Digital Twin Simulation Method and Device for Extracting Subsurface Brackish Water to Prevent Seawater Intrusion

Abstract

The present invention provides a digital twin simulation method and device for extracting underground saline water to prevent seawater intrusion, belonging to the technical field of seawater intrusion, and comprising the following steps: S1: Measuring corresponding values through a physical simulation device, and obtaining the corresponding physical basin information in nature by using a proportionality coefficient and conversion; S2: Constructing a digital twin model with multiple spatio-temporal scales and multiple dimensions; S3: Constructing a digital twin model; S4: Based on a cyclic cross-fusion mechanism, performing cyclic cross-recombination on data features, and performing fusion processing on the recombined features by using a mixture Gaussian process; S5: Running the digital twin model in a virtual environment to simulate the seawater intrusion situation under real conditions in nature, and realizing disaster monitoring within the physical basin range. The present invention makes full use of digital twin technology to achieve real-time synchronization between physical entities and virtual models, provides early warnings for possible geological disasters, and provides decision-making support for the orderly development and utilization of groundwater.

Description

Technical Field

[0001] The present invention belongs to the technical field of seawater intrusion, and specifically relates to a digital twin simulation method and device for extracting groundwater to prevent and control seawater intrusion. Background Art

[0002] Seawater intrusion into groundwater refers to the phenomenon that the groundwater level in coastal areas drops sharply due to excessive exploitation of groundwater, the hydrodynamic balance between seawater and freshwater is destroyed, and the salt-fresh water interface moves toward the land. Seawater intrusion increases the osmotic pressure of irrigation groundwater, soil salinization, and the scrapping of irrigation wells, resulting in a decrease in paddy field area, an increase in dry field area, a decrease in the area of ​​irrigated farmland, and an increase in wasteland area. Groundwater is an important source of water for agricultural irrigation, industry, mining, and cities. Once seawater intrusion occurs, it will have an adverse impact on human society. Therefore, the prevention and control of seawater intrusion into groundwater has many positive significances and far-reaching impacts.

[0003] With the continuous development of groundwater system research and the continuous improvement of the simulation degree of groundwater numerical models, the construction of more refined management models requires the development of advanced optimization technologies to solve the management problems of complex groundwater systems. Water resource management problems have evolved from traditional single-objective optimization to multi-objective optimization, and from deterministic optimization problems to stochastic optimization problems. The management models related to groundwater systems often have the characteristics of nonlinearity, non-convexity, randomness, high-dimensional decision variables and multi-objective variables. The existing seawater intrusion simulation methods have the defects of poor detection accuracy and poor timeliness, making it difficult to establish high-precision simulation models and difficult to promote and apply them on a large scale. Summary of the invention

[0004] In order to solve the above problems existing in the prior art, a digital twin simulation method and device for extracting groundwater to prevent seawater intrusion are provided.

[0005] The technical solution adopted by the present invention to solve its technical problem is:

[0006] This technical solution proposes a digital twin simulation method for extracting groundwater to prevent seawater intrusion, including the following steps:

[0007] S1: Build a physical simulation device for seawater intrusion, measure the corresponding values ​​through the physical simulation device, and use the proportional coefficient and conversion to obtain the corresponding physical basin information in nature;

[0008] S2: Feedback the physical watershed information to the watershed virtual twin, build an integrated sky-ground water conservancy perception network, and build a multi-dimensional digital twin model with multiple spatiotemporal scales based on the watershed virtual twin and the integrated sky-ground water conservancy perception network;

[0009] S3: Construct a digital twin model, which includes a basin information data model, a geographic information reference model, and a multi-dimensional visualization model. Extract multi-temporal scale and multi-dimensional data features from the original data in the data model according to unified data standards and business application requirements;

[0010] S4: Based on the cyclic cross-fusion mechanism, perform cyclic cross-recombination on the data features, use a mixture Gaussian process to fuse the recombined features, input the fused features into a calibration model for calibration, and calibrate the digital twin model through piecewise linear regression. The calibration process includes multiple rounds of simulation and comparison with actual data;

[0011] S5: Run the digital twin model in a virtual environment to simulate the seawater intrusion situation under real natural conditions, and combine the hydrological data collected by the sky-ground-integrated water conservancy sensing network to achieve the prediction of the underground reservoir storage capacity, storage variable, river channel infiltration volume, and exploitable volume in the physical basin, and at the same time achieve disaster monitoring within the physical basin range.

[0012] Preferably, in S2, the sky-ground-integrated water conservancy sensing network includes the hydrology, water resources, water conservancy projects, soil and water conservation, rivers and lakes, economic society, and ecological environment information of the physical basin. The sky-ground-integrated water conservancy sensing network can sense the digital terrain of the basin underlying surface, vegetation coverage, river channel obstruction buildings, floodplain occupation, hydrological cross-section, and underwater terrain of the river section.

[0013] Preferably, the multi-temporal scale and multi-dimensional digital twin model includes a macro scale, a meso scale, and a micro scale;

[0014] The macro scale includes vector map data, satellite remote sensing image maps, and DEM elevation map data;

[0015] The meso scale includes satellite remote sensing image maps, refined manual model data, oblique photography model data, and high-precision DEM data;

[0016] The micro scale includes water conservancy Internet of Things devices and reservoir dams. Through BIM technology, the establishment and detailed display of device models are realized.

[0017] Preferably, in S3, the basin information data model includes organizing and sorting different information based on knowledge graph technology, and constructing the association relationships, index relationships, and spatial relationships between physical basin entities through nodes and the logical relationships between nodes to form a data model and a knowledge graph, and integrating relevant data resources through the unified data model and knowledge graph;

[0018] The geographic information reference model includes establishing a set of method systems for standardizing the parsing and storage of complex multi-source geographic information. The standardization of geographic information includes the definition of geographic information interoperability, basic data types, modeling rules, and the semantics of real-world phenomena.

[0019] The multi-dimensional visualization model includes applying various spatial and temporal information in the physical world to construct a high-fidelity visual physical model. Through BIM technology, for the spatio-temporal characteristics of each element in the joint scheduling during the geographical distribution law and development process, it provides real-time rendering and visual presentation for simulation, and provides a multi-dimensional and multi spatio-temporal scale high-fidelity digital mapping for the physical space.

[0020] Preferably, in the above S4, the mixture Gaussian process realizes the weighted fusion of data, and the formula is as follows:

[0021] (1)

[0022] In the formula, F fused represents the fused feature, N is the total number of Gaussian components, w i represents the i th weight of the Gaussian component, represents the Gaussian distribution, F rec is the input feature, μ i , Σ i respectively represent the mean and covariance matrix of the i th Gaussian component.

[0023] Preferably, using the fused feature F fused as the input, the underground reservoir storage capacity, storage variable, river channel infiltration amount, and exploitable amount of the physical basin are predicted through a piecewise linear regression model to obtain the predicted output of the physical basin. The formula is as follows:

[0024] (2)

[0025] In the formula, y pred represents the predicted output, M is the total number of basis functions used in the model, a j represents the j th coefficient of the basis function, which is used to adjust the contribution degree of the basis function to the predicted value, b j represents the jThe bias term of the basis function, which is used to adjust the output of the basis function, F fused,j denotes the j th fused feature;

[0026] By calculating the error between the predicted output and the actual output, the calibration loss function is obtained:

[0027] (3)

[0028] In the formula, L cal denotes the calibration loss function, y real denotes the actual output, y pred denotes the predicted value;

[0029] Using the gradient descent algorithm to minimize the calibration loss function, the formula for updating the parameters of the digital twin model is as follows:

[0030] (4)

[0031] In the formula, θ and θ ´ respectively denote the parameters of the digital twin model before and after updating, η denotes the learning rate, ▽ θ L cal denotes the gradient of the calibration loss function with respect to the parameters of the digital twin model.

[0032] A digital twin simulation device for extracting subsurface saline water to prevent seawater intrusion, which is used to execute the above-mentioned digital twin simulation method for extracting subsurface saline water to prevent seawater intrusion. The physical simulation device includes a test sand box, and both ends of the test sand box are respectively connected with baffles with evenly distributed holes. The baffles divide the test sand box into a seawater chamber and a fresh water chamber on both sides, and a test chamber in the middle. The two sides of the test chamber are respectively connected to the seawater chamber and the fresh water chamber through the arranged holes. It also includes,

[0033] A constant head mechanism for keeping the water level in the seawater chamber balanced;

[0034] A pumping mechanism for pumping water from the test chamber;

[0035] A pressure measuring mechanism for measuring the pressure in the test chamber.

[0036] Preferably, the constant head mechanism includes a water tank and a water bucket. A partition board is fixedly connected inside the water tank, and the partition board divides the water tank into an overflow water tank and a constant water tank. The seawater chamber is connected to the constant water tank through a connector; the overflow water tank is connected to the water bucket through a connecting pipe, and a submersible pump is connected inside the water bucket. The submersible pump can pump the water in the water bucket into the constant water tank.

[0037] Preferably, a water stop plate for blocking the connector is slidably connected inside the connector. The water stop plate is connected to the connector through a tension spring. The seawater chamber is communicated with a fixed pipe, and a piston is slidably connected inside the fixed pipe. The piston is connected with a magnet that has an attracting effect on the water stop plate, and the magnetic force of the magnet is greater than the elastic force of the tension spring.

[0038] Preferably, a switch for controlling the operation of the submersible pump is connected inside the fixed pipe. The switch is located on one side of the piston. The piston is connected to the fixed pipe through a spring. The piston is fixedly connected with a sleeve, and the sleeve is threadedly connected with a screw rod. The screw rod is connected with the magnet. The water stop plate is connected with a one-way valve that only allows water to flow from the constant water tank into the seawater chamber.

[0039] Compared with the prior art, the present invention has the following advantages:

[0040] This application makes full use of digital twin technology to achieve real-time synchronization between physical entities and virtual models. The digital twin model is calibrated through piecewise linear regression. Using fusion features as input, through multiple rounds of simulation and comparison with actual data, the model parameters are calibrated to make it better match the actual situation, ensure the accuracy of the model, and can provide early warnings for possible geological disasters and provide decision-making support for the orderly development and utilization of groundwater.

[0041] The physical simulation device proposed in this application can adjust the water level height of the seawater chamber by adjusting the height of the water tank. When the seawater chamber rises, only when the seawater chamber rises a certain height and the magnet moves a sufficient distance, will the water stop plate open and the water in the seawater chamber flow into the constant water tank. When the seawater chamber drops, when it drops to a certain extent, the piston will drive the switch to work and the submersible pump will work to replenish water to the seawater chamber, preventing small-scale fluctuations from affecting the simulation of seawater intrusion, and at the same time keeping the seawater chamber at a sufficient and appropriate water level. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] The above and / or additional aspects and advantages of the present invention will become obvious and easy to understand from the following description of the embodiments in conjunction with the accompanying drawings, in which:

[0043] Figure 1 is the overall flowchart of the present invention;

[0044] Figure 2 is the overall structural schematic diagram of the present invention;

[0045] Figure 3 It is a schematic structural diagram of the joint part in the present invention.

[0046] Explanation of reference numerals in the drawings:

[0047] 1 test sand box; 2 water pumping mechanism; 3 constant head mechanism; 4 pressure measuring mechanism; 5 fresh water chamber; 6 sea water chamber; 7 baffle; 8 water tank; 9 water isolation plate; 10 water bucket; 11 submersible pump; 12 peristaltic pump; 13 piezometric tube; 14 scale paper; 15 test water chamber; 16 constant water tank; 17 overflow water tank; 18 fixed pipe; 19 joint; 20 water stop plate; 21 tension spring; 22 piston; 23 sleeve; 24 screw; 25 magnet; 26 spring; 27 switch. Specific implementation manners

[0048] The embodiments of the present invention will be described in detail below. The examples of the embodiments are shown in the drawings, where the same or similar reference numerals represent the same or similar elements or elements with the same or similar functions throughout. The embodiments described below with reference to the drawings are exemplary and are only used to explain the present invention and should not be construed as a limitation to the present invention.

[0049] As Figures 1 - 3 shown, the present embodiment proposes a digital twin simulation method for extracting underground brackish water to prevent seawater intrusion, including the following steps:

[0050] S1: Build a physical simulation device for seawater intrusion, measure corresponding values through the physical simulation device, and obtain the corresponding physical basin information in nature by using the proportional coefficient and conversion.

[0051] S2: Feed the physical basin information back to the basin virtual twin, build an integrated sky-ground-water conservancy perception network, and build a digital twin model with multiple spatio-temporal scales and multiple dimensions based on the basin virtual twin and the integrated sky-ground-water conservancy perception network.

[0052] S3: Build a digital twin model. The digital twin model includes a basin information data model, a geographic information reference model, and a multi-dimensional visualization model. Extract the data characteristics of the original data in the data model with multiple spatio-temporal scales and multiple dimensions according to the unified data standard and business application requirements.

[0053] S4: Based on the cyclic cross-fusion mechanism, perform cyclic cross-recombination on the data characteristics, use the mixture Gaussian process to fuse the recombined characteristics, input the fused characteristics into the calibration model for calibration, and calibrate the digital twin model through piecewise linear regression. The calibration process includes multiple rounds of simulation and comparison with actual data.

[0054] S5: Run the digital twin model in a virtual environment to simulate the seawater intrusion under real natural conditions, and combine the hydrological data collected by the sky-ground-integrated water conservancy sensing network to achieve the prediction of the storage capacity, storage variable, river infiltration volume, and exploitable volume of the underground reservoir in the physical basin. At the same time, disaster monitoring within the physical basin is realized.

[0055] In S2, the sky-ground-integrated water conservancy sensing network includes the hydrology, water resources, water conservancy projects, soil and water conservation, rivers and lakes, economic society, and ecological environment information of the physical basin. The sky-ground-integrated water conservancy sensing network can sense the digital terrain of the basin underlying surface, vegetation coverage, river water-blocking buildings, floodplain occupation, hydrological cross-section, and underwater terrain of river reaches.

[0056] Data is the most core element of digital twin. It originates from physical entities, operating systems, sensors, etc., covering simulation models, environmental data, physical object design data, maintenance data, operation data, etc., and runs through the entire process of the operation of physical objects. As a data storage platform, the digital twin body collects various types of raw data and then fuses and processes the data to drive the dynamic operation of each part of the simulation model, effectively reflecting each business process. Therefore, data is the blood of digital twin applications. Without multi-source fusion data, digital twin applications will lose their power source.

[0057] This application constructs a real-scene physical simulation device to finely reproduce the panorama and all-element situation of the basin, integrates the data resources of the existing information systems of each water conservancy department, combines the CIM base and the dynamic simulation deduction ability, organically combines information, technology, equipment, and water conservancy management requirements, covers multiple business fields such as basin overview, basin flood control monitoring, reservoir monitoring, and water conservancy dispatching, comprehensively empowers user business applications, and effectively improves the efficiency of cross-departmental decision-making and resource coordination in water conservancy.

[0058] The sky-ground-integrated water conservancy sensing network can collect the hydrology, water resources, water conservancy projects, soil and water conservation, rivers and lakes, economic society, and ecological environment information of the physical basin in a timely and accurate manner, comprehensively master the information such as the digital terrain of the basin underlying surface, vegetation coverage, river water-blocking buildings, floodplain occupation, hydrological cross-section, and underwater terrain of important river reaches, provide a basis for "going from reality to virtuality", and support the operation of the digital twin basin.

[0059] Real-time connection and interaction: It consists of a business network, an industrial control network, a government affairs network, the Internet, a sensor network, communication protocols, input and output devices, security guarantee facilities, and related technologies to achieve efficient connection and transmission, collaborative interactive control, and synchronous iterative optimization among the physical simulation device, virtual basin, digital empowerment service, twin basin data, and twin basin knowledge. Among them, the virtual basin is the digital mirror of the physical basin and the virtual twin body of the digital twin basin, and the physical simulation device is the physical twin body of the digital twin basin.

[0060] The digital twin model with multiple spatio-temporal scales and multiple dimensions includes the macro scale, the meso scale, and the micro scale;

[0061] The macro scale includes vector map data, satellite remote sensing image maps, and DEM elevation map data;

[0062] The meso scale includes satellite remote sensing images, refined manual model data, oblique photography model data, and high-precision DEM data;

[0063] The micro scale includes water conservancy Internet of Things devices and reservoir dams. Through BIM technology, the establishment and detailed display of device models are realized.

[0064] In S3, the basin information data model includes organizing and sorting out different information based on knowledge graph technology. Through nodes and the logical relationships between nodes, the association relationships, index relationships, and spatial relationships between physical basin entities are constructed to form a data model and a knowledge graph. Relevant data resources are integrated through the unified data model and knowledge graph;

[0065] The geographic information reference model includes establishing a set of standardized methods for parsing and storing complex multi-source geographic information. The standardization of geographic information includes the definition of geographic information interoperability, basic data types, modeling rules, and the semantics of real-world phenomena;

[0066] The multi-dimensional visualization model includes applying various spatial and temporal information in the physical world to construct a high-fidelity visual physical model. Through BIM technology, for the spatio-temporal characteristics of each element in the joint operation in the geographical distribution law and development process, real-time rendering and visualization presentation are provided for simulation, and a multi-dimensional, multi-spatio-temporal scale high-fidelity digital mapping is provided for the physical space.

[0067] All sensitive data must be desensitized before being used in the system. Data desensitization specifically includes steps such as terrain data desensitization, decryption processing of the longitude and latitude of water conservancy projects, desensitization processing of basic information such as personnel, and real-time coordinate conversion processing.

[0068] In S4, by enhancing the correlation between different modalities, the feature fusion becomes more complex and refined, improving the expression ability of the digital twin model. The recombined features are fused, and a mixture Gaussian process is used to achieve weighted fusion of the data to further improve the accuracy and reliability of the features. The formula is as follows:

[0069] (1)

[0070] In the formula, F fused represents the fused feature, N is the total number of Gaussian components, wi represents the weight of the i th Gaussian component, represents the Gaussian distribution, F rec is the input feature, μ i , Σ i respectively represent the mean and covariance matrix of the i th Gaussian component. The features of different modalities are comprehensively processed to obtain a more accurate and stable feature representation.

[0071] Using the fused feature F fused as the input, the underground reservoir storage capacity, storage variable, river channel infiltration volume, and exploitable volume of the physical watershed are predicted through a piecewise linear regression model to obtain the predicted output of the physical watershed. The formula is as follows:

[0072] (2)

[0073] In the formula, y pred represents the predicted output, M is the total number of basis functions used in the model, a j represents the coefficient of the j th basis function, which is used to adjust the contribution degree of the basis function to the predicted value, b j represents the bias term of the j th basis function, which is used to adjust the output of the basis function, F fused,j represents the j th fused feature;

[0074] By calculating the error between the predicted output and the actual output, the calibration loss function is obtained:

[0075] (3)

[0076] In the formula, L cal represents the calibration loss function, y real represents the actual output, y pred represents the predicted value;

[0077] Using the gradient descent algorithm to minimize the calibration loss function, the formula for updating the parameters of the digital twin model is as follows:

[0078] (4)

[0079] In the formula, θ and θ ' represent the digital twin model parameters before and after update respectively, η represents the learning rate, ▽ θ L cal represents the gradient of the calibration loss function with respect to the digital twin model parameters. By calibration, the model parameters are adjusted to better match the actual situation and ensure the accuracy of the model.

[0080] A digital twin simulation device for extracting subsurface brackish water to prevent seawater intrusion, which is used to execute the digital twin simulation method for extracting subsurface brackish water to prevent seawater intrusion. The physical simulation device includes a test sand box 1. Both ends of the test sand box 1 are respectively connected with baffles 7 with uniformly distributed holes. The baffle 7 divides the test sand box 1 into a seawater chamber 6 and a fresh water chamber 5 on both sides, and a test chamber 15 in the middle.

[0081] Both sides of the test chamber 15 are respectively connected to the seawater chamber 6 and the fresh water chamber 5 through the arranged holes. It also includes a constant head mechanism 3 for keeping the water level in the seawater chamber 6 balanced; a pumping mechanism 2 for pumping water from the test chamber 15; and a pressure measuring mechanism 4 for measuring the pressure in the test chamber 15.

[0082] The test sand box 1 is made of 1.2 cm thick acrylic transparent plate. The seepage phenomenon can be directly observed on one side, and piezometers or pressure gauges can be installed at corresponding positions on the other side to observe the change of liquid pressure during the test. The test chamber 15 has a specification of 160 cm in length, 10 cm in width, and 70 cm in height. The test chamber 15 is used to fill with washed quartz sand, and 1 cm * 1 cm scale grids are engraved on the front to record the seawater intrusion interface process line during the test.

[0083] The constant head mechanism 3 includes a water tank 10 with a specification of 20 cm in length, 8 cm in width, and 30 cm in height. A partition board 9 is fixedly connected inside the water tank 10. The partition board 9 is 15 cm high. The partition board 9 divides the water tank 10 into an overflow water tank 17 and a constant water tank 16. The constant water tank 16 is connected to the seawater chamber 6 through a connecting pipe, and the connecting pipe is specifically a PU pipe.

[0084] Outlet valves are arranged at intervals of 15 cm from bottom to top on the left side of the seawater chamber 6. The lowermost tap water valve is connected to the seawater chamber 6 through an 8 * 6 mm PU pipe to control the seawater level during the test. The seawater level can be adjusted by adjusting the height of the water tank 10.

[0085] The constant head mechanism 3 further includes a water bucket 10. The overflow water tank 17 is connected to the water bucket 10 through a connecting pipe. A submersible pump 11 with a head of 1.5 m is connected inside the water bucket 10. The submersible pump 11 can pump the water in the water bucket 10 into the fixed water tank 16. When the water in the fixed water tank 16 exceeds the fixed water level, the water flows automatically overflows to the left overflow water tank 17.

[0086] The length, width and height of the seawater chamber 6 and the fresh water chamber 5 are the same. The fresh water chamber 5 is used to simulate the boundary conditions towards the land. Outlet valves are arranged at intervals of 15 cm from top to bottom on the right side of the fresh water chamber 5 for flushing the artificial aquifer at the end of the test. A drain pipe is arranged at the 40 cm valve to maintain the fresh water level at 40 cm during the test.

[0087] The piezometric mechanism 4 includes a side pressure plate. The side pressure plate is a PVC white board with a length of 120 cm and a width of 100 cm. A scale paper 14 with a length of 100 cm is pasted on the white board. One group consists of 9 scale papers 14, which are divided into 4 groups. Small holes are provided above and below the side pressure plate for fixing piezometric tubes. The piezometric tubes are made of 6*4 mm acrylic glass tubes and are connected to the piezometric ports of the test water chamber 15. Compared with the traditional PVC tubes, the acrylic glass tubes are straight and easy to install, and the error caused by manual reading is much smaller than that of the traditional PVC tubes.

[0088] Piezometric ports and pumping wells are arranged on the back of the test water chamber 15 at intervals of 15 cm left and right and 10 cm up and down. The piezometric ports are connected to the piezometric tubes of the side pressure plate through 6*4 mm PU tubes. The pumping mechanism 2 includes a peristaltic pump 12. The pumping wells are connected to the peristaltic pump 12 through silicone tubes, and the aquifer in the test water chamber 15 can be pumped at a constant flow rate during the test.

[0089] Filter units that prevent quartz sand from passing through while allowing water to pass through are laid on the surface of the baffle 7 and the peripheral side of the test water chamber 15. The filter units are geotextiles or sieves. The baffle 7 has a specification of 10 cm in length, 10 cm in width and 10 cm in height. A 100-mesh stainless steel sieve is laid inside the back of the test water chamber 15 to prevent quartz sand from entering the plexiglass tubes of the piezometric plate and the pumping wells during the test.

[0090] The seawater chamber 6 is connected with a joint 19. The connecting pipe connected to the fixed water tank 16 is connected to the joint 19. A water stop plate 20 for plugging the joint 19 is slidably connected inside the joint 19. The water stop plate 20 is connected to the joint 19 through a tension spring 21. The tension spring 21 is located outside the joint 19. The tension spring 21 can be sleeved on the water stop plate 20. The upper end of the tension spring 21 is connected to the water stop plate 20, and the lower end is connected to the outer wall of the joint 19. Under the elastic force of the tension spring 21, the water stop plate 20 keeps the joint 19 in a plugged state.

[0091] The seawater chamber 6 is connected to a fixed pipe 18. A piston 22 is slidably connected inside the fixed pipe 18. The piston 22 is connected to a magnet 25 that attracts the water stop plate 20. The magnetic force of the magnet 25 is greater than the elastic force of the tension spring 21. The piston 22 is connected to the fixed pipe 18 through a spring 26.

[0092] Inside the fixed pipe 18, there is a switch 27 that can control the operation of the submersible pump 11. The switch 27 is located on one side of the piston 22. The piston 22 is fixedly connected to a sleeve 23. The sleeve 23 is threadedly connected to a screw rod 24. The screw rod 24 is connected to the magnet 25. The water stop plate 20 is connected to a one-way valve that only allows water to flow from the fixed water tank 16 into the seawater chamber 6. The one-way valve allows water to flow only in one direction.

[0093] When the water level in the fixed water tank 16 becomes higher, the water in the fixed water tank 16 can flow into the seawater chamber 6 through the one-way valve. When the water level in the seawater chamber 6 is high, due to the effect of the one-way valve, it will not flow into the fixed water tank 16 through the one-way valve, allowing water to flow only in one direction. Thus, the water in the seawater chamber 6 can flow better into the test water chamber 6, and the simulation of seawater intrusion can be better achieved.

[0094] Among them, the spring 26 is sleeved on the sleeve 23. The sleeve 23 is slidably connected to the fixed pipe 18. The right end of the spring 26 is connected to the piston 22, and the left end is connected to the fixed pipe 18. Under the elastic force of the spring 26, the piston 22 can move to the right end of the fixed pipe 18. At this time, the magnet 25 is located on one side of the water stop plate 20 and will not cooperate with the water stop plate 20. The upper end of the water stop plate 20 is an iron block, which is convenient for cooperation with the magnet 25.

[0095] The switch 27 can specifically be a proximity switch or a push switch. The switch 27 and the submersible pump 11 are connected to a controller. When it is a proximity switch, when the piston 22 moves to one side of the proximity switch, the proximity switch can receive a signal, and the submersible pump 11 can be controlled to operate through the controller. The switch 27 can be slidably connected to the fixed pipe 18. The switch 27 is connected to a small iron block, and the position of the switch 27 can be adjusted by an external magnet to drive the switch 27 to slide inside the fixed pipe 18.

[0096] A proximity switch is a position switch that can be operated without mechanical direct contact with moving parts. When the piston 22 approaches the sensing surface of the proximity switch to the operating distance, the switch can be actuated without mechanical contact and without applying any pressure, thereby driving the controller to provide a control instruction.

[0097] The usage method includes the following steps:

[0098] S1: Put the wet quartz sand into the test water chamber 15. While putting in the quartz sand, compact the quartz sand.

[0099] S2: Pour water into the test water chamber 15 and adjust the water level in the test water chamber 15 to the specified height.

[0100] S3: Put colored seawater into the water bucket 10, and start the submersible pump 11 to pump the colored seawater in the water bucket 10 into the fixed water tank 16. Use a water baffle to separate the seawater chamber 6 from the test water chamber 15, and replace the fresh water in the seawater chamber 6 with colored seawater. The water level height of the colored seawater is higher than the water level height of the test water chamber 15.

[0101] S4: Connect the test water chamber 15 to the pumping mechanism 2 and the pressure measuring mechanism 4 respectively. The pumping mechanism 2 works to pump water from the test water chamber 15. The pumping mechanism 2 adopts a point-like pumping method to conduct physical simulation tests of seawater intrusion under different pumping intensities and well layout methods.

[0102] S5: Use a soluble marker pen to depict the saltwater-freshwater interface on the test sand box 1, record the migration of the saltwater-freshwater interface at each stage during the test process, and draw the process line of the seawater intrusion interface during the test process.

[0103] In the simulation of seawater intrusion, when the water in the seawater chamber 6 tends to be stable, the water in the seawater chamber 6 pushes the piston 22 to move, compressing the spring 26. By rotating the screw rod 24, the distance between the magnet 25 and the water stop plate 20 at this time is adjusted. Through an external magnet, the position of the switch 27 in the fixed pipe 18 is adjusted, so as to determine the fluctuation range of the seawater chamber 6.

[0104] When the water level in the seawater chamber 6 increases, the water pressure increases, driving the piston 22 to move to the left. The piston 22 drives the magnet 25 to move. When the magnet 25 moves to one side of the water stop plate 20, it indicates that the water level height has exceeded the threshold at this time. The magnet 25 drives the water stop plate 20 to rise, and the water in the seawater chamber 6 can flow into the fixed water tank 16 through the joint 19.

[0105] When the water level in the seawater chamber 6 decreases, the water pressure decreases, and the spring 26 drives the piston 22 to move to the right. After the piston 22 and the switch 27 cooperate, the submersible pump 11 works. The submersible pump 11 pumps the water in the water bucket 10 into the fixed water tank 16, and then flows into the seawater chamber 6 to replenish water for the seawater chamber 6, preventing small fluctuations from affecting the simulation of seawater intrusion, and at the same time keeping the seawater chamber 6 always at a sufficient and appropriate water level.

[0106] When manufacturing the test sand box 1, a large amount of glue is needed for bonding, resulting in unclear toe angles of the observed seawater intrusion interface. The wall thickness of the test sand box 1 accounts for 1.2 cm at the bottom. At the same time, due to the wall effect (the wall effect means that since the gap between the filler and the tower wall is larger than the gap in the middle of the filler layer, the liquid is prone to flow towards the tower wall and affect the mass transfer effect), the water flow at the contact between the bottom of the sand box and the side wall of the sand box moves faster than that inside the aquifer. Therefore, when selecting data, the toe angle position and length of the seawater intrusion interface are read at a position 2 cm above the bottom of the test sand box 1. In addition, in order to prevent the quartz sand from swelling due to water saturation during the test and causing the sand box to deform and crack, detachable stainless steel frames are installed at the bottom and the main body of the test sand box 1 respectively.

[0107] In S1, in the test water chamber 15, while putting in quartz sand, it is compacted with a stainless steel plate (60 cm × 9.5 cm × 1 cm), and it is compacted as densely as possible to make it nearly in the state of the actual field aquifer. This test is for a single unconfined aquifer, and the thickness of the sand layer is 46 cm.

[0108] The refined quartz sand is used. The appearance of the refined quartz sand is white crystalline, and the particle size is 30 mesh - 40 mesh. Before putting it into the sand box, the quartz sand is repeatedly rinsed with flowing water to wash away the impurities and dust in the quartz sand and make its particle size distribution uniform. The quartz sand is screened with 30 - mesh, 35 - mesh, and 40 - mesh quartz sand sieves respectively.

[0109] In S2, since tap water itself contains a small amount of air, a large amount of air will also be brought into the test water chamber 15 during the process of discharging water into the test water chamber 15. Therefore, during the process of test water discharge, the water should enter the test water chamber 15 slowly with a small flow rate to prevent large air bubbles from entering during the test water discharge. After discharging water, let it stand for a while, and adjust the water level in the sand box until it stabilizes at 40 cm.

[0110] In S3, the fresh water in the seawater chamber 6 is replaced with colored seawater with a water level of 41 cm, and the seawater intrusion process is started. At 270 min, the fresh - salt water interface reaches a stable state for the first time (the piezometric head and the interface shape remain unchanged within 60 min).

[0111] Since Cl⁻ is the most stable major ion in seawater, in this test, seawater is prepared by adding test - pure NaCl and a dye to tap water. Test - pure NaCl is used in the test, with a concentration of 25 g / L; the dye used is carmine (trisodium 1 - (4 - sulfonato - 1 - naphthylazo) - 2 - hydroxy - 6,8 - naphthalenedisulfonate). The mutual interference between sodium chloride and carmine during the test can be ignored. And as a food dye, carmine has little pollution, strong coloring ability, does not react with NaCl and quartz sand, and is easy to wash. Therefore, carmine is used as the dye in this test. In order to ensure the full dissolution of the solute, a certain amount of colored seawater is prepared before the test for use in the test.

[0112] In S4, the test ends when the fresh-saline water interface reaches a stable state again (the piezometric head and the interface shape remain unchanged within 60 minutes).

[0113] Although the embodiments of the present invention have been shown and described, those of ordinary skill in the art can understand that various changes, modifications, substitutions, and variations can be made to these embodiments without departing from the principles and spirit of the present invention. The scope of the present invention is defined by the claims and their equivalents.

Claims

1. A digital twin simulation method for extracting groundwater to prevent seawater intrusion, characterized in that: The following steps are involved: S1: Build a physical simulation device for seawater intrusion, measure the corresponding values ​​through the physical simulation device, and use the proportional coefficient and conversion to obtain the corresponding physical basin information in nature; S2: Feedback the physical watershed information to the watershed virtual twin, build an integrated sky-ground water conservancy perception network, and build a multi-dimensional digital twin model with multiple spatiotemporal scales based on the watershed virtual twin and the integrated sky-ground water conservancy perception network; S3: Build a digital twin model, which includes a watershed information data model, a geographic information reference model, and a multi-dimensional visualization model. Extract data features of multiple spatiotemporal scales and dimensions from the original data in the digital twin model according to unified data standards and business application requirements; S4: The data features are cyclically cross-reorganized based on the cyclic cross-fusion mechanism, and the reorganized features are fused using a mixed Gaussian process. The fused features are input into the calibration model for calibration, and the digital twin model is calibrated through piecewise linear regression. The calibration process includes multiple rounds of simulation and actual data comparison. S5: Run the digital twin model in a virtual environment to simulate the seawater intrusion under real natural conditions. Combined with the hydrological data collected by the integrated sky-ground water conservancy sensing network, the underground reservoir capacity, storage volume, river infiltration and exploitable volume in the physical basin can be predicted, and disaster monitoring within the physical basin can be realized. In S4, the mixed Gaussian process realizes the weighted fusion of data, and the formula is as follows: （1） In the formula, F fused represents the fused features, N is the total number of Gaussian components, w i Indicates i The weights of the Gaussian components, represents a Gaussian distribution, F rec are input features, μ i , Σ i Respectively represent i The mean and covariance matrix of the Gaussian components; Use Fusion Features F fused As input, the groundwater reservoir capacity, storage variable, river infiltration and exploitable volume of the physical watershed are predicted by the piecewise linear regression model, and the predicted output of the physical watershed is obtained as follows: （2） In the formula, y pred represents the predicted output, M is the total number of basis functions used in the model, a j Indicates j The coefficients of the basis functions are used to adjust the contribution of the basis functions to the predicted values. b j Indicates j The bias term of the basis function is used to adjust the output of the basis function. F fused,j Indicates j The fused features; By calculating the error between the predicted output and the actual output, the calibration loss function is obtained: （3） In the formula, L cal represents the calibration loss function, y real Indicates the actual output, y pred represents the predicted value; Using the gradient descent algorithm to minimize the calibration loss function, the formula for the updated digital twin model parameters is as follows: （4） In the formula, θ and θ ´ respectively represent the parameters of the digital twin model before and after the update, η represents the learning rate, ▽ θ L cal Represents the gradient of the calibration loss function with respect to the digital twin model parameters.

2. The digital twin simulation method for extracting groundwater to prevent seawater intrusion according to claim 1 is characterized in that: In S2, the integrated sky-ground water conservancy perception network includes the hydrology, water resources, water conservancy projects, soil and water conservation, rivers and lakes, economic, social and ecological environmental information of the physical watershed. The integrated sky-ground water conservancy perception network can perceive the digital terrain of the underlying surface of the watershed, vegetation coverage, river water-blocking buildings, river beach occupation, large hydrological sections and underwater terrain of river sections.

3. The digital twin simulation method for extracting groundwater to prevent seawater intrusion according to claim 1 is characterized in that: The multi-dimensional digital twin model with multiple spatiotemporal scales includes macroscale, mesoscale and microscale; The macro scale includes vector map data, satellite remote sensing image maps and DEM elevation map data; The mesoscale includes satellite remote sensing images, refined manual model data, oblique photography model data and high-precision DEM data; The micro scale includes water conservancy IoT equipment and reservoir dams. Through BIM technology, the equipment model is established and detailed display is realized.

4. The digital twin simulation method for extracting groundwater to prevent seawater intrusion according to claim 1 is characterized in that: In S3, the watershed information data model includes organizing and sorting different information based on knowledge graph technology, building the association relationship, indicator relationship, and spatial relationship between physical watershed entities through nodes and logical relationships between nodes, forming a data model and knowledge graph, and integrating related data resources through a unified data model and knowledge graph; The geographic information reference model includes a set of standardized geographic information parsing and storage methods for complex multi-source geographic information. The standardization of geographic information includes the definition of geographic information interoperability, basic data types, modeling rules, and semantics of real-world phenomena. The multi-dimensional visualization model applies various spatial and temporal information of the physical world to build a high-fidelity visual physical model. Through BIM technology, it provides real-time rendering and visualization for simulation based on the temporal and spatial characteristics of the geographical distribution patterns and development process of various elements of joint scheduling, and provides high-fidelity digital mapping of physical space in multiple dimensions and time and space scales.

5. A digital twin simulation device for extracting underground salt water to prevent seawater intrusion, characterized in that: The digital twin simulation method for extracting underground salt water to prevent seawater intrusion according to any one of claims 1 to 4 is used to execute the physical simulation device, comprising a test sand box, wherein both ends of the test sand box are respectively connected to baffles with uniformly distributed holes, wherein the baffles divide the test sand box into a seawater chamber and a freshwater chamber located on both sides, and a test water chamber located in the middle, wherein both sides of the test water chamber are respectively connected to the seawater chamber and the freshwater chamber through holes arranged thereon, and further comprising: A constant head mechanism for maintaining the water level balance in the seawater chamber; A pumping mechanism for pumping water from the test water chamber; A pressure measuring mechanism for measuring the pressure of a test water chamber.

6. The digital twin simulation device for extracting underground salt water to prevent seawater intrusion according to claim 5 is characterized in that: The fixed water head mechanism includes a water tank and a water bucket. A baffle is fixedly connected inside the water tank. The baffle divides the water tank into an overflow tank and a fixed water tank. The seawater chamber is connected to the fixed water tank through a joint. The overflow tank is connected to the water bucket through a connecting pipe. A submersible pump is connected inside the water bucket. The submersible pump can pump the water in the bucket into the fixed water tank.

7. The digital twin simulation device for extracting underground salt water to prevent seawater intrusion according to claim 6 is characterized in that: A water stop plate for sealing the joint is slidably connected inside the joint, and the water stop plate is connected to the joint through a tension spring. The seawater chamber is connected to a fixed pipe, and a piston is slidably connected inside the fixed pipe. The piston is connected to a magnet that has an attractive effect on the water stop plate, and the magnetic force of the magnet is greater than the elastic force of the tension spring.

8. The digital twin simulation device for extracting underground salt water to prevent seawater intrusion according to claim 7 is characterized in that: A switch that can control the operation of the submersible pump is connected inside the fixed pipe, and the switch is located on one side of the piston. The piston is connected to the fixed pipe through a spring, and the piston is fixedly connected to a sleeve. The sleeve is threadedly connected to a screw, and the screw is connected to the magnet. The water stop plate is connected to a one-way valve that only allows water to flow from the fixed water tank into the seawater chamber.

Images