A smart early warning system for water conservancy systems based on digital twin technology

Abstract

This invention discloses an intelligent early warning system for water conservancy systems based on digital twin technology, belonging to the field of water conservancy early warning technology. By constructing a network topographic twin model Twin with the unified support of a standardized time-series dataset Dset and running it in real time, the system enables flood prediction to go beyond traditional single-point water accumulation monitoring. Instead, it combines the dynamic comparison of inflow demand load Qdem and outflow capacity Qsup to identify the most vulnerable areas in advance and generate a bottleneck set Bset. This effectively solves the problem of insufficient coverage of potential flood points in existing technologies. Furthermore, this system uses the bottleneck set Bset as the overflow starting point to calculate and generate a water accumulation depth grid Hmap and map it to form a water accumulation road segment set Aset. This allows the system to clearly indicate the location, depth level, and duration of water accumulation on specific road segments during rainstorms, achieving full-chain coverage from network carrying capacity assessment to road risk early warning.

Description

Technical Field

[0001] This invention relates to the field of water conservancy early warning technology, specifically to an intelligent early warning system for water conservancy systems based on digital twin technology. Background Technology

[0002] With the continuous integration of information technology and water conservancy engineering, digital twin technology has been gradually introduced into water conservancy systems to construct simulation models that allow for interaction between virtual and real environments. At a more specific level, urban drainage issues within water conservancy systems are a key application of digital twins, especially during rapid urbanization, where urban underground drainage networks are becoming crucial infrastructure for flood control and disaster reduction. In urban heavy rainfall scenarios, relying solely on traditional water level monitoring at water accumulation points is insufficient to accurately predict the development trend of urban flooding. However, digital twin technology allows for the reconstruction of the coupling relationship between the drainage network's carrying capacity and road terrain in virtual space, thereby enabling early warning of flooded road sections.

[0003] Currently, in urban drainage and flood control management, a common approach is to deploy water level gauges or water accumulation sensors on certain road sections, underpasses, or key areas to monitor the water depth at individual points in real time. While this method has some reference value in obtaining local water conditions, it has significant drawbacks: First, the coverage of monitoring points is limited, making it impossible to create a comprehensive water accumulation distribution map. Second, single-point data cannot reflect the overall carrying capacity of the drainage network, nor can it dynamically simulate the flow of rainwater between different pipe sections. Third, when extreme rainstorms occur, water accumulation points are often not detected in time during their initial formation, resulting in a significant lag in flood warnings. Because of these shortcomings, existing systems often only issue alerts when water accumulation has already occurred and caused traffic congestion, lacking predictive capabilities. Therefore, it is necessary to use digital twins to couple and model the drainage network and road topography, thereby enabling early prediction of potential water accumulation sections. Summary of the Invention

[0004] To address the shortcomings of existing technologies, this invention provides an intelligent early warning system for water conservancy systems based on digital twin technology, which solves the problems mentioned in the background section.

[0005] To achieve the above objectives, the present invention is implemented through the following technical solution: an intelligent early warning system for water conservancy systems based on digital twin technology, comprising a data acquisition and processing module, a pipeline topographic twin construction module, a load carrying capacity assessment module, and a water accumulation distribution simulation and early warning module;

[0006] The data acquisition and processing module connects to urban meteorological stations and rainfall monitoring equipment to form a unified, standardized time-series dataset Dset;

[0007] The pipeline terrain twin construction module is based on the standardized time-series dataset Dset. It establishes a virtual model corresponding to the real urban drainage system in a virtual environment, encapsulates the spatial coupling relationship between roads and pipelines within the virtual system, and obtains the pipeline terrain twin model Twin.

[0008] The load carrying capacity assessment module calls the standardized time-series dataset Dset and the pipeline terrain twin model Twin to calculate the inflow demand load Qdem and outflow capacity Qsup, and aggregates them to form the bottleneck set Bset;

[0009] The water accumulation distribution simulation and early warning module uses the bottleneck set Bset as the overflow starting point. Under the framework of the pipeline topographic twin model Twin, it performs water flow diffusion and confluence simulation, gradually calculates the accumulation process of surface water on road units, generates a water accumulation depth grid Hmap, and maps it to generate a water accumulation road section set Aset for early warning.

[0010] Preferably, the data acquisition and processing module includes a multi-source hydrological and geographic data access unit and a time-series data standardization processing unit;

[0011] The multi-source hydrological and geographic data access unit connects to data sources from different sources to form a raw dataset;

[0012] Specifically, this includes: real-time collection of rainfall intensity sequences Rint and rainfall distribution grid Rmap from urban meteorological stations and rainfall monitoring equipment;

[0013] The rainfall intensity sequence Rint is the rainfall sequence per unit time;

[0014] The rainfall distribution grid Rmap represents two-dimensional grid data of rainfall at different spatial grid points within the urban area;

[0015] The topographic elevation data Gdem and the road depression distribution data Dlow were retrieved from urban surveying data.

[0016] The terrain elevation data Gdem represents the elevation grid data of the urban area;

[0017] The road depression distribution data Dlow represents the set of road depressions and waterlogging-prone areas extracted from the terrain elevation data Gdem.

[0018] Access the drainage network topology data Tnet, pipe diameter parameter set Ddia, slope parameter set Sgra, roughness parameter set Nman, and inlet capacity Cinl from the municipal drainage department;

[0019] The drainage network topology data Tnet represents the spatial connection relationship data of pipe sections, nodes, and inspection wells in the urban drainage system.

[0020] The pipe diameter parameter set Ddia represents the set of inner diameter dimensions of each pipe segment in the urban drainage system;

[0021] The slope parameter set Sgra represents the ratio of the longitudinal elevation difference to the length of each pipe segment in the urban drainage system.

[0022] The roughness parameter set Nman represents the pipe inner wall resistance coefficient, reflecting the degree of frictional resistance to water flow;

[0023] The inlet capacity Cinl refers to the maximum flow rate that rainwater inlets and inspection wells can draw into the pipe network per unit time.

[0024] Simultaneously, the external river water level sequence Wext is imported by combining monitoring stations of adjacent rivers.

[0025] Preferably, after receiving the original dataset, the time-series data standardization processing unit performs preprocessing on various types of data, including time-series alignment, spatial unification, and missing data correction.

[0026] Preprocessing specifically includes:

[0027] By using a time series resampling method, the rainfall intensity sequence Rint, the rainfall distribution grid Rmap, and the external river water level sequence Wext are aligned hourly to ensure that the monitoring frequency is consistent with the calculation step size;

[0028] Then, the geographic information system coordinate projection and spatial overlay method is used to unify the terrain elevation data Gdem, the road depression distribution data Dlow, and the drainage pipe network topology data Tnet into the same coordinate system, so that the twin model can overlay spatial information.

[0029] Meanwhile, time-series interpolation and parameter inversion correction methods are used to perform missing interpolation and error correction on the pipe diameter parameter set Ddia, slope parameter set Sgra, roughness parameter set Nman, and inlet capacity Cinl.

[0030] After preprocessing, the original dataset is normalized using the min-max normalization method, which transforms data from different sources and with different physical dimensions into comparable dimensionless sequences, thus obtaining the standardized time series dataset Dset.

[0031] Preferably, the pipeline terrain twin construction module includes a spatial coupling modeling unit and a dynamic twin generation unit;

[0032] After receiving the standardized time-series dataset Dset, the spatial coupling modeling unit extracts the drainage network topology data Tnet and the terrain elevation data Gdem. It then uses a three-dimensional geographic information modeling method to reconstruct the spatial skeleton of the drainage network in the virtual environment, which corresponds to the real-world environment. Through a spatial registration process, it establishes a vertical mapping relationship between the ground elevation of the network nodes and the grid elevation points in the terrain elevation data Gdem, so that the network nodes can form a consistent geometric association with the road surface elevation.

[0033] In particular, when performing 3D modeling based on drainage network topology data Tnet, the inspection wells, rainwater inlets and drainage wells of the network are abstracted into discrete network node objects Nset, and a network node object Nset layer is formed inside them; then, based on the pipeline connection relationship, all pipe segments are abstracted into network pipe segment objects Eset, and a pipe segment layer is formed in the form of a set inside the drainage network spatial skeleton.

[0034] In this process, the road depression distribution data Dlow is combined as an additional attribute of the road unit and written into the model, so that the road terrain features can form a computable spatial coupling with the pipeline overflow point in the virtual environment to obtain a virtual hydraulic terrain model.

[0035] Preferably, based on the virtual hydraulic terrain model generated by the dynamic twin generation unit, the rainfall intensity sequence Rint and the rainfall distribution grid Rmap are used as dynamic boundary condition inputs, and the water level, flow velocity and flow rate changes of the pipeline nodes and road grid points are calculated hourly using the hydrodynamic numerical simulation method, so that the static virtual hydraulic terrain model can obtain the dynamic operating characteristics that evolve over time.

[0036] By combining the water flow exchange results generated during the simulation with the road depression distribution data (Dlow), the virtual hydraulic topography model quantifies the accumulation and dissipation process of overflow water in low-lying areas, thereby reflecting the bidirectional hydraulic interaction between the surface and the pipe network. Then, by receiving real-time data streams from field sensors and monitoring equipment, the simulation results of the virtual hydraulic topography model are dynamically corrected through the model state assimilation method, realizing state synchronization between the virtual environment and the real system. This results in a virtual hydraulic topography model endowed with dynamic calculation capabilities and a real-time correction mechanism.

[0037] After completing dynamic driving, overflow evolution and state correction, the virtual hydraulic topography model is instantiated into a sustainable digital twin, forming the pipeline topography twin model Twin.

[0038] Preferably, the load capacity assessment module includes an inflow demand calculation unit and a load capacity determination unit;

[0039] The inflow demand calculation unit calls the standardized time-series dataset Dset and the pipeline terrain twin model Twin, and extracts the rainfall intensity sequence Rint, rainfall distribution grid Rmap and terrain elevation data Gdem from them;

[0040] First, based on the topographic elevation data Gdem, the entire target warning area is decomposed into multiple water catchment units Hset using the water catchment zoning method. Each water catchment unit corresponds to an inlet or road water collection point in the pipeline network system, and is bound to the corresponding node in the pipeline network topographic twin model Twin through spatial mapping relationship.

[0041] Subsequently, by combining the rainfall intensity sequence Rint and the rainfall distribution grid Rmap, the runoff of multiple catchment units Hset is calculated hourly using the hydrological runoff calculation method, and the calculation results are assigned to the corresponding pipe network node objects Nset to form dynamic attributes;

[0042] All pipeline node objects Nset together constitute the inflow demand load Qdem in the pipeline terrain twin model Twin, and the inflow demand load Qdem is bound to the pipeline node object Nset layer in the form of a time series.

[0043] Preferably, after calling the standardized time-series dataset Dset and the pipeline topographic twin model Twin, the bearing capacity determination unit uses the pipe diameter parameter set Ddia, slope parameter set Sgra, roughness parameter set Nman, inlet capacity Cinl, and external river water level sequence Wext as inputs, and uses the Manning hydraulic calculation method to calculate the ultimate water conveyance capacity of each pipe segment hourly in the pipe segment layer, forming the outflow capacity Qsup bound to the pipe segment layer;

[0044] At the same time, the inflow demand load Qdem bound to the pipeline node object Nset layer is automatically transferred to the adjacent pipeline segment according to the topological relationship and the conservation relationship, so as to obtain the demand flow Qreq bound to the pipeline segment layer;

[0045] Subsequently, the ratio of flow capacity Qsup to demand flow Qreq is calculated and defined as the capacity ratio Rcap. Then, based on the capacity ratio Rcap, overload and potential overflow points are marked to generate the bottleneck set Bset.

[0046] Preferably, the bottleneck set Bset is generated using the following determination method:

[0047] When the load ratio Rcap < 1, the corresponding pipeline segment is in an overloaded state, and the Nset layer position of the pipeline node objects associated with both ends of the segment is marked as a potential overflow point. Then, all marked overloaded pipeline segments and potential overflow nodes are aggregated in the entire time series dimension to generate a bottleneck set Bset.

[0048] When the load capacity ratio Rcap≥1, the water conveyance capacity of the corresponding pipe section meets the demand during the time period, and no overload or potential overflow point is marked.

[0049] Preferably, the water accumulation distribution simulation and early warning module includes a water accumulation simulation unit and an early warning generation unit;

[0050] The water accumulation simulation unit starts with the overload and overflow nodes identified in the bottleneck set Bset. Under the dynamic operation environment of the pipeline topographic twin model Twin, it calls the road low-lying distribution data Dlow and uses the surface water flow diffusion calculation method to simulate the migration, accumulation and dissipation process of overflow water on the road unit hourly.

[0051] During the simulation, the channels and locations where water flows converge are determined based on the Dlow data of road depression distribution, thereby dynamically calculating the spatial distribution of surface water on the road surface and obtaining the water depth grid Hmap.

[0052] Preferably, after receiving the water depth grid Hmap, the early warning generation unit maps and matches the grid data with the road space units. Based on the water depth of each road unit, the unit adopts a risk level classification method to convert different water depth values ​​into corresponding water levels and integrates the water information in the time dimension.

[0053] A set of flooded road sections, Aset, is formed, which includes road signs, water accumulation levels, and corresponding time periods. The set of flooded road sections, Aset, identifies the affected road sections and the severity of water accumulation, and is used by the water conservancy system to issue communication content for issuing warnings of urban flooding risks.

[0054] This invention provides an intelligent early warning system for water conservancy systems based on digital twin technology, which has the following beneficial effects:

[0055] (1) By constructing a network topographic twin model Twin under the unified support of the standardized time series dataset Dset and running it in real time, the prediction of waterlogging is no longer limited to the traditional single-point water accumulation monitoring. Instead, it can combine the dynamic comparison of inflow demand load Qdem and outflow capacity Qsup to identify the most vulnerable areas in advance and generate a bottleneck set Bset. This effectively solves the problem of insufficient coverage of potential waterlogging points in the existing technology. At the same time, this system further uses the bottleneck set Bset as the overflow starting point to calculate and generate a water accumulation depth grid Hmap and map it to form a waterlogged road section set Aset. This allows the system to clearly indicate the water accumulation location, water accumulation depth level and duration of specific road sections when rainstorms occur, achieving full-chain coverage from network carrying capacity assessment to road risk warning.

[0056] (2) The calculated outflow capacity Qsup is compared segment by segment at the pipe segment level to generate a bottleneck set Bset containing overload points and potential overflow points. This mechanism enables the failure risk of the drainage system to be located to specific pipe segments and their associated network nodes before the water flow overflows the surface, thus avoiding the lag situation in traditional drainage management where problems are only discovered after obvious water accumulation on the surface.

[0057] (3) In the Twin network topographic twin model, starting from the overloaded or overflow nodes marked by the bottleneck set Bset, and combining the road low-lying distribution data Dlow, the migration and accumulation process of overflow water on road units is deduced hourly, generating a water depth grid Hmap that accurately describes the evolution of surface water accumulation, and further mapping it to the road level to form a set Aset of waterlogged road sections containing road identification, water accumulation level, and corresponding time period. Unlike the existing method of issuing fuzzy early warnings based on single-point water level monitoring, this system can directly convert the failure information inside the pipeline network into water accumulation early warning results for specific roads, thereby realizing the intuitive transmission from underground pipeline network risk to surface road risk. Attached Figure Description

[0058] Figure 1 This is a schematic diagram of a block diagram of an intelligent early warning system for water conservancy systems based on digital twin technology according to the present invention. Detailed Implementation

[0059] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention.

[0060] Example 1

[0061] This invention provides an intelligent early warning system for water conservancy systems based on digital twin technology. Please refer to [link / reference]. Figure 1 It includes a data acquisition and processing module, a pipeline topography twin construction module, a load carrying capacity assessment module, and a water accumulation distribution simulation and early warning module;

[0062] The data acquisition and processing module connects to urban meteorological stations and rainfall monitoring equipment to form a unified, standardized time-series dataset Dset;

[0063] The pipeline terrain twin construction module is based on the standardized time-series dataset Dset. It establishes a virtual model corresponding to the real urban drainage system in a virtual environment, encapsulates the spatial coupling relationship between roads and pipelines within the virtual system, and obtains the pipeline terrain twin model Twin.

[0064] The load carrying capacity assessment module calls the standardized time-series dataset Dset and the pipeline terrain twin model Twin to calculate the inflow demand load Qdem and outflow capacity Qsup, and aggregates them to form the bottleneck set Bset;

[0065] The water accumulation distribution simulation and early warning module uses the bottleneck set Bset as the overflow starting point. Under the framework of the pipeline topographic twin model Twin, it performs water flow diffusion and confluence simulation, gradually calculates the accumulation process of surface water on road units, generates a water accumulation depth grid Hmap, and maps it to generate a water accumulation road section set Aset for early warning.

[0066] In this embodiment, by constructing and running a network topographic twin model Twin in real time with the unified support of the standardized time-series dataset Dset, the flood prediction is no longer limited to traditional single-point waterlogging monitoring. Instead, it can combine the dynamic comparison of inflow demand load Qdem and outflow capacity Qsup to identify the most vulnerable areas in advance and generate a bottleneck set Bset, effectively solving the problem of insufficient coverage of potential flood points in existing technologies. Furthermore, this system uses the bottleneck set Bset as the overflow starting point to calculate and generate a waterlogging depth grid Hmap and map it to form a waterlogged road segment set Aset. This allows for the clear identification of the waterlogging location, waterlogging depth level, and duration of specific road segments during heavy rain, achieving full-chain coverage from network capacity assessment to road risk warning. Therefore, it not only overcomes the limitation of existing technologies that "can only monitor water accumulation that has already occurred", but also provides early warning information that has direct guiding significance for road traffic and emergency drainage before water accumulation occurs. For example, before water accumulation occurs on urban main roads, the risk level can be released through the Aset collection of water accumulation sections, providing a basis for traffic management departments to implement closure and drainage scheduling in advance, thereby achieving proactive prediction and precise intervention of urban flooding risks.

[0067] Example 2

[0068] Specifically: the data acquisition and processing module includes a multi-source hydrological and geographic data access unit and a time-series data standardization processing unit;

[0069] The multi-source hydrological and geographic data access unit connects to data sources from different sources to form a raw dataset;

[0070] Specifically, this includes: real-time collection of rainfall intensity sequences Rint and rainfall distribution grid Rmap from urban meteorological stations and rainfall monitoring equipment;

[0071] The rainfall intensity sequence Rint is a sequence of rainfall amounts per unit time, typically measured in millimeters per minute or millimeters per hour;

[0072] The rainfall distribution grid Rmap represents two-dimensional grid data of rainfall at different spatial grid points within the urban area, which is specifically obtained through meteorological radar inversion and rain gauge interpolation analysis.

[0073] The topographic elevation data Gdem and the road depression distribution data Dlow were retrieved from urban surveying data.

[0074] The terrain elevation data Gdem refers to the elevation grid data of the urban area, which is specifically obtained through urban surveying and mapping departments, remote sensing image data, and UAV surveying results.

[0075] The road depression distribution data Dlow represents the set of road depressions and waterlogging-prone areas extracted from the terrain elevation data Gdem, specifically extracted through GIS terrain analysis and DEM (Digital Elevation Model).

[0076] Access the drainage network topology data Tnet, pipe diameter parameter set Ddia, slope parameter set Sgra, roughness parameter set Nman, and inlet capacity Cinl from the municipal drainage department;

[0077] The drainage network topology data Tnet represents the spatial connection relationship data of pipe sections, nodes and inspection wells in the urban drainage system, which is obtained through municipal drainage design drawings and underground pipe network surveys.

[0078] The pipe diameter parameter set Ddia represents the set of inner diameter dimensions of each pipe segment in the urban drainage system, which is specifically obtained through urban drainage system design drawings, as-built data, and pipe inspection robots.

[0079] The slope parameter set Sgra represents the ratio of the longitudinal elevation difference to the length of each pipe segment in the urban drainage system, which is specifically obtained through drainage network surveying and urban drainage system as-built data extraction.

[0080] The roughness parameter set Nman represents the pipe inner wall resistance coefficient, reflecting the degree of frictional resistance to water flow. It is specifically defined by the design specifications of each pipe section in the urban drainage system, including different materials such as concrete, PVC, and steel pipes.

[0081] The inlet capacity Cinl refers to the maximum flow rate that rainwater inlets and manholes can draw into the pipe network per unit time. Its theoretical upper limit is constrained by the diameter of the pipe section connected to it (i.e., the pipe diameter parameter set Ddia). However, in actual operation, Cinl is also affected by factors such as the manhole opening area, the siltation of the grid, the road slope, and the water depth at the manhole. Therefore, the actual value of Cinl is usually less than the water conveyance capacity corresponding to Ddia.

[0082] Simultaneously, the external river water level sequence Wext is imported by combining the monitoring stations of adjacent rivers, specifically representing the hourly water level observation sequence of adjacent rivers or drainage sections.

[0083] After receiving the original dataset, the time-series data standardization processing unit performs preprocessing on various types of data, including time-series alignment, spatial unification, and missing data correction, to ensure data integrity, computability, and cross-module consistency.

[0084] Preprocessing specifically includes:

[0085] By using time series resampling methods to achieve a unified time step for data from different sources, we can ensure that all monitoring data can be directly compared and calculated at a unified time scale. We align the rainfall intensity series Rint, the rainfall distribution grid Rmap, and the external river water level series Wext hourly to make the monitoring frequency consistent with the calculation step size.

[0086] Then, the geographic information system coordinate projection and spatial overlay method is used to unify the terrain and pipeline data from different sources into the same coordinate system and grid resolution to ensure spatial consistency during twin modeling. The terrain elevation data Gdem, road depression distribution data Dlow, and drainage pipeline topology data Tnet are unified into the same coordinate system, so that twin modeling can overlay spatial information.

[0087] Meanwhile, time-series interpolation and parameter inversion correction methods are used to complete and correct missing data and outliers. Missing parameters Ddia, slope parameters Sgra, roughness parameters Nman, and inlet capacity Cinl are interpolated and error corrected to avoid the impact of single-point observation bias on the overall calculation.

[0088] After preprocessing, the original dataset is normalized using the min-max normalization method, which transforms data from different sources and with different physical dimensions into comparable dimensionless sequences, obtaining a standardized time-series dataset Dset. This dataset is then used as the sole result of cross-module interaction for the pipeline terrain twin construction module and the load-bearing assessment module to call, thereby ensuring the simplicity, traceability, and stability of the entire system's calculation process.

[0089] In this embodiment, the previously scattered multi-source data, including rainfall intensity sequence Rint, rainfall distribution grid Rmap, topographic elevation data Gdem, road depression distribution data Dlow, drainage network topology data Tnet, pipe diameter parameter set Ddia, slope parameter set Sgra, roughness parameter set Nman, inlet capacity Cinl, and external river water level sequence Wext, are unified into a standardized time-series dataset Dset. This effectively eliminates problems such as inconsistent monitoring frequencies, different coordinate systems, missing parameters, and errors. During heavy rain, traditional monitoring systems often struggle to directly integrate pipeline network calculations with external river scheduling because rainfall monitoring data is at the minute level and river water level data is at the hour level. However, this system, through time-series resampling and min-max normalization methods, makes data with different time steps and dimensions comparable within the same computational framework, thereby ensuring the stability of twin computation and seamless integration of cross-departmental data. As a result, municipal management departments can directly conduct urban flooding simulation and drainage scheduling with the support of a unified standardized time-series dataset Dset, avoiding delays or errors caused by differences in data formats and units in the past. For example, it can quickly compare the risk of water accumulation in low-lying areas of the city with the changing trend of water levels in external rivers at the beginning of a rainstorm, providing a real-time and executable data foundation for cross-departmental joint drainage.

[0090] Example 3

[0091] Specifically: the pipeline terrain twin construction module includes a spatially coupled modeling unit and a dynamic twin generation unit;

[0092] After receiving the standardized time-series dataset Dset, the spatial coupling modeling unit extracts the drainage network topology data Tnet and the terrain elevation data Gdem. It then uses a three-dimensional geographic information modeling method to reconstruct the spatial skeleton of the drainage network in the virtual environment, which corresponds to the real-world environment. Through a spatial registration process, it establishes a vertical mapping relationship between the ground elevation of the network nodes and the grid elevation points in the terrain elevation data Gdem, so that the network nodes can form a consistent geometric association with the road surface elevation.

[0093] In particular, when performing 3D modeling based on drainage network topology data Tnet, the inspection wells, rainwater inlets, and drainage wells of the network are abstracted into discrete network node objects Nset, and a network node object Nset layer is formed inside them; then, based on the pipe connectivity, all pipe segments are abstracted into network pipe segment objects Eset, and a pipe segment layer is formed in the form of a set inside the drainage network spatial skeleton, thereby establishing a two-layer structure of nodes and pipe segments in the drainage system twin;

[0094] In this process, the road depression distribution data Dlow is combined as an additional attribute of the road unit and written into the model, so that the road terrain features can form a computable spatial coupling with the overflow point of the pipeline network in the virtual environment, obtain a virtual hydraulic terrain model, and provide a basis for the interaction between terrain and pipeline network for dynamic twin computing.

[0095] It should be noted that: the network node object Nset in the model acts as a spatial inlet or exchange port, used to receive the inflow demand load calculated by the water catchment unit Hset, and to transfer water volume at the connection with the pipe segment; the pipe segment layer is used to represent the hydraulic bearing unit of each pipe in the drainage network. Each pipe segment object Eset is connected to the network node objects Nset at both ends and inherits its geometric parameters (including the pipe diameter parameter set Ddia, the slope parameter set Sgra, and the roughness parameter set Nman). At the same time, boundary conditions such as the inlet capacity Cinl and the external river water level sequence Wext can be superimposed. During operation, the pipe segment layer not only receives the inflow demand load Qdem transferred from the network node object Nset layer, but also stores the outflow capacity Qsup obtained by hydraulic calculation, thus becoming the direct calculation layer for bearing capacity determination and bottleneck identification.

[0096] Based on the virtual hydraulic terrain model generated by the dynamic twin generation unit, the rainfall intensity sequence Rint and the rainfall distribution grid Rmap are used as dynamic boundary condition inputs. The water level, flow velocity and flow rate changes of the pipeline nodes and road grid points are calculated hourly using the hydrodynamic numerical simulation method, so that the static virtual hydraulic terrain model can obtain the dynamic operation characteristics that evolve over time.

[0097] By combining the water flow exchange results generated during the simulation with the road depression distribution data (Dlow), the virtual hydraulic topography model quantifies the accumulation and dissipation process of overflow water in low-lying areas, thereby reflecting the bidirectional hydraulic interaction between the surface and the pipe network. Then, by receiving real-time data streams from field sensors and monitoring equipment, the simulation results of the virtual hydraulic topography model are dynamically corrected through the model state assimilation method, realizing state synchronization between the virtual environment and the real system. This results in a virtual hydraulic topography model endowed with dynamic calculation capabilities and a real-time correction mechanism.

[0098] After completing dynamic driving, overflow evolution and state correction, the virtual hydraulic topography model is instantiated into a sustainable digital twin, forming the pipeline topography twin model Twin. Through this process, the pipeline topography twin model Twin has the characteristics of "static spatial relationship + dynamic hydraulic driving + real-time state synchronization", realizing the step-by-step generation process from static virtual skeleton to dynamic operating twin, providing a calculable and traceable operating environment for the subsequent load bearing assessment module and water accumulation simulation and early warning module.

[0099] It needs to be clarified that:

[0100] Virtual hydraulic terrain model: At the static level, it is only used to describe the spatial coupling relationship between roads and pipelines, and is the skeleton framework in the digital environment;

[0101] Twin network topography twin model: Based on the virtual hydraulic-topography model, a dynamic simulation mechanism, monitoring data assimilation mechanism and operation instantiation mechanism are superimposed to generate an operational model that can evolve in real time with rainfall input and map the operational status of the real drainage system.

[0102] In this embodiment, supported by the standardized time-series dataset Dset, the drainage network topology data Tnet and the terrain elevation data Gdem are unified into a three-dimensional virtual environment, establishing a two-layer spatial skeleton composed of a network node object layer Nset and a pipe segment layer. Furthermore, driven by the rainfall intensity sequence Rint and the rainfall distribution grid Rmap, the virtual hydraulic terrain model possesses dynamic calculation capabilities. Combined with a model state assimilation method and real-time monitoring data, a twin model of the network terrain (Twin) capable of mapping the actual operational state is finally generated. This process not only ensures the geometric coupling between the drainage network and the road terrain but also enables the complete quantification of the hydraulic interaction between the network and the road, thus solving the common deficiency in existing drainage simulations where "static modeling results cannot be dynamically tracked." During actual heavy rain, when a storm drain in a city experiences localized flooding due to limited inlet capacity (Cinl), the Twin model of the pipeline network terrain in this system can reflect the dynamic changes in the water conveyance capacity of the pipeline section and the water catchment of low-lying points on the road in real time. This helps management departments identify risk locations before the floodwater spreads. For example, when signs of flooding appear on densely populated roads in commercial areas, drainage vehicles and temporary pumping stations can be deployed in advance, achieving a practical effect of shifting from post-event emergency response to pre-event intervention.

[0103] Example 4

[0104] Specifically: the load capacity assessment module includes an inflow demand calculation unit and a load capacity determination unit;

[0105] The inflow demand calculation unit calls the standardized time-series dataset Dset and the pipeline terrain twin model Twin, and extracts the rainfall intensity sequence Rint, rainfall distribution grid Rmap and terrain elevation data Gdem from them;

[0106] First, based on the topographic elevation data Gdem, the entire target warning area is decomposed into multiple water catchment units Hset using the water catchment zoning method. Each water catchment unit corresponds to an inlet or road water collection point in the pipeline network system, and is bound to the corresponding node in the pipeline network topographic twin model Twin through spatial mapping relationship.

[0107] Subsequently, by combining the rainfall intensity sequence Rint and the rainfall distribution grid Rmap, the runoff of multiple catchment units Hset is calculated hourly using the hydrological runoff calculation method, and the calculation results are assigned to the corresponding pipe network node objects Nset to form dynamic attributes;

[0108] All pipeline node objects Nset together constitute the inflow demand load Qdem in the pipeline terrain twin model Twin. The inflow demand load Qdem is bound to the pipeline node object Nset layer in the form of a time series and is used to compare the outflow capacity node by node in the subsequent load determination.

[0109] After calling the standardized time-series dataset Dset and the pipeline topography twin model Twin, the bearing capacity determination unit takes the pipe diameter parameter set Ddia, the slope parameter set Sgra, the roughness parameter set Nman, the inlet capacity Cinl, and the external river water level sequence Wext as inputs, and uses the Manning hydraulic calculation method to calculate the ultimate water conveyance capacity of each pipe segment hourly in the pipe segment layer, forming the outflow capacity Qsup bound to the pipe segment layer;

[0110] At the same time, the inflow demand load Qdem bound to the pipeline node object Nset layer is automatically transferred to the adjacent pipeline segment according to the topological relationship and the conservation relationship, so as to obtain the demand flow Qreq bound to the pipeline segment layer, so that the demand and capacity at each time step can be directly compared at the same spatial level.

[0111] Subsequently, the ratio of flow capacity Qsup to demand flow Qreq is calculated and defined as the capacity ratio Rcap. Then, based on the capacity ratio Rcap, overload and potential overflow points are marked to generate the bottleneck set Bset.

[0112] The bottleneck set Bset is generated using the following criteria:

[0113] When the load ratio Rcap < 1, the corresponding pipeline segment is in an overloaded state, and the Nset layer position of the pipeline node objects associated with both ends of the segment is marked as a potential overflow point. Then, all marked overloaded pipeline segments and potential overflow nodes are aggregated in the entire time series dimension to generate a bottleneck set Bset.

[0114] When the load capacity ratio Rcap≥1, the water conveyance capacity of the corresponding pipe section meets the demand during the time period, and no overload or potential overflow point is marked.

[0115] In this embodiment, the rainfall intensity sequence Rint, rainfall distribution grid Rmap, and topographic elevation data Gdem are transformed into inflow demand load Qdem, which is bound to the pipeline node object Nset layer. This load is then used to generate demand flow Qreq under topological relationships. This demand flow is then compared segment by segment at the pipe segment level with the outflow capacity Qsup calculated based on the pipe diameter parameter set Ddia, slope parameter set Sgra, roughness parameter set Nman, inlet capacity Cinl, and external river water level sequence Wext. This generates a bottleneck set Bset containing overload points and potential overflow points. This mechanism allows the failure risk of the drainage system to be located at specific pipe segments and their associated pipeline nodes before water overflows the surface, thus avoiding the delayed situation in traditional drainage management where problems are only discovered after significant surface water accumulation. In practical applications, for example, when an old drainage network in a certain area cannot withstand the runoff of rainwater due to a small pipe diameter parameter set Ddia, this system can directly mark the pipe section as overloaded by determining the load-bearing ratio Rcap when the water level is still below the wellhead, and generate a bottleneck set Bset. This allows the municipal department to deploy pumping equipment or start diversion measures in advance, thereby significantly reducing the risk of the entire road being forced to close due to sudden overflow of the pipe section.

[0116] Example 5

[0117] Specifically: the water accumulation distribution simulation and early warning module includes a water accumulation simulation unit and an early warning generation unit;

[0118] The water accumulation simulation unit starts with the overload and overflow nodes identified in the bottleneck set Bset. Under the dynamic operation environment of the pipeline topographic twin model Twin, it calls the road low-lying distribution data Dlow and uses the surface water flow diffusion calculation method to simulate the migration, accumulation and dissipation process of overflow water on the road unit hourly.

[0119] During the simulation, the channels and locations where water flows converge are determined based on the road low-lying distribution data Dlow, thereby dynamically calculating the spatial distribution of surface water on the road surface and obtaining the water depth grid Hmap. The water depth grid Hmap records the water depth values ​​at different locations at each time step and serves as a direct input for subsequent risk warnings.

[0120] After receiving the water depth grid Hmap, the early warning generation unit maps and matches the grid data with the road space units. Based on the water depth of each road unit, the unit adopts a risk level classification method to convert different water depth values ​​into corresponding water levels and integrates the water information in the time dimension.

[0121] The risk level classification method is based on the water depth values ​​in the water depth grid Hmap, and sets multiple threshold levels according to the degree of impact on road traffic and pedestrian safety.

[0122] When the water depth is ≤0.10m, it is classified as low-level water accumulation and converted to water accumulation level I, indicating that vehicles and pedestrians can still pass, but caution is advised.

[0123] When the water depth is between 0.10m and 0.30m, it is classified as medium-level waterlogging and converted to waterlogging level II, indicating that small vehicles are restricted from passing and pedestrians are at risk of safety hazards.

[0124] When the water depth is between 0.30m and 0.50m, it is classified as high-level waterlogging, which is converted to waterlogging level III, indicating that most vehicles will have difficulty passing and traffic control measures need to be implemented.

[0125] When the water depth is greater than 0.50m, it is classified as severe waterlogging, which is converted to waterlogging level IV, indicating that the road is completely unusable and needs to be closed urgently and emergency drainage should be carried out.

[0126] A set of flooded road sections, Aset, is formed, which includes road signs, water accumulation levels, and corresponding time periods. The set of flooded road sections, Aset, identifies the affected road sections and the severity of water accumulation, and is used by the water conservancy system to issue communication content for issuing warnings of urban flooding risks.

[0127] In this embodiment, in the Twin network topographic twin model, starting with the overloaded or overflow nodes marked by the bottleneck set Bset, and combining the road low-lying distribution data Dlow, the migration and accumulation process of overflow water on road units is simulated hourly, generating a water depth grid Hmap that accurately describes the evolution of surface water accumulation. This Hmap is then further mapped to the road level, forming a set Aset of waterlogged road sections containing road identification, water accumulation level, and corresponding time period. Unlike existing methods that can only issue fuzzy warnings based on single-point water level monitoring, this system can directly transform network failure information into specific road water accumulation warning results, thus achieving a direct transmission of "risk from underground pipelines to surface roads." In real-world applications, when heavy rainfall causes water to gradually deepen in low-lying sections of urban main roads, this system can not only provide a detailed spatial distribution through the water depth grid Hmap, but also clearly mark the water level and possible duration of the waterlogged sections in the waterlogged section set Aset. This allows traffic management departments to issue detour guidance or implement temporary closures in advance before the road surface becomes severely flooded, thus avoiding situations where vehicles are trapped and traffic is completely paralyzed.

[0128] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and variations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A smart early warning system for water conservancy systems based on digital twin technology, characterized in that: It includes a data acquisition and processing module, a pipeline topography twin construction module, a load carrying capacity assessment module, and a water accumulation distribution simulation and early warning module; The data acquisition and processing module connects to urban meteorological stations and rainfall monitoring equipment to form a unified, standardized time-series dataset Dset; The pipeline terrain twin construction module is based on the standardized time-series dataset Dset. It establishes a virtual model corresponding to the real urban drainage system in a virtual environment, encapsulates the spatial coupling relationship between roads and pipelines within the virtual system, and obtains the pipeline terrain twin model Twin. The load carrying capacity assessment module calls the standardized time-series dataset Dset and the pipeline terrain twin model Twin to calculate the inflow demand load Qdem and outflow capacity Qsup, and aggregates them to form the bottleneck set Bset; The load capacity assessment module includes an inflow demand calculation unit and a load capacity determination unit; After calling the standardized time-series dataset Dset and the pipeline topography twin model Twin, the bearing capacity determination unit takes the pipe diameter parameter set Ddia, the slope parameter set Sgra, the roughness parameter set Nman, the inlet capacity Cinl, and the external river water level sequence Wext as inputs, and uses the Manning hydraulic calculation method to calculate the ultimate water conveyance capacity of each pipe segment hourly in the pipe segment layer, forming the outflow capacity Qsup bound to the pipe segment layer; At the same time, the inflow demand load Qdem bound to the pipeline node object Nset layer is automatically transferred to the adjacent pipeline segment according to the topological relationship and the conservation relationship, so as to obtain the demand flow Qreq bound to the pipeline segment layer; Subsequently, the ratio of flow capacity Qsup to demand flow Qreq is calculated and defined as the capacity ratio Rcap. Then, based on the capacity ratio Rcap, overload and potential overflow points are marked to generate the bottleneck set Bset. The bottleneck set Bset is generated using the following criteria: When the load-bearing ratio Rcap < 1, the corresponding pipeline segment is in an overloaded state, and the Nset layer position of the pipeline node objects associated with both ends of the segment is marked as a potential overflow point. Subsequently, all marked overloaded pipe segments and potential overflow nodes are aggregated along the entire time series to generate a bottleneck set Bset; When the load-bearing ratio Rcap≥1, the water conveyance capacity of the corresponding pipe section meets the demand during the time period, and no overload or potential overflow point is marked. The water accumulation distribution simulation and early warning module uses the bottleneck set Bset as the overflow starting point. Under the framework of the pipeline topographic twin model Twin, it performs water flow diffusion and confluence simulation, gradually calculates the accumulation process of surface water on road units, generates a water accumulation depth grid Hmap, and maps it to generate a water accumulation road section set Aset for early warning.

2. The intelligent early warning system for water conservancy systems based on digital twin technology according to claim 1, characterized in that: The data acquisition and processing module includes a multi-source hydrological and geographic data access unit and a time-series data standardization processing unit. The multi-source hydrological and geographic data access unit connects to data sources from different sources to form a raw dataset; Specifically, this includes: real-time collection of rainfall intensity sequences Rint and rainfall distribution grid Rmap from urban meteorological stations and rainfall monitoring equipment; The rainfall intensity sequence Rint is the rainfall sequence per unit time; The rainfall distribution grid Rmap represents two-dimensional grid data of rainfall at different spatial grid points within the urban area; The topographic elevation data Gdem and the road depression distribution data Dlow were retrieved from urban surveying data. The terrain elevation data Gdem represents the elevation grid data of the urban area; The road depression distribution data Dlow represents the set of road depressions and waterlogging-prone areas extracted from the terrain elevation data Gdem. Access the drainage network topology data Tnet, pipe diameter parameter set Ddia, slope parameter set Sgra, roughness parameter set Nman, and inlet capacity Cinl from the municipal drainage department; The drainage network topology data Tnet represents the spatial connection relationship data of pipe sections, nodes, and inspection wells in the urban drainage system. The pipe diameter parameter set Ddia represents the set of inner diameter dimensions of each pipe segment in the urban drainage system; The slope parameter set Sgra represents the ratio of the longitudinal elevation difference to the length of each pipe segment in the urban drainage system. The roughness parameter set Nman represents the pipe inner wall resistance coefficient, reflecting the degree of frictional resistance to water flow; The inlet capacity Cinl refers to the maximum flow rate that rainwater inlets and inspection wells can draw into the pipe network per unit time. Simultaneously, the external river water level sequence Wext is imported by combining monitoring stations of adjacent rivers.

3. The intelligent early warning system for water conservancy systems based on digital twin technology according to claim 2, characterized in that: After receiving the original dataset, the time-series data standardization processing unit performs preprocessing on various types of data, including time-series alignment, spatial unification, and missing data correction. Preprocessing specifically includes: By using a time series resampling method, the rainfall intensity sequence Rint, the rainfall distribution grid Rmap, and the external river water level sequence Wext are aligned hourly to ensure that the monitoring frequency is consistent with the calculation step size; Then, the geographic information system coordinate projection and spatial overlay method is used to unify the terrain elevation data Gdem, the road depression distribution data Dlow, and the drainage pipe network topology data Tnet into the same coordinate system, so that the twin model can overlay spatial information. Meanwhile, time-series interpolation and parameter inversion correction methods are used to perform missing interpolation and error correction on the pipe diameter parameter set Ddia, slope parameter set Sgra, roughness parameter set Nman, and inlet capacity Cinl. After preprocessing, the original dataset is normalized using the min-max normalization method, which transforms data from different sources and with different physical dimensions into comparable dimensionless sequences, thus obtaining the standardized time series dataset Dset.

4. The intelligent early warning system for water conservancy systems based on digital twin technology according to claim 3, characterized in that: The pipeline terrain twin construction module includes a spatially coupled modeling unit and a dynamic twin generation unit; After receiving the standardized time-series dataset Dset, the spatial coupling modeling unit extracts the drainage network topology data Tnet and the terrain elevation data Gdem. It then uses a three-dimensional geographic information modeling method to reconstruct the spatial skeleton of the drainage network in the virtual environment, which corresponds to the real-world environment. Through a spatial registration process, it establishes a vertical mapping relationship between the ground elevation of the network nodes and the grid elevation points in the terrain elevation data Gdem, so that the network nodes can form a consistent geometric association with the road surface elevation. In particular, when performing 3D modeling based on drainage network topology data Tnet, the inspection wells, rainwater inlets and drainage wells of the network are abstracted into discrete network node objects Nset, and a network node object Nset layer is formed inside them; then, based on the pipeline connection relationship, all pipe segments are abstracted into network pipe segment objects Eset, and a pipe segment layer is formed in the form of a set inside the drainage network spatial skeleton. In this process, the road depression distribution data Dlow is combined as an additional attribute of the road unit and written into the model, so that the road terrain features can form a computable spatial coupling with the pipeline overflow point in the virtual environment to obtain a virtual hydraulic terrain model.

5. The intelligent early warning system for water conservancy systems based on digital twin technology according to claim 4, characterized in that: Based on the virtual hydraulic terrain model generated by the dynamic twin generation unit, the rainfall intensity sequence Rint and the rainfall distribution grid Rmap are used as dynamic boundary condition inputs. The water level, flow velocity and flow rate changes of the pipeline nodes and road grid points are calculated hourly using the hydrodynamic numerical simulation method, so that the static virtual hydraulic terrain model can obtain the dynamic operation characteristics that evolve over time. By combining the water flow exchange results generated during the simulation with the road depression distribution data (Dlow), the virtual hydraulic topography model quantifies the accumulation and dissipation process of overflow water in low-lying areas, thereby reflecting the bidirectional hydraulic interaction between the surface and the pipe network. Then, by receiving real-time data streams from field sensors and monitoring equipment, the simulation results of the virtual hydraulic topography model are dynamically corrected through the model state assimilation method, realizing state synchronization between the virtual environment and the real system. This results in a virtual hydraulic topography model endowed with dynamic calculation capabilities and a real-time correction mechanism. After completing dynamic driving, overflow evolution and state correction, the virtual hydraulic topography model is instantiated into a sustainable digital twin, forming the pipeline topography twin model Twin.

6. The intelligent early warning system for water conservancy systems based on digital twin technology according to claim 5, characterized in that: The inflow demand calculation unit calls the standardized time-series dataset Dset and the pipeline terrain twin model Twin, and extracts the rainfall intensity sequence Rint, rainfall distribution grid Rmap and terrain elevation data Gdem from them; First, based on the topographic elevation data Gdem, the entire target warning area is decomposed into multiple water catchment units Hset using the water catchment zoning method. Each water catchment unit corresponds to an inlet or road water collection point in the pipeline network system, and is bound to the corresponding node in the pipeline network topographic twin model Twin through spatial mapping relationship. Subsequently, by combining the rainfall intensity sequence Rint and the rainfall distribution grid Rmap, the runoff of multiple catchment units Hset is calculated hourly using the hydrological runoff calculation method, and the calculation results are assigned to the corresponding pipe network node objects Nset to form dynamic attributes; All pipeline node objects Nset together constitute the inflow demand load Qdem in the pipeline terrain twin model Twin, and the inflow demand load Qdem is bound to the pipeline node object Nset layer in the form of a time series.

7. The intelligent early warning system for water conservancy systems based on digital twin technology according to claim 6, characterized in that: The water accumulation distribution simulation and early warning module includes a water accumulation simulation unit and an early warning generation unit; The water accumulation simulation unit starts with the overload and overflow nodes identified in the bottleneck set Bset. Under the dynamic operation environment of the pipeline topographic twin model Twin, it calls the road low-lying distribution data Dlow and uses the surface water flow diffusion calculation method to simulate the migration, accumulation and dissipation process of overflow water on the road unit hourly. During the simulation, the channels and locations where water flows converge are determined based on the Dlow data of road depression distribution, thereby dynamically calculating the spatial distribution of surface water on the road surface and obtaining the water depth grid Hmap.

8. The intelligent early warning system for water conservancy systems based on digital twin technology according to claim 7, characterized in that: After receiving the water depth grid Hmap, the early warning generation unit maps and matches the grid data with the road space units. Based on the water depth of each road unit, the unit adopts a risk level classification method to convert different water depth values ​​into corresponding water levels and integrates the water information in the time dimension. A set of flooded road sections, Aset, is formed, which includes road signs, water accumulation levels, and corresponding time periods. The set of flooded road sections, Aset, identifies the affected road sections and the severity of water accumulation, and is used by the water conservancy system to issue communication content for issuing warnings of urban flooding risks.

Images