Water conservancy digital twin information management system and method

Abstract

The application discloses a water conservancy digital twin information management system and method, the method comprises the following steps: constructing a hydrograph structure of a target hydrological region at M observation timestamps; using the hydrograph structure of the M observation timestamps as a supervised training sample to train a hydrological twin model; acquiring a real-time hydrograph of the target hydrological region; using the real-time hydrograph as an input feature of the hydrological twin model to output future hydrographs of L future timestamps; and performing information management of the target hydrological region based on the future hydrographs of the L future timestamps; the application truly reflects the trend and conduction of water flow in the current hydrological region, avoids distortion of the expression of the hydraulic relationship caused by a fixed topology or an experience weight, and thus improves the information level of water conservancy management.

Description

Technical Field

[0001] This invention relates to the field of water conservancy information management, and specifically to a water conservancy digital twin information management method. Background Technology

[0002] Current water conservancy information systems are mostly based on fixed river network topologies to construct hydrological monitoring networks. The connections between nodes are predefined by geographic information, and the edge directions and weights remain unchanged during operation. Such static graph structures cannot reflect changes in actual water flow conditions—for example, the reversal of flow direction during floods due to water level fluctuations or gate / pump scheduling. The model still uses the original upstream and downstream relationships, resulting in distorted predictions.

[0003] Some methods attempt to incorporate dynamic data such as water level and flow rate to adjust edge weights, but weight calculations typically rely solely on the reciprocal of distance or parameter correlation, failing to incorporate fundamental hydraulic principles. Water level difference, as the driving force of water flow, flow velocity, reflecting flow activity, and geographical distance, characterizing energy loss, collectively determine hydraulic conduction capacity. However, current technologies have failed to integrate these factors into a unified edge strength metric.

[0004] Therefore, existing hydrological maps generally suffer from problems such as unchangeable flow direction and lack of physical basis for weights, making it difficult to support high-level information-based management of water conservancy twins. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this invention provides a water conservancy digital twin information management method, which solves the technical problems mentioned in the background by constructing a hydraulic transmission distance that reflects the actual difficulty of water flow transmission.

[0006] To achieve the above objectives, the present invention provides the following technical solution:

[0007] In a first aspect, the present invention provides a water conservancy digital twin information management method, comprising the following steps:

[0008] Construct a hydrological map structure for the target hydrological region at M observation timestamps;

[0009] The hydrological twin model is trained using the hydrological map structure with M observation timestamps as supervised training samples.

[0010] Obtain real-time hydrological maps of the target hydrological area;

[0011] Using real-time hydrological maps as input features for the hydrological twin model, the model outputs future hydrological maps with L future timestamps.

[0012] Information management of target hydrological areas based on future hydrological maps with L future timestamps;

[0013] In some specific embodiments, the hydrological map structure of the target hydrological region at M observation timestamps is constructed, including:

[0014] Obtain J hydrological observation vectors for N observation nodes at M observation timestamps; where each hydrological observation vector is associated with its observation node and observation timestamp, and J = N × M;

[0015] Among N observation nodes, N1 valid observation nodes are identified; where N1≤N;

[0016] The node features of effective observation nodes are defined as hydrological observation vectors, and directed connections are established between effective observation nodes.

[0017] Calculate the edge weights of the directed connections;

[0018] Traverse the N1 valid observation nodes, repeatedly establish directed edges and calculate edge weights until a hydrological map structure containing N1 valid observation nodes is constructed.

[0019] In some specific embodiments, obtaining J hydrological observation vectors from N observation nodes at M observation timestamps includes:

[0020] Select N observation nodes for the target hydrological area on the electronic map;

[0021] Assign M uniform observation timestamps to N observation nodes on the time axis;

[0022] For any given observation timestamp, obtain its Class G hydrological observation parameters across N observation nodes;

[0023] Based on the G-type hydrological observation parameters of N observation nodes, construct N hydrological observation vectors corresponding to the observation timestamps;

[0024] Iterate through M observation timestamps until J hydrological observation vectors are constructed for N observation nodes at M observation timestamps.

[0025] In some specific embodiments, based on the G-type hydrological observation parameters of N observation nodes, N hydrological observation vectors corresponding to the observation timestamps are constructed, including:

[0026] In the G-class hydrological observation parameters, select the target class observation parameters and obtain the hydrological observation parameters of the target class observation parameters at M observation timestamps;

[0027] Normalize the hydrological observation parameters of the M observation timestamps to obtain the first hydrological observation feature of the target class observation parameters at the M observation timestamps;

[0028] Traverse the G-type hydrological observation parameters and repeatedly select the target type observation parameters until G×M first hydrological observation features are obtained;

[0029] Select the target timestamp sequentially from the M observation timestamps;

[0030] Obtain the Class G hydrological observation parameters and their corresponding first hydrological observation features at the target timestamp;

[0031] Match the hydrological weights and splicing sequence numbers of Class G hydrological observation parameters in a pre-constructed hydrological prior knowledge base;

[0032] The hydrological weights of the Class G hydrological observation parameters are multiplied by their first hydrological observation features to generate the Class G second hydrological observation features.

[0033] The G second hydrological observation features are spliced ​​together according to the splicing sequence number of their hydrological observation parameters to construct the hydrological observation vector corresponding to the target timestamp;

[0034] Traverse N observation nodes until N hydrological observation vectors corresponding to the observation timestamps are constructed.

[0035] In some specific embodiments, identifying N² valid observation nodes out of N observation nodes includes:

[0036] Choose any two observation nodes from the N observation nodes as a node pair;

[0037] Determine whether the node pairs have hydrological connectivity on the electronic map;

[0038] If hydrological connectivity exists, the node pair is marked as a candidate node pair;

[0039] Obtain the geographic coordinates of the two observation nodes in the candidate node pair;

[0040] Calculate the geographical distance between candidate node pairs based on the geographical coordinates;

[0041] If the geographical distance does not exceed the set distance threshold, the candidate node pair will be marked as a valid node pair.

[0042] Traverse N observation nodes and repeatedly mark the valid node pairs until P valid node pairs are obtained;

[0043] Split the P valid node pairs into 2×P observation nodes;

[0044] Duplicate observation nodes are removed from the 2×P observation nodes to obtain N1 valid observation nodes;

[0045] In some specific embodiments, establishing directed connections between valid observation nodes includes:

[0046] Obtain the water levels of N1 valid observation nodes in the hydrological observation parameters;

[0047] Based on the water level in the hydrological observation parameters, compare the relative water levels of any two valid observation nodes among the N1 valid observation nodes;

[0048] The effective observation nodes on the side with the relatively higher water level are marked as upstream nodes, and the other side is marked as downstream nodes;

[0049] Starting from the upstream node, establish directed edges leading to the downstream node;

[0050] Traverse the N1 valid observation nodes and repeatedly establish directed connections until N2 directed connections are obtained.

[0051] In some specific embodiments, calculating the edge weight of the directed connection edge includes:

[0052] Extract the water level, flow velocity, and corresponding geographical distance from the hydrological observation parameters of N1 valid observation nodes;

[0053] Based on the water level, flow velocity and geographical distance of N1 effective observation nodes in the hydrological observation parameters, calculate the hydraulic transmission distance of N2 directed connecting edges;

[0054] Normalize the hydraulic transmission distance of the N2 directed connection edges to generate the corresponding N2 edge weights.

[0055] In some specific embodiments, the hydrological map structure with M observation timestamps is used as supervised training samples to train the hydrological twin model, including:

[0056] Encode the M observation timestamps into M ordered, incremental time numbers;

[0057] Based on M ordered and increasing time numbers, M hydrological map structures are arranged in ascending order to obtain a hydrological map sequence.

[0058] The hydrological map sequence is divided into K hydrological map subsequences using a sliding window method, with a window length of L and a step size of 1; where K = M - L + 1.

[0059] The first L1 hydrographic structures of the hydrographic subsequence are used as input features, and the last L2 hydrographic structures are used as supervision labels to construct the supervised training samples; where L = L1 + L2.

[0060] K supervised training samples corresponding to K hydrological map subsequences are input into the DCRNN model, and the model is iteratively forward propagated and backward gradient updated to obtain a hydrological twin model.

[0061] This invention provides a water conservancy digital twin information management method, which has the following beneficial effects:

[0062] This invention, at each observation timestamp, identifies upstream nodes based on the measured water level relationships of valid observation nodes, marking nodes with higher water levels as upstream and those with lower levels as downstream, and establishing directed edges connecting upstream to downstream. Simultaneously, it calculates the hydraulic conduction distance by combining the geographical distance, water level difference, and average flow velocity between these node pairs, and normalizes this distance to use as the edge weight for the corresponding directed edge. In the resulting hydrological map structure, the edge direction dynamically changes with the water level; a smaller edge weight indicates easier hydraulic conduction, accurately reflecting the direction and conduction of water flow in the hydrological region at the current moment. This avoids distortions in the expression of hydraulic relationships caused by fixed topology or empirical weights, thereby improving the informatization level of water conservancy management.

[0063] Secondly, the present invention provides a water conservancy digital twin information management system for executing the water conservancy digital twin information management method according to any one of claims 1 to 8, characterized in that the system comprises:

[0064] The hydrological map construction module is used to construct the hydrological map structure of the target hydrological region at M observation timestamps.

[0065] The twin model training module is used to train the hydrological twin model by using the hydrological map structure with M observation timestamps as supervised training samples.

[0066] The real-time hydrology module is used to acquire real-time hydrological maps of the target hydrological area;

[0067] The Future Hydrology module is used to take real-time hydrological maps as input features for the hydrological twin model and output future hydrological maps with L future timestamps.

[0068] The hydrological management module is used for information management of target hydrological areas based on future hydrological maps with L future timestamps.

[0069] Compared with the prior art, the beneficial effects of the water conservancy digital twin information management system of the present invention are the same as those of the water conservancy digital twin information management method described above, so they will not be repeated here. Attached Figure Description

[0070] Figure 1 This is a flowchart illustrating a water conservancy digital twin information management method according to the present invention.

[0071] Figure 2 This is a schematic diagram illustrating the construction process of the hydrographic map structure described in this invention;

[0072] Figure 3 This is a schematic diagram of the selection process for effective observation nodes as described in this invention;

[0073] Figure 4 This is a schematic diagram of the edge weight generation process described in this invention;

[0074] Figure 5 This is a structural block diagram of a water conservancy digital twin information management system according to the present invention. Detailed Implementation

[0075] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0076] Example 1: Please refer to Figures 1 to 4 This invention provides a water conservancy digital twin information management method, comprising the following steps:

[0077] S1. Construct the hydrological map structure of the target hydrological region at M observation timestamps;

[0078] Specifically, the hydrographic structure represents the hydraulic topological relationship and dynamic hydrological state between multiple observation nodes in the target hydrological area. Each observation node corresponds to a hydraulic facility or hydrological section. The node features include multi-dimensional hydrological observation parameters at the corresponding timestamp. The directed connection edges between nodes represent the direction of water flow, and the edge weights reflect the hydraulic conduction intensity.

[0079] S2. Use the hydrological map structure with M observation timestamps as supervised training samples to train the hydrological twin model;

[0080] Specifically, in this embodiment, the initial model of the hydrological twin model can be a spatio-temporal graph neural network (STGNN); it captures spatial hydraulic correlations through graph convolutional layers and models the temporal evolution of node states through recurrent neural networks or temporal convolutional layers.

[0081] Preferably, the spatiotemporal graph neural network can be the DCRNN model; it performs diffusion convolution operation on the directed connection edges in the hydrological graph structure by constructing forward and backward random walk transition matrices, and combines gated recurrent units to model the state evolution of the time series of node features.

[0082] S3. Obtain real-time hydrological maps of the target hydrological area;

[0083] S4. Use the real-time hydrological map as the input feature of the hydrological twin model, and output the future hydrological map with L future timestamps;

[0084] S5. Information management of the target hydrological area based on future hydrological maps with L future timestamps;

[0085] The information management system refers to selectively implementing flood control scheduling plans, water supply optimization instructions, or engineering inspection tasks based on predicted abnormal water levels, flow rates, or water quality nodes in future hydrological maps, and then pushing these to the water conservancy business management system for execution.

[0086] In this embodiment, by constructing a hydrological map structure that includes the direction of water flow and the intensity of hydraulic transmission, and training a hydrological twin model based on it, the model can predict the water level, flow rate or water quality status at multiple future timestamps based on the real-time hydrological map, and perform hydraulic information management such as flood control scheduling plans, water supply optimization instructions or engineering inspection tasks accordingly.

[0087] Specifically, in this embodiment, step S1 includes:

[0088] S1-1. Obtain J hydrological observation vectors for N observation nodes at M observation timestamps; where each hydrological observation vector is associated with its observation node and observation timestamp, and J=N×M;

[0089] Specifically, the hydrological observation vector is used to characterize the normalized fusion features of multi-source hydrological observation parameters of the corresponding observation node at a specific observation timestamp. It can be used as the feature input of nodes in the hydrological map structure and support the hydrological twin model to supervise the temporal evolution of hydrological state.

[0090] S1-2. Among N observation nodes, identify N1 valid observation nodes; where N1≤N;

[0091] S1-3. Define the node characteristics of effective observation nodes as hydrological observation vectors and establish directed connection edges between effective observation nodes.

[0092] S1-4. Calculate the edge weights of the directed connections;

[0093] S1-5. Traverse the N1 valid observation nodes, repeatedly establish directed edges and calculate edge weights until a hydrological map structure containing N1 valid observation nodes is constructed.

[0094] In this embodiment, effective observation nodes with hydrological connectivity are selected from the original observation nodes, and their normalized and fused hydrological observation vectors are used as node features. Directed connections are constructed using edge weights that reflect the hydraulic conduction intensity. The resulting hydrological graph structure not only retains the real hydraulic topology but also has node and edge representations that can be used for time-series characterization.

[0095] Furthermore, step S1-1 specifically includes:

[0096] S1-1-1. Select N observation nodes for the target hydrological area on the electronic map;

[0097] S1-1-2. Assign M uniform observation timestamps to N observation nodes on the time axis;

[0098] Specifically, the unified M observation timestamps are a set of identical time points, and all observation nodes synchronously acquire hydrological observation parameters at these M observation timestamps.

[0099] S1-1-3. For any given observation timestamp, obtain its Class G hydrological observation parameters at N observation nodes;

[0100] Specifically, the observation node refers to a fixed hydrological monitoring location deployed within the target hydrological area, which corresponds to a water conservancy facility or a natural hydrological section; the Class G hydrological observation parameters are multi-source heterogeneous parameters, which include: water level, flow rate, flow velocity, water quality parameters, gate opening degree and pump status of any observation node at the corresponding observation timestamp.

[0101] S1-1-4. Based on the G-type hydrological observation parameters of N observation nodes, construct N hydrological observation vectors corresponding to the observation timestamps.

[0102] S1-1-5. Traverse the M observation timestamps until constructing J hydrological observation vectors for N observation nodes at the M observation timestamps.

[0103] In this embodiment, by selecting a fixed hydrological monitoring location on an electronic map and simultaneously collecting water level, flow rate, flow velocity, water quality parameters, gate opening degree, and pump status at M unified time points, J hydrological observation vectors corresponding one-to-one with the observation nodes and observation timestamps are constructed.

[0104] Furthermore, step S1-1-4 specifically includes:

[0105] S1-1-4-1. In the G-class hydrological observation parameters, select the target class observation parameters and obtain the hydrological observation parameters of the target class observation parameters at M observation timestamps.

[0106] S1-1-4-2. Normalize the hydrological observation parameters of the M observation timestamps to obtain the first hydrological observation feature of the target class observation parameters at the M observation timestamps.

[0107] S1-1-4-3. Traverse the G-type hydrological observation parameters and repeatedly select the target type observation parameters until G×M first hydrological observation features are obtained.

[0108] S1-1-4-4. Select the target timestamp sequentially from the M observation timestamps;

[0109] S1-1-4-5. Obtain the Class G hydrological observation parameters and their corresponding first hydrological observation features at the target timestamp;

[0110] S1-1-4-6, matching the hydrological weights and splicing sequence numbers of Class G hydrological observation parameters in a pre-constructed hydrological prior knowledge base;

[0111] Specifically, the hydrological prior knowledge base stores the hydrological weights and splicing numbers corresponding to each of the G types of hydrological observation parameters. The hydrological weights represent the relative importance of the hydrological observation parameters in water conservancy scheduling decisions, and the splicing numbers define the characteristic arrangement position of the hydrological observation parameters in the hydrological observation vector. The hydrological weights and splicing numbers are generated offline based on historical flood scheduling records.

[0112] S1-1-4-7. Multiply the hydrological weights of the Class G hydrological observation parameters with its first hydrological observation feature to generate the Class G second hydrological observation feature.

[0113] S1-1-4-8. For the G second hydrological observation features, feature splicing is performed according to the splicing sequence number of their hydrological observation parameters to construct the hydrological observation vector corresponding to the target timestamp;

[0114] S1-1-4-9. Traverse N observation nodes until N hydrological observation vectors corresponding to the observation timestamps are constructed.

[0115] In this embodiment, by normalizing various hydrological observation parameters, combining them with hydrological weights generated based on historical scheduling records, and integrating them into a fixed-dimensional vector according to a preset splicing sequence number, the multi-source heterogeneous parameters of each observation node at any time stamp are uniformly expressed as an ordered, weighted, and normalized hydrological observation vector.

[0116] Furthermore, step S1-2 specifically includes:

[0117] S1-2-1. Among the N observation nodes, select any two observation nodes as a node pair;

[0118] S1-2-2. Determine whether the node pair has a hydrological connectivity relationship on the electronic map;

[0119] Specifically, the hydrological connectivity relationship means that two observation nodes are located in the same river channel, the same reservoir area, or are directly connected through gates, pumping stations, or water conveyance channels.

[0120] S1-2-3. If hydrological connectivity exists, mark the node pair as a candidate node pair.

[0121] S1-2-4. Obtain the geographic coordinates of the two observation nodes in the candidate node pair;

[0122] S1-2-5. Calculate the geographical distance between candidate node pairs based on the geographical coordinates.

[0123] The formula for calculating the geographical distance is:

[0124] ;

[0125] in, Geographic distance is represented by calculating the Euclidean distance between the two observation nodes in a candidate node pair; , (), , () represent the Cartesian coordinates of the two observation nodes in the candidate node pair on the electronic map;

[0126] S1-2-6. If the geographical distance does not exceed the set distance threshold, then mark the candidate node pair as a valid node pair.

[0127] S1-2-7. Traverse N observation nodes and repeatedly mark the valid node pairs until P valid node pairs are obtained.

[0128] S1-2-8. Split the P valid node pairs into 2×P observation nodes;

[0129] S1-2-9. Delete duplicate observation nodes from the 2×P observation nodes to obtain N1 valid observation nodes;

[0130] Specifically, the repeated observation node refers to the same observation node that is included in multiple valid node pairs.

[0131] In this embodiment, node pairs are screened by combining the hydrological connectivity and geographical distance constraints on the electronic map, and the screening results are deduplicated. The final N1 valid observation nodes are all monitoring points that have actual water flow paths to each other and are reasonably spaced, thus eliminating isolated or weakly correlated invalid nodes.

[0132] Furthermore, steps S1-3 specifically include:

[0133] S1-3-1. Obtain the water levels of N1 valid observation nodes in the hydrological observation parameters;

[0134] S1-3-2. Based on the water level in the hydrological observation parameters, compare the relative water levels of any two valid observation nodes among the N1 valid observation nodes;

[0135] Specifically, the relative water level refers to the relationship between the water level values ​​of any two valid observation nodes at the same observation timestamp, that is, the water level of one observation node is higher than the water level of the other observation node.

[0136] S1-3-3: Mark the effective observation node on the side with the relatively higher water level as the upstream node, and mark the other side as the downstream node;

[0137] S1-3-4. Starting from the upstream node, establish directed edges towards the downstream node;

[0138] S1-3-5. Traverse N1 valid observation nodes and repeatedly establish directed connection edges until N2 directed connection edges are obtained.

[0139] In this embodiment, by comparing the water levels between effective observation nodes at the same observation timestamp, the node with the higher water level is designated as upstream and the node with the lower water level as downstream, and a directed connection edge from upstream to downstream is established accordingly, so that the direction of the connection edge in the hydrological map structure is consistent with the actual water flow direction.

[0140] Furthermore, steps S1-4 specifically include:

[0141] S1-4-1. Extract the water level, flow velocity, and corresponding geographical distance from the hydrological observation parameters of N1 effective observation nodes;

[0142] S1-4-2. Based on the water level, flow velocity, and geographical distance of N1 effective observation nodes in the hydrological observation parameters, calculate the hydraulic transmission distance of N2 directed connecting edges.

[0143] The formula for calculating the hydraulic conduction distance is:

[0144] ;

[0145] in, This represents the hydraulic transmission distance between the i-th valid observation node and the j-th valid observation node. This represents the geographical distance between the i-th valid observation node and the j-th valid observation node. and These represent the water levels at the corresponding observation timestamps for the i-th and j-th valid observation nodes, respectively. This represents the average flow velocity of the i-th and j-th effective observation nodes; that is, the flow velocities of each effective observation node at the corresponding observation timestamp are summed and averaged. This is a preset regularization constant used to avoid zero denominators; its value range is [value range missing]. ;

[0146] Furthermore, geographical distance leads to longer water propagation time and greater energy loss, thus increasing the difficulty of conduction. Therefore, geographical distance, as the numerator, is directly proportional to the conduction distance; while water level difference... The water level difference is the main driving force of water flow. The greater the water level difference, the stronger the driving force, and the easier the conduction. The shorter the hydraulic conduction distance, the easier the conduction. The average flow velocity reflects the flow intensity on that path. The higher the flow velocity, the more active the water flow, the faster the response, the higher the conduction efficiency, and the shorter the hydraulic conduction distance, indicating smoother conduction.

[0147] S1-4-3. Normalize the hydraulic transmission distance of the N2 directed connection edges to generate the corresponding N2 edge weights.

[0148] In this step, the normalization is preferably achieved by using Min-Max Normalization, which linearly maps all hydraulic conduction distances to the interval [0,1] and takes their complement as the edge weights.

[0149] In this embodiment, the hydraulic transmission distance of each directed connection edge is calculated by combining geographical distance, water level difference and average flow velocity, and then normalized to an edge weight. The smaller the edge weight value, the stronger the hydraulic connection between the two points, and the better it reflects the difficulty of actual water flow transmission.

[0150] Specifically, in this embodiment, the training steps of the hydrological twin model are as follows:

[0151] S2-1. Encode the M observation timestamps into M ordered, incremental time numbers;

[0152] Specifically, the encoding refers to sorting the M observation timestamps in chronological order and then assigning them consecutive integers from 1 to M as time sequence numbers.

[0153] S2-2. Based on M ordered and increasing time numbers, arrange the M hydrological map structures in ascending order to obtain the hydrological map sequence.

[0154] S2-3. Divide the hydrological map sequence into K hydrological map subsequences using a sliding window method, with a window length of L and a step size of 1; where K = M - L + 1;

[0155] S2-4. Using the first L1 hydrological map structures of the hydrological map subsequence as input features and the last L2 hydrological map structures as supervision labels, construct the supervised training samples; where L = L1 + L2.

[0156] S2-5. Input the K supervised training samples corresponding to the K hydrological map subsequences into the DCRNN model, and iterate the forward propagation and backward gradient update to obtain the hydrological twin model.

[0157] In this embodiment, by dividing the hydrological map structure arranged in chronological order into input and label pairs using a sliding window, and using the DCRNN model to learn its spatiotemporal evolution, the trained hydrological twin model can predict the hydrological map structure at subsequent L2 times based on the hydrological map structure at consecutive L1 times.

[0158] Example 2: See Figure 5 This invention also provides a water conservancy digital twin information management system, which is used to implement the above-described method embodiments; details already described will not be repeated. The terms "module," "unit," and "subunit," etc., used below refer to combinations of software and / or hardware that achieve a predetermined function. Although the system described in the following embodiments is preferably implemented in software, hardware implementation, or a combination of software and hardware, is also possible and contemplated.

[0159] like Figure 5 As shown, Figure 5 This is a structural block diagram of a water conservancy digital twin information management system according to the present invention. The system includes:

[0160] The hydrological map construction module is used to construct the hydrological map structure of the target hydrological region at M observation timestamps.

[0161] The twin model training module is used to train the hydrological twin model by using the hydrological map structure with M observation timestamps as supervised training samples.

[0162] The real-time hydrology module is used to acquire real-time hydrological maps of the target hydrological area;

[0163] The Future Hydrology module is used to take real-time hydrological maps as input features for the hydrological twin model and output future hydrological maps with L future timestamps.

[0164] The hydrological management module is used for information management of target hydrological areas based on future hydrological maps with L future timestamps.

[0165] In the above system, a hydrological map structure is constructed through a hydrological map construction module, a hydrological twin model is trained through a twin model training module, a target real-time hydrological map is obtained through a real-time hydrological module, a future hydrological map with L future timestamps is output through a future hydrological module, and information management of the target hydrological area is carried out through a hydrological management module. This solves the problems that hydrological maps generally have, such as unchangeable flow direction and no physical basis for weights.

[0166] The above embodiments can be implemented, in whole or in part, by software, hardware, firmware, or any other combination thereof. When implemented using software, the above embodiments can be implemented, in whole or in part, as a computer program product. The computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer program are loaded or executed on a computer, all or part of the processes or functions described in the embodiments of this application are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired (e.g., infrared, wireless, microwave, etc.) means.

[0167] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without any creative effort.

[0168] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments or some parts of the embodiments.

[0169] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any changes or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application.

Claims

1. A water conservancy digital twin information management method, characterized in that, include: Construct a hydrological map structure for the target hydrological region at M observation timestamps; The hydrological twin model is trained using the hydrological map structure with M observation timestamps as supervised training samples. Obtain real-time hydrological maps of the target hydrological area; Using real-time hydrological maps as input features for the hydrological twin model, the model outputs future hydrological maps with L future timestamps. Information management of target hydrological areas is carried out based on future hydrological maps with L future timestamps.

2. The water conservancy digital twin information management method according to claim 1, characterized in that, Construct the hydrological map structure of the target hydrological region at M observation timestamps, including: Obtain J hydrological observation vectors for N observation nodes at M observation timestamps; where each hydrological observation vector is associated with its observation node and observation timestamp, and J = N × M; Among N observation nodes, N1 valid observation nodes are identified; where N1≤N; The node features of effective observation nodes are defined as hydrological observation vectors, and directed connections are established between effective observation nodes. Calculate the edge weights of the directed connections; Traverse the N1 valid observation nodes, repeatedly establish directed edges and calculate edge weights until a hydrological map structure containing N1 valid observation nodes is constructed.

3. The water conservancy digital twin information management method according to claim 1, characterized in that, Obtain J hydrological observation vectors from N observation nodes at M observation timestamps, including: Select N observation nodes for the target hydrological area on the electronic map; Assign M uniform observation timestamps to N observation nodes on the time axis; For any given observation timestamp, obtain its Class G hydrological observation parameters across N observation nodes; Based on the G-type hydrological observation parameters of N observation nodes, construct N hydrological observation vectors corresponding to the observation timestamps; Iterate through M observation timestamps until J hydrological observation vectors are constructed for N observation nodes at M observation timestamps.

4. The water conservancy digital twin information management method according to claim 2, characterized in that, Based on the G-type hydrological observation parameters of N observation nodes, construct N hydrological observation vectors corresponding to the observation timestamps, including: In the G-class hydrological observation parameters, select the target class observation parameters and obtain the hydrological observation parameters of the target class observation parameters at M observation timestamps; Normalize the hydrological observation parameters of the M observation timestamps to obtain the first hydrological observation feature of the target class observation parameters at the M observation timestamps; Traverse the G-type hydrological observation parameters and repeatedly select the target type observation parameters until G×M first hydrological observation features are obtained; Select the target timestamp sequentially from the M observation timestamps; Obtain the Class G hydrological observation parameters and their corresponding first hydrological observation features at the target timestamp; Match the hydrological weights and splicing sequence numbers of Class G hydrological observation parameters in a pre-constructed hydrological prior knowledge base; The hydrological weights of the Class G hydrological observation parameters are multiplied by their first hydrological observation features to generate the Class G second hydrological observation features. The G second hydrological observation features are spliced ​​together according to the splicing sequence number of their hydrological observation parameters to construct the hydrological observation vector corresponding to the target timestamp; Traverse N observation nodes until N hydrological observation vectors corresponding to the observation timestamps are constructed.

5. The water conservancy digital twin information management method according to claim 2, characterized in that, The identification of N² valid observation nodes out of N observation nodes includes: Choose any two observation nodes from the N observation nodes as a node pair; Determine whether the node pairs have hydrological connectivity on the electronic map; If hydrological connectivity exists, the node pair is marked as a candidate node pair; Obtain the geographic coordinates of the two observation nodes in the candidate node pair; Calculate the geographical distance between candidate node pairs based on the geographical coordinates; If the geographical distance does not exceed the set distance threshold, the candidate node pair will be marked as a valid node pair. Traverse N observation nodes and repeatedly mark the valid node pairs until P valid node pairs are obtained; Split the P valid node pairs into 2×P observation nodes; By deleting duplicate observation nodes from the 2×P observation nodes, N1 valid observation nodes are obtained.

6. The water conservancy digital twin information management method according to claim 2, characterized in that, Establish directed connections between valid observation nodes, including: Obtain the water levels of N1 valid observation nodes in the hydrological observation parameters; Based on the water level in the hydrological observation parameters, compare the relative water levels of any two valid observation nodes among the N1 valid observation nodes; The effective observation nodes on the side with the relatively higher water level are marked as upstream nodes, and the other side is marked as downstream nodes; Starting from the upstream node, establish directed edges leading to the downstream node; Traverse the N1 valid observation nodes and repeatedly establish directed connections until N2 directed connections are obtained.

7. The water conservancy digital twin information management method according to claim 2, characterized in that, Calculating the edge weight of the directed connection includes: Extract the water level, flow velocity, and corresponding geographical distance from the hydrological observation parameters of N1 valid observation nodes; Based on the water level, flow velocity and geographical distance of N1 effective observation nodes in the hydrological observation parameters, calculate the hydraulic transmission distance of N2 directed connecting edges; Normalize the hydraulic transmission distance of the N2 directed connection edges to generate the corresponding N2 edge weights.

8. The water conservancy digital twin information management method according to claim 2, characterized in that, The hydrological twin model is trained using the hydrological map structure with M observation timestamps as supervised training samples, including: Encode the M observation timestamps into M ordered, incremental time numbers; Based on M ordered and increasing time numbers, M hydrological map structures are arranged in ascending order to obtain a hydrological map sequence. The hydrological map sequence is divided into K hydrological map subsequences using a sliding window method, with a window length of L and a step size of 1; where K = M - L + 1. The first L1 hydrographic structures of the hydrographic subsequence are used as input features, and the last L2 hydrographic structures are used as supervision labels to construct the supervised training samples; where L = L1 + L2. K supervised training samples corresponding to K hydrological map subsequences are input into the DCRNN model, and the model is iteratively forward propagated and backward gradient updated to obtain a hydrological twin model.

9. A water conservancy digital twin information management system, used to execute the water conservancy digital twin information management method according to any one of claims 1 to 8, characterized in that, The system includes: The hydrological map construction module is used to construct the hydrological map structure of the target hydrological region at M observation timestamps. The twin model training module is used to train the hydrological twin model by using the hydrological map structure with M observation timestamps as supervised training samples. The real-time hydrology module is used to acquire real-time hydrological maps of the target hydrological area; The Future Hydrology module is used to take real-time hydrological maps as input features for the hydrological twin model and output future hydrological maps with L future timestamps. The hydrological management module is used for information management of target hydrological areas based on future hydrological maps with L future timestamps.

Images