0036] Based on the calculation results of the surface runoff, the surface water depth and inundation range of the target area are output as the surface inundation result.

[0037] Optionally, obtaining the discrete coordinates of the surface drainage facilities in the target area includes:

[0038] Acquire real-world images with geolocation and attitude metadata;

[0039] A target detection model is used to identify surface drainage facilities from the real-scene image and output detection boxes;

[0040] Based on the preset installation height and field of view of the camera that collects the real-scene images, the pixel position of the detection frame is converted into the depression angle and horizontal deflection angle of the surface drainage facility relative to the camera.

[0041] Based on the depression angle and the horizontal deflection angle, combined with the camera's geographical location and heading angle, the horizontal distance and absolute azimuth angle of the surface drainage facility relative to the camera are calculated.

[0042] By performing a spatial vector inverse mapping based on the horizontal distance and the absolute azimuth, the discrete coordinates of each of the surface drainage facilities are calculated.

[0043] Optionally, before constructing the one-dimensional hydrodynamic model based on the approximate drainage network, the method further includes:

[0044] The effective surface grid cells in the target area, after removing invalid and water areas, are assigned to the nearest node according to the principle of proximity, so that each node is allocated the corresponding surface catchment area.

[0045] Obtain the first digital surface model of the target area;

[0046] The first digital surface model is resampled to a set resolution to obtain the second digital surface model;

[0047] The second digital surface model is used to identify depressions, and the identified virtual depressions are filled to obtain the third digital surface model;

[0048] Based on the third digital surface model, the water area is processed in a preset manner to obtain topographic data used to drive the two-dimensional surface hydrodynamic model.

[0049] A device for reconstructing drainage pipe networks and simulating urban flooding in areas without pipe network data includes:

[0050] The acquisition unit is used to acquire the discrete coordinates of the surface drainage facilities in the target area, as well as the road vector data of the target area;

[0051] A grouping unit is used to project the discrete coordinates onto the road centerline in the road vector data, and to group the discrete coordinates based on the projection position and lateral attributes.

[0052] The drainage network reconstruction unit is used to merge the grouped discrete coordinates into nodes according to the distribution spacing and generate connecting pipe segments. At road intersections, it stitches the segments together based on proximity relationships to reconstruct an approximate drainage network for hydraulic simulation.

[0053] A one-dimensional hydrodynamic simulation unit is used to construct a one-dimensional hydrodynamic model based on the approximate drainage network and extract water exchange data between the one-dimensional hydrodynamic model and the surface.

[0054] The two-dimensional surface hydrodynamic simulation unit is used to drive the two-dimensional surface hydrodynamic model based on the water exchange data and output the surface inundation results of the target area.

[0055] An electronic device, comprising: a memory and a processor;

[0056] The memory is used to store at least one instruction;

[0057] The processor is used to execute at least one instruction to implement the above-described method for reconstructing drainage networks and simulating urban flooding in areas without network data.

[0058] As can be seen from the above technical solution, the present invention discloses a method and related device for reconstructing drainage pipe networks and simulating urban flooding in areas without pipe network data. The method involves obtaining discrete coordinates of surface drainage facilities in the target area, as well as road vector data of the target area. The discrete coordinates are projected onto the road centerline in the road vector data. Based on the projection position and lateral attributes, the discrete coordinates are grouped. The grouped discrete coordinates are merged into nodes according to their distribution spacing to generate connecting pipe segments. At road intersections, the nodes are stitched together based on proximity relationships to reconstruct an approximate drainage pipe network for hydraulic simulation. A one-dimensional hydrodynamic model is constructed based on the approximate drainage pipe network. Water exchange data between the one-dimensional hydrodynamic model and the surface is extracted. Based on the water exchange data, a two-dimensional surface hydrodynamic model is driven, and the surface inundation results for the target area are output. This invention maps the discrete coordinates of surface drainage facilities in the target area to road vector data and reconstructs an approximate drainage network with a connected topology. It transforms the distribution characteristics of surface facilities into a network structure solvable by a hydraulic model, thus establishing a numerical simulation foundation for the drainage system without relying on any real network data. This overcomes the bottleneck of traditional methods that cannot conduct refined simulations due to the lack of network data. Furthermore, by using a one-dimensional hydrodynamic model to simulate runoff generation, confluence, and unsteady flow in the reconstructed network, it accurately reflects the control effect of the actual spatial distribution of drainage facilities on runoff convergence and discharge processes, overcoming the limitations of existing equivalent processing methods. The assumption of spatial homogeneity makes it impossible to characterize the non-uniformity of drainage systems. By solving the two-dimensional evolution process of pipe network overflow on the surface through a two-dimensional surface hydrodynamic model, the diffusion path and range of water accumulation under topographic constraints can be accurately simulated, making up for the shortcomings of simplified methods in dynamically tracking the water receding process. The above links constitute a complete extrapolation chain from surface facility data to urban flooding results, enabling the simulation results to truly reflect the formation, diffusion and receding dynamics of local water accumulation points determined by the spatial layout of drainage facilities. This solves the technical problem that existing technologies cannot accurately identify high-risk points of urban flooding due to ignoring the spatial non-uniformity of drainage facilities. Attached Figure Description

[0059] 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 only embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on the published drawings without creative effort.

[0060] Figure 1 This is a flowchart of a method for reconstructing drainage pipe networks and simulating urban flooding in areas without pipe network data, as disclosed in an embodiment of the present invention.

[0061] Figure 2This is a schematic diagram of a leak-free pipeline network reconstruction disclosed in this invention, wherein (a) is the street scene detection landing point, (b) is the rainwater inlet node obtained by clustering and fusion, (c) the rainwater inlet node is projected onto the road centerline and its side is determined, (d) the same side is merged to obtain the inspection well node and connect the main pipe and the branch pipe, and (e) adjacent branches are stitched together at the intersection to form a connected network.

[0062] Figure 3 This is a schematic diagram illustrating the principle of unidirectional coupling between the one-dimensional drainage network model and the two-dimensional surface water dynamic model disclosed in this invention.

[0063] Figure 4 This is a schematic diagram of the principle of monocular vision geometric inverse calculation disclosed in this invention, wherein (a) is a side view showing the geometric relationship of obtaining the horizontal object distance from the target depression angle; (a) is a schematic diagram of the image plane showing the geometric relationship of obtaining the horizontal deflection angle from the horizontal pixel offset of the detection box, and then combining it with the heading angle to obtain the absolute azimuth angle;

[0064] Figure 5 This is a schematic diagram of the structure of a drainage network reconstruction and flood simulation device for areas without pipe network data, as disclosed in an embodiment of the present invention.

[0065] Figure 6 This is a schematic diagram of the structure of an electronic device disclosed in an embodiment of the present invention;

[0066] The attached figures are labeled as follows: 1—Camera; 2—Camera installation height; 3—Drainage facility target; 4—Camera line of sight; 5—Dip angle; 6—Horizontal object distance; 7—Image frame; 8—Image optical axis; 9—Target detection box; 10—Road centerline; 11—Street scene detection landing point; 12—Drain inlet node; 13—Inspection well node; 14—Main pipe; 15—Branch pipe; 16—Seam pipe; 17—One-dimensional drainage network; 18—Node overflow; 19—Overflow point source; 20—Two-dimensional surface grid; 21—Point source inflow grid; 22—Surface runoff. Detailed Implementation

[0067] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0068] See Figure 1 The present invention discloses a flowchart of a method for reconstructing drainage pipe networks and simulating urban flooding in areas without pipe network data. The method includes:

[0069] Step S101: Obtain the discrete coordinates of the surface drainage facilities in the target area, as well as the road vector data of the target area.

[0070] The target area is the urban geographic area to be analyzed for urban flooding simulation. Surface drainage facilities refer to structures installed on the ground surface to collect and discharge rainwater, such as curb drains and manhole covers. Discrete coordinates are the planar location coordinates of surface drainage facilities in a unified geographic coordinate system, such as WGS84 or a local projected coordinate system. Road vector data describes the geometry and topology of roads within the target area, including information such as the geometric path of the road centerline, road class, and road width.

[0071] Data acquisition methods include: using vehicles equipped with street view acquisition devices to travel along roads and capture real-world images with geographic location and attitude metadata; then using monocular visual geometric inverse calculation technology to calculate the geographic coordinates of surface drainage facilities; or directly calling the urban geographic information database interface to read the labeled point data. Road vector data can be obtained from open-source platforms or extracted from basic geographic information data released by urban surveying and mapping departments. The aforementioned discrete coordinates and road vector data constitute the spatial basis for subsequent drainage network reconstruction.

[0072] Step S102: Project the discrete coordinates onto the road centerline in the road vector data, and group the discrete coordinates based on the projection position and lateral attributes.

[0073] The road centerline is a core geometric element characterizing the road's orientation. Projecting discrete coordinates onto the road centerline involves calculating the perpendicular from each discrete coordinate point to the nearest road centerline; this perpendicular is the projection point, and its mileage along the centerline reflects the longitudinal relative position of the facility. Lateral attributes refer to the spatial orientation of surface drainage facilities relative to the road centerline, which can be either to the left or right.

[0074] In practice, the process begins by matching each discrete coordinate to the nearest road centerline and calculating the projection point and its mileage. Next, based on the relative vector direction between the discrete coordinates and the projection point, and the digitized direction of the road centerline, it is determined whether the facility is located on the left or right side of the road. Finally, discrete coordinates on the same road segment and side are sorted according to the mileage of their projection points. This grouping transforms scattered facility locations into an ordered sequence along the road, providing structured input for topology construction.

[0075] Step S103: The grouped discrete coordinates are merged into nodes according to the distribution spacing and connected pipe segments are generated. At road intersections, they are stitched together based on proximity relationships to reconstruct an approximate drainage network for hydraulic simulation.

[0076] Nodes represent confluence points or connection points in an approximate drainage network; in this embodiment, they are manhole nodes. Connecting pipe segments are virtual pipes connecting nodes, simulating rainwater flow channels. The approximate drainage network is a virtual model reconstructed based on the distribution of surface facilities and road morphology, with a topological structure and hydraulic characteristics approximating the real underground pipe network.

[0077] The merging process is as follows: For discrete coordinate sequences on the same side of the same road, the projection distance between adjacent coordinates is calculated; if the distance is less than a preset merging threshold, the corresponding discrete coordinates are merged into clusters. Subsequently, based on the centroid position of the cluster group, a manhole node is generated by offsetting a preset distance along a direction perpendicular to the road centerline towards the centerline to simulate the actual position of the manhole.

[0078] The preset merging threshold is dynamically determined based on road grade or historical statistics, for example, 20 to 50 meters. The preset distance is determined according to actual needs, for example, 0.5 meters for main roads and 0.3 meters for secondary roads.

[0079] After generating nodes, adjacent manhole nodes are sequentially connected along the road centerline to form the main pipe that undertakes the main transportation function; at the same time, the original discrete coordinates are connected to the corresponding manhole nodes to form branch pipes that collect surface rainwater. For spatially adjacent but unconnected branches at road intersections, a disjoint-set data structure algorithm is used to stitch together pipe network branches with a spatial distance less than a preset stitching threshold, eliminating isolated branches and making the approximate drainage pipe network a fully connected network with topological connectivity that can be directly used for hydraulic simulation.

[0080] Step S104: Construct a one-dimensional hydrodynamic model based on the approximate drainage network, and extract the water exchange data between the one-dimensional hydrodynamic model and the surface.

[0081] The one-dimensional hydrodynamic model simulates unsteady flow within a pipe network based on Saint-Venant's equations. When constructing the one-dimensional hydrodynamic model, an approximate drainage pipe network is used as the physical framework, and hydraulic property parameters are assigned to nodes and pipe segments. Node parameters include ground elevation and bottom elevation; pipe segment parameters include pipe diameter, Manning roughness coefficient, and slope.

[0082] The ground elevation is obtained by interpolation from the digital surface model. The bottom elevation is the ground elevation minus the preset burial depth, such as 1.5 meters to 2.5 meters.

[0083] Pipe diameter is set according to road grade or catchment area, for example, main pipes are 600 mm to 1200 mm, and branch pipes are 300 mm to 500 mm.

[0084] The Manning roughness factor is determined based on actual needs; for example, 0.013 for concrete pipes.

[0085] After the one-dimensional hydrodynamic model is constructed, design rainfall is applied to simulate the rainwater generation, inflow, and transport processes. Node water levels are monitored in real time; when a node's water level exceeds the surface elevation, an overflow is identified, and the overflow flow rate at that node is extracted as water exchange data. Overflow flow rate characterizes the amount of water that cannot be discharged from the pipe network in a timely manner and overflows to the surface; it is a key coupling variable connecting the one-dimensional pipe network model and the two-dimensional surface model.

[0086] Step S105: Drive a two-dimensional surface hydrodynamic model based on the water exchange data, and output the surface inundation results of the target area.

[0087] A two-dimensional surface hydrodynamic model is used to simulate surface water diffusion based on two-dimensional shallow water equations, solving for the velocity and depth fields. Water exchange data is used as input boundary conditions when driving the model.

[0088] In practice, the overflow flow is used as the sole water source. The geographic coordinates of the overflow nodes are mapped to the corresponding grid cells of a two-dimensional surface grid model, serving as the point source inflow boundary condition. The two-dimensional surface grid model divides the target area into regular or irregular grids with topographic elevation information. The two-dimensional shallow water equations are solved using the finite volume method or the finite difference method to simulate the dynamic process of surface water flowing from the overflow point to surrounding low-lying areas. Finally, the water depth values ​​of each grid cell and the set of areas where the water depth exceeds a preset threshold (e.g., 0.15 meters) are extracted, representing the surface water depth and inundation range, respectively, thus forming the surface inundation result.

[0089] In summary, this invention maps the discrete coordinates of surface drainage facilities in the target area to road vector data and reconstructs an approximate drainage network with a connected topology. This transforms the distribution characteristics of surface facilities into a network structure solvable by a hydraulic model, thus establishing a numerical simulation foundation for the drainage system without relying on any real network data. This overcomes the bottleneck of traditional methods that cannot conduct refined simulations due to the lack of network data. Furthermore, by using a one-dimensional hydrodynamic model to simulate runoff generation, confluence, and unsteady flow in the reconstructed network, it accurately reflects the control effect of the actual spatial distribution of drainage facilities on runoff convergence and discharge processes, overcoming the limitations of existing equivalent methods. The previous method, due to its assumption of spatial homogeneity, could not characterize the non-uniform characteristics of the drainage system. By solving the two-dimensional evolution process of the overflow from the pipe network on the surface using a two-dimensional surface hydrodynamic model, it is possible to accurately simulate the diffusion path and range of water accumulation under topographic constraints, thus making up for the shortcomings of the simplified method in dynamically tracking the water receding process. The above links constitute a complete extrapolation chain from surface facility data to the result of urban flooding, enabling the simulation results to truly reflect the formation, diffusion and receding dynamics of local water accumulation points determined by the spatial layout of drainage facilities. This solves the technical problem that the existing technology cannot accurately identify high-risk points of urban flooding because it ignores the spatial non-uniformity of drainage facilities.

[0090] In one embodiment, step S102 may specifically include:

[0091] Project each discrete coordinate onto the centerline of the road to which it belongs, and calculate the mileage position of each projection point along the centerline of the road.

[0092] The lateral attribute is determined based on the deviation of each discrete coordinate relative to the road centerline;

[0093] Discrete coordinates belonging to the same road and having the same lateral attributes are sorted according to their mileage positions.

[0094] This embodiment establishes a logical relationship between facility locations and road spatial topology by mapping discrete facility points to a road network. For the discrete coordinates of surface drainage facilities within the target area obtained in the preceding steps, the road entity to which each coordinate point is attached is first determined. By calculating the Euclidean distance between the discrete coordinates and the centerlines of surrounding roads, each discrete coordinate is assigned to the nearest road centerline, thereby establishing the dependency relationship between the facility and the road.

[0095] After establishing the hierarchical relationships, the perpendicular projection algorithm is used to calculate the projection points of each discrete coordinate on the centerline of its respective road. This projection process transforms any coordinate point on the two-dimensional plane into a definite position on the one-dimensional linear space of the road, thereby eliminating lateral deviation interference caused by facility positioning errors. Based on this, the mileage position of each projection point along the road centerline is further calculated. The mileage position is usually represented by a linearly accumulated distance value along the road direction, with the starting point of the road as the zero point; its accuracy depends on the resolution of the road vector data. By transforming discrete coordinates into mileage positions, the facility points, originally randomly distributed on the two-dimensional plane, are abstracted into an ordered one-dimensional sequence along the road direction, providing a basic spatial index for subsequent node grouping by spacing.

[0096] This embodiment constructs a unit direction vector originating from the projection point and extending along the road centerline, as well as a vector originating from the projection point and pointing towards the original discrete coordinates. By calculating the cross product sign or relative azimuth angle of these two vectors, it can be determined whether the discrete coordinates are located to the left or right of the road centerline. The lateral attribute is defined with reference to the road travel direction; that is, when viewed along the direction of increasing road mileage, the facility point is located to the left or right of the travel direction. By determining the lateral attribute, facility points belonging to the same road but distributed on both sides are strictly distinguished, avoiding the incorrect merging and connection of facilities on both sides of the road in subsequent pipeline reconstruction. This ensures that the reconstructed drainage pipeline network maintains consistency with the actual road layout in terms of topology, i.e., each road has independent drainage branches on both sides.

[0097] The sequencing operation establishes the spatial order of facility points laid along the road, which directly corresponds to the physical connection order of the drainage network. Through sequencing, scattered facility points are organized into a sequence with clear flow direction characteristics, enabling subsequent steps to identify adjacent nodes, calculate node spacing, and then perform node merging and pipe segment connection operations based on this sequence. If merging is performed directly without sequencing, it may lead to disordered or looped pipe network connections, violating the one-dimensional hydrodynamic model's pre-defined requirements for the directionality of water flow. Therefore, sequencing based on mileage location is a necessary prerequisite to ensure that the reconstructed approximate drainage network has correct topological connectivity and hydraulic feasibility.

[0098] In summary, this invention projects discrete coordinates containing spatial errors onto regular road centerlines and rigorously groups them using lateral attributes, ultimately forming an ordered spatial sequence through mileage sorting. This road-anchored spatial organization effectively eliminates random noise and positional biases in the original data. Even without pipeline network data, it constructs a logically rigorous and engineering-intuitive pipeline network framework, providing a reliable data foundation for subsequently generating high-quality approximate drainage networks and conducting urban flooding simulations.

[0099] In one embodiment, the process of merging grouped discrete coordinates into nodes according to their distribution spacing and generating connecting pipe segments, then stitching them together at road intersections based on proximity relationships to reconstruct an approximate drainage network for hydraulic simulation, specifically includes:

[0100] Among discrete coordinates belonging to the same road and having the same lateral attributes, those with a projection distance between adjacent discrete coordinates that is less than the merging threshold are grouped into one cluster group.

[0101] Based on the centroid position of the discrete coordinates within the cluster group, a manhole node is generated by offsetting a preset distance towards the road centerline along a direction perpendicular to the road centerline.

[0102] Adjacent manhole nodes are connected sequentially along the centerline of the road to form a main pipeline;

[0103] Connect each discrete coordinate with the corresponding inspection well node to form a branch pipe;

[0104] At road intersections, a disjoint-set data structure algorithm is used to stitch together spatially adjacent but unconnected branches of the pipe network, so that the approximate drainage pipe network forms a connected network.

[0105] Furthermore, after using a disjoint-set data structure algorithm to stitch together spatially adjacent but disconnected pipe network branches, a fallback mechanism can be applied to ensure pipe network connectivity. Specifically, each connected branch in the approximate drainage pipe network is identified, and a drainage outlet is determined for each branch. If no outlet node located near the boundary of the target area exists in the current connected branch, the node with the lowest surface elevation within that branch is taken as the drainage outlet, ensuring that each connected branch has a drainage path. This fallback mechanism avoids the reconstructing of subnetworks with internally trapped water and no drainage path, guaranteeing the integrity of the boundary conditions in the one-dimensional hydrodynamic model.

[0106] It should be noted that all geometric and attribute parameters required for reconstructing the approximate drainage network, including the offset distance of manhole nodes, the diameter of main and branch pipes, and the roughness and burial depth of each pipe segment, are derived solely from the following three types of data sources: first, discrete coordinates and their spatial distribution characteristics obtained from real-scene image recognition and geometric inverse calculation; second, road grades included in publicly available road vector data; and third, preset engineering default values. These parameters do not involve any actual drainage network data for the target area, thus meeting the prerequisite of no leakage.

[0107] In this embodiment, when processing the discrete coordinates sorted by road affiliation and lateral attributes in the preceding steps, the influence of street view image positioning errors and the randomness of facility distribution means that adjacent points may have excessively close spacing. Directly constructing the pipeline network based on the original points would result in a number of nodes far exceeding the actual engineering requirements and increase the redundancy of subsequent hydraulic calculations. Therefore, spatial clustering of the discrete coordinates is performed based on a set merging threshold. This merging threshold is used to determine whether two adjacent discrete coordinates should be considered facilities serving the same catchment area. When the projection distance between adjacent discrete coordinates along the road centerline is less than the merging threshold, these two coordinates are merged into the same cluster group. This processing method effectively compresses the data size while preserving the facility distribution pattern, laying the foundation for subsequently generating pipeline network nodes that conform to the actual scale of the project.

[0108] Generating manhole nodes is a core step in constructing the pipeline network topology. Essentially, it transforms clustered surface facility locations into underground pipeline nodes near the road centerline. First, the centroid of all discrete coordinates within a cluster is calculated; this centroid represents the geometric center of the cluster in planar space. Then, starting from this centroid, a spatial offset is made along a geometric direction perpendicular to the road centerline. The offset is controlled by a preset distance, typically determined based on the road grade; for example, 8 meters for main roads and 4 meters for branch roads, to simulate the actual burial locations of manholes on roads of different grades. This centroid-based, vertically offset generation method not only eliminates the random positioning error of the original discrete coordinates but also precisely places the manhole nodes at locations representing the drainage starting point of the catchment area, ensuring the engineering rationality of the pipeline network's starting point.

[0109] Main pipes form the backbone of the drainage network, undertaking the primary function of water transport. After the manhole nodes are generated, the connection relationships between the nodes are established according to the projected mileage sequence of each manhole node along the road centerline. For manhole nodes on the same road, adjacent manhole nodes are connected end-to-end by line segments according to their arrangement along the road direction. These connected line segments are defined as main pipes. The routing of the main pipes strictly follows the direction of the road centerline, thus reflecting the physical law of the gravity flow drainage system being arranged along the road slope. By connecting them sequentially along the road centerline, a main drainage channel network covering the entire road network is constructed, providing a path for surface runoff to reach downstream outlets.

[0110] Branch pipes are responsible for collecting and transporting scattered surface runoff to the main pipe system. After the main pipe network is constructed, the connection between the original discrete coordinates and their respective manhole nodes is further established. Specifically, for each discrete coordinate group merged into a certain cluster, it is used as the starting point, and the corresponding manhole node of that cluster is used as the ending point to establish a connecting line segment, which is defined as the branch pipe. Branch pipes are usually short and extend perpendicular to the main pipe, representing the physical channel through which rainwater inlets or surface drainage facilities guide surface runoff into the underground main pipe network. By connecting each discrete coordinate with its corresponding manhole node, the capillary structure of the pipe network is improved, realizing the construction of a complete water flow path from specific surface facility points to the main pipe network channel.

[0111] See Figure 2 The diagram shows a non-leaking pipeline network reconstruction, where (a) is the street view detection landing point, (b) is the rainwater inlet node obtained by clustering and fusion, (c) the rainwater inlet node is projected onto the road centerline and its side is determined, (d) the same side is merged to obtain the inspection well node and connect the main pipe and the branch pipe, and (e) adjacent branches are stitched together at the intersection to form a connected network.

[0112] like Figure 2 As shown in (a), the street view detection points 11 appear as a dense set of points clustered around the same actual facility location.

[0113] like Figure 2 As shown in (b), after spatial clustering and fusion, multiple detection points belonging to the same actual facility location are merged into a single rainwater inlet node 12, eliminating data redundancy caused by multi-view detection.

[0114] like Figure 2 As shown in (c), after generating the rainwater inlet node 12, each rainwater inlet node 12 is assigned to the nearest road centerline, the projection position of each rainwater inlet node 12 on the road centerline is calculated, and its location on the left or right side of the road is determined according to its deviation relative to the road centerline.

[0115] like Figure 2 As shown in (d), rainwater inlet nodes 12 belonging to the same road and located on the same side are sorted and merged according to their projection positions along the road direction to generate manhole nodes 13; then, adjacent manhole nodes 13 are connected sequentially along the road centerline to form a main pipe 14, and each rainwater inlet node 12 is connected to its corresponding manhole node 13 to form a branch pipe 15.

[0116] like Figure 2 As shown in (e), at the road intersection, the disjoint-set data structure algorithm is used to stitch together the spatially adjacent but not yet connected branches of the pipeline network on each road. By generating stitching pipe 16, the adjacent branches are merged and connected, so that the approximate drainage pipeline network forms a fully connected network.

[0117] After completing the connection of trunk and branch pipes within a single road, since the pipe networks of each road are often independently constructed, there may be multiple pipe network branches that are spatially adjacent but lack topological connections at road intersections. To address this issue, this embodiment introduces a disjoint-set data structure (DFS) algorithm for topological stitching at road intersections. The DFS algorithm is an efficient data structure suitable for handling merging and querying non-overlapping nodes. First, the endpoints of all pipe network branches are scanned to identify neighboring node pairs whose spatial distance is less than a preset stitching threshold. Then, the DFS algorithm is used to merge node pairs belonging to different connected branches but satisfying the spatial proximity condition into the same set, and a connecting pipe segment is generated between them. Through this stitching process, the originally independent pipe network branches are connected, realizing the transformation of the entire regional pipe network from an isolated structure to a single connected network, ensuring smooth water flow throughout the entire network, and meeting the basic requirements of hydraulic simulation for pipe network connectivity.

[0118] In one embodiment, before grouping discrete coordinates belonging to the same road and having the same lateral attributes into a single cluster group where the projection distance between adjacent discrete coordinates is less than a merging threshold, the method further includes:

[0119] Spatial clustering and fusion are performed on the discrete coordinates of the surface drainage facilities that belong to the category of rainwater inlets, and multiple discrete coordinates located around the same actual facility location are merged into one rainwater inlet node.

[0120] Statistical analysis of the spacing distribution characteristics between adjacent rainwater inlet nodes after fusion;

[0121] The merging threshold is determined based on the spacing distribution characteristics.

[0122] During street view image acquisition and processing, factors such as shooting angle, depth of field differences, and image overlap often lead to repeated detection of storm drain facilities at the same physical location. This results in the discrete coordinates obtained through geometric inverse calculation not appearing as a single precise point in geographic space, but rather as a dense set of points clustered around the actual facility location. Directly using these redundant and fluctuating discrete coordinates for pipeline network reconstruction can easily misclassify the same physical facility as multiple independent nodes, thereby compromising the accuracy of the pipeline network topology. Therefore, spatial clustering and fusion operations are necessary to merge multiple discrete coordinates that are geographically adjacent and belong to the same actual facility into a single representative storm drain node.

[0123] In practical implementation, a density-based noise-based spatial clustering algorithm can be used to perform the fusion process. This algorithm searches for density-connected core points in two-dimensional space by setting neighborhood radius parameters and minimum neighborhood point number parameters, and clusters discrete coordinate points that can be connected together through density reachability into the same cluster.

[0124] For each formed cluster, the centroid position of all discrete coordinates within the cluster is calculated, and this centroid position is determined as the final geographic coordinates of the fused storm drain node. Through the above processing, the originally discrete and repetitive detection points are transformed into a set of storm drain nodes with unique geographic identifiers and relatively accurate locations, effectively eliminating positioning jitter and data redundancy caused by multi-view observations, and providing a high-quality data foundation for subsequent pipeline topology construction.

[0125] After obtaining a deduplicated set of storm drain nodes through spatial clustering and fusion, it is necessary to quantitatively analyze the spatial distribution density of these nodes to obtain statistical information reflecting the layout pattern of drainage facilities in the target area. Specifically, firstly, the sequence of storm drain nodes belonging to the road is traversed along the road centerline; secondly, the distance between adjacent storm drain nodes is calculated based on the projected mileage position of each node on the road; subsequently, the distance values ​​of all adjacent storm drain nodes in the area are statistically analyzed to obtain distance statistics reflecting the storm drain layout density of the area, including the average, median, or quantile values ​​of the distance. These statistics constitute the distance distribution characteristics, used to characterize the density and typical interval of storm drain layout in the current area, and provide a quantitative basis for subsequently determining the merging threshold.

[0126] In summary, this invention addresses the data heterogeneity problem in reconstructing a connected pipe network from discrete, noisy detection coordinates through a combination of techniques: spatial clustering of multi-view detection results and back-calculation of merging thresholds using spacing statistical features. Clustering first avoids the generation of pseudo-nodes, then statistics provide a quantitative benchmark at the regional scale, and the final determined merging threshold allows subsequent merging operations to smooth positioning errors while reflecting the distribution patterns of facilities. This ensures that the reconstructed approximate drainage pipe network possesses physical authenticity and engineering usability in terms of topology and spatial morphology.

[0127] In one embodiment, the process of generating a manhole node by offsetting a preset distance towards the road centerline along a direction perpendicular to the road centerline based on the centroid position of discrete coordinates within the cluster group specifically includes:

[0128] Obtain the centroid position of the discrete coordinates within the cluster group;

[0129] The corresponding offset distance is determined based on the road classification of the road in the road vector data;

[0130] The centroid position is shifted by the offset distance along a direction perpendicular to the road centerline towards the road centerline to obtain the planar coordinates of the manhole node.

[0131] Spatial statistical calculations are performed on the discrete coordinates of multiple rainwater inlets merged within the same cluster to obtain the geometric center point representing the centroid of the distribution of this group of discrete coordinates in a two-dimensional plane. Specifically, the horizontal and vertical coordinate values ​​of all discrete coordinates included in the cluster are summed, and then divided by the total number of discrete coordinates in the group to calculate the centroid coordinate point that represents the overall distribution of the facilities in the group. The physical significance of this centroid location is to eliminate random jitter caused by positioning errors or viewing angle deviations of a single detection point, providing a more robust starting point as a spatial reference for the subsequent generation of inspection well nodes.

[0132] Road classification is an attribute indicator that categorizes urban roads by function and construction standards; for example, roads can be classified as arterial roads, secondary arterial roads, and local roads. Given the significant differences in road width, drainage pipe diameter, and facility layout specifications among different road classifications, manholes typically require specific setback distances based on road width. A mapping table or configuration strategy is constructed that includes different road classifications and their corresponding offset distances. This configuration strategy is set based on the general length requirements of storm drain connection pipes in municipal engineering.

[0133] In this embodiment, the translation process is achieved through vector operations. Specifically, coordinate increments determined by the offset distance and direction vector are superimposed on the original horizontal and vertical coordinates of the centroid to generate new coordinate values. These newly generated coordinate values ​​are then used as the final planar coordinates of the manhole node. By offsetting the centroid towards the road centerline, the reconstructed manhole node position is placed within a reasonable range of the actual road drainage facilities. This corrects the lateral deviation of the roadside discrete coordinates and ensures the geometric accuracy of the pipeline topology, thus providing a spatial node foundation consistent with engineering practice for subsequent one-dimensional hydrodynamic simulations.

[0134] In one embodiment, step S104 may specifically include:

[0135] The designed rainfall is applied to the one-dimensional hydrodynamic model to simulate the generation, confluence, and unsteady flow transport process within the approximate drainage network.

[0136] When the water level at a node in the one-dimensional hydrodynamic model exceeds the surface elevation of that node, an overflow is determined to have occurred and the overflow flow rate of that node is extracted.

[0137] The overflow flow rate is used as the water exchange data between the one-dimensional hydrodynamic model and the surface.

[0138] In this embodiment, rainfall is designed as the system's input excitation, determining the intensity and temporal distribution characteristics of the water input throughout the simulation process. This designed rainfall is not randomly selected data, but rather a standard rainfall process line with a clear return period, derived from historical meteorological statistical patterns and rainstorm intensity formulas for a specific region.

[0139] By applying designed rainfall to a one-dimensional hydrodynamic model, the generation, runoff, and unsteady transport processes within a pipe network are simulated. The generation and runoff process refers to the physical process by which surface runoff, after deducting initial losses such as infiltration and depression, is distributed through catchment areas and ultimately flows into the nodes of an approximate drainage pipe network. The unsteady transport process refers to the one-dimensional fluid motion of water entering the pipe network under the influence of gravity and pressure, propelling it along the network over time. This process is typically described by solving the one-dimensional Saint-Venant equations, which reflect the dramatic fluctuations and evolution of flow rate and water level within the pipes over time.

[0140] In a one-dimensional hydrodynamic model based on an approximate drainage network, the surface elevation of a node is considered a physical threshold for determining whether the pipeline system has sufficient drainage capacity. This value is typically obtained by sampling using a digital surface model and represents the actual surface elevation at the node's location. The node water level is a dynamically changing simulated variable, characterizing the real-time filling height of water within the node's manhole.

[0141] The logic for determining overflow is based on the physical connectivity between the hydraulic state inside the one-dimensional model and the external surface space. When the simulated node water level is numerically strictly greater than the surface elevation of that node, it means that the water head inside the pipe is higher than the ground, the hydraulic connectivity is broken, and the water flow will no longer be constrained by the well wall and will overflow outward. At this time, it is determined that an overflow has occurred.

[0142] Overflow flow rate forms a bridge connecting the underground pipe network simulation and the surface inundation simulation. Its temporal resolution is consistent with the calculation step size of the one-dimensional simulation, and its spatial location strictly corresponds to the geographical coordinates of each node in the approximate drainage pipe network. By designating the overflow flow rate as the unique water exchange data, a clear, unidirectional data transmission channel is established, providing a standardized data input basis for subsequent steps to inject it as a point source into the two-dimensional surface hydrodynamic model.

[0143] In one embodiment, the process of driving a two-dimensional surface hydrodynamic model based on the water exchange data and outputting the surface inundation results for the target area includes:

[0144] The overflow flow rate is used as the sole water source of the two-dimensional surface hydrodynamic model, and mapped to two-dimensional surface grid cells according to geographic coordinates, serving as the point source inflow boundary condition of the two-dimensional surface hydrodynamic model.

[0145] The two-dimensional surface hydrodynamic model is driven by the point source inflow boundary conditions to solve the two-dimensional shallow water equation to simulate the surface runoff process and obtain the solution results of the surface runoff.

[0146] Based on the calculation results of the surface runoff, the surface water depth and inundation range of the target area are output as the surface inundation result.

[0147] In this embodiment, the overflow flow of each overflow node in the one-dimensional model is mapped to the corresponding grid cell of the two-dimensional surface hydrodynamic model according to its geographical coordinates, and used as the point source inflow boundary condition of the grid cell. The two-dimensional surface hydrodynamic model itself does not directly receive rainfall input, and its only water source is the overflow flow of each overflow node in the one-dimensional model.

[0148] The two-dimensional shallow water equations are a set of partial differential equations describing free surface flow, derived from the conservation of mass and momentum. Their physical essence lies in neglecting vertical acceleration and assuming a hydrostatic pressure distribution along the water depth direction, thus simplifying the three-dimensional flow into a two-dimensional planar flow. The model discretizes the study area into regular grid cells, using the generated point source inflow boundary conditions as source terms substituted into the governing equations for solution. This solution process simulates the flow state of surface water under gravity-driven, topographically obstructed, and structurally blocked conditions, tracing the spatiotemporal evolution of water diffusion, convergence, and receding on the surface. The two-dimensional shallow water equations are discretized using numerical discretization schemes such as the finite volume method or the finite difference method, and iterative calculations are used to obtain the hydraulic elements of each grid cell at different times, ultimately yielding the solution results for surface runoff.

[0149] The surface inundation calculation results encompass the hydraulic state information of all grid cells within the simulation area throughout the entire simulation period. The extracted surface water depth refers to the thickness of the water layer formed on a two-dimensional surface grid cell at a specific moment, reflecting the severity of waterlogging at that location. The inundation range refers to the geographical area comprised of grid cells where the water depth exceeds a specific threshold (e.g., curb height), used to quantify the spatial impact range of urban flooding events. Post-processing the calculation results generates output data or visualization maps containing the above information, thus obtaining complete surface inundation results. This result directly characterizes the distribution of urban flooding risk caused by drainage network overflow in the target area under design rainfall conditions, providing a quantitative basis for urban flood control decision-making and risk assessment.

[0150] See Figure 3 The present invention discloses a schematic diagram of the principle of unidirectional coupling between a one-dimensional drainage network model and a two-dimensional surface model, illustrating the water transfer relationship where overflow flow is injected into the two-dimensional grid as a point source and activates surface runoff. Figure 3As shown, the one-dimensional drainage network 17 consists of nodes and pipe segments. During the simulation, when the water level at a node in the one-dimensional drainage network 17 exceeds the surface elevation, the node overflows, forming node overflow 18. The overflow flow rate corresponding to node overflow 18 is the water exchange data transferred from the one-dimensional model to the two-dimensional model. The overflow flow rate of node overflow 18 is not directly distributed in the form of area source, but is injected into the point source inflow grid 21 in the two-dimensional surface grid 20 after being mapped by geographic coordinates. The point source inflow grid 21 is a two-dimensional surface grid cell containing the geographic coordinates of the overflow node. The overflow flow rate serves as the point source inflow boundary condition for this grid cell, driving the two-dimensional surface hydrodynamic model to start solving the surface water flow movement. After receiving the point source inflow boundary condition, the two-dimensional surface hydrodynamic model activates the simulation of surface flooding 22, solves the diffusion path and range of surface water flow under terrain constraints, and finally outputs the surface inundation results. The entire water transfer process is a one-way transfer from the one-dimensional drainage network 17 to the two-dimensional surface grid 20. The surface water level of the two-dimensional model is not fed back to the one-dimensional model, which reflects the one-way coupled water transfer relationship.

[0151] In summary, this invention establishes a unidirectional coupling structure based on overflow flow between the one-dimensional hydrodynamic model and the two-dimensional surface hydrodynamic model. This unidirectional coupling structure uses pipeline overflow as the core driving force for surface runoff, avoiding the numerical instability problems that may arise from complex feedback calculations in two-dimensional coupling. It also accurately reproduces the localized characteristics of urban flooding determined by the spatial distribution of drainage facilities. With precise point source input, the two-dimensional surface hydrodynamic model can realistically reflect the constraints of gravity and topography on surface water flow, thereby transforming abstract pipeline overflow data into intuitive and quantifiable surface water depth and inundation range, achieving an effective mapping from the hydraulic state of underground pipelines to surface disaster scenarios.

[0152] In one embodiment, step S101 may specifically include:

[0153] Acquire real-world images with geolocation and attitude metadata;

[0154] A target detection model is used to identify surface drainage facilities from the real-scene image and output detection boxes;

[0155] Based on the preset installation height and field of view of the camera that collects the real-scene images, the pixel position of the detection frame is converted into the depression angle and horizontal deflection angle of the surface drainage facility relative to the camera.

[0156] Based on the depression angle and the horizontal deflection angle, combined with the camera's geographical location and heading angle, the horizontal distance and absolute azimuth angle of the surface drainage facility relative to the camera are calculated.

[0157] By performing a spatial vector inverse mapping based on the horizontal distance and the absolute azimuth, the discrete coordinates of each of the surface drainage facilities are calculated.

[0158] In practical applications, after obtaining the discrete coordinates of various surface drainage facilities, considering that a smaller depression angle results in a larger horizontal distance and that the back-calculation results are more sensitive to depression angle errors, a maximum effective distance threshold is preset to control the positioning error of distant targets. In one implementation, this maximum effective distance threshold is determined based on the positioning accuracy requirements of monocular visual geometric back-calculation, for example, it can be set to 50 meters. Detection points whose calculated horizontal distance exceeds this maximum effective distance threshold are removed and do not participate in subsequent pipeline reconstruction, thereby ensuring that the facility locations involved in the modeling have reliable positioning accuracy.

[0159] Real-world imagery typically originates from street view data collected along roads within the target area, acquired via vehicle-mounted panoramic cameras or mobile acquisition devices. Geographic location refers to the latitude and longitude coordinates of the camera lens center at the acquisition point, while attitude metadata encompasses parameters such as the camera's heading, pitch, and roll angles at the time of capture. This metadata is usually stored in the image's exchange image file format header or in an associated external metadata file. By acquiring real-world imagery with complete spatiotemporal labels, a mapping benchmark is established between the image's two-dimensional pixel space and the real geographic three-dimensional space. This ensures that subsequent geometric inverse calculations have an absolute spatial reference, thus providing the necessary geometric constraints for inversely deducing the true location of ground targets from the image plane.

[0160] The object detection model extracts features and performs classification regression on the input real-world imagery, outputting bounding boxes for surface drainage facilities on the image. These bounding boxes are typically represented by pixel coordinates, such as the coordinates of the top-left and bottom-right corners of the smallest bounding rectangle containing the target, or the coordinates of the rectangle's center point, width, and height. Furthermore, the object detection model outputs corresponding category labels and confidence scores to identify the type of facility identified and the reliability of the result. This automated recognition process transforms unstructured image data into structured target location information, enabling efficient screening and localization of drainage facilities in massive amounts of street view imagery, avoiding the inefficiency and high cost of manual image-by-image interpretation.

[0161] In the vertical direction, using the camera's preset installation height and field of view, the vertical offset of the geometric center relative to the image optical axis (usually the vertical line of the image center) is calculated; combined with the camera's vertical field of view, this vertical offset is converted into the depression angle of the surface drainage facility relative to the camera lens, which reflects the angle by which the target point deviates from the horizontal plane in the vertical profile.

[0162] In the horizontal direction, the horizontal offset of the detection box's geometric center relative to the image's central axis is calculated. Combined with the camera's horizontal field of view, this is converted into the horizontal deflection angle of the surface drainage facility relative to the camera's line of sight. This horizontal deflection angle represents the angle by which the target point deviates from the camera's aiming direction in the horizontal plane. Through this conversion, the absolute pixel position, which depends on imaging resolution, is transformed into a relative angular quantity that does not change with resolution. This eliminates the differences caused by different camera sensor sizes and provides standardized geometric input for subsequent distance calculation.

[0163] Geometric constraints are used to transform angular parameters into physical quantities describing the polar coordinate position of the target. For the horizontal object distance calculation, a right-angled trigonometric relationship is constructed based on the depression angle and the camera's preset installation height. The calculation formula is as follows:

[0164] ;

[0165] In the formula, The horizontal object distance between the surface drainage facility and the camera is the projected distance from the target point to the center of the camera on the ground plane. The preset installation height for the camera that captures real-world images is a known constant, such as 2.50 meters. Let be the angle of depression of the target relative to the camera. This formula uses trigonometric relationships between sides and angles to inversely calculate the horizontal distance from the height and the angle of depression.

[0166] In polar coordinates, object distance alone is insufficient to determine the target's planar position; the target's absolute orientation relative to true north must also be determined. First, by calculating the deviation ratio of the horizontal pixel coordinates of the detection box center relative to the image's central axis, and combining this with the camera's horizontal field of view, the horizontal deflection angle of the target relative to the camera's viewing axis is calculated.

[0167]

[0168] In the formula, It is a horizontal deflection angle; The horizontal pixel coordinates of the center of the detection box; The total width of the image in horizontal pixels. The horizontal center of the image; The horizontal field of view of the camera is 90 degrees in one implementation.

[0169] The calculation of the absolute azimuth angle is based on the superposition of the heading angle and the horizontal deflection angle, and the calculation formula is as follows:

[0170] ;

[0171] In the formula, This is the absolute azimuth of the surface drainage facility relative to true north. The absolute heading angle at the time of camera acquisition is derived from the attitude metadata of the real-world image; It is a horizontal deflection angle; To take the modulus of the angle results, ensure that the results fall within the effective azimuth range of 0 to 360 degrees.

[0172] In obtaining horizontal object distance with absolute azimuth Then, starting from the camera's absolute geographic coordinates, the absolute latitude and longitude coordinates of the target are calculated by performing a spatial vector inverse mapping using the spherical distance transformation relationship:

[0173] Lat_obj=Lat_cam+(d·cosθ / R)·(180 / π);

[0174] Lon_obj=Lon_cam+(d·sinθ / (R·cos(Lat_cam·π / 180)))·(180 / π);

[0175] In the formula, Lat_obj is the absolute latitude of the target, Lon_obj is the absolute longitude of the target; Lat_cam is the absolute latitude of the camera, Lon_cam is the absolute longitude of the camera; R is the average radius of the Earth.

[0176] Through the above calculations, the distance and orientation of the target relative to the camera in the polar coordinate system were obtained, and a complete spatial vector from the camera position to the target position was constructed.

[0177] See Figure 4 This is a schematic diagram of the principle of monocular vision geometric inverse calculation provided by the present invention, wherein (a) is a side view showing the geometric relationship of obtaining the horizontal object distance from the target depression angle; (a) is a schematic diagram of the image plane showing the geometric relationship of obtaining the horizontal deflection angle from the horizontal pixel offset of the detection box, and then combining it with the heading angle to obtain the absolute azimuth angle.

[0178] like Figure 4 As shown in (a), camera 1 is mounted at a fixed height 2 above the data acquisition vehicle, and its lens optical axis is usually parallel to the horizontal plane. When the camera photographs the drainage facility target 3 on the side of the road, a depression angle 5 is formed between the camera's line of sight 4 and the horizontal plane where camera 1 is located. Since the camera mounting height 2 is a known constant, the horizontal distance 6 between the drainage facility target 3 and camera 1 can be calculated using the tangent function of the depression angle 5, based on the side-angle relationship of a right triangle. Figure 4 (b) in the diagram is a schematic diagram of the image plane, showing the geometric relationship between calculating the horizontal deflection angle from the horizontal pixel offset of the detection box and then combining it with the heading angle to calculate the absolute azimuth angle. Figure 4As shown in (b) of the image, in image 7, the image optical axis 8 corresponds to the camera's viewing axis direction, and the geometric center of the target detection box 9 has a lateral offset relative to the image optical axis 8. Combining this with the camera's horizontal field of view, this lateral offset is converted into a horizontal deflection angle. This will further reduce the horizontal deflection angle. With camera heading angle By performing angle superposition calculations, the absolute azimuth of the target relative to true north N is obtained. .

[0179] In summary, this embodiment achieves high-precision geographic coordinates of surface drainage facilities in a target area at low cost by relying solely on monocular street view imagery and without relying on on-site mapping using a Global Navigation Satellite System (GNSS). These discrete coordinates constitute the basic spatial data for subsequent drainage network reconstruction in areas lacking pipe network data.

[0180] In one embodiment, before constructing a one-dimensional hydrodynamic model based on the approximate drainage network, the following may also be included:

[0181] The effective surface grid cells in the target area, after removing invalid and water areas, are assigned to the nearest node according to the principle of proximity, so that each node is allocated the corresponding surface catchment area.

[0182] Obtain the first digital surface model of the target area;

[0183] The first digital surface model is resampled to a set resolution to obtain the second digital surface model;

[0184] The second digital surface model is used to identify depressions, and the identified virtual depressions are filled to obtain the third digital surface model;

[0185] Based on the third digital surface model, the water area is processed in a preset manner to obtain topographic data used to drive the two-dimensional surface hydrodynamic model.

[0186] To ensure that the grid cells participating in the calculation have effective runoff generation capacity, surface grid cells corresponding to invalid regions and water bodies need to be removed beforehand. Invalid regions include areas with no data due to missing data, cloud cover, or numerical anomalies; water bodies include permanent water bodies such as rivers, lakes, and large flood storage areas. Since these areas typically have independent runoff generation and collection mechanisms and do not directly flow into drainage network nodes, they are removed from the set to be assigned, and the remaining cells are the effective surface grid cells. For each effective surface grid cell, the Euclidean distance between its center point and each network node is calculated, and it is uniquely assigned to the nearest node based on the principle of minimum distance. After all grid cells have been assigned, the total number of surface grid cells covered by each node is counted, and this number is multiplied by the actual physical area of ​​a single grid cell to determine the surface catchment area assigned to that node.

[0187] The first digital surface model not only includes ground elevation information but also incorporates height data from surface structures, vegetation, and other attachments, enabling it to accurately reflect the roughness characteristics and three-dimensional structure of the urban surface. Data sources for the first digital surface model can include data products generated through technologies such as satellite remote sensing stereo image pairs, UAV aerial survey lidar point clouds, or airborne synthetic aperture radar interferometry. This type of data is typically stored in raster format, with each raster value corresponding to the elevation information of a specific geographical location. As the foundational data for two-dimensional surface hydrodynamic simulation, the accuracy and resolution of the first digital surface model directly determine the reliability of subsequent simulation results.

[0188] The original first digital surface model often possesses extremely high spatial resolution. Directly applying it to two-dimensional hydrodynamic simulations would generate a huge computational load, significantly reducing simulation efficiency. Conversely, if the resolution is too low, it would be difficult to characterize the guiding effect of urban micro-topography such as road camber and sidewalk elevation differences on water flow paths. Therefore, the first digital surface model is resampled to a set resolution (e.g., 1 meter, 2 meters, or 5 meters) that matches the grid scale of the two-dimensional surface hydrodynamic model, thus generating a second digital surface model. The resampling algorithm can employ conventional raster processing techniques such as bicubic convolution interpolation, nearest neighbor interpolation, or bilinear interpolation.

[0189] The core of terrain smoothing for the second digital surface model lies in identifying and correcting virtual depressions. Due to factors such as data acquisition errors, the openwork structures within urban buildings (e.g., atriums of high-rise buildings), or spaces under bridges, digital elevation models often contain localized depressions, which appear as enclosed areas where water cannot flow naturally—virtual depressions. Without processing, the two-dimensional surface hydrodynamic model will generate false permanent water accumulation in these areas, severely interfering with the realism of the flooding simulation results. A hydrological algorithm based on flow direction analysis is used for depression identification. By calculating the flow direction of each grid cell, it tracks the confluence and accumulation areas where water cannot flow to the boundary or lower elevation points, identifying these as virtual depressions. After identification, a filling operation is performed, raising the elevation values ​​of the grid cells within the virtual depressions so that the elevation of their outlets is level with the lowest point of the depression's edge, thereby eliminating terrain depressions and constructing a continuous slope where water flow can evolve naturally. The elevation data after filling constitutes the third digital surface model.

[0190] Based on the third digital surface model, water bodies undergo specific processing to generate the final topographic driving data. Although water bodies were removed in the aforementioned catchment area allocation step, rivers, lakes, and other water bodies still occupy specific geographic spaces at the topographic data level. Directly using the original elevation data could lead to instability in the two-dimensional model calculations due to the difference between the water surface elevation and the riverbed elevation. Pre-defined processing methods include uniformly correcting the elevation values ​​of water bodies to their normal water level elevation, or assigning specific Manning roughness coefficients based on the water body type. After this processing, the output topographic data not only includes corrected urban land elevation information but also integrates water body parameters adapted to the characteristics of the hydrodynamic model. This topographic data directly serves as the topographic boundary condition driving the two-dimensional surface hydrodynamic model, ensuring a robust topographic foundation for the model when handling the interaction between land runoff and water bodies.

[0191] This invention provides high-quality data input for one-dimensional and two-dimensional coupled urban flooding simulation in areas without pipe network data by refining the allocation of surface catchment areas, resampling digital surface models, filling depressions, and processing water bodies. The catchment area proximity allocation mechanism ensures the accuracy of runoff generation calculation sources, while the preprocessing of topographic data effectively eliminates noise and defects in the original data. The synergistic effect of these two methods strongly supports the subsequent accurate characterization of urban surface inundation processes by hydrodynamic models.

[0192] Based on the above discussion, it can be seen that the present invention has the following beneficial effects compared with the prior art:

[0193] The present invention has the following beneficial effects:

[0194] This invention does not rely on any real drainage network data. All network geometry and attribute parameters are derived solely from discrete coordinates and their spatial distribution characteristics, road vector data, and preset engineering default values. This makes it truly applicable to areas without network data, avoiding the problem of contradicting the premise of no data application due to the implicit use of real network information.

[0195] This invention uses the road centerline as the geometric anchor point. By projecting the coordinates of discrete facilities onto the road centerline and determining their lateral attributes, orderly merging them into manhole nodes according to their projection positions along the road, and then connecting the main pipes and branch pipes along the road and stitching them together at road intersections using a disjoint-set data structure algorithm, it can robustly reconstruct an approximate drainage network with correct connectivity topology that can be directly solved by a one-dimensional hydrodynamic model, even when there are positioning errors at the facility locations. This overcomes the difficulty of automatically networking discrete points.

[0196] This invention connects the reconstructed approximate drainage network with a two-dimensional surface hydrodynamic model in a unidirectional coupling manner. Only the design rainfall is applied to the one-dimensional hydrodynamic model, and the overflow of the network is used as the only point source input of the two-dimensional surface hydrodynamic model. This makes the surface inundation result driven by the spatial distribution of drainage facilities, thereby being able to characterize the local water accumulation points determined by the heterogeneous distribution of drainage facilities.

[0197] The data sources relied upon by this invention are real-scene images, publicly available road vector data, and digital surface models. These are all low-cost, readily available, and up-to-date data sources, which makes this method highly scalable and practical for engineering applications. It can serve as a practical tool for quickly identifying high-risk points for urban flooding.

[0198] Corresponding to the above method embodiments, the present invention also discloses a device for reconstructing drainage pipe networks and simulating urban flooding in areas without pipe network data.

[0199] See Figure 5 The present invention discloses a structural schematic diagram of a device for reconstructing drainage pipe networks and simulating urban flooding in areas without pipe network data. The device includes:

[0200] The acquisition unit 201 is used to acquire the discrete coordinates of the surface drainage facilities in the target area, as well as the road vector data of the target area;

[0201] Grouping unit 202 is used to project the discrete coordinates onto the road centerline in the road vector data, and to group the discrete coordinates based on the projection position and lateral attributes;

[0202] The drainage network reconstruction unit 203 is used to merge the grouped discrete coordinates into nodes according to the distribution spacing and generate connecting pipe segments. At road intersections, it stitches the segments together based on proximity relationships to reconstruct an approximate drainage network for hydraulic simulation.

[0203] The one-dimensional hydrodynamic simulation unit 204 is used to construct a one-dimensional hydrodynamic model based on the approximate drainage network and extract the water exchange data between the one-dimensional hydrodynamic model and the surface.

[0204] The two-dimensional surface hydrodynamic simulation unit 205 is used to drive the two-dimensional surface hydrodynamic model based on the water exchange data and output the surface inundation results of the target area.

[0205] In summary, this invention maps the discrete coordinates of surface drainage facilities in the target area to road vector data and reconstructs an approximate drainage network with a connected topology. This transforms the distribution characteristics of surface facilities into a network structure solvable by a hydraulic model, thus establishing a numerical simulation foundation for the drainage system without relying on any real network data. This overcomes the bottleneck of traditional methods that cannot conduct refined simulations due to the lack of network data. Furthermore, by using a one-dimensional hydrodynamic model to simulate runoff generation, confluence, and unsteady flow in the reconstructed network, it accurately reflects the control effect of the actual spatial distribution of drainage facilities on runoff convergence and discharge processes, overcoming the limitations of existing equivalent methods. The previous method, due to its assumption of spatial homogeneity, could not characterize the non-uniform characteristics of the drainage system. By solving the two-dimensional evolution process of the overflow from the pipe network on the surface using a two-dimensional surface hydrodynamic model, it is possible to accurately simulate the diffusion path and range of water accumulation under topographic constraints, thus making up for the shortcomings of the simplified method in dynamically tracking the water receding process. The above links constitute a complete extrapolation chain from surface facility data to the result of urban flooding, enabling the simulation results to truly reflect the formation, diffusion and receding dynamics of local water accumulation points determined by the spatial layout of drainage facilities. This solves the technical problem that the existing technology cannot accurately identify high-risk points of urban flooding because it ignores the spatial non-uniformity of drainage facilities.

[0206] In one embodiment, the grouping unit 202 can be specifically used for:

[0207] Project each discrete coordinate onto the centerline of the road to which it belongs, and calculate the mileage position of each projection point along the centerline of the road.

[0208] The lateral attribute is determined based on the deviation of each discrete coordinate relative to the road centerline;

[0209] Discrete coordinates belonging to the same road and having the same lateral attributes are sorted according to their mileage positions.

[0210] In one embodiment, the drainage network reconfiguration unit 203 can be specifically used for:

[0211] Among discrete coordinates belonging to the same road and having the same lateral attributes, those with a projection distance between adjacent discrete coordinates that is less than the merging threshold are grouped into one cluster group.

[0212] Based on the centroid position of the discrete coordinates within the cluster group, a manhole node is generated by offsetting a preset distance towards the road centerline along a direction perpendicular to the road centerline.

[0213] Adjacent manhole nodes are connected sequentially along the centerline of the road to form a main pipeline;

[0214] Connect each discrete coordinate with the corresponding inspection well node to form a branch pipe;

[0215] At road intersections, a disjoint-set data structure algorithm is used to stitch together spatially adjacent but unconnected branches of the pipe network, so that the approximate drainage pipe network forms a connected network.

[0216] In one embodiment, the drainage network reconfiguration unit 203 can also be used for:

[0217] Before merging adjacent discrete coordinates belonging to the same road and having the same lateral attributes into a cluster group, spatial clustering and fusion are performed on the discrete coordinates belonging to the storm drain category in the surface drainage facilities, and multiple discrete coordinates located around the same actual facility location are merged into a storm drain node.

[0218] Statistical analysis of the spacing distribution characteristics between adjacent rainwater inlet nodes after fusion;

[0219] The merging threshold is determined based on the spacing distribution characteristics.

[0220] In one embodiment, the drainage network reconfiguration unit 203 can be specifically used for:

[0221] Obtain the centroid position of the discrete coordinates within the cluster group;

[0222] The corresponding offset distance is determined based on the road classification of the road in the road vector data;

[0223] The centroid position is shifted by the offset distance along a direction perpendicular to the road centerline towards the road centerline to obtain the planar coordinates of the manhole node.

[0224] In one embodiment, the one-dimensional hydrodynamic simulation unit 204 can be specifically used for:

[0225] The designed rainfall is applied to the one-dimensional hydrodynamic model to simulate the generation, confluence, and unsteady flow transport process within the approximate drainage network.

[0226] When the water level at a node in the one-dimensional hydrodynamic model exceeds the surface elevation of that node, an overflow is determined to have occurred and the overflow flow rate of that node is extracted.

[0227] The overflow flow rate is used as the water exchange data between the one-dimensional hydrodynamic model and the surface.

[0228] In one embodiment, the two-dimensional surface hydrodynamic simulation unit 205 can be specifically used for:

[0229] The overflow flow rate is used as the sole water source of the two-dimensional surface hydrodynamic model, and mapped to two-dimensional surface grid cells according to geographic coordinates, serving as the point source inflow boundary condition of the two-dimensional surface hydrodynamic model.

[0230] The two-dimensional surface hydrodynamic model is driven by the point source inflow boundary conditions to solve the two-dimensional shallow water equation to simulate the surface runoff process and obtain the solution results of the surface runoff.

[0231] Based on the calculation results of the surface runoff, the surface water depth and inundation range of the target area are output as the surface inundation result.

[0232] In one embodiment, the acquisition unit 201 can be specifically used for:

[0233] Acquire real-world images with geolocation and attitude metadata;

[0234] A target detection model is used to identify surface drainage facilities from the real-scene image and output detection boxes;

[0235] Based on the preset installation height and field of view of the camera that collects the real-scene images, the pixel position of the detection frame is converted into the depression angle and horizontal deflection angle of the surface drainage facility relative to the camera.

[0236] Based on the depression angle and the horizontal deflection angle, combined with the camera's geographical location and heading angle, the horizontal distance and absolute azimuth angle of the surface drainage facility relative to the camera are calculated.

[0237] By performing a spatial vector inverse mapping based on the horizontal distance and the absolute azimuth, the discrete coordinates of each of the surface drainage facilities are calculated.

[0238] In one embodiment, the drainage network reconstruction and flood simulation device for areas without pipe network data may further include:

[0239] The node allocation unit is used to assign the effective surface grid cells in the target area after removing invalid areas and water areas to the nearest node according to the principle of proximity, so that each node is allocated the corresponding surface catchment area before constructing a one-dimensional hydrodynamic model based on the approximate drainage network.

[0240] The first digital surface model acquisition unit is used to acquire the first digital surface model of the target area;

[0241] The second digital surface model acquisition unit is used to resample the first digital surface model to a set resolution to obtain the second digital surface model.

[0242] The depression identification unit is used to identify depressions in the second digital surface model and fill in the identified virtual depressions to obtain the third digital surface model.

[0243] The terrain data determination unit is used to process the water area according to a preset method based on the third digital surface model to obtain terrain data for driving the two-dimensional surface hydrodynamic model.

[0244] It should be noted that for the specific working principles of each component in the device embodiment, please refer to the corresponding section of the method embodiment, which will not be repeated here.

[0245] Corresponding to the above embodiments, such as Figure 6 As shown, the present invention also provides a schematic diagram of the structure of an electronic device, which may include: a processor 1 and a memory 2;

[0246] The processor 1 and memory 2 communicate with each other via communication bus 3.

[0247] Processor 1, for executing at least one instruction;

[0248] Memory 2 is used to store at least one instruction;

[0249] Processor 1 may be a central processing unit (CPU), an application-specific integrated circuit (ASIC), or one or more integrated circuits configured to implement embodiments of the present invention.

[0250] Memory 2 may include high-speed RAM memory, and may also include non-volatile memory, such as at least one disk storage device.

[0251] In this embodiment, processor 1 executes at least one instruction to implement the steps shown in the method embodiment for reconstructing drainage pipe networks and simulating urban flooding in areas without pipe network data.

[0252] Finally, it should be noted that in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0253] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on the differences from other embodiments. The same or similar parts between the various embodiments can be referred to each other.

[0254] The above description of the disclosed embodiments enables those skilled in the art to make or use the invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the invention. Therefore, the invention is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for reconstructing drainage pipe networks and simulating urban flooding in areas lacking pipe network data, characterized in that, include: Obtain the discrete coordinates of the surface drainage facilities in the target area, as well as the road vector data of the target area; The discrete coordinates are projected onto the road centerline in the road vector data, and the discrete coordinates are grouped based on the projection position and lateral attributes. The grouped discrete coordinates are merged into nodes according to the distribution interval and connected pipe segments are generated. At road intersections, they are stitched together based on proximity to reconstruct an approximate drainage network for hydraulic simulation. A one-dimensional hydrodynamic model is constructed based on the approximate drainage network, and the water exchange data between the one-dimensional hydrodynamic model and the surface is extracted. Based on the water exchange data, a two-dimensional surface hydrodynamic model is driven to output the surface inundation results of the target area.

2. The method for reconstructing drainage pipe networks and simulating urban flooding in areas without pipe network data as described in claim 1, characterized in that, The step of projecting the discrete coordinates onto the road centerline in the road vector data, and grouping the discrete coordinates based on the projection position and lateral attributes, includes: Project each discrete coordinate onto the centerline of the road to which it belongs, and calculate the mileage position of each projection point along the centerline of the road. The lateral attribute is determined based on the deviation of each discrete coordinate relative to the road centerline; Discrete coordinates belonging to the same road and having the same lateral attributes are sorted according to their mileage positions.

3. The method for reconstructing drainage pipe networks and simulating urban flooding in areas without pipe network data as described in claim 2, characterized in that, The process of merging the grouped discrete coordinates into nodes according to their distribution spacing and generating connecting pipe segments, then stitching them together at road intersections based on proximity relationships, reconstructs an approximate drainage network for hydraulic simulation, including: Among discrete coordinates belonging to the same road and having the same lateral attributes, those with a projection distance between adjacent discrete coordinates that is less than the merging threshold are grouped into one cluster group. Based on the centroid position of the discrete coordinates within the cluster group, a manhole node is generated by offsetting a preset distance towards the road centerline along a direction perpendicular to the road centerline. Adjacent manhole nodes are connected sequentially along the centerline of the road to form a main pipeline; Connect each discrete coordinate with the corresponding inspection well node to form a branch pipe; At road intersections, a disjoint-set data structure algorithm is used to stitch together spatially adjacent but unconnected branches of the pipe network, so that the approximate drainage pipe network forms a connected network.

4. The method for reconstructing drainage pipe networks and simulating urban flooding in areas without pipe network data as described in claim 3, characterized in that, Before merging discrete coordinates belonging to the same road and having the same lateral attributes into a single cluster group, where the projection distance between adjacent discrete coordinates is less than the merging threshold, the process further includes: Spatial clustering and fusion are performed on the discrete coordinates of the surface drainage facilities that belong to the category of rainwater inlets, and multiple discrete coordinates located around the same actual facility location are merged into one rainwater inlet node. Statistical analysis of the spacing distribution characteristics between adjacent rainwater inlet nodes after fusion; The merging threshold is determined based on the spacing distribution characteristics.

5. The method for reconstructing drainage pipe networks and simulating urban flooding in areas without pipe network data as described in claim 3, characterized in that, The step of generating manhole nodes by offsetting a preset distance towards the road centerline along a direction perpendicular to the road centerline based on the centroid position of the discrete coordinates within the cluster group includes: Obtain the centroid position of the discrete coordinates within the cluster group; The corresponding offset distance is determined based on the road classification of the road in the road vector data; The centroid position is shifted by the offset distance along a direction perpendicular to the road centerline towards the road centerline to obtain the planar coordinates of the manhole node.

6. The method for reconstructing drainage pipe networks and simulating urban flooding in areas without pipe network data as described in claim 1, characterized in that, The step of constructing a one-dimensional hydrodynamic model based on the approximate drainage network and extracting water exchange data between the one-dimensional hydrodynamic model and the surface includes: The designed rainfall is applied to the one-dimensional hydrodynamic model to simulate the generation, confluence, and unsteady flow transport process within the approximate drainage network. When the water level at a node in the one-dimensional hydrodynamic model exceeds the surface elevation of that node, an overflow is determined to have occurred and the overflow flow rate of that node is extracted. The overflow flow rate is used as the water exchange data between the one-dimensional hydrodynamic model and the surface.

7. The method for reconstructing drainage pipe networks and simulating urban flooding in areas without pipe network data as described in claim 6, characterized in that, The two-dimensional surface hydrodynamic model driven by the water exchange data outputs the surface inundation results for the target area, including: The overflow flow rate is used as the sole water source of the two-dimensional surface hydrodynamic model, and mapped to two-dimensional surface grid cells according to geographic coordinates, serving as the point source inflow boundary condition of the two-dimensional surface hydrodynamic model. The two-dimensional surface hydrodynamic model is driven by the point source inflow boundary conditions to solve the two-dimensional shallow water equation to simulate the surface runoff process and obtain the solution results of the surface runoff. Based on the calculation results of the surface runoff, the surface water depth and inundation range of the target area are output as the surface inundation result.

8. The method for reconstructing drainage pipe networks and simulating urban flooding in areas without pipe network data as described in claim 1, characterized in that, The process of obtaining the discrete coordinates of the surface drainage facilities in the target area includes: Acquire real-world images with geolocation and attitude metadata; A target detection model is used to identify surface drainage facilities from the real-scene image and output detection boxes; Based on the preset installation height and field of view of the camera that collects the real-scene images, the pixel position of the detection frame is converted into the depression angle and horizontal deflection angle of the surface drainage facility relative to the camera. Based on the depression angle and the horizontal deflection angle, combined with the camera's geographical location and heading angle, the horizontal distance and absolute azimuth angle of the surface drainage facility relative to the camera are calculated. By performing a spatial vector inverse mapping based on the horizontal distance and the absolute azimuth, the discrete coordinates of each of the surface drainage facilities can be calculated.

9. The method for reconstructing drainage pipe networks and simulating urban flooding in areas without pipe network data as described in claim 1, characterized in that, Before constructing the one-dimensional hydrodynamic model based on the approximate drainage network, the following steps are also included: The effective surface grid cells in the target area, after removing invalid and water areas, are assigned to the nearest node according to the principle of proximity, so that each node is allocated the corresponding surface catchment area. Obtain the first digital surface model of the target area; The first digital surface model is resampled to a set resolution to obtain the second digital surface model; The second digital surface model is used to identify depressions, and the identified virtual depressions are filled to obtain the third digital surface model; Based on the third digital surface model, the water area is processed in a preset manner to obtain topographic data used to drive the two-dimensional surface hydrodynamic model.

10. A device for reconstructing drainage pipe networks and simulating urban flooding in areas lacking pipe network data, characterized in that, include: The acquisition unit is used to acquire the discrete coordinates of the surface drainage facilities in the target area, as well as the road vector data of the target area; A grouping unit is used to project the discrete coordinates onto the road centerline in the road vector data, and to group the discrete coordinates based on the projection position and lateral attributes. The drainage network reconstruction unit is used to merge the grouped discrete coordinates into nodes according to the distribution spacing and generate connecting pipe segments. At road intersections, it stitches the segments together based on proximity relationships to reconstruct an approximate drainage network for hydraulic simulation. A one-dimensional hydrodynamic simulation unit is used to construct a one-dimensional hydrodynamic model based on the approximate drainage network and extract water exchange data between the one-dimensional hydrodynamic model and the surface. The two-dimensional surface hydrodynamic simulation unit is used to drive the two-dimensional surface hydrodynamic model based on the water exchange data and output the surface inundation results of the target area.

11. An electronic device, characterized in that, The electronic device includes: a memory and a processor; The memory is used to store at least one instruction; The processor is used to execute the at least one instruction to implement the method for reconstructing drainage networks and simulating urban flooding in areas without network data as described in any one of claims 1 to 9.

Images