A method and system for constructing a natural-artificial coupled virtual water network digital twin model

Abstract

The application discloses a natural-artificial coupling virtual water network digital twin model construction method and system, the method comprising: obtaining historical monitoring data of natural water network and artificial water network; preprocessing the historical monitoring data to obtain a standardized data baseplate; constructing a natural water network model and an artificial water network model based on the standardized data baseplate of the natural water network; and constructing a natural-artificial coupling virtual model based on the natural water network model and the artificial water network model through a graph theory algorithm. The application realizes effective integration of natural water network and artificial water network data, provides a comprehensive and accurate data basis for subsequent model construction, helps to improve comprehensive cognition and analysis capability of the entire water network system, enables the constructed model to comprehensively and accurately simulate the natural and artificial water network system, realizes real-time interaction and sharing of information such as water flow and water quality between the natural water network and the artificial water network, and provides a powerful tool for rational allocation and management of water resources.

Description

Technical Field

[0001] The present application relates to the field of computer technology, and in particular to a method and system for constructing a digital twin model of a natural-artificial coupled virtual water network. Background Art

[0002] With the rapid development of urbanization and industrialization, the rational allocation and management of water resources has become increasingly critical. Natural and artificial water networks are intertwined within the water resources system, jointly fulfilling important functions such as water transportation, distribution, and utilization. However, water network management methods have exposed numerous problems when faced with the complex coupling of natural and artificial water networks.

[0003] Existing technologies are dedicated to solving these problems, such as using the Preissmann four-point implicit difference format to transform the St. Venant equations, constructing multiple internal compatibility condition equations, calculating the water flow state of the water network and the flow of the regulating gate, and optimizing the model by comparing and verifying with physical model tests and prototype observation experimental data.

[0004] However, these technologies still have limitations. Regarding data fusion, the heterogeneity of natural and artificial water network data is not fully addressed, failing to encompass all data formats and sources, impacting the integrity and accuracy of data integration. During model construction, the complex dynamic interactions between natural and artificial water networks are not sufficiently detailed, which can easily lead to model deviations when simulating extreme or complex conditions. Finally, regarding real-time performance, the model's update speed still cannot meet practical requirements, making it difficult to accurately grasp and quickly respond to the real-time status of the water network. Summary of the Invention

[0005] The purpose of this application is to provide a method and system for constructing a natural-artificial coupled virtual water network digital twin model to solve the above technical problems.

[0006] To achieve the above objectives, this application provides the following solutions:

[0007] In a first aspect, the present application provides a method for constructing a digital twin model of a natural-artificial coupled virtual water network, the method comprising:

[0008] Acquiring historical monitoring data of natural and artificial water networks, wherein the historical monitoring data of the natural water network includes hydrological data and water quality data collected by a distributed hydrological sensor array, and the historical monitoring data of the artificial water network includes pipe network structure data, environmental parameters, and control parameters collected by a SCADA system;

[0009] Using a spatiotemporal alignment algorithm to synchronously preprocess the historical monitoring data to obtain a standardized data base;

[0010] Based on the standardized data base of the natural water network, a natural water network model is constructed through a hydrodynamic model;

[0011] Based on the standardized data base of the artificial water network, an artificial water network model is constructed through an LSTM neural network and a fluid dynamics model;

[0012] Based on the natural water network model and the artificial water network model, a natural-artificial coupling virtual model is constructed by a graph theory algorithm. The natural-artificial coupling virtual model is used to characterize the interaction relationship between the artificial water network and the natural water network.

[0013] In one embodiment, constructing a natural-artificial coupling virtual model based on the natural water network model and the artificial water network model by a graph theory algorithm includes:

[0014] Using graph theory algorithms, we identify the connection nodes between the natural and artificial water network models and establish a data exchange channel.

[0015] Establishing the natural-artificial coupling virtual model between the artificial water network and the natural water network according to the data interaction channel;

[0016] A bidirectional interactive interface of the natural-artificial coupling virtual model is configured, wherein the bidirectional interactive interface is used to represent the interactive state of the natural water network and the artificial water network.

[0017] In one embodiment, identifying the connection nodes between the natural water network model and the artificial water network model using a graph theory algorithm includes:

[0018] Traverse all node information in the artificial water network model and the natural water network model;

[0019] Compare all node information in the artificial water network model and the natural water network model. If the nodes in the natural water network model and the nodes in the artificial water network graph meet the connection point identification rules, they are determined as the connection nodes between the natural water network model and the artificial water network model.

[0020] In one embodiment, establishing the natural-artificial coupling virtual model between the artificial water network and the natural water network according to the data interaction channel includes:

[0021] A coupling matrix is ​​established according to the data interaction channel as the natural-artificial coupling virtual model. The coupling matrix is ​​used to characterize the water flow direction, flow conductivity and water quality influencing factors in the water network.

[0022] In one embodiment, configuring the bidirectional interactive interface of the natural-artificial coupling virtual model includes:

[0023] Configure a boundary condition transfer module and a feedback control module. The boundary condition transfer module is used to input the outflow / inflow parameters of the natural water network into the boundary nodes of the artificial water network. The feedback control module is used to reversely input the valve opening and pump station start and stop instructions of the artificial water network into the natural water network model.

[0024] In one embodiment, the coupling matrix is ​​a three-dimensional tensor structure;

[0025] The coupling matrix is ​​expressed as follows:

[0026]

[0027] in, i Indicates the node number in the natural water network, j Represents the node number in the artificial water network. If K 1 ij When it is a non-zero value, it indicates a node in the natural water network i Nodes with artificial water networks j There is a connection relationship; if K 1 ij If it is zero, it means that there is no connection between the two nodes;

[0028] Among them, the first dimension represents the topological connection relationship between natural water network and artificial water network;

[0029] The second dimension represents the flow conductivity coefficient, which is used to characterize the efficiency of transmitting water flow from a node in the natural water network to the corresponding node in the artificial water network;

[0030] The third dimension represents water quality influencing factors, which include dissolved oxygen diffusion coefficient and pollutant attenuation rate.

[0031] In one embodiment, the standardized data base of the artificial water network is used to construct an artificial water network model through an LSTM neural network and a fluid dynamics model, including:

[0032] Inputting the standardized data base of the artificial water network into the constructed initial model, training the initial model so that the initial model outputs an initial prediction result, the initial prediction result including a predicted value of the start-up and shutdown probability of the pump stations in the artificial water network;

[0033] Based on the initial prediction result and the actual pump station start and stop status label, the difference between the predicted value and the true value is calculated, and the corresponding loss function is calculated based on the difference;

[0034] According to the loss function, the parameters of the initial model are updated, and the above training steps are repeated so that the loss function meets the set threshold.

[0035] In one embodiment, the method for constructing the natural-artificial coupled virtual water network digital twin model further includes:

[0036] Obtain real-time monitoring data of artificial and natural water networks;

[0037] Inputting the historical monitoring data into the constructed natural-artificial coupling virtual model to obtain a prediction result;

[0038] Calculating the root mean square error between the real-time monitoring data and the prediction result;

[0039] Based on the root mean square error, the weight coefficient of the natural-artificial coupling virtual model is dynamically adjusted through a parameter sensitivity analysis algorithm until a preset convergence threshold is reached.

[0040] In one embodiment, the method for constructing the natural-artificial coupled virtual water network digital twin model further includes:

[0041] Get real-time monitoring data stream;

[0042] Processing the monitoring data stream in real time based on edge computing nodes;

[0043] Based on the processed monitoring data stream, the digital twin corresponding to the natural-artificial coupled virtual water network model is calibrated online through an incremental model correction algorithm.

[0044] In a third aspect, the present application provides a system for constructing a digital twin model of a natural-artificial coupled virtual water network, comprising:

[0045] An acquisition module is used to acquire historical monitoring data of natural water networks and artificial water networks, wherein the historical monitoring data of natural water networks includes hydrological data and water quality data collected by a distributed hydrological sensor array, and the historical monitoring data of artificial water networks includes pipe network structure data, environmental parameters, and control parameters collected by a SCADA system;

[0046] A preprocessing module, configured to synchronously preprocess the historical monitoring data using a spatiotemporal alignment algorithm to obtain a standardized data base;

[0047] A first construction module is configured to construct a natural water network model based on the standardized data base of the natural water network and through a hydrodynamic model;

[0048] The second construction module is used to construct an artificial water network model based on the standardized data base of the artificial water network through an LSTM neural network and a fluid dynamics model;

[0049] A coupling module is used to construct a natural-artificial coupling virtual model based on the natural water network model and the artificial water network model through a graph theory algorithm, and the natural-artificial coupling virtual model is used to characterize the interaction relationship between the artificial water network and the natural water network.

[0050] According to the specific embodiments provided in this application, the following technical effects are disclosed:

[0051] This application provides a method and system for constructing a digital twin model of a natural-artificial coupled virtual water network. It collects multi-source heterogeneous data through a distributed hydrological sensor array and a SCADA system, and uses a spatiotemporal alignment algorithm to unify coordinates and synchronize time to generate a standardized data base. This achieves effective integration of natural water network and artificial water network data, provides a comprehensive and accurate data foundation for subsequent model construction, and helps to improve the comprehensive understanding and analysis capabilities of the entire water network system.

[0052] On this basis, this application uses a standardized data base to construct a virtual model of the natural water network, which can more realistically simulate the flow movement and water quality changes in the natural water network. It also combines the LSTM neural network with fluid dynamics equations to construct a pipe network control model for the artificial water network, improving the accuracy and intelligence level of artificial water network control. Finally, through graph theory algorithms, the results of the constructed natural water network and artificial water network virtual models are coupled to construct a natural-artificial coupled virtual model, namely a natural-artificial coupled virtual water network digital twin model. This natural-artificial coupled virtual model can achieve comprehensive and accurate simulation of natural and artificial water network systems, providing a powerful tool for the rational allocation and management of water resources. It also realizes real-time interaction and sharing of water flow, water quality and other information between natural and artificial water networks. In other words, the model can accurately reflect the dynamic interaction between the two, and can more comprehensively consider various factors when making water resource management decisions, improving the scientificity and accuracy of decision-making. BRIEF DESCRIPTION OF THE DRAWINGS

[0053] Figure 1 A flowchart of a method for constructing a digital twin model of a natural-artificial coupled virtual water network according to some embodiments of the present application;

[0054] Figure 2 A schematic diagram illustrating a coupling matrix tensor visualization in some embodiments of the present application;

[0055] Figure 3 This is a schematic diagram of the LSTM neural network structure in some embodiments of the present application;

[0056] Figure 4 This is a diagram of the edge computing node deployment architecture in some embodiments of the present application;

[0057] Figure 5A schematic diagram of the structure of a system for constructing a virtual model of an artificial-natural coupled water network in some embodiments of the present application;

[0058] Figure 6 A schematic structural diagram of a processing device provided in some embodiments of the present application. DETAILED DESCRIPTION

[0059] The following will be combined with the drawings in the embodiments of this application to clearly and completely describe the technical solutions in the embodiments of this application. Obviously, the embodiments described are only part of the embodiments of this application, not all of the embodiments. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of this application.

[0060] In order to make the above-mentioned purposes, features and advantages of the present application more obvious and easy to understand, the present application is further described in detail below with reference to the accompanying drawings and specific implementation methods.

[0061] It can be understood that in this application, in order to improve the scientificity and accuracy of the construction of the water network digital twin model to achieve more reasonable allocation and management of water resources. By enriching and integrating source data, finely depicting the dynamic interaction relationship between natural and artificial water networks, and combining real-time data dynamic updates, the richness of source data when building the model, the accuracy of the constructed model in simulating the dynamic interaction between natural and artificial water networks, and the precise grasp and rapid responsiveness of the real-time status of natural and artificial water networks are ultimately ensured.

[0062] Specifically, this application first collects heterogeneous data from multiple sources through a distributed hydrological sensor array and SCADA system. It then uses a spatiotemporal alignment algorithm to unify coordinates and synchronize time, generating a standardized data base. This process effectively integrates data from both natural and artificial water networks, providing a comprehensive and accurate data foundation for subsequent model construction and improving comprehensive understanding and analysis of the entire water network system.

[0063] Furthermore, a three-dimensional distributed hydrological and hydrodynamic coupling model of the natural water network was constructed, integrating terrain-driven algorithms and water quality transport equations to more realistically simulate water flow and water quality changes within the natural water network. A pipe network control model for the artificial water network was constructed by combining an LSTM neural network with fluid dynamics equations. This model incorporates valve control logic, pump station scheduling strategies, and hydraulic response characteristics, improving the accuracy and intelligence of artificial water network control. The collaborative construction of these two models enables a comprehensive and accurate simulation of the natural-artificial water network system, providing a powerful tool for the rational allocation and management of water resources. Using graph theory algorithms to identify connecting nodes and establish a coupling matrix, a bidirectional interactive interface was established, enabling real-time interaction and sharing of flow, water quality, and other information between the natural and artificial water networks. This enables the model to accurately reflect the dynamic interaction between the two, allowing for a more comprehensive consideration of various factors in water resource management decisions, improving the scientific nature and accuracy of decision-making.

[0064] In order to better understand the method for constructing the natural-artificial coupled virtual water network digital twin model provided in this application, it is described in detail below with reference to the accompanying drawings.

[0065] S110, obtaining historical monitoring data of natural water networks and artificial water networks, wherein the historical monitoring data of the natural water network includes hydrological data and water quality data collected by a distributed hydrological sensor array, and the historical monitoring data of the artificial water network includes pipe network structure data, environmental parameters, and control parameters collected by a SCADA system.

[0066] Specifically, in the embodiment of the present application, in order to construct a natural-artificial coupled virtual water network model, historical monitoring data of the natural water network and the artificial water network can be collected first as training data.

[0067] That is, for the source data of the natural water network, the hydrological, meteorological, topographic and water quality data of the natural water network can be collected through a distributed hydrological sensor array.

[0068] Among them, terrain data is used to implement the subsequent terrain-driven algorithm, and hydrological data (flow, water level), meteorological data (precipitation, evaporation) and water quality data (pollutant concentration, dissolved oxygen content) are used as the initial input conditions and boundary conditions of the model.

[0069] For the source data of the artificial water network, the pipe network structure, flow, pressure and control parameters of the artificial water network can be obtained through the SCADA system.

[0070] S120: Synchronously preprocess the historical monitoring data using a spatiotemporal alignment algorithm to obtain a standardized data base.

[0071] Specifically, in some embodiments of the present application, a spatiotemporal alignment algorithm can be used to coordinate unify and time synchronize heterogeneous data to generate a standardized data base.

[0072] Specifically, the Kriging spatial interpolation algorithm can be used to compensate for missing values ​​in historical monitoring data from natural water networks. Linear timestamp resampling can be used to achieve millisecond-level time synchronization for historical monitoring data from artificial water networks. Finally, a coordinate transformation matrix can be used to unify the data formats of the geographic coordinate system and the engineering coordinate system, resulting in a corresponding standardized data base.

[0073] It can be understood that in the embodiment of the present application, after the training data of the natural water network and the artificial water network are obtained through the above steps, a natural water network virtual model and an artificial water network virtual model can be constructed based on the configured algorithm.

[0074] That is, S130, based on the standardized data base of the natural water network, constructs a natural water network model through a hydrodynamic model.

[0075] Specifically, after obtaining the standardized base plate data of the natural water network through the above steps, a three-dimensional distributed natural water network model of the natural water network, namely, a natural water network model, can be constructed based on the data base plate and by using the principles of hydrodynamics.

[0076] In practice, this natural water network model integrates a terrain-driven algorithm and a water mass transport equation. Specifically, the terrain-driven algorithm and the water mass transport equation can be constructed separately, and then these two can be coupled with the hydrodynamic model to construct the natural water network model.

[0077] Among them, the construction process of the terrain-driven algorithm is to determine the initial flow direction and flow velocity distribution of the water flow based on the terrain data using the terrain-driven algorithm.

[0078] For example, the terrain slope and slope direction are calculated using digital elevation model (DEM) data. Based on this information, the initial movement direction of water in the natural water network is determined according to the principle that water always flows from high to low.

[0079] Incorporating terrain slope information into the hydrodynamic equations influences the acceleration and deceleration of water flow. In the hydrodynamic equations, terrain slope is a key parameter that affects the gravity term of the water flow, thereby influencing the magnitude and direction of the water flow.

[0080] The process of constructing this water transport equation involves establishing an equation based on the law of conservation of mass to describe the transport and transformation of water components within a stream. For dissolved pollutants, the transport equation typically includes convection, diffusion, and source / sink terms. Convection terms describe the migration of pollutants along the stream, diffusion terms account for molecular and turbulent diffusion of pollutants within the stream, and source / sink terms represent the processes by which pollutants are generated and removed, such as chemical reactions and biodegradation.

[0081] Finally, the constructed water transport equation is coupled with the hydrodynamic model. After the hydrodynamic model calculates water velocity and water level information, these results are used as inputs to the water transport equation to calculate the concentration distribution changes of pollutants in the water network.

[0082] In practice, numerical methods such as the finite difference method, the finite element method, or the finite volume method can be used to solve the hydrodynamic equations. These methods discretize the continuous hydrodynamic equations in space and time, transforming them into a solvable algebraic system of equations.

[0083] According to the actual boundary conditions of the natural water network (such as the flow rate and water level at the inflow boundary, and the water level or flow rate at the outflow boundary), the discretized set of equations is solved to obtain hydrodynamic parameters such as water flow velocity and water level at different times and locations.

[0084] Furthermore, the model can be verified and adjusted. This means using historical monitoring data to verify the constructed model and compare the differences between the model simulation results and actual observations, such as water level, flow rate, and water quality concentration. If there is a significant deviation between the simulation results and the actual data, the cause is analyzed and the model parameters are adjusted until the model simulation results better match the actual situation.

[0085] It can be understood that the output of the natural water network model constructed in the embodiment of the present application may specifically include hydrodynamic parameters, hydrodynamic parameters and terrain-related derivative parameters.

[0086] Hydrodynamic parameters include flow velocity (in the x, y, and z directions), water level, flow rate, and hydraulic gradient. These parameters specifically reflect the motion and energy changes of water flows in natural water networks and are crucial for analyzing flow stability, scouring, and sedimentation.

[0087] This water quality parameter primarily outputs the concentration distribution of various pollutants, such as chemical oxygen demand (COD), biochemical oxygen demand (BOD), ammonia nitrogen, heavy metal ion concentrations, and dissolved oxygen concentration. These parameters can intuitively demonstrate the water quality status of natural water networks, providing key data support for water resource quality assessment and water pollution control.

[0088] While terrain data is the model's input, the model also outputs terrain-related derivative parameters during operation, such as flow path length and catchment area. These parameters provide a deeper understanding of the interaction between natural water networks and terrain, and are of great value in studying watershed hydrological processes.

[0089] It can be understood that in the embodiment of the present application, the natural water network model is constructed through the hydrodynamic model, which can accurately simulate and output the hydrodynamic parameters, hydrodynamic parameters and terrain-related derivative parameters, etc., providing a basis for the construction of a natural-artificial coupled virtual water network model.

[0090] S140, based on the standardized data base of artificial water networks, constructs an artificial water network model through LSTM neural network and fluid dynamics model.

[0091] Specifically, in this embodiment, the obtained standardized data base of the artificial water network can be trained and processed by using the LSTM neural network and the fluid dynamics equation to construct a virtual model of the artificial water network.

[0092] It can be understood that the constructed artificial water network virtual model may specifically include valve control logic, pump station scheduling strategy and hydraulic response characteristics.

[0093] Among them, the valve control logic: during peak water usage, in order to ensure the water supply to key areas, the valve control logic will automatically adjust the opening of relevant valves according to the water usage priority of each area, giving priority to ensuring water supply to key areas; when a certain section of the pipeline network has abnormal pressure, the valve will automatically adjust the opening according to the preset logic to maintain the pipeline network pressure stable and prevent pipe bursts or water shortages.

[0094] Valve control logic algorithms include the proportional-integral-differential (PID) control algorithm based on feedback control, which adjusts the valve opening through proportional, integral, and differential operations according to the deviation between the actual pressure and the set pressure; and the fuzzy control algorithm, which fuzzifies parameters such as pressure and flow, makes inference decisions based on the fuzzy rule base, and thus realizes intelligent control of the valve.

[0095] Pump station scheduling strategy: During the nighttime low water consumption period, the operating power of some pumps can be appropriately reduced or some pumps can be stopped according to the water level of the water network to save energy; when there is a sudden increase in water demand, the backup pumps can be quickly started according to the predetermined scheduling strategy to ensure the normal water supply of the water network.

[0096] Pump station scheduling strategies include rule-based scheduling strategies, which control the start and stop of water pumps according to pre-set rules such as water level and flow thresholds; and optimization scheduling strategies, such as using optimization methods such as linear programming and genetic algorithms, with minimum energy consumption and maximum water supply satisfaction rate as objective functions to solve the optimal water pump operation combination and scheduling plan.

[0097] Hydraulic response characteristics: When a valve is suddenly closed, the pressure upstream and downstream will rise and fall rapidly, and the flow rate will also change accordingly. The change process and amplitude of this pressure and flow rate are the embodiment of the hydraulic response characteristics.

[0098] It's also understandable that hydraulic response characteristics vary under different water network structures and operating conditions. Studying these characteristics can help engineers better understand the network's operating patterns, predict the potential impact of regulatory measures, and optimize valve control logic and pump station scheduling strategies. In practical applications, hydraulic simulation software is often used to simulate and analyze the hydraulic response under different regulation schemes, generating hydraulic response characteristic data that provides a scientific basis for water network regulation.

[0099] In practice, the training methods of LSTM neural networks can include:

[0100] S141, inputting the standardized data base of the artificial water network into the constructed initial model, training the initial model so that the initial model outputs an initial prediction result, which includes a prediction value of the start-up and shutdown probability of the pump stations in the artificial water network.

[0101] S142: Based on the initial prediction result and the actual pump station start / stop status label, the difference between the predicted value and the true value is obtained, and the corresponding loss function is calculated based on the difference.

[0102] S143, updating the parameters of the initial model according to the loss function, and repeating the above training steps so that the loss function meets the set threshold.

[0103] Specifically, if Figure 2 As shown in the figure, the input layer receives standardized data base, such as the pressure gradient of the artificial water network and the time series data of valve opening; the hidden layer sets a bidirectional recurrent unit containing a forget gate and an output gate; the output layer generates a predicted value of the start and stop probability of the pump station, and optimizes the network weights jointly through the back propagation algorithm and the residual of the fluid dynamics equation.

[0104] The specific training process of LSTM neural network:

[0105] Data preparation: Collect historical time series data on pressure gradients and valve openings in the artificial water network and divide the data into training and test sets. Normalize the data to be in the range [0, 1] or [-1, 1] to accelerate model training convergence.

[0106] Initialize model parameters: Initialize the weights and bias parameters of the LSTM neural network, and set hyperparameters such as the number of hidden layer neurons and the bidirectional recurrent unit structure.

[0107] Forward propagation: The training set data is input into the LSTM neural network in sequence. After the input layer receives the data, the data is processed in the bidirectional recurrent unit of the hidden layer. The memory unit is updated through the operations of the forget gate, input gate, and output gate. Finally, the predicted value of the start and stop probability of the pump station is generated in the output layer.

[0108] Calculate the loss function: Compare the predicted value with the actual pump station start and stop status label, and use the cross entropy loss function or mean square error loss function to calculate the difference between the predicted value and the true value.

[0109] Backpropagation and parameter update: The backpropagation algorithm is used to calculate the gradient of the loss function with respect to the weights and biases of each layer of the neural network. Combined with the residual of the fluid dynamics equation, the weight and bias parameters are updated through the optimizer (such as Adam, SGD, etc.) to reduce the loss function value.

[0110] Model evaluation and iteration: Evaluate model performance on the test set, such as calculating metrics like accuracy and recall. If the model performance does not meet expectations, continue training iterations, repeating the above steps until the model converges or the preset number of training rounds is reached.

[0111] S150, based on the natural water network model and the artificial water network model, construct a natural-artificial coupling virtual model, i.e., a natural-artificial coupling virtual water network digital twin model, through a graph theory algorithm. The natural-artificial coupling virtual model is used to characterize the interaction relationship between the artificial water network and the natural water network.

[0112] Specifically, after constructing the virtual models of the artificial water network and the natural water network through the above steps, we can identify the connection points of the two virtual models and set up a two-way interactive interface to finally construct an artificial-natural coupled virtual water network model, that is, a natural-artificial coupled virtual water network digital twin model, to characterize the interactive relationship between the two.

[0113] Optionally, in some embodiments, this step may specifically include the following steps:

[0114] S151, using graph theory algorithms to identify the connection nodes between the natural water network model and the artificial water network model, and establish a data exchange channel;

[0115] S152, establishing a natural-artificial coupling virtual model between the artificial water network and the natural water network according to the data exchange channel;

[0116] S153: Configuring a bidirectional interactive interface of a natural-artificial coupling virtual model, wherein the bidirectional interactive interface is used to represent the interactive state of the natural water network and the artificial water network.

[0117] Specifically, we can first use graph theory algorithms to identify the connection node information of the natural water network model and the artificial water network model, and then determine the connection points between the two through the identification results. Then, we can associate the identified connection points with each other based on factors such as geographical location, that is, associate the connection points between the natural water network and the artificial water network to obtain a group of mutually associated connection points. Finally, we can establish a data interaction channel based on the association results (a data interaction channel is formed between the two associated connection points), and then establish a coupling matrix containing water flow direction, flow conductivity coefficient and water quality influencing factors based on the established data interaction channel, that is, construct a virtual model of the artificial-natural coupled water network.

[0118] Finally, a two-way interactive interface can be set for the constructed coupling matrix, that is, a boundary condition transfer module and a feedback control module can be configured.

[0119] Among them, the boundary condition transfer module inputs the outflow / inflow parameters of the natural water network into the boundary nodes of the artificial water network, and the feedback control module reversely inputs the valve opening and pump station start and stop instructions of the artificial water network into the natural water network model.

[0120] Optionally, in some embodiments, S131 may specifically include the following steps:

[0121] S01, traverse all node information in the artificial water network model and the natural water network model.

[0122] S02, comparing all node information in the artificial water network model and the natural water network model, if the node in the natural water network model and the node in the artificial water network diagram meet the connection point identification rules, then determine it as the connection node between the natural water network model and the artificial water network model.

[0123] In practice, if a river in a natural water network is directly connected to a section of a pipe in an artificial water network, and there is a possibility of water exchange, then the graph vertices corresponding to the two connected positions may be connection points; or when a pump station in an artificial water network draws water from or discharges water to the natural water network, the vertices corresponding to the pump station location and the natural water network location connected to it can also be regarded as connection points.

[0124] When identifying connection points, all node information in both the artificial and natural water network models can be traversed. This means starting from a vertex in the natural water network graph, the adjacent vertices are accessed along the edges, and during this access, the vertex information in the artificial water network graph is compared. If a vertex in the natural water network graph meets the connection point identification criteria (e.g., physical proximity and interchangeable water flows), it is identified as a connection point.

[0125] Similarly, you can also start from the starting vertex and expand the search outward layer by layer, and also determine whether it is a connection point based on the rules.

[0126] In practical applications, it can also be combined with geographic information system (GIS) data to more intuitively assist in determining the location of connection points and improve the accuracy and efficiency of identification.

[0127] Furthermore, after identifying the connection points between the artificial water network and the natural water network, a coupling matrix can be constructed based on the associated connection point group as the natural-artificial coupling virtual model. The coupling matrix is ​​used to characterize the water flow direction, flow conductivity and water quality influencing factors in the water network.

[0128] Optionally, in some embodiments, the coupling matrix may be a three-dimensional tensor structure.

[0129] Then Figure 3 As shown, the constructed coupling matrix can be expressed as:

[0130]

[0131] in, i Indicates the node number of the natural water network, j Indicates the node number of the artificial water network; when K 1 ij Here we only consider the value of the first dimension, not the specific values ​​of other dimensions) When it is non-zero, it means the node of the natural water network i Nodes with artificial water networks j There is a connection, which may be a direct physical connection, such as the connection between a river and a pipeline; if K 1 ij If is zero, it means that there is no connection between the two nodes. It is the basic structure of the entire coupling matrix, which determines which nodes can interact with each other in terms of water flow, flow rate and water quality information, and builds a framework for the subsequent information transmission in the two dimensions. Figure 3 K11 shown indicates that there is some connection between node 1 of the natural water network and node 1 of the artificial water network.

[0132] Among them, Figure 3 As shown, the first dimension represents the topological connection relationship between the natural water network and the artificial water network, such as including three natural water network nodes and two artificial water network nodes.

[0133] The second dimension represents the flow conductivity coefficient, which directly reflects the efficiency with which a node in a natural water network transmits water to a corresponding node in an artificial water network. For example, a high flow conductivity coefficient between connected nodes indicates that, under conditions such as the same water level difference, the river can efficiently transport water to the pipeline; a low coefficient indicates poor water transfer efficiency. For example, for urban water supply systems that draw water from natural sources, a high flow conductivity coefficient ensures a stable and sufficient water supply; a low coefficient can lead to water shortages, affecting residents' access to water.

[0134] Different areas correspond to different flow conduction coefficients.

[0135] Its value is calculated by the Manning formula and the pipe diameter parameter. The calculation formula is:

[0136]

[0137] in, K ij is the flow conduction coefficient from node i to j, n is the Manning roughness coefficient, A ij is the cross-sectional area of ​​water flow, R ij is the hydraulic radius, S ij is the hydraulic slope.

[0138] In practice, A in the above Manning formula ij 、R ij The specific measurement method can be to use the ADCP velocity profiler to measure the velocity distribution data of the water flow. By measuring the flow velocity at different depths, the average flow velocity V of the water section can be calculated.

[0139] At the same time, the cross-sectional area A of the water flow is determined by combining the location information of the measurement point and the geometric shape of the channel or pipe. ij For example, for a rectangular channel, A ij =width × water depth.

[0140] For circular pipes, A ij =π×(pipe diameter / 2)^2.

[0141] Hydraulic radius R ij According to the cross-sectional area A ij and wet periphery P ij Calculated, that is, R ij =A ij / P ij .

[0142] For rectangular channels, P ij =2×water depth+width.

[0143] For circular pipes, P ij =π×pipe diameter.

[0144] The third dimension records water quality influencing factors, including dissolved oxygen diffusion coefficient and pollutant attenuation rate, where the pollutant attenuation rate satisfies:

[0145]

[0146] Where C is the pollutant concentration, k is the reaction rate constant, and m is the reaction order, which are determined by fitting the on-site monitoring data.

[0147] Different regions correspond to different water quality influencing factors.

[0148] It can be understood that by constructing a three-dimensional distributed hydrological and hydrodynamic coupling model of the natural water network through the above-mentioned embodiment, integrating the terrain-driven algorithm and the water quality transport equation, it is possible to more realistically simulate the water flow movement and water quality change process in the natural water network. In addition, the LSTM neural network and the fluid dynamics equation are combined to construct a pipe network control model for the artificial water network, which includes valve control logic, pump station scheduling strategy and hydraulic response characteristics, thereby improving the accuracy and intelligence level of artificial water network control. The collaborative construction of the two models realizes a comprehensive and accurate simulation of the natural-artificial water network system, providing a powerful tool for the rational allocation and management of water resources. By using graph theory algorithms to identify connection nodes and establish a coupling matrix, and setting a two-way interactive interface, real-time interaction and sharing of water flow, water quality and other information between the natural water network and the artificial water network are realized.

[0149] Furthermore, in some embodiments of the present application, in order to improve the accuracy of the model output, the constructed model can also be verified and optimized online.

[0150] That is, in some embodiments, the method may further include the following steps:

[0151] S001, obtain real-time monitoring data of artificial and natural water networks;

[0152] S002, inputting the historical monitoring data into the constructed natural-artificial coupling virtual model to obtain a prediction result;

[0153] S003, calculating the root mean square error between the real-time monitoring data and the prediction result;

[0154] S004: Based on the root mean square error, dynamically adjust the weight coefficient of the natural-artificial coupling virtual model through a parameter sensitivity analysis algorithm until a preset convergence threshold is reached.

[0155] Specifically, when performing online verification and optimization in this embodiment, real-time monitoring data can be obtained first, and then the historical hydrological data collected in the above S110 can be used to drive the model operation, calculate the root mean square error between the measured data and the simulation results, and dynamically adjust the model weight coefficient through parameter sensitivity analysis until the preset convergence threshold is reached.

[0156] The historical monitoring data acquired in S110, collected through the distributed hydrological sensor array and SCADA system, accumulates over time and constitutes a significant component of historical data. This data, after undergoing spatial and temporal alignment and other processing to generate a standardized data base, is used in this step to drive the model and facilitate self-adjustment and optimization.

[0157] The parameter sensitivity analysis may specifically include:

[0158] The Sobol index method is used to identify key parameters that contribute more than 5% to the root mean square error in the hydrological and hydrodynamic coupling model. The total sensitivity index is calculated as:

[0159]

[0160] Among them, S Ti is the overall sensitivity index of parameter i, represents the variance of all parameters except i, E Xi is the expected value of parameter i; Var(Y) is the variance of model output Y;

[0161] When all parameters except i are fixed, the conditional variance after taking the expectation of parameter i is calculated by Monte Carlo sampling;

[0162] The weight coefficients of key parameters are adjusted by Bayesian optimization algorithm, and the objective function is defined as:

[0163]

[0164] Where θ is the parameter to be optimized, θ0 is the initial parameter value, α and β are balance factors, and the range of values ​​is α∈[0.6,0.8] and β∈[0.2,0.4]. After each iteration, verify whether the model convergence rate is improved by more than 15%.

[0165] is the L2 norm of the parameter change.

[0166] It can be understood that after multiple experimental verifications, in this model, when the balance factors α=0.7 and β=0.3, the model fitting accuracy and parameter stability can be better balanced.

[0167] The number of parameter sampling times when calculating the Sobol index: 10,000 Monte Carlo samplings are used to calculate the Sobol index. During the calculation process, a large number of parameter samples are randomly generated, and the total sensitivity index S is calculated based on the model output results. Ti , identify key parameters in the hydrological and hydrodynamic coupling model that contribute more than 5% to the root mean square error.

[0168] Furthermore, in some embodiments of the present application, in order to improve the accuracy of model construction, the monitoring data stream can be processed in real time based on the edge computing node, and the digital twin composed of the natural water network model and the artificial water network model can be calibrated online using an incremental model correction algorithm.

[0169] That is, the method may further comprise the following steps:

[0170] S005, obtaining real-time monitoring data stream;

[0171] S006, processing the monitoring data stream in real time based on the edge computing node;

[0172] S007: Based on the processed monitoring data stream, online calibration is performed on the digital twin corresponding to the natural-artificial coupled virtual water network model through an incremental model correction algorithm.

[0173] Specific, combined Figure 4 As shown in the figure, incremental twin updates can first process monitoring data streams in real time based on edge computing nodes, and then use incremental model correction algorithms to perform online calibration on the digital twins composed of natural water network models and artificial water network models, and synchronously update the spatiotemporal distribution map of the virtual water network in the three-dimensional geographic information engine. The engine can integrate GIS and BIM technologies to achieve multi-level visualization.

[0174] In practice, the steps for real-time processing of monitoring data streams based on edge computing nodes are as follows:

[0175] In urban water supply systems, distributed hydrological sensors monitor water levels, flow rates, and water quality in natural water networks. Artificial water networks use SCADA systems to collect data such as pressure, flow rates, valve openings, and pump station operating status at each node in the network. These sensors and data collection devices continuously generate large amounts of monitoring data streams. Edge computing nodes access data from various sensors and systems, aggregating this dispersed monitoring data.

[0176] The collected raw data often contains noise, missing values, and inconsistent data formats. The data preprocessing module built into the edge computing node cleans, removes noise, and fills in missing values ​​on the aggregated data. For example, abnormal fluctuations in water quality data are identified and corrected by setting reasonable thresholds. Missing values ​​in artificial water network flow data are filled using linear interpolation or machine learning-based methods. Furthermore, data in different formats is unified and converted into a standard format suitable for subsequent processing to ensure data accuracy and consistency.

[0177] Real-time analysis and decision-making functions of pre-processed data can quickly respond to abnormal situations in water network operation and ensure the safe and stable operation of the water network.

[0178] While edge computing nodes can perform most data processing and decision-making locally, they still need to interact with the cloud for some complex analytical tasks and large-scale data storage. Edge computing nodes upload initially processed and analyzed data to the cloud, which leverages its powerful computing resources for deeper data analysis and mining, such as long-term trend analysis and predictive model training. The cloud can also distribute updated model parameters or optimized decision-making strategies to edge computing nodes, enabling them to continuously optimize their processing capabilities and decision-making capabilities.

[0179] Virtual Water Network Spatiotemporal Distribution Map: The 3D geographic information engine integrates GIS and BIM technologies to present a visual 3D spatiotemporal distribution map of the virtual water network. This map, updated in real time, allows managers and decision makers to understand the network's operational status, including real-time flow direction, flow rate, and water quality.

[0180] The three-dimensional atlas map provides a visual representation of the distribution of pipe network pressure in different areas. Areas with excessively high or low pressure are highlighted with different colors or symbols, allowing managers to promptly identify problems and take action. This visual monitoring significantly improves awareness of the water network's operational status and facilitates rapid decision-making. Based on the changing trends in water flow velocity and flow rate shown in the atlas, it is possible to predict whether certain areas will face water shortages during peak water usage periods, allowing for the development of water allocation plans in advance. Furthermore, the atlas can be used for simulation analysis during water network planning and renovation, assessing the impact of different options on network operation and selecting the optimal one.

[0181] The incremental model correction algorithm uses a sliding window mechanism:

[0182] The window length is 24 hours for monitoring data streams, and each additional hour of data triggers a local parameter calibration;

[0183] During the calibration process, the verified model parameters are frozen, and only the target parameters whose error change rate exceeds the threshold within the window period are adjusted.

[0184] The incremental model correction algorithm uses a sliding window mechanism with a window length of 24 hours of monitoring data stream. Every additional hour of data triggers a local parameter calibration. Let t represent the time step, θ represent the model parameters, e(t) represent the error at time t, θ(t) represent the updated parameters at time t, η represent the learning rate, and β represent the momentum factor (usually in the range [0,1]). The iterative formula is as follows:

[0185]

[0186]

[0187] Where v(t) is the parameter update rate. In each iteration, the parameters are updated based on the error gradient at the current moment and the velocity at the previous moment. During calibration, the verified model parameters are frozen, and only the target parameters whose error change rate exceeds the threshold during the window period are adjusted.

[0188] On the other hand, the present application also provides a natural-artificial coupled virtual water network model construction system.

[0189] like Figure 5 As shown, the system includes:

[0190] Acquisition module 210, for acquiring historical monitoring data of natural water networks and artificial water networks, wherein the historical monitoring data of natural water networks includes hydrological data and water quality data collected by a distributed hydrological sensor array, and the historical monitoring data of artificial water networks includes pipe network structure data, environmental parameters, and control parameters collected by a SCADA system;

[0191] A preprocessing module 220 is used to synchronously preprocess the historical monitoring data using a spatiotemporal alignment algorithm to obtain a standardized data base;

[0192] A first construction module 230 is configured to construct a natural water network model based on the standardized data base of the natural water network and using a hydrodynamic model;

[0193] A second construction module 240 is configured to construct an artificial water network model based on the standardized data base of the artificial water network by using an LSTM neural network and a fluid dynamics model;

[0194] The coupling module 250 is used to construct a natural-artificial coupling virtual model based on the natural water network model and the artificial water network model through a graph theory algorithm. The natural-artificial coupling virtual model is used to characterize the interaction between the artificial water network and the natural water network.

[0195] Optionally, in some embodiments of the present application, the artificial-natural coupled water network virtual model construction system provided by the coupling module is specifically used to:

[0196] Using graph theory algorithms, we identify the connection nodes between the natural and artificial water network models and establish a data exchange channel.

[0197] Establishing the natural-artificial coupling virtual model between the artificial water network and the natural water network according to the data interaction channel;

[0198] A bidirectional interactive interface of the natural-artificial coupling virtual model is configured, wherein the bidirectional interactive interface is used to represent the interactive state of the natural water network and the artificial water network.

[0199] Optionally, in some embodiments of the present application, the artificial-natural coupled water network virtual model construction system provided by the coupling module is specifically used to:

[0200] Traverse all node information in the artificial water network model and the natural water network model;

[0201] Compare all node information in the artificial water network model and the natural water network model. If the nodes in the natural water network model and the nodes in the artificial water network graph meet the connection point identification rules, they are determined as the connection nodes between the natural water network model and the artificial water network model.

[0202] Optionally, in some embodiments of the present application, the artificial-natural coupled water network virtual model construction system provided by the coupling module is specifically used to:

[0203] A coupling matrix is ​​constructed based on the established data interaction channel as the natural-artificial coupling virtual model. The coupling matrix is ​​used to characterize the water flow direction, flow conductivity and water quality influencing factors in the water network.

[0204] Optionally, in some embodiments of the present application, the artificial-natural coupled water network virtual model construction system provided by the coupling module is specifically used to:

[0205] Configure a boundary condition transfer module and a feedback control module. The boundary condition transfer module is used to input the outflow / inflow parameters of the natural water network into the boundary nodes of the artificial water network. The feedback control module is used to reversely input the valve opening and pump station start and stop instructions of the artificial water network into the natural water network model.

[0206] Optionally, in the artificial-natural coupled water network virtual model construction system provided in some embodiments of the present application, the coupling matrix is ​​a three-dimensional tensor structure;

[0207] The coupling matrix is ​​expressed as follows:

[0208]

[0209] in, i Indicates the node number in the natural water network, j Represents the node number in the artificial water network. If K 1 ij When it is a non-zero value, it indicates a node in the natural water network i Nodes with artificial water networks j There is a connection relationship; if K 1 ij If it is zero, it means that there is no connection between the two nodes;

[0210] Among them, the first dimension represents the topological connection relationship between natural water network and artificial water network;

[0211] The second dimension represents the flow conductivity coefficient, which is used to characterize the efficiency of transmitting water flow from a node in the natural water network to the corresponding node in the artificial water network;

[0212] The third dimension represents water quality influencing factors, which include dissolved oxygen diffusion coefficient and pollutant attenuation rate.

[0213] Optionally, in the artificial-natural coupled water network virtual model construction system provided in some embodiments of the present application, the third construction module is specifically used to:

[0214] The standardized data base based on the artificial water network, through the LSTM neural network and the fluid dynamics model, constructs the artificial water network model, including:

[0215] Inputting the standardized data base of the artificial water network into the constructed initial model, training the initial model so that the initial model outputs an initial prediction result, the initial prediction result including a predicted value of the start-up and shutdown probability of the pump stations in the artificial water network;

[0216] Based on the initial prediction result and the actual pump station start and stop status label, calculate the difference between the predicted value and the true value, and calculate the corresponding loss function based on the difference;

[0217] According to the loss function, the parameters of the initial model are updated, and the above training steps are repeated so that the loss function meets the set threshold.

[0218] Optionally, the artificial-natural coupled water network virtual model construction system provided in some embodiments of the present application further includes a verification and optimization module 160, which is specifically used to:

[0219] Obtain real-time monitoring data of artificial and natural water networks;

[0220] Inputting the historical monitoring data into the constructed natural-artificial coupling virtual model to obtain a prediction result;

[0221] Calculating the root mean square error between the real-time monitoring data and the prediction result;

[0222] Based on the root mean square error, the weight coefficient of the natural-artificial coupling virtual model is dynamically adjusted through a parameter sensitivity analysis algorithm until a preset convergence threshold is reached.

[0223] Optionally, the artificial-natural coupled water network virtual model construction system provided in some embodiments of the present application further includes an updating module 170, which is specifically configured to:

[0224] Get real-time monitoring data stream;

[0225] Processing the monitoring data stream in real time based on edge computing nodes;

[0226] Based on the processed monitoring data stream, the digital twin corresponding to the natural-artificial coupled virtual water network model is calibrated online through an incremental model correction algorithm.

[0227] In an exemplary embodiment, a processing device is further provided, which includes at least a memory and a processor. The memory stores a computer program, and the processor implements the steps in the above method embodiments when executing the computer program.

[0228] The above-mentioned processing device is a data collection device, which can be a server or a terminal. Its internal exemplary structure is as follows: Figure 6 As shown. The processing device includes a processor, a memory, an input / output interface (Input / Output, abbreviated as I / O) and a communication interface. The processor, memory and input / output interface are connected via a system bus, and the communication interface is connected to the system bus via the input / output interface. The processor of the processing device is used to provide computing and control capabilities. The memory of the processing device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and computer program in the non-volatile storage medium. The database of the processing device is used to store the data and programs involved in the aforementioned embodiments. The input / output interface of the processing device is used to exchange information between the processor and an external device. The communication interface of the processing device is used to communicate with an external terminal via a network connection. When the computer program is executed by the processor, it implements the method for constructing a digital twin model of a natural-artificial coupled virtual water network in the aforementioned embodiments.

[0229] Those skilled in the art will understand that Figure 6 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the processing device to which the solution of the present application is applied. The specific processing device may include more or fewer components than shown in the figure, or combine certain components, or have a different component arrangement.

[0230] In an exemplary embodiment, a computer-readable storage medium is provided, storing a computer program. When the computer program is executed by a processor, the steps in the above-mentioned method embodiments are implemented.

[0231] In an exemplary embodiment, a computer program product is provided, including a computer program. When the computer program is executed by a processor, the steps in the above method embodiments are implemented.

[0232] It can be understood that the method and equipment for constructing a digital twin model of a natural-artificial coupled virtual water network provided in the embodiments of the present application collect multi-source heterogeneous data through a distributed hydrological sensor array and a SCADA system, and use a spatiotemporal alignment algorithm to unify coordinates and synchronize time to generate a standardized data base, thereby achieving effective integration of natural water network and artificial water network data, providing a comprehensive and accurate data foundation for subsequent model construction, and helping to improve the comprehensive understanding and analysis capabilities of the entire water network system.

[0233] On this basis, this application uses a standardized data base to construct a virtual model of the natural water network, which can more realistically simulate the flow movement and water quality changes in the natural water network; and combines the LSTM neural network with fluid dynamics equations to construct a pipe network control model for the artificial water network, improving the accuracy and intelligence level of artificial water network control. Finally, through graph theory algorithms, the results of the constructed natural water network and artificial water network virtual models are coupled to construct a natural-artificial coupled virtual model. This natural-artificial coupled virtual model can achieve a comprehensive and accurate simulation of the natural and artificial water network systems, providing a powerful tool for the rational allocation and management of water resources. It realizes the real-time interaction and sharing of water flow, water quality and other information between the natural water network and the artificial water network. In other words, the model can accurately reflect the dynamic interactive relationship between the two, and can more comprehensively consider various factors when making water resource management decisions, improving the scientificity and accuracy of the decisions.

[0234] Those skilled in the art will appreciate that all or part of the processes in the above-described method embodiments, or the functions of the modules in the above-described determination device, can be implemented by instructing the relevant hardware through a computer program. The computer program can be stored in a non-volatile computer-readable storage medium. When executed, the computer program can include the processes of the above-described method embodiments. Any reference to a memory, database, or other medium used in the embodiments provided herein may include at least one of non-volatile and volatile memory. Non-volatile memory may include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory may include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM may be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM).

[0235] The databases involved in the various embodiments provided herein may include at least one of a relational database and a non-relational database. Non-relational databases may include, but are not limited to, distributed databases based on blockchains. The processors involved in the various embodiments provided herein may include, but are not limited to, general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic units, data processing logic units based on quantum computing, and the like.

[0236] The technical features of the above embodiments can be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0237] This document uses specific examples to illustrate the principles and implementation methods of this application. The description of the above examples is only intended to help understand the method and core concept of this application. At the same time, for those skilled in the art, based on the concept of this application, there may be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be understood as limiting this application.

Claims

1. A method for constructing a digital twin model of a natural-artificial coupled virtual water network, characterized in that: The method comprises: Acquiring historical monitoring data of natural water networks and artificial water networks, wherein the historical monitoring data of natural water networks includes hydrological data and water quality data collected by a distributed hydrological sensor array, and the historical monitoring data of artificial water networks includes pipe network structure data, environmental parameters, and control parameters collected by a SCADA system; Using a spatiotemporal alignment algorithm to synchronously preprocess the historical monitoring data to obtain a standardized data base; Based on the standardized data base of the natural water network, a natural water network model is constructed through a hydrodynamic model; Based on the standardized data base of the artificial water network, an artificial water network model is constructed through an LSTM neural network and a fluid dynamics model; Using graph theory algorithms, we identify the connection nodes between the natural and artificial water network models and establish a data exchange channel. Establishing the natural-artificial coupling virtual model between the artificial water network and the natural water network according to the data interaction channel, wherein the natural-artificial coupling virtual model is used to characterize the interaction relationship between the artificial water network and the natural water network; Configuring a bidirectional interactive interface of the natural-artificial coupling virtual model, wherein the bidirectional interactive interface is used to represent the interactive state of the natural water network and the artificial water network; The step of establishing the natural-artificial coupling virtual model between the artificial water network and the natural water network according to the data interaction channel includes: Establishing a coupling matrix based on the data interaction channel as the natural-artificial coupling virtual model, wherein the coupling matrix is ​​used to characterize the water flow direction, flow conductivity and water quality influencing factors in the water network; The coupling matrix is ​​a three-dimensional tensor structure, and is expressed as follows: ； in, i Indicates the node number in the natural water network, j Represents the node number in the artificial water network. If K 1 ij When it is a non-zero value, it indicates a node in the natural water network i Nodes with artificial water networks j There is a connection relationship; if K 1 ij If it is zero, it means that there is no connection between the two nodes; Among them, the first dimension represents the topological connection relationship between natural water network and artificial water network; The second dimension represents the flow conductivity coefficient, which is used to characterize the efficiency of transmitting water flow from a node in the natural water network to the corresponding node in the artificial water network; The third dimension represents water quality influencing factors, which include dissolved oxygen diffusion coefficient and pollutant attenuation rate.

2. The method for constructing a digital twin model of a natural-artificial coupled virtual water network according to claim 1 is characterized in that: The method of identifying the connection nodes between the natural water network model and the artificial water network model by using a graph theory algorithm includes: Traverse all node information in the artificial water network model and the natural water network model; All node information in the artificial water network model and the natural water network model is compared. If the nodes in the natural water network model and the nodes in the artificial water network model meet the connection point identification rules, they are determined as the connection nodes between the natural water network model and the artificial water network model.

3. The method for constructing a digital twin model of a natural-artificial coupled virtual water network according to claim 1 is characterized in that: The bidirectional interactive interface for configuring the natural-artificial coupling virtual model includes: Configure a boundary condition transfer module and a feedback control module. The boundary condition transfer module is used to input the outflow / inflow parameters of the natural water network into the boundary nodes of the artificial water network. The feedback control module is used to reversely input the valve opening and pump station start and stop instructions of the artificial water network into the natural water network model.

4. The method for constructing a digital twin model of a natural-artificial coupled virtual water network according to any one of claims 1 to 3, characterized in that: The artificial water network model is constructed based on the standardized data base of the artificial water network through the LSTM neural network and the fluid dynamics model, including: Inputting the standardized data base of the artificial water network into the constructed initial model, training the initial model so that the initial model outputs an initial prediction result, the initial prediction result including a predicted value of the start-up and shutdown probability of the pump stations in the artificial water network; Based on the initial prediction result and the actual pump station start and stop status label, the difference between the predicted value and the true value is calculated, and the corresponding loss function is calculated based on the difference; According to the loss function, the parameters of the initial model are updated, and the above training steps are repeated so that the loss function meets the set threshold.

5. The method for constructing a digital twin model of a natural-artificial coupled virtual water network according to any one of claims 1 to 3, characterized in that: The method further comprises: Obtain real-time monitoring data of artificial and natural water networks; Inputting the historical monitoring data into the constructed natural-artificial coupling virtual model to obtain a prediction result; Calculating the root mean square error between the real-time monitoring data and the prediction result; Based on the root mean square error, the weight coefficient of the natural-artificial coupling virtual model is dynamically adjusted through a parameter sensitivity analysis algorithm until a preset convergence threshold is reached.

6. The method for constructing a digital twin model of a natural-artificial coupled virtual water network according to any one of claims 1 to 3, characterized in that: The method further comprises: Get real-time monitoring data stream; Processing the monitoring data stream in real time based on edge computing nodes; Based on the processed monitoring data stream, the digital twin corresponding to the natural-artificial coupled virtual water network model is calibrated online through an incremental model correction algorithm.

7. A system for constructing a digital twin model of a natural-artificial coupled virtual water network, characterized in that: include: An acquisition module is used to acquire historical monitoring data of natural water networks and artificial water networks, wherein the historical monitoring data of natural water networks includes hydrological data and water quality data collected by a distributed hydrological sensor array, and the historical monitoring data of artificial water networks includes pipe network structure data, environmental parameters, and control parameters collected by a SCADA system; A preprocessing module, configured to synchronously preprocess the historical monitoring data using a spatiotemporal alignment algorithm to obtain a standardized data base; A first construction module is configured to construct a natural water network model based on the standardized data base of the natural water network and through a hydrodynamic model; The second construction module is used to construct an artificial water network model based on the standardized data base of the artificial water network through an LSTM neural network and a fluid dynamics model; A coupling module is configured to identify connection nodes between the natural water network model and the artificial water network model through a graph theory algorithm and establish a data interaction channel; establish the natural-artificial coupling virtual model between the artificial water network and the natural water network based on the data interaction channel, wherein the natural-artificial coupling virtual model is used to characterize the interaction relationship between the artificial water network and the natural water network; and configure a bidirectional interaction interface of the natural-artificial coupling virtual model, wherein the bidirectional interaction interface is used to characterize the interaction status between the natural water network and the artificial water network; The step of establishing the natural-artificial coupling virtual model between the artificial water network and the natural water network according to the data interaction channel includes: Establishing a coupling matrix based on the data interaction channel as the natural-artificial coupling virtual model, wherein the coupling matrix is ​​used to characterize the water flow direction, flow conductivity and water quality influencing factors in the water network; The coupling matrix is ​​a three-dimensional tensor structure, and is expressed as follows: ； in, i Indicates the node number in the natural water network, j Represents the node number in the artificial water network. If K 1 ij When it is a non-zero value, it indicates a node in the natural water network i Nodes with artificial water networks j There is a connection relationship; if K 1 ij If it is zero, it means that there is no connection between the two nodes; Among them, the first dimension represents the topological connection relationship between natural water network and artificial water network; The second dimension represents the flow conductivity coefficient, which is used to characterize the efficiency of transmitting water flow from a node in the natural water network to the corresponding node in the artificial water network; The third dimension represents water quality influencing factors, which include dissolved oxygen diffusion coefficient and pollutant attenuation rate.

Images