Inspection well overflow point forecasting and early warning method and system based on digital twinning

Abstract

The invention discloses an inspection well overflow point forecasting and early warning method and system based on digital twinborn. The system comprises a city sensing module, a digital twinborn body module, an inspection well forecasting module and a flood risk early warning module. The method comprises the following steps: firstly, perfecting a city knowledge graph to construct a city perception module; constructing a digital twin city; constructing an inspection well forecasting module; and constructing a flood risk early warning module. Based on a digital twin platform, through an Internet of Things network and a data flow processing framework, an urban hydrological hydrodynamic model is coupled, parameters of the coupled model are dynamically calibrated, and an inspection well is finely simulated; the method comprehensively displays the multi-temporal-spatial-scale urban development state flood inundation condition, tracks and reflects the actual operation state of an urban drainage system, dynamically evaluates the risk level and carries out real-time early warning, improves the urban emergency management response efficiency, and provides support for smooth development of urban flood control and drainage work.

Description

Technical Field

[0001] The present invention relates to the field of urban flood numerical simulation, and in particular to a method and system for forecasting and warning overflow points of inspection wells based on digital twins. Background Art

[0002] Against the backdrop of global warming and rapid urbanization, rainfall and flood patterns in urban areas are constantly changing. Hydrological and hydrodynamic models based on numerical simulation models have been widely used to simulate urban flood processes under extreme rainfall. Compared with natural watersheds, cities are not simple catchment areas. The complex underlying surface in cities makes urban flood simulation more uncertain. There are three main aspects: first, the uncertainty of model input. During the monitoring process of rainfall data and river flow data, monitoring equipment failures often lead to data anomalies and data loss. Underground pipe network data is difficult to monitor due to the large amount of information. Second, the uncertainty of model structure. The urban underlying surface is complex and diverse. The addition of dense buildings, roads and underground pipe networks has changed the way of natural runoff and the hydraulic exchange conditions are complex. Finally, the uncertainty of model parameters. The initial parameters of the model are mainly set by empirical values. The consistency of these parameters with the measured data directly affects the accuracy of the model results. In model calculation, model parameter calibration can reduce the difference between simulated values ​​and measured values, further improve the simulation accuracy, and thus objectively affect the overflow of manholes and the distribution of urban flooding. At present, parameter calibration aims to minimize the error between the flood process and the measured process, and optimizes the algorithm. This method simulates that the model parameters do not change during the rainfall and flood process, and systematically ignores the dynamic change process of the urban underlying surface. It is difficult to overcome the problem that the historical data simulation is good but the performance is poor in the prediction process. After the construction and calibration of the traditional urban flood model, it is difficult to adjust the model according to the development and changes of the city. It is impossible to realize the visualization of the city status at any spatial scale based on the actual monitoring situation and future meteorological forecast data. It is difficult to accurately and efficiently reflect the dynamic changes in the city and centrally manage urban flood disasters. Summary of the invention

[0003] The purpose of the present invention is to address the limitations and shortcomings of existing urban flood simulation models in capturing the dynamic characteristics of urban changes. The present invention proposes a method and system for forecasting and warning of manhole overflow points based on digital twin technology. This method solves the problem that traditional models cannot accurately reflect the current and future status of the city in a timely manner due to differences in temporal and spatial scales. By integrating a real-time transmission perception module, accurate correction of the urban flood model is achieved, thereby refining the manhole simulation and the response speed of the model. In addition, the invention normalizes the flood disaster risk based on the overflow water level of the manhole, dynamically grades the flood risk and issues early warning decisions, providing a scientific basis and technical support for the formulation of urban flood control and drainage strategies.

[0004] In order to achieve the above object, the technical solution adopted by the present invention is as follows:

[0005] A method for predicting and warning overflow points of inspection wells based on digital twins includes the following steps:

[0006] S100, first improve the city knowledge graph and build the city perception module;

[0007] S200, building a digital twin city;

[0008] S300, constructing an inspection well prediction module;

[0009] S400. Construct flood risk warning module.

[0010] In some specific embodiments, step S100 specifically includes the following steps:

[0011] Step S110, real-time dynamic data collection of the city: collect pipe network attribute data, water quality attribute data, rainfall monitoring data, water conservancy project facility monitoring data, and social media behavior data in the city, and transmit them back to the digital city database through sensors every 5 minutes, and dynamically refresh the database;

[0012] Step S120, city historical static data collection: static data collection includes collecting geographic information data and underground pipe network system data; underground drainage pipe network system data includes inspection well node information: inspection well node bottom elevation, well depth, coordinate information; also includes: inspection rainwater pipe information: rainwater pipe starting node, ending node, starting burial depth, ending burial depth, pipe diameter, pipe section length, pipe section roughness information; drainage outlet information: drainage outlet coordinates, elevation, drainage type information.

[0013] In some specific embodiments, step S200 specifically includes the following steps:

[0014] Based on the digital city database, key features are extracted from urban geospatial data, infrastructure data, and social media data, including time series features, geospatial distribution features, and flood features. Statistical analysis and artificial intelligence methods are integrated to identify representative overflow scenario simulations from the extracted key features as typical patterns of flood disasters. The representative patterns are combined with urban hydrological and hydrodynamic models, and the three-dimensional digital city inspection well overflow process is constructed and rendered through BIM or GIS, including the onset, accumulation, and receding stages of overflow, to form a digital city resource pool.

[0015] In some specific embodiments, step S300 specifically includes the following steps:

[0016] Step S310, constructing a one-dimensional underground pipe network model: based on the historical underground drainage pipe network data in the city, extracting the node bottom elevation, well depth, coordinate information, the starting node, ending node, starting buried depth, ending buried depth, pipe diameter, pipe section length, pipe roughness of the rainwater pipe, the coordinates, elevation, and drainage type information of the drain outlet from the underground drainage pipe network data, and storing them in the hydrological elements of the drainage pipe network model to construct a one-dimensional underground pipe network model;

[0017] Step S320, constructing a two-dimensional surface hydrodynamic model: after raising the elevation of the building area in the urban area by 20m and reducing the corresponding elevation of the river area by 5m, assign the corrected terrain elevation data in the study area to the grid, and define the elevation value in each unit as a constant value; assign an initial roughness value to each grid in the study area according to the land use type data, and store the roughness value at the center of the grid unit to generate a .slf file; set the boundary of the two-dimensional surface hydrodynamic model with the outer frame of the study area, which can be set as a free outflow boundary or a closed boundary to generate a .cli file, and construct a two-dimensional surface hydrodynamic model;

[0018] Step S330, constructing a coupled hydrological and hydrodynamic model: ensuring that the one-dimensional underground pipe network model and the two-dimensional surface hydrodynamic model are in the same coordinate system, determining that each inspection well node in the underground drainage pipe network has a corresponding two-dimensional surface grid unit, defining the computing nodes involved in water exchange and the two-dimensional surface grid as coupling nodes and coupling units respectively, and constructing an urban flood model.

[0019] In some specific embodiments, step S300 further includes the following steps:

[0020] Step S340: Use the weather forecast data for different time periods in the future as model input, determine the model simulation time period, analyze the results to determine the overflow situation of the inspection well, extract the water level characteristics of the overflow point, and combine the digital twin module to complete the digital twin deduction of the inspection well. Dynamically display whether there will be flood disasters in the city in the future predicted time period, as well as the flood inundation situation and the flow velocity, flow rate, flow direction, and inundation range of the inspection well overflow point, providing a three-dimensional and intuitive impact for urban flood warning decision-making.

[0021] In some specific embodiments, step S400 further includes the following steps:

[0022] Step S410: extract the time series characteristics of the inspection well forecast results and the waterlogging situation in the digital city database to build a city safety database, and normalize the data in the city safety database to divide the urban flood risk level;

[0023] The normalization process is as follows:

[0024]

[0025] Where: Y is the normalized urban safety warning data, ranging from [0,1], X is the historically collected urban safety warning data, X min is the minimum value of historically collected urban safety warning data, X max The maximum value of the historically collected urban safety warning data;

[0026] When the flooded water level approaches or exceeds the low-risk water level, a level one warning will be issued; when the flooded water level reaches the medium-risk water level, a level two warning will be issued; if the overflow water level reaches or exceeds the high-risk water level, a level three warning will be issued. The warning information includes detailed information such as the affected area, affected depth, water accumulation time, water accumulation location, water level, and water accumulation area; subsequently, the measured water accumulation data will be transmitted back through the digital twin module to determine whether it conforms to the actual situation, record the warning information, and refresh the city safety database on a rolling basis.

[0027] In some specific embodiments, the continuity equation of the one-dimensional hydrodynamic model is as follows:

[0028]

[0029] The momentum equation of the one-dimensional hydrodynamic model is as follows:

[0030]

[0031] In the formula, x is the distance, t is the time, A is the cross-sectional area of ​​the water flow, Q is the flow rate, H is the water head, g is the gravitational acceleration, S f is the friction ratio;

[0032] The continuity equation of the two-dimensional hydrodynamic model is as follows:

[0033]

[0034] The momentum equation of the two-dimensional hydrodynamic model is as follows:

[0035]

[0036] In the formula, y is the second distance, u is the flow velocity in the direction, v is the flow velocity in the y direction, h is the water depth, Z is the water level, and v e The effective viscosity coefficient, F x is the bottom friction in the x direction, F y is the bottom friction in the y direction, div(·)w is the general formula of the divergence sign, is the gradient operator.

[0037] In some specific embodiments, three measured rainfall data are extracted from the digital twin database every 8 hours to calibrate and verify the parameters of the one-dimensional underground pipe network model, and the measured rainfall time period is selected to be at least 2 hours and at most 48 hours. Three indicators are used to evaluate the relative error between the simulation data and the measured data, and are set as the objective function. The first objective function is the Nash efficiency coefficient as follows:

[0038]

[0039] Where: Qo and Qs are the measured flow and simulated flow respectively; is the average value of the measured flow; n is the total number of numerical points; the value range of NSE is (-∞, 1], the closer NSE is to 1, the better the fitting effect is; in the process of parameter calibration, the NSE value must be greater than 0.5;

[0040] The second objective function is the absolute error calculation method of the peak flow as follows:

[0041]

[0042] Where: Ro and Rs are the measured peak flow and simulated peak flow at the i-th time node respectively. During the calibration process, the RE value should not be greater than 25%;

[0043] The third objective function is the relative error calculation method of the peak time as follows:

[0044] AE=|T s -T o

[0045] Where: To and Ts are the measured peak time and simulated peak time of the i-th time node respectively; the AE value should not be greater than 1h during the parameter calibration process; when the error range of the three objective functions is met during the parameter calibration process, the calibration is stopped, otherwise the step is repeated until the requirements are met.

[0046] In some specific embodiments, the flow rate expression of the overflow point is as follows:

[0047]

[0048] Where: u is the flow velocity in the x direction, v is the flow velocity in the y direction;

[0049] The end point coordinates of the velocity arrow are the node coordinates of the grid in the study area. The expressions of the start and end point coordinates are as follows:

[0050]

[0051]

[0052]

[0053]

[0054] In the formula: x2 is the abscissa of the end point, y2 is the ordinate of the end point, x3 is the abscissa of the starting point of the next flow velocity arrow, y3 is the corresponding ordinate, x4 is the abscissa of the end point of the next flow velocity arrow, y4 is its ordinate, and α is the angle between the arrow direction and the x direction.

[0055] A digital twin-based inspection well overflow point prediction and early warning system based on the same concept includes:

[0056] The city perception module is used to transmit infrastructure information throughout the city, covering both static and dynamic information levels;

[0057] The digital twin module integrates multi-source data to build a virtual model of the digital city, synchronizes the physical city in real time, and provides visualization and interactive functions to achieve comprehensive monitoring and high-precision simulation of the city's operating status;

[0058] Inspection well prediction module, used for calculation of hydrological and hydrodynamic coupling model, uses global thinking to solve the problems of different spatiotemporal scales and mismatch of simulation elements between sub-models in the coupling model;

[0059] The flood risk warning module is used to extract the time series characteristics of the inspection well forecast results and the waterlogging conditions in the digital city database to build a city safety database, normalize the data in the city safety database, and divide the city flood risk level.

[0060] The beneficial effects of the present invention are as follows: the present invention discloses a method and system for forecasting and warning overflow points of manholes based on digital twins, including the following steps: first, improve the urban knowledge graph to construct an urban perception module; construct a digital twin city; construct a manhole forecasting module; and construct a flood risk warning module. The present invention is based on a digital twin platform, and through an Internet of Things network and a data stream processing framework, couples the urban hydrological and hydrodynamic model, dynamically calibrates the coupled model parameters, and simulates the manholes in a refined manner; comprehensively displays the flood and inundation conditions of urban development status at multiple spatiotemporal scales, tracks and reflects the actual operating status of the urban drainage system, dynamically evaluates the risk level and real-time warning, improves the efficiency of urban emergency management response, and provides support for the smooth implementation of urban flood control and drainage work. BRIEF DESCRIPTION OF THE DRAWINGS

[0061] Figure 1 It is a flow chart of a method for predicting and warning overflow points of inspection wells based on digital twins of the present invention;

[0062] Figure 2It is a flow chart of another embodiment of a method for predicting and warning overflow points of inspection wells based on digital twins of the present invention;

[0063] Figure 3 It is a structural diagram of a digital twin-based inspection well overflow point forecasting and early warning system of the present invention;

[0064] Figure 4 It is a simulation effect diagram of the pipe network model based on the original calibration parameter method and the dynamic real-time calibration parameters of the digital twin according to the present invention;

[0065] Figure 5 It is a schematic diagram of the model results based on the original method and the dynamic real-time adjustment of the model results by the digital twin. DETAILED DESCRIPTION

[0066] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings. It should be understood that the specific implementation methods described herein are only used to explain the present invention and are not used to limit the present invention.

[0067] Reference Figure 1 , Figure 2 , Figure 3 , Figure 4 and Figure 5 A method for predicting and warning overflow points of inspection wells based on digital twins is shown, comprising the following steps:

[0068] S100. First, improve the city knowledge graph and build the city perception module.

[0069] S200. Build a digital twin city.

[0070] S300, constructing an inspection well prediction module.

[0071] S400. Construct flood risk warning module.

[0072] In some specific embodiments, city data perception is an important foundation for the construction of digital twin cities. Step S100 specifically includes the following steps:

[0073] Step S110, real-time dynamic data collection of the city: collect the pipe network attribute data, water quality attribute data, rainfall monitoring data, water conservancy project facility monitoring data (river section flow, gate water level, pump station water level, etc.), and social media behavior data (resident feedback, behavior data, and media news records of waterlogging in the city), and transmit them back to the digital city database through sensors every 5 minutes, and dynamically refresh the database.

[0074] Step S120, city historical static data collection: static data collection includes geographic information data and underground pipe network system data; underground drainage pipe network system data includes inspection well node information: inspection well node bottom elevation, well depth, coordinate information; also includes: inspection rainwater pipe information: rainwater pipe starting node, end node, starting buried depth, end buried depth, pipe diameter, pipe section length, pipe section roughness information; drainage outlet information: drainage outlet coordinates, elevation, drainage type information. For areas without pipe network data, the road network is generalized and empirically valued, and the initial state of the underground drainage pipe network system is set. Subsequently, the pipe network information is dynamically adjusted in real time through the digital city resource pool and the coupling model dynamic parameter calibration.

[0075] Ensure that the above-mentioned basic information of the city is collected in the same coordinate system, clean and integrate the abnormal data in the collected data set, use the reasoning engine (Apache Jena, Droo ls) to perform rule and semantic reasoning to build the city knowledge graph and improve the digital city database.

[0076] It should be noted here that static data collection consists of geographic information data and underground pipe network system data. Geographic information data is collected at the smallest grid point scale through satellite remote sensing images (STRM, ASTER) or drone aerial photography, including digital elevation information, building distribution, road network distribution, river system, etc.; underground pipe network data is obtained through underwater robot detection.

[0077] Static data is mainly collected through satellite remote sensing images. For example, terrain elevation data uses STRM satellite to obtain DEM with a resolution of 12.5m×12.5m, land use remote sensing data comes from the "Gaofen-1" satellite with a spatial resolution of 16m, and detailed underground drainage network data includes inspection well node information.

[0078] In some specific embodiments, step S200 specifically includes the following steps:

[0079] The digital twin city is based on the digital city database, extracting key features from urban geospatial data, infrastructure data, and social media data, including time series features (such as rainfall, manhole water level change trends), geospatial distribution features (such as waterlogging inundation location, inundation flow direction, inundation depth), and flood characteristics (such as urban topography, land use, and soil moisture). By integrating statistical analysis and artificial intelligence methods, representative overflow scenario simulations are identified from the extracted key features as typical patterns of flood disasters (such as inundation distribution characteristics under specific terrain, and overflow evolution under specific return period design rainfall conditions). The representative patterns are combined with the urban hydrological and hydrodynamic model, and the process of manhole overflow in a three-dimensional digital city is constructed and rendered through BIM or GIS, including the beginning, accumulation, and ebb stages of overflow, to form a digital city resource pool.

[0080] In some specific embodiments, step S300 specifically includes the following steps:

[0081] Step S310, constructing a one-dimensional underground pipe network model: based on the historical underground drainage pipe network data in the city, extracting the node bottom elevation, well depth, coordinate information, the starting node, ending node, starting burial depth, ending burial depth, pipe diameter, pipe section length, pipe roughness of the rainwater pipe, the coordinates, elevation, and drainage type information of the drain outlet in the obtained underground drainage pipe network data are stored in the hydrological elements of the drainage pipe network model to construct a one-dimensional underground pipe network model.

[0082] In this embodiment, according to the literature, the sensitivity parameters in the one-dimensional underground pipe network model mainly include the Manning coefficient of the permeable area (N-Prev), the Manning coefficient of the impermeable area (N-Imprev), the depression depth of the permeable area (Dstore-Prev), the depression depth of the impermeable area (Dstore-Imprev), the pipe section roughness (Manning), the maximum infiltration rate (Max-Rate), the minimum infiltration rate (Min-Rate), the permeability attenuation coefficient (Decay), and the drying coefficient (Drying-Time), and the parameter range is continuously adjusted according to different regions, where N -Imprev generally ranges from 0.011 to 0.024, N-Imprev generally ranges from 0.05 to 0.8, Dstore-Imprev generally ranges from 0.013 to 0.25, Dstore-Prev generally ranges from 0.25 to 0.5, Max-Rate generally ranges from 20 to 100, Min-Rate generally ranges from 0.5 to 20, Decay generally ranges from 2 to 5, drying time generally ranges from 2 to 10, and pipe section roughness coefficient generally ranges from 0.01 to 0.5. In this example, we assign them 0.02, 0.21, 0.23, 0.3, 70, 3, 3, 10, and 0.012 respectively as the initial values ​​of the parameters.

[0083] Step S320, construct a two-dimensional surface hydrodynamic model: use a grid division tool to spatially discretize the study area, and the grid unit size should not be greater than 10m, such as SMS, BlueKenue software, etc. In order to conform to the actual terrain in the urban area. After the elevation of the building area in the urban area is increased by 20m, and the corresponding elevation of the river area is reduced by 5m, the corrected terrain elevation data in the study area is assigned to the grid, and the elevation value in each unit is defined as a constant value; according to the land use type data, an initial roughness value is assigned to each grid in the study area, and the roughness value is stored at the center of the grid unit to generate a .slf file; the boundary of the two-dimensional surface hydrodynamic model is set with the outer frame of the study area, which can be set as a free outflow boundary or a closed boundary to generate a .cli file to construct a two-dimensional surface hydrodynamic model.

[0084] Step S330, construct a coupled hydrological and hydrodynamic model: ensure that the one-dimensional underground pipe network model and the two-dimensional surface hydrodynamic model are in the same coordinate system, determine that each inspection well node in the underground drainage pipe network has a corresponding surface two-dimensional grid unit, define the computing nodes involved in water exchange and the surface two-dimensional grid as coupling nodes and coupling units respectively, and construct an urban flood model. That is, when the vertex coordinates of the inspection well node are located in the surface grid unit, a spatial connection between the two is established. Extract the overflow point time series calculated by the one-dimensional underground pipe network model. And set the corresponding simulation duration and simulation parameters, generate a .cas parameter file, and complete the construction of the urban flood model. Select the location information of the flood-prone points monitored in the same rainfall time period as step S320, the depth of water accumulation and the area of ​​water accumulation data to calibrate and verify the roughness value of land use in the two-dimensional surface hydrodynamic model. Use the confusion matrix as the objective function to calculate the relative error, and the confusion matrix evaluation calculation method is as follows:

[0085]

[0086]

[0087]

[0088]

[0089] In the formula: TP (True Positive) means that there is water accumulation in reality, which is consistent with the simulation results; FP (False Negative) means that there is no water accumulation in reality and the model simulation is wrong; FN (False Negative) means that there is water accumulation in reality, but the model simulation is wrong; TN (True Negative) means that there is no water accumulation in reality and the simulation results. Among them, TP and TN both indicate that the model results are consistent with the actual situation, while FN and FP indicate that the simulation results are inconsistent with the actual situation, and are recorded as S respectively. TP , S TN , S FP and S FN . Use R ACC , R TPR , R PPV The performance of the model is comprehensively evaluated by using F1, which represents accuracy, sensitivity, precision and F1 score. The closer the F1 value is to 1, the better the model effect is. The global optimal solution is sought between the staged correction and the control section error correction in the hydrological runoff process and the hydrodynamic evolution process, and the iterative optimization is carried out in each period so that the model parameters and model results are constantly close to the measured values, and a flood simulation coupling model suitable for the corresponding urban area is constructed.

[0090] In some specific embodiments, step S300 further includes the following steps:

[0091] Step S340, using the weather forecast data of different time periods in the future as model input, determining the model simulation time period, analyzing the results to determine the overflow of the inspection well, extracting the water level characteristics of the overflow point, and combining the digital twin module to complete the digital twin deduction of the inspection well, dynamically displaying whether there will be flood disasters in the city in the future forecast time period, as well as the flood inundation situation and the flow velocity, flow rate, flow direction, and inundation range of the inspection well overflow point, providing a three-dimensional and intuitive impact for urban flood warning decision-making. The results of the overflow point are stored in the nodes of the two-dimensional surface grid, and the attributes of the overflow point can be read to directly obtain the water depth, flow velocity in the x and y directions, and the flow direction of the overflow point. The water depth can be displayed according to the overflow point coordinates of the study area, and the overflow point flow velocity can be synthesized using the flow velocity in the x and y directions. During the flow velocity display process, it is described by the size of the arrow, and the direction of the arrow is the flow direction.

[0092] In some specific embodiments, step S400 further includes the following steps:

[0093] Step S410, flood risk warning: extract the time series characteristics of the inspection well forecast results (overflow point location, overflow volume time series, overflow evolution path, overflow flow rate, etc.) and the waterlogging situation in the digital city database to build a city safety database, and normalize the data in the city safety database to divide the city flood risk level;

[0094] The normalization process is as follows:

[0095]

[0096] Where: Y is the normalized urban safety warning data, ranging from [0,1], X is the historically collected urban safety warning data, X min is the minimum value of historically collected urban safety warning data, X max The maximum value of the historically collected urban safety warning data;

[0097] When the flooded water level approaches or exceeds the low-risk water level, a level one warning will be issued; when the flooded water level reaches the medium-risk water level, a level two warning will be issued; if the overflow water level reaches or exceeds the high-risk water level, a level three warning will be issued. The warning information includes detailed information such as the affected area, affected depth, water accumulation time, water accumulation location, water level, and water accumulation area; subsequently, the measured water accumulation data will be transmitted back through the digital twin module to determine whether it conforms to the actual situation, record the warning information, and refresh the city safety database on a rolling basis.

[0098] In some specific embodiments, the continuity equation of the one-dimensional hydrodynamic model is as follows:

[0099]

[0100] The momentum equation of the one-dimensional hydrodynamic model is as follows:

[0101]

[0102] In the formula, x is the distance, t is the time, A is the cross-sectional area of ​​the water flow, Q is the flow rate, H is the water head, g is the gravitational acceleration, S f is the friction ratio;

[0103] The continuity equation of the two-dimensional hydrodynamic model is as follows:

[0104]

[0105] The momentum equation of the two-dimensional hydrodynamic model is as follows:

[0106]

[0107] In the formula, y is the second distance, u is the flow velocity in the direction, v is the flow velocity in the y direction, h is the water depth, Z is the water level, and v e The effective viscosity coefficient, F x is the bottom friction in the x direction, F y is the bottom friction in the y direction, div(·)w is the general formula of the divergence sign, is the gradient operator.

[0108] In some specific embodiments, three measured rainfall data are extracted from the digital twin database every 8 hours to calibrate and verify the parameters of the one-dimensional underground pipe network model, and the measured rainfall time period is selected to be at least 2 hours and at most 48 hours. Three indicators are used to evaluate the relative error between the simulation data and the measured data, and are set as the objective function. The first objective function is the Nash efficiency coefficient as follows:

[0109]

[0110] Where: Qo and Qs are the measured flow and simulated flow respectively; is the average value of the measured flow; n is the total number of numerical points; the value range of NSE is (-∞, 1], the closer NSE is to 1, the better the fitting effect is; in the process of parameter calibration, the NSE value must be greater than 0.5;

[0111] The second objective function is the absolute error calculation method of the peak flow as follows:

[0112]

[0113] Where: Ro and Rs are the measured peak flow and simulated peak flow at the i-th time node respectively. During the calibration process, the RE value should not be greater than 25%;

[0114] The third objective function is the relative error calculation method of the peak time as follows:

[0115] AE=|T s -T o

[0116] Where: To and Ts are the measured peak time and simulated peak time of the i-th time node respectively; the AE value should not be greater than 1h during the parameter calibration process; when the error range of the three objective functions is met during the parameter calibration process, the calibration is stopped, otherwise the steps are repeated until the requirements are met. The dynamic real-time parameter calibration method of the present invention is similar to the previous parameter calibration effect as shown in Figure 4 As shown, the maximum Nash efficiency coefficient (NSE) obtained by the previous parameter calibration method is 0.825, and the maximum Nash efficiency coefficient (NSE) of the parameter calibration method of the present invention that dynamically and real-time returns data is 0.961, and the simulation effect is significantly improved.

[0117] In some specific embodiments, the flow rate expression of the overflow point is as follows:

[0118]

[0119] Where: u is the flow velocity in the x direction, v is the flow velocity in the y direction;

[0120] The end point coordinates of the velocity arrow are the node coordinates of the grid in the study area. The expressions of the start and end point coordinates are as follows:

[0121]

[0122]

[0123]

[0124]

[0125] In the formula: x2 is the abscissa of the end point, y2 is the ordinate of the end point, x3 is the abscissa of the starting point of the next flow velocity arrow, y3 is the corresponding ordinate, x4 is the abscissa of the end point of the next flow velocity arrow, y4 is its ordinate, and α is the angle between the arrow direction and the x direction.

[0126] A digital twin-based inspection well overflow point prediction and early warning system based on the same concept includes:

[0127] The city perception module is used to transmit infrastructure information throughout the city, covering both static and dynamic information levels.

[0128] In this embodiment, the city perception module is responsible for transmitting infrastructure information within the entire city, covering static and dynamic information levels. Static basic information uses satellite remote sensing technology to obtain large-scale urban geographic spatial data, including urban digital elevation information, building distribution, road network distribution, river system distribution, water conservancy project infrastructure distribution, land use data, sponge city facility distribution and actual urban underground pipe network system distribution (if there is no pipe network data, the initial pipe network data can be replaced by generalized road network). Dynamic basic information is transmitted by sensors in real time every 5 minutes, including inspection well attribute data, pipe segment attribute data, water quality attribute data, rainfall monitoring data, water conservancy project facility monitoring data (river section flow, gate water level, pump station water level, etc.) and social media behavior data collected by web crawlers (resident feedback, behavior data and media news records of water accumulation in the city). Through the urban Internet of Things, a large amount of effective data is transmitted in real time to form an urban knowledge map, dynamically integrate urban basic data, characterize the relationship between urban underlying surface characteristics, physical mechanisms, and hydrological and water resource elements, assist in flood risk decision-making, and roll out digital city databases and urban safety databases.

[0129] The digital twin module is used to integrate multi-source data, build a virtual model of the digital city, synchronize the physical city in real time, and provide visualization and interactive functions to achieve comprehensive monitoring and high-precision simulation of the city’s operating status.

[0130] In this embodiment, the digital twin module is the core module of the digital twin city platform, which consists of a digital city database and a digital twin city. The module integrates multi-source data, builds a digital city virtual model, synchronizes the physical city in real time, and provides visualization and interactive functions, achieving comprehensive monitoring and high-precision simulation of the city's operating status.

[0131] The digital city database extracts structured data from the urban knowledge graph for data cleaning and integration, unifies the data, converts the data format and stores it in the cloud, and establishes a data management mechanism, including data backup, recovery, permissions, etc. The digital twin city is based on the digital city database, extracting key features from urban geospatial data, infrastructure data, and social media data, including time series features (such as rainfall, manhole water level change trend), geospatial distribution features (such as waterlogging location, flooding direction, flooding depth), and flood features (such as urban topography, land use, and soil moisture). By integrating statistical analysis, artificial intelligence and other methods, representative overflow scenario simulations are identified from the extracted key features as typical patterns of flood disasters (such as flooding distribution characteristics under specific terrain, overflow evolution process under specific return period design rainfall conditions). The representative pattern is combined with the urban hydrological and hydrodynamic model, and the process of manhole overflow in a three-dimensional digital city is constructed and rendered through BIM or GIS, including the beginning, accumulation, and retreat of the overflow, forming a digital city resource pool.

[0132] The inspection well prediction module is used for the calculation of the hydrological and hydrodynamic coupling model. It uses a global approach to solve the problems of different temporal and spatial scales and mismatched simulation elements between sub-models in the coupling model.

[0133] In this embodiment, the core work of the inspection well forecast module is the calculation of the hydrological and hydrodynamic coupling model, and the global approach is used to solve the problems of different spatiotemporal scales and mismatched simulation elements between sub-models in the coupling model. The historical, real-time and future rainfall data in the digital city resource pool are extracted and combined with rainfall scenarios of different medium and long recurrence periods to drive the coupled hydrological and hydrodynamic model. Combined with the urban knowledge map, the key elements of the physical process of water exchange are explored, and the influence mechanism and mathematical transmission relationship between elements under different spatiotemporal coupling boundaries are analyzed to complete the coupling of hydrological and hydrodynamic models of multi-scale spatiotemporal elements. The measured inspection well water level data and water accumulation depth and water accumulation area corresponding to three rainfalls in the digital city database are extracted every 8 hours to dynamically calibrate and verify the coupling model. The meteorological forecast data for different time periods in the future are used as the model input, the model simulation time period is determined, the results are analyzed to determine the overflow of the inspection well, the water level characteristics of the overflow point are extracted, and the digital twin module is combined to complete the digital twin deduction of the inspection well.

[0134] The flood risk warning module is used to extract the time series characteristics of the inspection well forecast results and the waterlogging conditions in the digital city database to build a city safety database, normalize the data in the city safety database, and divide the city flood risk level.

[0135] In this embodiment, the flood risk warning is to extract the time series characteristics of the inspection well forecast results (overflow point location, overflow volume time series, overflow evolution path, overflow flow rate, etc.) and the waterlogging situation in the digital city database to build a city safety database, and normalize the data in the city safety database to divide the urban flood risk level. When the flooding water level approaches or exceeds the low-risk water level, a first-level warning is issued; when the flooding water level reaches the medium-risk water level, a second-level warning is issued; if the overflow water level reaches or exceeds the high-risk water level, a third-level warning is issued, and the warning information includes detailed information such as the affected area, the affected depth, the waterlogging time, the waterlogging location, the waterlogging level, and the waterlogging area. Subsequently, the measured waterlogging data is transmitted back through the digital twin module to determine whether it conforms to the actual situation, record the warning information, and refresh the city safety database in a rolling manner.

[0136] By adopting the above technical solution disclosed in the present invention, the following beneficial effects are obtained:

[0137] The present invention discloses a method and system for forecasting and warning overflow points of manholes based on digital twins, including the following steps: first, improve the urban knowledge graph to construct an urban perception module; construct a digital twin city; construct a manhole forecasting module; and construct a flood risk warning module. Based on a digital twin platform, the present invention couples the urban hydrological and hydrodynamic model through an Internet of Things network and a data stream processing framework, dynamically calibrates the coupled model parameters, and performs refined simulation of manholes; comprehensively displays the flood and inundation conditions of urban development status at multiple spatiotemporal scales, tracks and reflects the actual operating status of the urban drainage system, dynamically evaluates the risk level and real-time warning, improves the efficiency of urban emergency management response, and provides support for the smooth implementation of urban flood control and drainage work.

[0138] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principle of the present invention. These improvements and modifications should also be considered as the scope of protection of the present invention.

Claims

1. A digital twin-based inspection well overflow point forecasting and early warning method, characterized in that: The following steps are involved: S100, first improve the city knowledge graph and build the city perception module; S200, building a digital twin city; S300, constructing an inspection well prediction module; S400. Construct a flood risk warning module.

2. The digital twin-based inspection well overflow point forecasting and early warning method according to claim 1 is characterized in that: The step S100 specifically includes the following steps: Step S110, real-time dynamic data collection of the city: collect pipe network attribute data, water quality attribute data, rainfall monitoring data, water conservancy project facility monitoring data, and social media behavior data in the city, and transmit them back to the digital city database through sensors every 5 minutes, and dynamically refresh the database; Step S120, city historical static data collection: the static data collection includes collecting geographic information data and underground pipe network system data; the underground drainage pipe network system data includes inspection well node information: inspection well node bottom elevation, well depth, coordinate information; also includes: inspection rainwater pipe information: the starting node, ending node, starting burial depth, ending burial depth, pipe diameter, pipe section length, pipe section roughness information of the rainwater pipe; drainage outlet information: drainage outlet coordinates, elevation, drainage type information.

3. The digital twin-based inspection well overflow point forecasting and early warning method according to claim 2 is characterized in that: The step S200 specifically includes the following steps: Based on the digital city database, key features are extracted from urban geospatial data, infrastructure data, and social media data, including time series features, geospatial distribution features, and flood features. Statistical analysis and artificial intelligence methods are integrated to identify representative overflow scenario simulations from the extracted key features as typical patterns of flood disasters. The representative patterns are combined with urban hydrological and hydrodynamic models, and the process of manhole overflow in a three-dimensional digital city is constructed and rendered through BIM or GIS, including the start, accumulation, and receding stages of the overflow, to form a digital city resource pool.

4. The digital twin-based inspection well overflow point forecasting and early warning method according to claim 3 is characterized in that: The step S300 specifically includes the following steps: Step S310, constructing a one-dimensional underground pipe network model: based on the historical underground drainage pipe network data in the city, extracting the node bottom elevation, well depth, coordinate information, the starting node, ending node, starting buried depth, ending buried depth, pipe diameter, pipe section length, pipe roughness of the rainwater pipe, the coordinates, elevation, and drainage type information of the drain outlet from the underground drainage pipe network data, and storing them in the hydrological elements of the drainage pipe network model to construct a one-dimensional underground pipe network model; Step S320, constructing a two-dimensional surface hydrodynamic model: after raising the elevation of the building area in the urban area by 20m and reducing the corresponding elevation of the river area by 5m, assign the corrected terrain elevation data in the study area to the grid, and define the elevation value in each unit as a constant value; assign an initial roughness value to each grid in the study area according to the land use type data, and store the roughness value at the center of the grid unit to generate a .slf file; set the boundary of the two-dimensional surface hydrodynamic model with the outer frame of the study area, which can be set as a free outflow boundary or a closed boundary to generate a .cli file, and construct a two-dimensional surface hydrodynamic model; Step S330, constructing a coupled hydrological and hydrodynamic model: ensuring that the one-dimensional underground pipe network model and the two-dimensional surface hydrodynamic model are in the same coordinate system, determining that each inspection well node in the underground drainage pipe network has a corresponding two-dimensional surface grid unit, defining the computing nodes involved in water exchange and the two-dimensional surface grid as coupling nodes and coupling units respectively, and constructing an urban flood model.

5. The digital twin-based inspection well overflow point forecasting and early warning method according to claim 4 is characterized in that: The step S300 specifically further includes the following steps: Step S340: Use the weather forecast data for different time periods in the future as model input, determine the model simulation time period, analyze the results to determine the overflow situation of the inspection well, extract the water level characteristics of the overflow point, and combine the digital twin module to complete the digital twin deduction of the inspection well. Dynamically display whether there will be flood disasters in the city in the future predicted time period, as well as the flood inundation situation and the flow velocity, flow rate, flow direction, and inundation range of the inspection well overflow point, providing a three-dimensional and intuitive impact for urban flood warning decision-making.

6. The digital twin-based inspection well overflow point forecasting and early warning method according to claim 5 is characterized in that: The step S400 specifically further includes the following steps: Step S410: extracting the time series characteristics of the inspection well forecast results and the waterlogging conditions in the digital city database to build a city safety database, and normalizing the data in the city safety database to divide the city flood risk level; The normalization process is as follows: Where: Y is the normalized urban safety warning data, ranging from [0,1], X is the historically collected urban safety warning data, X min is the minimum value of historically collected urban safety warning data, X max The maximum value of the historically collected urban safety warning data; When the flooded water level approaches or exceeds the low-risk water level, a level one warning will be issued; when the flooded water level reaches the medium-risk water level, a level two warning will be issued; if the overflow water level reaches or exceeds the high-risk water level, a level three warning will be issued. The warning information includes detailed information such as the affected area, affected depth, water accumulation time, water accumulation location, water level, and water accumulation area; subsequently, the measured water accumulation data will be transmitted back through the digital twin module to determine whether it conforms to the actual situation, record the warning information, and refresh the city safety database on a rolling basis.

7. The digital twin-based inspection well overflow point forecasting and early warning method according to claim 3 is characterized in that: The continuity equation of the one-dimensional hydrodynamic model is as follows: The momentum equation of the one-dimensional hydrodynamic model is as follows: In the formula, x is the distance, t is the time, A is the cross-sectional area of ​​the water flow, Q is the flow rate, H is the water head, g is the gravitational acceleration, S f is the friction ratio; The continuity equation of the two-dimensional hydrodynamic model is as follows: The momentum equation of the two-dimensional hydrodynamic model is as follows: In the formula, y is the second distance, u is the flow velocity in the direction, v is the flow velocity in the y direction, h is the water depth, Z is the water level, and v e The effective viscosity coefficient, F x is the bottom friction in the x direction, F y is the bottom friction in the y direction, div(·)w is the general formula of the divergence sign, is the gradient operator.

8. The digital twin-based inspection well overflow point forecasting and early warning method according to claim 3 is characterized in that: In the digital twin database, three measured rainfall data are extracted every 8 hours to calibrate and verify the parameters of the one-dimensional underground pipe network model. The shortest time period of measured rainfall is 2 hours and the longest time is no more than 48 hours. Three indicators are used to evaluate the relative error between the simulated data and the measured data, and set as the objective function. The first objective function is the Nash efficiency coefficient as follows: Where: Qo and Qs are the measured flow and simulated flow respectively; is the average value of the measured flow; n is the total number of numerical points; the value range of NSE is (-∞, 1], the closer NSE is to 1, the better the fitting effect is; in the process of parameter calibration, the NSE value must be greater than 0.5; The second objective function is the absolute error calculation method of the peak flow as follows: Where: Ro and Rs are the measured peak flow and simulated peak flow at the i-th time node respectively. During the calibration process, the RE value should not be greater than 25%; The third objective function is the relative error calculation method of the peak time as follows: AE=|T s -T o | Where: To and Ts are the measured peak time and simulated peak time of the i-th time node respectively; the AE value should not be greater than 1h during the parameter calibration process; when the error range of the three objective functions is met during the parameter calibration process, the calibration is stopped, otherwise the step is repeated until the requirements are met.

9. The digital twin-based inspection well overflow point forecasting and early warning method according to claim 3 is characterized in that: The flow rate expression of the overflow point is as follows: Where: u is the flow velocity in the x direction, v is the flow velocity in the y direction; The end point coordinates of the velocity arrow are the node coordinates of the grid in the study area. The expressions of the start and end point coordinates are as follows: In the formula: x2 is the abscissa of the end point, y2 is the ordinate of the end point, x3 is the abscissa of the starting point of the next flow velocity arrow, y3 is the corresponding ordinate, x4 is the abscissa of the end point of the next flow velocity arrow, y4 is its ordinate, and α is the angle between the arrow direction and the x direction.

10. A digital twin-based inspection well overflow point forecasting and warning system, characterized in that: include: The city perception module is used to transmit infrastructure information throughout the city, covering both static and dynamic information levels; The digital twin module integrates multi-source data to build a virtual model of the digital city, synchronizes the physical city in real time, and provides visualization and interactive functions to achieve comprehensive monitoring and high-precision simulation of the city's operating status; Inspection well prediction module, used for calculation of hydrological and hydrodynamic coupling model, uses global thinking to solve the problems of different spatiotemporal scales and mismatch of simulation elements between sub-models in the coupling model; The flood risk warning module is used to extract the time series characteristics of the inspection well forecast results and the waterlogging conditions in the digital city database to build a city safety database, normalize the data in the city safety database, and divide the city flood risk level.

Images