A Substation Flood Simulation Method Based on Real-Scene 3D Feature Recognition and Model Coupling

Abstract

This invention provides a substation flooding simulation method based on real-scene 3D feature recognition and model coupling, comprising: acquiring real-scene 3D data of the substation and its catchment area; extracting topographic elevation, water-blocking boundaries, drainage facilities, pipeline elements, and key equipment area boundaries; delineating potential waterlogged depressions and identifying hydraulic saddle points; constructing an inundation topology map based on potential waterlogged depressions and hydraulic saddle points; establishing a one-dimensional drainage model and a two-dimensional surface runoff model; forming a coupled dynamic model through model coupling ports; generating candidate rainfall processes; tracking surface runoff waves outside the substation, runoff waves within the substation, and backwater response waves from underground pipelines; generating wave overlap records and topology boundary violation records; switching local mesh templates based on topology boundary violation records; tracking water propagation paths; selecting the most dangerous rainfall process based on the comprehensive risk value of the key equipment area; backtracking key hydraulic saddle points; and verifying hydraulic blocking combinations.

Description

Technical Field

[0001] This invention relates to the field of flood prevention simulation technology for power facilities, and in particular to a method for simulating flooding in substations based on real-scene 3D feature recognition and model coupling. Background Technology

[0002] Substations are critical infrastructure in the power system. Some substations are located in low-lying urban areas or areas with limited drainage capacity. Under conditions of short-term heavy rainfall, continuous rainfall, and backflow from external pipelines, they are prone to backflow of water from outside the station, runoff on the ground inside the station, and poor drainage of pump pools, which in turn threaten the operational safety of the transformer area and control equipment area.

[0003] Existing methods for simulating substation flooding typically establish a hydrodynamic model based on topographic elevation, building distribution, drainage network, and design rainfall events. This model is then coupled with a one-dimensional drainage model to a two-dimensional surface runoff model to calculate water depth, inundation range, flow velocity, and direction. However, conventional digital elevation models struggle to fully capture local elevation changes at low points of perimeter walls, entrance / exit thresholds, local road elevations, areas around storm drains, and pump pool edges. While these elevation differences may be small, they can directly determine whether external water enters the substation. Existing methods usually treat these locations as ordinary topographic units or fixed water-blocking boundaries, failing to reflect the changes in water propagation paths after localized flooding.

[0004] Meanwhile, existing methods mostly use fixed design rainfall processes for simulation, focusing on analyzing total rainfall and peak rainfall intensity, while rarely considering the concentrated superposition of external surface runoff, internal ground runoff, and underground pipe network backwater response reaching the same local location at different times. Such superposition can cause local water levels to rapidly exceed thresholds, gaps in walls, or road high points, forming new water propagation paths. Therefore, this invention proposes a substation flooding simulation method based on real-scene 3D feature recognition and model coupling.

[0005] The information disclosed in the background section is only intended to enhance the understanding of the background of this disclosure, and therefore may contain prior art information that is not common knowledge to those skilled in the art. Summary of the Invention

[0006] The purpose of this invention is to address the shortcomings of existing technologies by providing a substation flood simulation method based on real-scene 3D feature recognition and model coupling, thereby solving the technical problems mentioned in the background section.

[0007] To achieve the above objectives, the present invention provides the following technical solution: A substation flood simulation method based on real-scene 3D feature recognition and model coupling includes the following steps: S1. Obtain real-world 3D data of the substation and catchment area, extract topographic elevation, water-blocking boundary, drainage facilities, pipeline elements and key equipment area boundary, delineate potential waterlogged depressions, and identify hydraulic saddle points that control the connectivity of adjacent potential waterlogged depressions. S2. Construct an inundation topology map based on potential waterlogged depressions and hydraulic saddles, establish a one-dimensional drainage model and a two-dimensional surface runoff model covering the in-station surface area and the out-of-station waterlogging area, set up model coupling ports, and form a coupled dynamic model. S3. Generate candidate rainfall processes under preset rainfall constraints, and track surface runoff waves outside the station, runoff waves inside the station and the top support response waves of underground pipe network based on the coupled dynamic model to generate water wave overlap records and topology boundary crossing records. S4. Update the flooded topological edge state according to the topological edge over-limit record, switch the corresponding local mesh template, calculate the local overflow and trace the water propagation path to the critical equipment area. S5. Based on the comprehensive risk value of the key equipment area corresponding to each candidate rainfall process, select the most dangerous rainfall process, trace back the key hydraulic saddle points, generate and verify the hydraulic blocking combination, and output the strategy optimization results and the verification results of the newly added water accumulation propagation path.

[0008] S1 specifically includes: acquiring UAV oblique photography data of substations and catchment areas, 3D laser scanning point cloud data, drainage network as-built drawings, drainage pipeline survey results, equipment ledgers, substation general layout plan, and on-site measurement data; performing coordinate unification, outlier removal, and feature extraction to form a unified real-scene 3D feature set; progressively increasing the set water level based on the global terrain model and local fine terrain model; performing connectivity region search on surface units not isolated by water-blocking boundaries to form a set of potential waterlogged depressions; searching for passable paths for adjacent potential waterlogged depressions; determining control elevations, hydraulic saddle points, and effective flow widths; and establishing a set of local terrain detail windows for use by S2.

[0009] S2 specifically includes: setting potential waterlogged depressions as waterlogging nodes, setting hydraulic saddle points as inundation topological edges connecting adjacent waterlogging nodes, generating an inundation topology map and topological edge mapping relationship; establishing a one-dimensional drainage model and a two-dimensional surface runoff model based on drainage facility element set, pipe network element set, global terrain model, local fine terrain model and water-blocking boundary set, pre-generating and embedding local mesh templates; mapping rainwater inlets, inspection wells with surface overflow openings, pump pool openings connected to the surface and drainage outlets within the model range as model coupling ports, calculating and exchanging water volume according to water level relationship, forming a coupled dynamic model, and retaining the topological edge state switching interface and local mesh template switching interface.

[0010] S3 specifically includes: reading the regional rainfall intensity formula, local rainfall station historical data, and project flood control standards; adjusting the rainfall peak occurrence time, the ratio of preceding and following rainfall peaks, and the rainfall peak interval based on the basic design rainfall process; performing total rainfall normalization and peak rainfall intensity verification to form a set of candidate rainfall processes; inputting each candidate rainfall process into the coupled dynamic model; tracking the external water source volume, internal water source volume, and underground pipe network backwater source volume according to the coupled time slice; calculating the local water level of each hydraulic saddle point and forming a water wave arrival record; identifying the concentrated superposition of three types of water waves within the overlapping time window; and generating water wave overlap records and topological edge violation records respectively based on whether the local water level reaches the control elevation.

[0011] S4 specifically includes: reading topological edge exceedance records, local water level time series, and flooded topology map; updating the flooded topological edge state based on control elevation, receding water level difference, and continuous coupling time slices; generating topological state switching events; calling local mesh templates based on topological state switching events; switching between blocking state templates and overflow state templates; calculating local overflow volume based on local water levels on both sides of the hydraulic saddle point, control elevation, and effective overflow width; constructing a directed topology map based on the direction of local overflow volume; tracing the water propagation path from water source nodes to associated nodes in the critical equipment area; recording the first water accumulation time, maximum water accumulation depth, water accumulation duration, and the proportion of water from the three types of sources; and forming candidate rainfall risk records.

[0012] S5 specifically includes: reading the candidate rainfall process set, water wave overlap record, and candidate rainfall risk record; calculating the comprehensive risk value of the key equipment area based on the first water accumulation time, maximum water accumulation depth, water accumulation duration, and equipment importance coefficient; screening the most dangerous rainfall process; tracing back along the water accumulation propagation path corresponding to the most dangerous rainfall process to candidate key hydraulic saddle points; performing counterfactual verification on each candidate key hydraulic saddle point and its combination; generating hydraulic blocking combinations and strategy optimization models that meet the strategy optimization constraints; inputting the strategy verification rainfall set into each strategy optimization model; calculating the residual risk value; verifying the newly added water accumulation propagation path and necessary drainage functions; and outputting the strategy optimization results that meet the risk control objectives.

[0013] The beneficial effects of this invention are as follows: This invention extracts terrain elevation, water-blocking boundaries, drainage facilities, pipeline elements, and key equipment area boundaries based on real-world 3D data. It identifies hydraulic saddle points that control the connectivity of adjacent potential waterlogged depressions, and can preserve details such as low points of walls, entrance and exit thresholds, and local high points of roads, thereby improving the matching degree between urban flooding simulation and actual conditions.

[0014] This invention dynamically couples a one-dimensional drainage model with a two-dimensional surface runoff model to calculate the water exchange at storm drains, manholes, pump pool openings, and drainage outlets. This reflects surface inflow, pipe network backflow, and reverse overflow processes, improving the accuracy of water accumulation simulation both inside and outside the station. By tracking surface runoff waves outside the station, ground runoff waves inside the station, and underground pipe network backflow response waves, and identifying localized water level exceedances caused by the concentrated superposition of these three types of waves, it can detect dangerous urban flooding conditions that are difficult to cover during fixed design rainfall events.

[0015] This invention, by switching local grid templates based on hydraulic saddle point exceedance results, calculates local overflow and tracks water propagation paths, thus clearly identifying the source, propagation direction, and key channels of water entering critical equipment areas, improving the targeting of risk location. Through screening the most dangerous rainfall processes, counterfactual verification of key hydraulic saddle points, and combined verification of hydraulic blocking measures, it outputs local elevation or blocking measures that meet risk control objectives, reducing ineffective modifications and preventing the formation of new water propagation paths after local adjustments. Attached Figure Description

[0016] Figure 1 This is a schematic diagram of the substation flood simulation method based on real-scene 3D feature recognition and model coupling according to the present invention; Figure 2 This is a schematic diagram of real-scene 3D feature recognition and hydraulic saddle point extraction in an embodiment of the present invention; Figure 3 This is a diagram showing the coupling framework between the flooded topology and the model in an embodiment of the present invention. Figure 4 This is a schematic diagram of three types of water wave overlap and local grid template switching in an embodiment of the present invention. Detailed Implementation

[0017] 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.

[0018] Example: Figures 1 to 4 As shown in the figure, this embodiment provides a substation flooding simulation method based on real-scene 3D feature recognition and model coupling, including the following steps: S1. Obtain real-world 3D data of the substation and catchment area, extract topographic elevation, water-blocking boundary, drainage facilities, pipeline elements and key equipment area boundary, delineate potential waterlogged depressions, and identify hydraulic saddle points that control the connectivity of adjacent potential waterlogged depressions. S2. Construct an inundation topology map based on potential waterlogged depressions and hydraulic saddles, establish a one-dimensional drainage model and a two-dimensional surface runoff model covering the in-station surface area and the out-of-station waterlogging area, set up model coupling ports, and form a coupled dynamic model. S3. Generate candidate rainfall processes under preset rainfall constraints, and track surface runoff waves outside the station, runoff waves inside the station and the top support response waves of underground pipe network based on the coupled dynamic model to generate water wave overlap records and topology boundary crossing records. S4. Update the flooded topological edge state according to the topological edge over-limit record, switch the corresponding local mesh template, calculate the local overflow and trace the water propagation path to the critical equipment area. S5. Based on the comprehensive risk value of the key equipment area corresponding to each candidate rainfall process, select the most dangerous rainfall process, trace back the key hydraulic saddle points, generate and verify the hydraulic blocking combination, and output the strategy optimization results and the verification results of the newly added water accumulation propagation path.

[0019] S1 specifically includes the following sub-steps: S110. Acquire UAV oblique photography data, 3D laser scanning point cloud data, drainage network as-built drawings, drainage pipeline survey results, equipment ledgers, substation general layout plan and on-site measurement data for substations and catchment areas.

[0020] UAV oblique photogrammetry data is used to acquire the overall topography, building outlines, and land use types of the catchment area; 3D laser scanning point cloud data is used to acquire the location of the perimeter wall, the top elevation of the perimeter wall, the bottom elevation of the perimeter wall, the entrance and exit thresholds, road edges, local low points of the road, the top elevation of the storm drain grate, and the foundation elevation of key equipment; drainage network as-built drawings, drainage pipeline survey results, and field measurement data are used to acquire the pipeline route, pipe diameter, pipeline node location, pipeline node elevation, pump pool outline, pump pool bottom elevation, effective volume of the pump pool, and drainage outlet location; equipment ledgers and substation general layout plans are used to determine the boundaries of key equipment areas.

[0021] Critical equipment areas refer to transformer areas, switchgear areas, control equipment areas, or other pre-designated facility areas that require water accumulation risk assessment.

[0022] Underground pipe network data is not inferred from surface imagery. For pipe sections missing from the as-built drawings of the drainage pipe network or inconsistent with the actual site, supplementary data is obtained through manhole measurements, storm drain measurements, and drainage outlet verification. For storm drains, manholes, and drainage outlets, spatial coordinates, top elevation, bottom elevation, opening dimensions, and their connection relationships with pipe network nodes are recorded respectively.

[0023] Using the field measurement control points as a reference, all types of data are transformed into a unified spatial coordinate system:

[0024] In the formula, Let be the three-dimensional coordinates of the i-th point to be transformed in the original data coordinate system; Let be the three-dimensional coordinates of the i-th point to be transformed in the unified spatial coordinate system; The coordinate rotation matrix; This is the coordinate translation vector. After the transformation, the horizontal position difference and elevation difference of the corresponding control points are checked. In one specific embodiment, the horizontal position difference is no greater than 0.05m, and the elevation difference is no greater than 0.03m.

[0025] Outlier removal from 3D laser-scanned point clouds:

[0026] In the formula, The average distance between the i-th point and a preset number of adjacent points; This is the mean of all average distances; is the standard deviation of the total average distance. When the above formula is satisfied, the corresponding point is deleted. The point cloud after removing outliers is then classified into ground points, building points, and wall points.

[0027] Based on the processed data, a global terrain model covering the substation and catchment area is constructed, and local fine-grained terrain models are built for the perimeter wall, entrance / exit thresholds, local low points on roads, the area around storm drains, and the edge of pumping stations. The global terrain model is used to represent the overall elevation distribution and overall flow direction, using a 2m grid. The local fine-grained terrain model is used to preserve local elevation abrupt changes, and its grid side length is smaller than the minimum effective width of the area to be identified for elevation abrupt changes. In a specific embodiment, the entrance / exit threshold width is 0.80m, the top elevation of the threshold is 25.46m, and the lowest point elevation of the road is 25.38m. A local fine-grained terrain model with a 0.05m grid is generated for a 10m radius around the threshold to preserve a 0.08m elevation difference.

[0028] The outer contour of the building and the location of the continuous wall are defined as the water-blocking boundary. The global terrain model, the local detailed terrain model, the water-blocking boundary set, the drainage facility element set, the pipe network element set, the key equipment area boundary set, and the land use type set are summarized to form a unified real-scene 3D element set, which is used by S120 to delineate potential waterlogged areas and by S220 to configure terrain, boundary, and drainage facility parameters.

[0029] S120. Potential waterlogged depressions refer to independent surface water accumulation areas that, under the constraint of water-blocking boundaries, can remain independent until the control elevation of the adjacent hydraulic saddle point is reached. The global terrain model and the local fine-grained terrain model are converted into surface units with adjacency relationships; for the raster model, an 8-neighborhood connection method is used; for unstructured triangular meshes, a common-edge connection method is used. The outer contours of buildings and the locations of continuous walls are written into the water-blocking boundary set; entrance thresholds, local gaps in walls, local high points on roads, and the edges of pumping stations are not set as permanent water-blocking boundaries, but rather their actual elevations are retained, allowing them to form water propagation channels after the water level reaches the corresponding elevation.

[0030] The set water level is increased step by step, from low to high:

[0031] In the formula, Let be the set of surface units that can be covered by water at a given water level h; D is the surface calculation region; B is the set of water-blocking boundaries; and q is the surface unit. q represents the elevation of the surface unit; h represents the set water level.

[0032] A seed-filling algorithm or a breadth-first search (BFS) algorithm is used to perform connectivity searches on surface units that meet the elevation conditions and are not isolated by water-blocking boundaries. When a connectivity region appears for the first time, it is registered as an initial water-filling unit. When the coverage area expands after the set water level rises but still does not connect with other initial water-filling units, the coverage area of ​​the initial water-filling unit is updated. When two initial water-filling units connect for the first time, the independent numbers of the two initial water-filling units are retained, and the initial connection water level, contact boundary, and initial connection position are recorded.

[0033] In one specific embodiment, the lowest elevation of the low-lying area outside the station road is 25.20m, and the lowest elevation of the inner side of the station entrance / exit is 25.31m. When the set water level rises to 25.46m, the two initial water accumulation units connect for the first time at the entrance / exit threshold. At this point, the low-lying area outside the station road and the inner side of the station entrance / exit are respectively registered as potential water accumulation depressions, and 25.46m is recorded as the initial connection water level.

[0034] The potential waterlogged depressions are aggregated by their numbers, surface unit numbers, lowest ground elevation, maximum boundary range, adjacent potential waterlogged depression numbers, first connection water level, contact boundary, and first connection location, forming a set of potential waterlogged depressions, which is then output to S130 and S210.

[0035] S130, a hydraulic saddle point refers to the local location on a passable path connecting two adjacent potential waterlogged areas, where the minimum water level required to control the flow of water from one potential waterlogged area to the other is determined; the control elevation refers to the minimum water level that water needs to reach to pass through the corresponding hydraulic saddle point.

[0036] For each group of adjacent potential waterlogged depressions, calculate the control elevation:

[0037] In the formula, The control elevation between potential waterlogged depression a and potential waterlogged depression b; Let be the set of passable paths from potential waterlogged depression a to potential waterlogged depression b; be one of these passable paths; q be the surface unit on the path. Let be the elevation of surface unit q. When hydraulic saddle point e corresponds to potential water-filled depressions a and b, then... Recorded as For future use This indicates the control elevation of the hydraulic saddle point e.

[0038] Passable paths must not cross the outer contours of buildings or continuous perimeter walls; paths are retained when they pass through entrance / exit thresholds, partial gaps in walls, local high points on roads, or the edges of pumping stations. When multiple paths share the same control elevation, hydraulic saddle point records are generated for each. For continuous surface units corresponding to the control elevation, the center point of the location with the minimum passable width is determined as the hydraulic saddle point, and the effective flow width is recorded.

[0039] Establish a local topographic detail window centered on each hydraulic saddle point, extending it along the flowable direction of water to two adjacent potential waterlogged areas, and vertically to completely cover intersecting wall sections, threshold sections, road edge sections, or pump pool edge sections; when the window side length is less than 10m, it is truncated to 10m. Summarize the hydraulic saddle point number, adjacent potential waterlogged area number, spatial coordinates, control elevation, flowable direction, effective flow width, field element type, and local topographic detail window number to form a hydraulic saddle point set; summarize the window number, window boundary, local fine topographic unit, and connection relationship with adjacent global topographic units to form a local topographic detail window set.

[0040] The hydraulic saddle point set is output to S210, S330, S410 and S520; the local terrain detail window set is output to S220 and S420.

[0041] S2 specifically includes the following sub-steps: S210: Read the set of potential waterlogged depressions output by S120 and the set of hydraulic saddle points output by S130. Set each potential waterlogged depression as a waterlogging node; a waterlogging node is a topological node corresponding to one potential waterlogged depression, used to record the spatial extent and waterlogging state of that potential waterlogged depression. Set each hydraulic saddle point as a flooded topological edge; a flooded topological edge is a topological connection corresponding to one hydraulic saddle point, used to represent a possible water propagation channel between two adjacent potential waterlogged depressions.

[0042] Based on the two potential waterlogged depressions corresponding to the hydraulic saddle points, the inundation topology edges are connected to the corresponding two waterlogged nodes to form an inundation topology map. For each inundation topology edge, the following information is recorded: inundation topology edge number, hydraulic saddle point number, starting waterlogged node number, ending waterlogged node number, hydraulic saddle point spatial coordinates, control elevation, effective flow width, flow direction, and local topographic detail window number.

[0043] Each submerged topology edge is set to an inactive state and an active state. The inactive state indicates that the corresponding hydraulic saddle point has not yet formed cross-edge water propagation; the active state indicates that the corresponding hydraulic saddle point has formed cross-edge water propagation. The initial state of each submerged topology edge is set to inactive, and is subsequently updated by S410 based on the local water level time series, control elevation, receding water level difference, and the number of continuous coupling time slices.

[0044] Based on the spatial coordinates, control elevation, and local topographic detail window number of each hydraulic saddle point, a topological edge mapping relationship is generated to be written into the two-dimensional surface flood model. This topological edge mapping relationship is used to determine the local triangular mesh edges or local triangular mesh cells corresponding to the spatial coordinates of the hydraulic saddle points as topological edge mapping cells after the two-dimensional surface flood model is established in S220. A topological edge mapping cell refers to a local mesh location used to receive the state transition of the submerged topological edge.

[0045] Output the flooded topology graph and topology edge mapping relationship to S220, S410 and S430.

[0046] A one-dimensional drainage model is established based on the drainage facility element set and pipe network element set output from S220 and S110. The one-dimensional drainage model is constructed using the SWMM (Sudden Rainstorm Flood Management Model), and includes storm drains, manholes with surface overflow openings, pipe network nodes, pipe sections, pump sumps, pumps, and drainage outlets. The coordinates of storm drains, the elevation of the storm drain grate top, and the opening size are derived from field measurements and point cloud results; the locations of pipe network nodes, the bottom elevation of nodes, the start and end positions of pipe sections, pipe diameter, length, and slope are derived from drainage network as-built drawings, drainage pipeline survey results, and field verification results; the bottom elevation of pump sumps, effective volume, starting water level, stopping water level, pump rated flow rate, and drainage outlet location are derived from equipment ledgers and field measurement data.

[0047] Runoff generation parameters for sub-catchments are set based on land use type, impermeability, and slope. The runoff curve number (CN) is a parameter used to characterize runoff generation capacity under different surface conditions. Hardened roads, hard surfaces around buildings, and green areas are assigned corresponding CN values, which are then verified based on measured rainfall, water level, and water accumulation data.

[0048] The water flow at the network nodes in a one-dimensional drainage model satisfies:

[0049] In the formula, Let n be the water storage volume of the pipeline node n at time t; Let n be the flow rate of the pipe segment flowing into node n of the pipeline network; This refers to the flow rate of the pipe segment exiting from pipe network node n; This represents the runoff volume entering the pipeline network node n from the corresponding sub-catchment area; The exchange flow entering or exiting the pipeline node n through the model coupling port; This refers to the flow rate discharged by the water pump.

[0050] The water pump uses start-stop water level control. When the water level in the pump pool reaches or exceeds the start-up water level, the water pump switches to running mode; when the water level in the pump pool drops to or below the stop water level, the water pump switches to stopping mode. The start-up water level is higher than the stop water level to avoid repeated start-stop cycles between adjacent calculation times.

[0051] Based on the global terrain model, local fine terrain model, water-blocking boundary set, and land use type set output by S110, a two-dimensional surface runoff model covering the substation's internal surface area and external water-blocking area is established. The two-dimensional surface runoff model uses TELEMAC-2D to construct an unstructured triangular mesh. The building outlines and continuous wall locations are set as inaccessible boundaries; entrance / exit thresholds, local gaps in the walls, local road elevations, and pump pool edges are retained as local mesh edges with true elevations; and Manning coefficients are mapped according to land use types.

[0052] Based on the set of local terrain detail windows output by S130, a set of local mesh templates is pre-generated. Each local mesh template includes a blocking state template and a flow state template. The blocking state template indicates that the corresponding inundated topological edge is inactive, and the local boundary corresponding to the hydraulic saddle point does not allow cross-edge water exchange; the flow state template indicates that the corresponding inundated topological edge is active, and the local boundary corresponding to the hydraulic saddle point calculates the local flow rate according to the control elevation and effective flow width. The local mesh templates are embedded into the two-dimensional surface runoff model, and the connection relationship between them and adjacent global triangular mesh cells is preserved.

[0053] Based on the topology edge mapping relationship output by S210, the local triangular mesh edges or local triangular mesh elements corresponding to the spatial coordinates of the hydraulic saddle point are determined as topology edge mapping elements. This allows the local boundary to switch between the blocking state template and the overflow state template according to the submerged topology edge state, thus completing dynamic modeling.

[0054] Output the one-dimensional drainage model, the two-dimensional surface runoff model, and the local mesh template set to S230 and S420.

[0055] S230. Map the rainwater inlets, inspection wells with surface overflow openings, pump pool openings connected to the surface, and drainage outlets within the coverage area of ​​the two-dimensional surface runoff model in the one-dimensional drainage model to the corresponding triangular mesh elements to form a model coupling port set. Ordinary underground pipe network nodes do not have model coupling ports.

[0056] A model coupling port refers to the water exchange interface between a one-dimensional drainage model and a two-dimensional surface runoff model. Each model coupling port records the port number, port type, corresponding facility number, triangular mesh element number, surface elevation, storm drain grate top elevation or manhole overflow outlet elevation, effective opening area, effective flow width, design flow rate, and current exchange flow rate.

[0057] The switching throughput of the model-coupled port is determined by the following formula:

[0058] In the formula, Let be the switching traffic at model coupling port c at time t; This indicates that surface water has entered the pipe network; This indicates that water in the pipe network is overflowing backwards to the ground surface; The flow rate of the weir; This refers to the outflow rate of surface water when it enters the pipe network. The outflow rate from the orifice when water in the pipeline overflows back to the ground surface; Design flow rate for model coupling port c; This corresponds to the water level of the triangular grid unit on the Earth's surface. This refers to the water level at the corresponding pipeline node; This refers to the elevation of the top of the rainwater grate or the elevation of the overflow outlet of the inspection well.

[0059] in:

[0060] In the formula, The weir flow coefficient at the model coupling port c; The effective overcurrent width of the model coupling port c.

[0061]

[0062] In the formula, Let be the orifice outflow coefficient of the model coupling port c; denoted as c, where c is the effective opening area of ​​the model coupling port; g is the gravitational acceleration.

[0063]

[0064] In the formula, the meaning of each letter is consistent with the aforementioned definition.

[0065] Set up coupling time slices and calculate the exchanged water volume within each coupling time slice: In the formula, Let be the amount of water exchanged at the model coupling port c within the m-th coupling time slice; This represents the start time of the m-th coupling time slice; This represents the coupling time slice length. The amount of water exchanged in each settlement does not exceed the available water volume within the corresponding surface triangular mesh cell or pipe network node. The water exchange volume is settled only once.

[0066] When the water storage capacity of the pump pool is expressed by a one-dimensional drainage model, the pump discharge volume is only deducted from the pump pool storage capacity. The water exchange between the pump pool and the two-dimensional surface runoff model is handled through the model coupling port corresponding to the pump pool opening. When the drainage outlet is within the coverage area of ​​the two-dimensional surface runoff model, the equivalent amount of water discharged by the pump is written into the inflow source term of the triangular mesh cell corresponding to the drainage outlet; when the drainage outlet is outside the model range, no inflow source term is written into the two-dimensional surface runoff model.

[0067] The flood topology map, one-dimensional drainage model, two-dimensional surface runoff model, local mesh template set, model coupling port set, water volume exchange settlement rules, topology edge state switching interface and local mesh template switching interface are combined into a coupled dynamic model and output to S320-S430.

[0068] S3 specifically includes the following sub-steps: S310. Read the regional rainfall intensity formula, local historical rainfall station data, project flood control standards, and basic design rainfall process. The basic design rainfall process refers to the time series of design rainfall events generated according to the rainfall return period, rainfall duration, and Chicago rainfall pattern; the candidate rainfall process refers to the rainfall time series formed after adjusting the rainfall peak occurrence time, the ratio of preceding and following rainfall peaks, and the interval between rainfall peaks while keeping the rainfall return period and total design rainfall constant.

[0069] The duration of rainfall is determined based on the runoff time in the catchment area, and discrete time slices of rainfall are set according to the time resolution of the rainfall data. Based on the basic design rainfall process, the total design rainfall is divided into two sub-rainfall processes according to the ratio of the preceding and following rainfall peaks. The peak occurrence times of the two sub-rainfall processes are adjusted, and time shifts are performed according to the interval between rainfall peaks. The two sub-rainfall processes are then superimposed, and the total rainfall is normalized and the peak rainfall intensity is checked.

[0070] In one specific embodiment, the ratio of rainfall peak occurrence locations is set to 0.3, 0.5, and 0.7; the ratio of consecutive rainfall peaks is set to 1:1, 2:1, and 1:2; the interval between rainfall peaks is set to 10 min, 20 min, and 30 min; and the length of the rainfall discrete time slice is set to 5 min. These parameter combinations are used to generate multiple sets of initial perturbation rainfall sequences.

[0071] Candidate rainfall events are normalized to total rainfall using the following formula:

[0072] In the formula, Let the rainfall intensity of the k-th candidate rainfall process be the rainfall intensity in the j-th discrete time slice. The rainfall intensity before normalization; The total design rainfall is used for all candidate rainfall events; J is the total number of discrete time slices of rainfall. is the length of the discrete time slice of rainfall.

[0073] Candidate rainfall events with peak rainfall intensity exceeding the project's preset upper limit are deleted, forming a candidate rainfall event set. The candidate rainfall event set should include at least the candidate rainfall event number, rainfall return period, total design rainfall, rainfall duration, rainfall discrete time slice length, rainfall intensity corresponding to each rainfall discrete time slice, rainfall peak occurrence time, ratio of consecutive rainfall peaks, and rainfall peak interval.

[0074] The simulation period for each candidate rainfall event includes the rainfall duration and the drainage continuation period. The drainage continuation period refers to the time during which the coupled dynamic model continues to run after the rainfall stops, used to record the continued propagation of water inside and outside the station, the back pressure on pipe network nodes, and the continuous drainage process of the pumping station. The set of candidate rainfall events is output to S320.

[0075] S320. For each candidate rainfall event, S320-S430 are executed cyclically according to the coupled time slices. Within each coupled time slice, S320 first updates the three types of source water volume and local water level, and then S330 determines whether water wave overlap records and topological edge violation records are formed. When a topological edge violation record is formed, S410 updates the flooded topological edge state, and S420 switches the corresponding local mesh template. Subsequently, S430 calculates the water propagation state of the next coupled time slice until the simulation of the corresponding candidate rainfall event ends.

[0076] The three types of water waves include surface runoff waves outside the station, runoff waves within the station's ground level, and backwater response waves from underground pipe networks. Surface runoff waves outside the station refer to the water volume changes formed by rainfall in the surrounding drainage area and propagating along the surface to the vicinity of the hydraulic saddle point. Runoff waves within the station's ground level refer to the water volume changes formed by rainfall within the station's ground level and propagating along the surface to the vicinity of the hydraulic saddle point. Backwater response waves from underground pipe networks refer to the water volume changes in the one-dimensional drainage model caused by overflowing pipes or rising water levels at pipe network nodes, which are then released back into the two-dimensional surface runoff model through the model coupling port. Backwater response waves from underground pipe networks do not represent water hammer processes. Only after the reverse overflow volume is written into the two-dimensional surface runoff model can it participate in the local water level calculation at the hydraulic saddle point.

[0077] For each triangular grid cell in the two-dimensional surface runoff model, the water volume within the cell is divided into three types of water sources:

[0078] In the formula, Let q be the total water volume within the triangular mesh cell at time t; Water volume originating from outside the station; The water volume originating from within the station; The water volume is supplied by the underground pipe network.

[0079] Each water volume retains only one source identifier. When surface water enters the pipe network through the model coupling port, the corresponding source water volume is deducted from the corresponding triangular grid cell. When this water volume overflows back to the surface through the model coupling port, it is uniformly converted into underground pipe network top-supported source water volume, and the original source identifier is no longer retained. When water volume exchanges occur between adjacent triangular grid cells, the outflow water volume is allocated according to the proportion of the three types of source water volume in the outflow cell to the total water volume, and is written into the adjacent triangular grid cells respectively.

[0080] For each hydraulic saddle point, the three types of water sources within the corresponding topological edge mapping cell are read, and the local water level is calculated:

[0081] In the formula, Let be the local water level of the topological edge mapping cell corresponding to the hydraulic saddle point e at time t; This refers to the ground elevation of the corresponding topological edge mapping unit; , and These are the external water source volume, the internal water source volume, and the water source volume supported by the underground pipe network within the topological edge mapping unit, respectively. This represents the effective area of ​​the topological edge mapping unit.

[0082] When the water depth contribution and local flow rate of a certain type of water source both reach a preset threshold and are maintained for two consecutive coupling time slices, the moment when the condition is first met is determined as the corresponding water wave arrival time. The preset thresholds include an effective arrival water depth threshold and an effective arrival flow rate threshold, used to exclude instantaneous minor wetting caused by local grid interpolation. In a specific embodiment, the effective arrival water depth threshold is set to 0.005m. For each group of candidate rainfall processes and each hydraulic saddle point, the arrival times of the three types of water waves, the three types of water depth contributions, the local water level time series, and the non-arrival markers are recorded to form a water wave arrival record, which is output to S330 and S410.

[0083] S330: Read the water wave arrival record and local water level time series generated by S320. Within each coupled time slice, extract the overlapping time window backward from the current time. When the surface runoff water wave outside the station, the runoff water wave inside the station, and the backwater response wave from the underground pipe network all contribute water volume reaching the effective arrival water depth threshold and the effective arrival flow threshold within the overlapping time window, it is determined that the three types of water waves have concentrated superposition at the corresponding hydraulic saddle point, and a water wave overlap record is generated.

[0084] The overlap time window is set to an integer multiple of the coupling time slice length. In one specific embodiment, the coupling time slice length is set to 10 s, and the overlap time window is set to 60 s. The water wave overlap record includes at least the candidate rainfall event number, the hydraulic saddle point number, the arrival time of the three types of water waves, the water depth contribution of the three types, the overlap time window, the time of concentrated superposition, and the local water level after concentrated superposition.

[0085] Regardless of whether a water wave overlap record is generated, the local water level time series of the corresponding hydraulic saddle point is read. When the local water level first reaches or exceeds the control elevation, a topological edge exceedance record is generated. The topological edge exceedance record includes at least the candidate rainfall event number, hydraulic saddle point number, submerged topological edge number, the time of the first exceedance, the local water level at the time of the first exceedance, whether it was triggered by the concentrated superposition of the three types of water waves, the topological edge mapping unit number, and the local terrain detail window number.

[0086] If any one of the three types of water waves fails to reach the corresponding hydraulic saddle point before the simulation ends, no water wave overlap record is generated, but it is still determined whether a topological edge violation record is generated. This processing is used to retain cases where hydraulic saddle point violation is directly caused by a single source of water volume.

[0087] Output the water wave overlap record set to S410, S510 and S530; output the topological edge over-limit record set to S410.

[0088] S4 specifically includes the following sub-steps: S410 reads the topology edge violation record set and water wave overlap record set output by S330, the local water level time series output by S320, the submerged topology map output by S210, and the hydraulic saddle point set output by S130. Topology state switching events refer to the state update records formed when a submerged topology edge switches from an inactive state to an active state, or from an active state back to an inactive state, during the operation of the coupled dynamic model.

[0089] For each submerged topological edge, the local water level at the corresponding hydraulic saddle point is read according to the coupling time slice, and the state of the topological edge is determined:

[0090] In the formula, This represents the state of the submerged topological edge e within the m-th coupling time slice; Indicates an active state; Indicates an inactive state; This represents the local water level at hydraulic saddle point e. The control elevation of hydraulic saddle point e; This is the difference in water level during the decline; The number of consecutive coupling time slices required to determine the activation state; The number of consecutive coupling time slices required to determine the inactive state; This represents the topological edge state within the previous coupling time slice. The receding water level difference is used to prevent the topological edges from repeatedly switching when the local water level fluctuates near the control elevation. In one specific embodiment, the receding water level difference is set to 0.01m. and All are set to 2.

[0091] When the state of a submerged topological edge changes, a topological state transition event is generated. Each topological state transition event includes at least the candidate rainfall event number, the topological state transition event number, the submerged topological edge number, the hydraulic saddle point number, the topological edge mapping cell number, the local terrain detail window number, the topological edge state before and after the state transition, the state transition time, the local water level at the time of the state transition, the control elevation, the navigable direction of water flow, and the corresponding water wave overlap record number. The set of topological state transition events is output to S420 and S430.

[0092] S420 reads the set of topology state switching events output by S410 and the set of local mesh templates output by S220. Based on the local terrain detail window number in the topology state switching event, it locates the corresponding local mesh template. The local mesh template has a pre-embedded two-dimensional surface runoff model in S220, and triangular meshing is not repeated during operation.

[0093] When a submerged topology edge switches from an inactive state to an active state, the corresponding flow state template is invoked; when a submerged topology edge returns from an active state to an inactive state, the corresponding blocking state template is invoked. The blocking state template does not allow cross-edge water exchange at the corresponding hydraulic saddle point; the flow state template calculates the local flow rate based on the local water level, control elevation, and effective flow width on both sides of the hydraulic saddle point.

[0094] When embedding a local mesh template into a 2D surface runoff model for the first time, the volume of water accumulation within the local terrain detail window should be kept constant according to the following formula:

[0095] In the formula, This is the local terrain detail window corresponding to hydraulic saddle point e; and These represent the areas of mesh elements q before and after the local mesh template is embedded; and These represent the water depth of the grid cells q before and after the local grid template is embedded.

[0096] The local overcurrent of an active flooded topology edge is calculated using the following formula:

[0097] In the formula, Let be the local overflow of the submerged topological edge e at time t; This serves as an indicator of water flow direction; when water flows from the initial water accumulation node to the final water accumulation node... When flowing in the opposite direction, ; This refers to the local overcurrent coefficient. The effective flow width is g; g is the acceleration due to gravity. This refers to the local water level on the upstream side; This refers to the local water level on the downstream side. To control the elevation.

[0098] The local mesh template switching result and local overflow rate are output to S430. The local overflow rate is written as an inflow source term into the triangular mesh cell of the two-dimensional surface runoff model on the downstream side of the flooded topology edge in the next coupled time slice, and as an outflow sink term into the triangular mesh cell on the upstream side.

[0099] S430. Within each coupling time slice, the flooded topology graph is converted into a directed topology graph based on the local overflow direction of each flooded topology edge. Flooded topology edges that are active and whose absolute local overflow value reaches a preset path tracing flow threshold are identified as valid propagation edges.

[0100] The water accumulation propagation path refers to an ordered topological path that starts from the water source node, connects water accumulation nodes sequentially along the effective propagation edge, and finally reaches the associated node of the critical equipment area. The water source node is the water accumulation node that receives water from external sources, internal sources, or water from underground pipe networks and serves as the starting point of the path; the associated node of the critical equipment area is the water accumulation node that intersects with the boundary of the critical equipment area or contains grid cells of the critical equipment area.

[0101] Starting from each water source node, the water-collecting nodes in the directed topology graph are traversed level by level using either the Depth-First Search (DFS) or Breadth-First Search (BFS) algorithm, following the direction of the effective propagation edges.

[0102] When the traversal result reaches the associated node of the critical equipment area, record the water accumulation node number and the flooded topology edge number passed in sequence. When multiple water accumulation propagation paths arrive at the same critical equipment area at the same time, they are retained separately; when the same water accumulation propagation path remains valid in multiple consecutive coupling time slices, they are merged into one path duration period.

[0103] For each critical equipment area, calculate the water depth using the following formula:

[0104] In the formula, Let r be the water depth in the critical equipment area at time t; This refers to the set of triangular mesh elements corresponding to the critical equipment area r. Let q be the water depth of the triangular mesh element.

[0105] When the water depth in the critical equipment area reaches the risk water depth threshold for two consecutive coupled time slices, the moment when the condition is first met is recorded as the first water accumulation moment; the maximum water depth and water accumulation duration within the simulation period are recorded simultaneously. In one specific embodiment, the risk water depth threshold is set to 0.01m.

[0106] For each critical equipment area and each type of water source, calculate the cumulative inflow percentage:

[0107] In the formula, The percentage of cumulative inflow water corresponding to source type s in the critical equipment area r; This is the cumulative water volume of source type s entering the critical equipment area r during the simulated time period; It is a collection of source types, including sources outside the station, sources inside the station, and sources supported by underground pipelines.

[0108] For each group of candidate rainfall events, a set of candidate rainfall risk records is formed. Each candidate rainfall risk record includes at least the candidate rainfall event number, critical equipment area number, initial water accumulation time, maximum water accumulation depth, water accumulation duration, water accumulation propagation path number, water accumulation source node number, water accumulation node numbers included in the path, inundation topological edge number, activation time of each inundation topological edge, local flow direction, cumulative flow volume, corresponding water wave overlap record number, and the proportion of water volume from the three types of sources.

[0109] When the water depth in all critical equipment areas is below the preset residual water depth threshold, and no new topological edge violation record is formed within three consecutive coupled time slices, the simulation of the corresponding candidate rainfall process ends. The candidate rainfall risk record set is output to S510, S520, and S530.

[0110] S5 specifically includes the following sub-steps: S510 reads the candidate rainfall process set output by S310, the water wave overlap record set output by S330, the candidate rainfall risk record set output by S430, and the critical equipment area boundary set, critical equipment foundation elevation, and equipment ledger formed by S110. Based on the equipment ledger and substation operation and management data, an equipment importance coefficient is set for each critical equipment area; based on the project's flood control standards, critical equipment foundation elevation, and equipment operation requirements, a risk control water depth is set for each critical equipment area.

[0111] The most dangerous rainfall event is defined as the candidate rainfall event that causes water accumulation in at least one critical equipment area and has the highest overall risk value. To ensure comparability among the candidate rainfall events, a common evaluation duration is set for all candidate rainfall events. The evaluation duration covers the rainfall duration and the preset drainage continuation period.

[0112] For the k-th candidate rainfall event and the critical equipment area r, the comprehensive risk value is calculated according to the following formula:

[0113] In the formula, Let r be the comprehensive risk value of the critical equipment area r during the k-th candidate rainfall process; The equipment importance coefficient for critical equipment area r; This refers to the moment of initial flooding. The evaluation duration is used for all candidate rainfall events; This is the maximum water depth. To control the water depth for risk management; This refers to the duration of water accumulation. , and The evaluation weights are the initial water accumulation time, the maximum water accumulation depth, and the duration of water accumulation, respectively, and the sum of the three is 1. When no water accumulation occurs in the critical equipment area, the corresponding comprehensive risk value is recorded as 0.

[0114] The overall risk value of the k-th candidate rainfall event is determined according to the following formula:

[0115] In the formula, Let be the overall risk value of the k-th candidate rainfall event; This is a collection of critical equipment.

[0116] The candidate rainfall process with the highest overall risk value is identified as the most dangerous rainfall process. When the overall risk values ​​are the same, the maximum water depth, the time of the first water accumulation, and the duration of water accumulation are compared in that order. The most dangerous rainfall process number, overall risk value, number of the critical equipment area with the highest risk, number of water propagation path, number of hydraulic saddle point, number of water wave overlap record, and the proportion of water volume from the three sources are summarized to form the most dangerous rainfall process record, which is then output to S520 and S530.

[0117] S520. Read the record of the most dangerous rainfall process and the corresponding water propagation path. Backtrack the flooded topology edges along the water propagation path from the critical equipment area to the water source node, and extract the hydraulic saddle points corresponding to each flooded topology edge. Identify the hydraulic saddle points contained in all water propagation paths pointing to the critical equipment area during the most dangerous rainfall process as candidate critical hydraulic saddle points.

[0118] For each candidate critical hydraulic saddle point, while keeping other model parameters unchanged, the corresponding inundated topological edge is temporarily set to an inactive state, the most dangerous rainfall process is rerun, and the risk reduction contribution value is calculated:

[0119] In the formula, The contribution value for risk reduction of candidate key hydraulic saddle point e; This represents the overall risk value without any adjustment measures taken. This represents the overall risk value after the corresponding flooded topological edge remains inactive.

[0120] When a single candidate critical hydraulic saddle point can meet the risk control objective, the candidate critical hydraulic saddle point is determined as a critical hydraulic saddle point.

[0121] When a single candidate key hydraulic saddle point cannot meet the risk control objective, a full permutation combination or heuristic search algorithm is used to select candidate key hydraulic saddle points from different water propagation paths to form hydraulic blocking combinations, and counterfactual verification is performed on each hydraulic blocking combination.

[0122] Hydraulic saddle points within hydraulic blocking combinations that meet risk control objectives are identified as critical hydraulic saddle points. This approach is used to avoid overlooking water accumulation in critical equipment areas caused by multiple parallel water propagation paths.

[0123] For key hydraulic saddle points, local elevation or local blocking measures are generated. Local elevation measures refer to the treatment method of raising the control elevation of the hydraulic saddle point; local blocking measures refer to the treatment method of keeping the corresponding submerged topological edge in an inactive state within a preset water level range by setting up water-blocking structures.

[0124] The local elevation amount is determined according to the following formula:

[0125] In the formula, The local elevation of hydraulic saddle point e; This represents the maximum local water level on the upstream side of hydraulic saddle point e during the most dangerous rainfall process. For safety margin; This is the original control elevation. The adjusted control elevation is... In one specific embodiment, the safety margin is set to 0.05m.

[0126] Local blocking measures are only used for partial gaps in the perimeter wall or low points of non-essential passageways, and not for storm drains, drainage outlets, pumping stations, or entrances and exits where passage must be maintained. Strategy optimization constraints are established based on the substation master plan, maintenance access requirements, drainage facility records, and on-site measurement data. Hydraulic blocking combinations that satisfy the strategy optimization constraints are written into a copy of the coupled dynamic model to form a strategy optimization model set, which is then output to S530.

[0127] S530. Construct a strategy verification rainfall set. The strategy verification rainfall set includes the basic design rainfall process, the most dangerous rainfall process, and other candidate rainfall processes that can cause water accumulation in any critical equipment area. Input the strategy verification rainfall set into each strategy optimization model in sequence, recalculate the comprehensive risk value, the time of first water accumulation, the maximum water accumulation depth, the duration of water accumulation, and the water accumulation propagation path for all critical equipment areas, and check whether new topological edge violation records and new water accumulation propagation paths are formed.

[0128] For the strategy optimization model u, the residual risk value is calculated according to the following formula:

[0129] In the formula, The residual risk value corresponding to the strategy optimization model u; To validate the rainfall set for the strategy; This is a collection of key equipment areas; To verify the comprehensive risk value of critical equipment area r during rainfall in the k-th strategy optimization model u.

[0130] The corresponding hydraulic blocking combination is determined as a valid strategy optimization result only when the residual risk value does not exceed the allowable risk value set by the project, no new water propagation path is formed pointing to other critical equipment areas, the necessary drainage function is not damaged, and the strategy optimization constraints are met.

[0131] When multiple hydraulic barrier combinations meet the risk control objectives, priority should be given to combinations with fewer critical hydraulic saddle points, smaller total local elevation, and less impact on drainage facilities and maintenance access. For hydraulic barrier combinations not selected, the residual risk value, the reason for not being selected, and the corresponding new water propagation path should be retained for review by engineering personnel.

[0132] The final output includes the most dangerous rainfall process, key hydraulic saddle points, water propagation paths, effective hydraulic blocking combinations, local elevation, local blocking locations, comprehensive risk values ​​before and after strategy optimization, residual risk values, and verification results of newly added water propagation paths. These results are then overlaid onto a real-world 3D model for use in substation flood risk assessment, local renovation design, and early warning threshold configuration.

[0133] All the above formulas are performed using dimensionless numerical calculations; the relevant formulas are based on empirical models that approximate the real situation, obtained through extensive data collection and software simulation fitting. The preset parameters and thresholds involved in the formulas can be conventionally set and adjusted by those skilled in the art according to the physical constraints of the actual application scenario.

[0134] Those skilled in the art will recognize that the modules and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0135] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

[0136] In conclusion, the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for simulating substation flooding based on real-scene 3D feature recognition and model coupling, characterized in that, Includes the following steps: S1. Obtain real-world 3D data of the substation and catchment area, extract topographic elevation, water-blocking boundary, drainage facilities, pipeline elements and key equipment area boundary, delineate potential waterlogged depressions, and identify hydraulic saddle points that control the connectivity of adjacent potential waterlogged depressions. S2. Construct an inundation topology map based on potential waterlogged depressions and hydraulic saddles, establish a one-dimensional drainage model and a two-dimensional surface runoff model covering the in-station surface area and the out-of-station waterlogging area, set up model coupling ports, and form a coupled dynamic model. S3. Generate candidate rainfall processes under preset rainfall constraints, and track surface runoff waves outside the station, runoff waves inside the station and the top support response waves of underground pipe network based on the coupled dynamic model to generate water wave overlap records and topology boundary crossing records. S4. Update the flooded topological edge state based on the topological edge over-limit record, switch the corresponding local mesh template, calculate the local overflow, and trace the water propagation path to the critical equipment area.

2. The substation flood simulation method based on real-scene 3D feature recognition and model coupling as described in claim 1, characterized in that, Also includes: S5. Based on the comprehensive risk value of the key equipment area corresponding to each candidate rainfall process, select the most dangerous rainfall process, trace back the key hydraulic saddle points, generate and verify the hydraulic blocking combination, and output the strategy optimization results and the verification results of the newly added water accumulation propagation path.

3. The substation flood simulation method based on real-scene 3D feature recognition and model coupling as described in claim 1, characterized in that, S1 specifically includes: Acquire UAV oblique photography data of substations and catchment areas, 3D laser scanning point cloud data, drainage network as-built drawings, drainage pipeline survey results, equipment ledgers, substation general layout plan and on-site measurement data, and perform coordinate unification, outlier removal and feature extraction to form a unified real-scene 3D feature set; Based on the global terrain model and the local fine terrain model, the set water level is gradually increased, and a connected region search is performed on the surface units that are not isolated by water-blocking boundaries to form a set of potential waterlogged depressions; For adjacent potential waterlogged depressions, search for passable paths, determine the control elevation, hydraulic saddle point location and effective flow width, and establish a set of local terrain detail windows for S2 to call.

4. The substation flood simulation method based on real-scene 3D feature recognition and model coupling as described in claim 1, characterized in that, S2 specifically includes: Potential waterlogged depressions are set as waterlogged nodes, and hydraulic saddle points are set as inundation topological edges connecting adjacent waterlogged nodes, generating an inundation topology graph and topological edge mapping relationship; Based on the set of drainage facility elements, the set of pipe network elements, the global terrain model, the local fine terrain model, and the set of water-blocking boundaries, a one-dimensional drainage model and a two-dimensional surface runoff model are established, and local grid templates are pre-generated and embedded.

5. The substation flood simulation method based on real-scene 3D feature recognition and model coupling according to claim 4, characterized in that, Also includes: The rainwater inlets, inspection wells with surface overflow openings, pump pool openings connected to the surface, and drainage outlets within the model range are mapped as model coupling ports. The water volume is calculated and exchanged according to the water level relationship to form a coupled dynamic model, while retaining the topology edge state switching interface and the local mesh template switching interface.

6. The substation flooding simulation method based on real-scene 3D feature recognition and model coupling as described in claim 1, characterized in that, S3 specifically includes: Read the regional rainfall intensity formula, local historical data of rain stations and project flood control standards, adjust the time of rainfall peak occurrence, the ratio of previous and subsequent rainfall peaks and the interval between rainfall peaks based on the basic design rainfall process, perform total rainfall normalization and peak rainfall intensity verification, and form a set of candidate rainfall processes; Each candidate rainfall process is input into the coupled dynamic model. The external water source, the internal water source, and the water source from the underground pipe network are tracked according to the coupled time slice. The local water level of each hydraulic saddle point is calculated and a water wave arrival record is generated. Three types of concentrated superposition of water waves are identified within the overlapping time window, and water wave overlap records and topological edge violation records are generated respectively based on whether the local water level reaches the control elevation.

7. The substation flood simulation method based on real-scene 3D feature recognition and model coupling according to claim 1, characterized in that, S4 specifically includes: Read the topology edge over-limit records, local water level time series and flooded topology map, update the flooded topology edge status according to the control elevation, backwater level difference and continuous coupling time slice, and generate topology state switching events; The local mesh template is invoked based on the topology state switching event, switching between the blocking state template and the overflow state template, and the local overflow is calculated according to the local water level on both sides of the hydraulic saddle point, the control elevation, and the effective overflow width.

8. The substation flooding simulation method based on real-scene 3D feature recognition and model coupling according to claim 7, characterized in that, Also includes: A directed topology graph is constructed based on the direction of local overflow. The water propagation path from the water source node to the associated nodes in the critical equipment area is tracked. The time of the first water accumulation, the maximum water depth, the duration of water accumulation, and the proportion of water volume from the three types of sources are recorded to form a candidate rainfall risk record.

9. The substation flooding simulation method based on real-scene 3D feature recognition and model coupling according to claim 2, characterized in that, S5 specifically includes: Read the candidate rainfall process set, water wave overlap record and candidate rainfall risk record, calculate the comprehensive risk value of the key equipment area based on the first water accumulation time, maximum water accumulation depth, water accumulation duration and equipment importance coefficient, and screen the most dangerous rainfall process; By tracing back along the water propagation path corresponding to the most dangerous rainfall process, candidate key hydraulic saddle points are identified. Counterfactual verification is performed on each candidate key hydraulic saddle point and its combination to generate hydraulic blocking combinations and strategy optimization models that satisfy strategy optimization constraints.

10. The substation flood simulation method based on real-scene 3D feature recognition and model coupling according to claim 9, characterized in that, Also includes: Input the rainfall set for strategy verification into each strategy optimization model, calculate the residual risk value, verify the newly added waterlogging propagation path and necessary drainage functions, and output the strategy optimization results that meet the risk control objectives.

Images