Drainage system prediction data determination method and device, equipment and storage medium

Abstract

The invention discloses a drainage system prediction data determination method and device, equipment and a storage medium, and the method comprises the steps: obtaining drainage system design data of a target region, and determining a graph structure model corresponding to the target region according to the drainage system design data; establishing a physical simulation model corresponding to the target area according to the drainage system design data, and determining a training data set according to the physical simulation model and the graph structure model; and training the first graph structure neural network model based on the training data set to obtain a second graph neural network model, and determining drainage system prediction data corresponding to the target area according to the second graph neural network model. Based on the technical scheme, the graph structure is used as a basic data expression form, the graph neural network model capable of covering elements of the whole process is constructed, overall proxy and substitution of a traditional multi-physical model combination system are achieved through the model, and the determination efficiency and prediction accuracy of prediction data are improved.

Description

Technical Field

[0001] The present invention relates to the field of computer technology, and particularly to a method, apparatus, device, and storage medium for determining prediction data of a drainage system. Background Art

[0002] With the acceleration of the urbanization process, the construction and operation and maintenance of urban drainage systems face increasingly complex challenges. To achieve the full-process simulation of urban rainfall runoff, hydraulic conveyance, water quality evolution, and finally discharge into receiving water bodies, traditional methods often use multiple professional physical models for subsystem modeling and complete the linkage simulation of the entire system through data interaction between models.

[0003] However, in practical applications, there are significant differences among models in terms of spatial structure, time step, calculation method, and data format. The coupling between models usually relies on artificial rules or customized interfaces and is difficult to automatically adapt to the topological adjustment requirements brought about by the structural evolution of urban drainage systems. In addition, to ensure the credibility of simulation results, all models must be calibrated with monitoring data, and this process itself has the characteristics of high threshold, high computational cost, and strong dependence. Even so, the multi-model system after calibration still requires long-term operation and fine scheduling during the prediction stage and cannot meet the growing scenario requirements of real-time prediction and high-frequency simulation. Summary of the Invention

[0004] The present invention provides a method, apparatus, device, and storage medium for determining prediction data of a drainage system. Using a graph structure as the basic data expression form, a graph neural network model that can cover all process elements is constructed, and through this model, the overall agency and substitution of the traditional multi-physical model combination system are realized, thereby solving the technical problems of fragmented structure, split system response, and low operation efficiency in existing modeling methods.

[0005] According to one aspect of the present invention, there is provided a method for determining prediction data of a drainage system, which includes:

[0006] Obtain the drainage system design data of the target area, and determine a graph structure model corresponding to the target area according to the drainage system design data;

[0007] Establish a physical simulation model corresponding to the target area according to the drainage system design data, and determine a training data set according to the physical simulation model and the graph structure model;

[0008] Train a first graph structure neural network model based on the training data set to obtain a second graph neural network model, and determine the drainage system prediction data corresponding to the target area according to the second graph neural network model, where the first graph neural network model is a graph neural network constructed according to the graph structure model.

[0009] According to another aspect of the present invention, a device for determining drainage system prediction data is provided, which includes:

[0010] A graph structure model determination module, configured to obtain the drainage system design data of a target area, and determine a graph structure model corresponding to the target area according to the drainage system design data;

[0011] A data set determination module, configured to establish a physical simulation model corresponding to the target area according to the drainage system design data, and determine a training data set according to the physical simulation model and the graph structure model;

[0012] A prediction data determination module, configured to train a first graph neural network model based on the training data set to obtain a second graph neural network model, and determine drainage system prediction data corresponding to the target area according to the second graph neural network model, where the first graph neural network model is a graph neural network constructed according to the graph structure model.

[0013] According to another aspect of the present invention, an electronic device is provided, and the electronic device includes:

[0014] At least one processor; and

[0015] A memory communicatively connected to the at least one processor; wherein,

[0016] The memory stores a computer program executable by the at least one processor, and the computer program is executed by the at least one processor so that the at least one processor can execute the method for determining drainage system prediction data according to any embodiment of the present invention.

[0017] According to another aspect of the present invention, a computer-readable storage medium is provided, and the computer-readable storage medium stores computer instructions, and the computer instructions are used to implement the method for determining drainage system prediction data according to any embodiment of the present invention when executed by a processor.

[0018] The technical solution of the embodiment of the present invention obtains the drainage system design data of the target area, determines the graph structure model corresponding to the target area according to the drainage system design data, establishes the physical simulation model corresponding to the target area according to the drainage system design data, determines the training data set according to the physical simulation model and the graph structure model, trains the first graph structure neural network model based on the training data set to obtain the second graph neural network model, and determines the drainage system prediction data corresponding to the target area according to the second graph neural network model. Based on the above technical solution, taking the graph structure as the basic data expression form, a graph neural network model that can cover all process elements is constructed, and the overall proxy and replacement of the traditional multi-physical model combination system are realized through this model, improving the determination efficiency and prediction accuracy of the prediction data.

[0019] It should be understood that the content described in this part is not intended to identify the key or important features of the embodiments of the present invention, nor is it used to limit the scope of the present invention. Other features of the present invention will become easily understood through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following will briefly introduce the drawings required for the description of the embodiments. Obviously, the following drawings are only some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0021] Figure 1 is a schematic flowchart of a method for determining drainage system prediction data provided by an embodiment of the present invention;

[0022] Figure 2 is a schematic flowchart of a method for determining drainage system prediction data provided by an embodiment of the present invention;

[0023] Figure 3 is a block diagram of the structure of a device for determining drainage system prediction data provided by an embodiment of the present invention;

[0024] Figure 4 is a schematic diagram of the structure of an electronic device provided by an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0025] To enable those skilled in the art to better understand the solution of the present invention, the following will clearly and completely describe the technical solution in the embodiments of the present invention with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative work shall fall within the protection scope of the present invention.

[0026] It should be noted that the terms "first", "second", etc. in the specification and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that such data used in this way can be interchanged under appropriate circumstances, so that the embodiments of the present invention described here can be implemented in an order different from those illustrated or described here. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusion. For example, a process, method, system, product or device including a series of steps or units does not necessarily have to be limited to those steps or units clearly listed, but may include other steps or units not clearly listed or inherent to these processes, methods, products or devices.

[0027] Embodiment 1

[0028] Figure 1 is a schematic flowchart of a method for determining prediction data of a drainage system provided by an embodiment of the present invention. This embodiment is applicable to the situation of determining prediction data of a drainage system. This method can be executed by a device for determining prediction data of a drainage system. The device for determining prediction data of a drainage system can be implemented in the form of hardware and / or software. The device for determining prediction data of a drainage system can be configured in an electronic device, and the electronic device can be a terminal device or a server. As Figure 1 shown, the method includes:

[0029] S110. Obtain the drainage system design data of the target area, and determine a graph structure model corresponding to the target area according to the drainage system design data.

[0030] Among them, the target area can be an area where the state of the drainage system needs to be predicted, such as a city or an area divided according to preset rules. The drainage system design data can be understood as the design-related data of the drainage system corresponding to the target area, such as design drawings, maximum drainage volume and other related data. The graph structure model can be understood as a graphical model for each element and its connection relationship in the drainage pipe network, nodes, and treatment facilities.

[0031] Specifically, extract the drainage system design data of the target area from the database, and then determine the graph structure model corresponding to the target area according to the drainage system design data of the target area. Exemplarily, it can be to obtain design drawings and documents including parameters such as drainage network design drawings, node layout drawings, pipe diameters, slopes, materials, historical data such as past drainage system operation records, maintenance records, fault reports, sensor data, such as data of water level, flow rate, flow velocity, etc. real-time monitored by sensors installed in the drainage system, and meteorological data, such as meteorological information such as rainfall, evaporation, temperature, etc., for analyzing the load of the drainage system, and then clean, transform and integrate the collected data to form a unified data set, and construct a graph structure model corresponding to the target area according to the collected drainage system design data. For example, use a graph in graph theory to represent the drainage system, with nodes representing each element of the drainage system and edges representing the connection relationships between the elements. A directed graph or an undirected graph can be selected, which is determined according to the actual operation of the drainage system. For example, if the water flow in the drainage system has a direction, a directed graph is selected.

[0032] Based on the above technical solution, the determining the graph structure model corresponding to the target area according to the drainage system design data includes: determining the nodes of the graph structure model based on the drainage system design data and node types; determining the connection relationships between each pair of nodes according to the drainage system design data, and determining the edges of the graph structure model according to the connection relationships; establishing the graph structure model based on the nodes of the graph structure model and the edges of the graph structure model.

[0033] Among them, the node type can be the node type set for subsystems in the drainage system, and the node types include source area nodes, drainage network nodes, treatment facility nodes, and receiving water body nodes. The connection relationship can be the physical connection relationship or the logical connection relationship between each graph node.

[0034] Specifically, nodes of the graph structure model are determined from the drainage system design data according to the node types, and edges of the graph structure model are determined according to the connection relationships between the nodes. Then, a graph structure model is established based on the edges and nodes of the graph structure model. Exemplarily, through GIS, the catchment area within the target area is identified to determine the catchment source area nodes. Drainage network nodes are determined according to the intersection points and turning points of the pipelines. Treatment facility nodes are determined according to the locations of facilities such as sewage treatment plants and pumping stations, and the final discharge point of the drainage system is determined as the receiving water body node. Furthermore, according to the pipeline directions and connection relationships, the edges between the nodes are determined. It can also be to determine the edges between the facility nodes and other nodes according to the inlet and outlet positions of the facilities, and determine the edges between the discharge nodes and the pipeline network or facility nodes. Then, all the determined nodes are added to the node set, all the determined edges are added to the edge set, and attributes are assigned to each edge. According to the relationships between the nodes and edges, a graph structure model is constructed. For example, an adjacency matrix or an adjacency list can be used to represent the graph structure.

[0035] On the basis of the above technical solution, after establishing the graph structure model based on the nodes and edges of the graph structure model, the following steps are further included: determining node structure parameters and node status parameters corresponding to the nodes of the graph structure model according to the drainage system design data; determining edge structure attributes corresponding to the edges of the graph structure model according to the drainage system design data; and adding the node structure parameters, the node status parameters, and the edge structure attributes to the graph structure model.

[0036] Among them, the node structure parameters are used to reflect the physical characteristics corresponding to each node in the subsystems of the drainage system. The node status parameters are used to reflect the real-time operating status of the nodes. The edge structure attributes reflect the connection characteristics between the nodes in the drainage system.

[0037] Specifically, node structure parameters and node state parameters corresponding to the nodes of each graph structure model, as well as edge structure attributes corresponding to the edges of each graph structure model, are extracted from the drainage system design data, and the extracted parameters are correspondingly added to the graph structure model. Exemplarily, the parameters of nodes corresponding to different drainage system subsystems are different. For example, the node structure parameters corresponding to the source water area node include: area: the projected area of the water area, which is used to calculate the runoff; the impervious rate reflects the surface runoff capacity; the elevation is used to determine the water flow direction and the confluence time; the node structure parameters corresponding to the drainage pipe network node include: pipe diameter: the cross-sectional diameter of the pipe, which affects the flow capacity, slope: the longitudinal slope of the pipe, which determines the water flow velocity, roughness coefficient: Manning coefficient, which characterizes the inner wall roughness of the pipe; the node structure parameters corresponding to the treatment facility node include: treatment scale: the maximum water treatment volume per unit time. Treatment efficiency: pollutant removal rate. Residence time: the average residence time of water flow in the facility; the node structure parameters corresponding to the receiving water body node include: water body capacity: the maximum water storage capacity of the water body. Water quality standard: such as COD, BOD emission limits, etc. The node state parameters may include water level: reflecting the water storage capacity and flooding risk of the node; flow rate: the amount of water passing through the node per unit time; water quality indicators: such as pH value, dissolved oxygen concentration; operation state of the facility: such as start-stop of the pump station, failure of the treatment equipment, etc. The structural attributes of the edge may include length, directionality, connection type: such as combined flow, split flow, overflow. Hydraulic attributes, such as resistance coefficient, flow rate-water level relationship, and topological attributes, such as connectivity (whether nodes are directly connected), path weight, such as water flow time, energy consumption cost. Furthermore, after obtaining the above parameters, an independent attribute table can be created for each node and edge, including parameter name, value, unit, and update time. An attribute graph model (such as Neo4j) can also be used to store the parameters as label attributes of nodes or edges.

[0038] The technical solution provided by the embodiment of the present invention integrates node structure parameters, state parameters, and edge structure attributes into the graph structure model, realizing a comprehensive upgrade from static topology to dynamic attributes. It not only enhances the physical meaning and practical interpretability of the model, but also provides more refined tool support for system simulation, risk assessment, and optimization decision-making.

[0039] S120. Establish a physical simulation model corresponding to the target area according to the drainage system design data, and determine a training data set according to the physical simulation model and the graph structure model.

[0040] Among them, the physical simulation model can be understood as a hydraulic model used to simulate the drainage system. The training data set can be a set of data used to train the neural network model, which can include a training set and a validation set.

[0041] Specifically, collect the drainage system design drawings, topographic maps, meteorological station data, etc. of the target area, select the SWMM software to establish a physical simulation model, set the model parameters according to the drainage system design data, then run the model to generate simulation results, extract features such as node water levels and flows from the physical simulation model, and represent the features in vector form. When determining the training objective as predicting the node water level, generate a large amount of simulation data and divide it into a training set, a validation set, and a test set. Add graph structure information, such as the connection relationships between nodes and edges, to the training dataset.

[0042] Based on the above technical solution, establish a physical simulation model corresponding to the target area according to the drainage system design data, including: determining the drainage subsystem of the target area according to the drainage system design data, and obtaining the historical drainage system data corresponding to the target area; establishing a subsystem physical simulation model corresponding to the drainage subsystem with the historical drainage system data as a constraint condition; using the subsystem physical simulation model as the physical simulation model corresponding to the target area.

[0043] Among them, the drainage subsystem can be understood as a component of the drainage system in the target area. For example, it can include: a catchment subsystem, a pipe network subsystem, a treatment subsystem, and a receiving subsystem. The historical drainage system data can be the data generated during the historical operation of the drainage system in the target area. The subsystem physical simulation model can be a simulation model corresponding to each drainage subsystem.

[0044] Specifically, according to the design data, the drainage system of the target area is divided into sub-catchment subsystems, such as surface catchment areas, pipe network subsystems, such as pipes, inspection wells, pumping stations, treatment subsystems, such as sewage treatment facilities, and receiving subsystems. The topological connection relationships and spatial distribution characteristics of each subsystem are determined, and historical rainfall event data, pipe network operation monitoring data, such as historical water levels and historical flows, receiving water body water quality data, and geographical information data, such as DEM, land use, etc. of the target area are collected. Through data cleaning, missing value filling, and spatio-temporal alignment, a standardized data set is formed. For the sub-catchment subsystem: the SCS-CN or Green-Ampt model is used to simulate runoff generation, and the runoff is calculated by combining the land use type and impervious rate. For the pipe network subsystem: the SWMM model is used to simulate water flow transmission, considering the pipe roughness coefficient, slope, and inspection well overflow threshold. For the treatment subsystem: the activated sludge model or biochemical reaction kinetic equation is used to simulate the sewage treatment efficiency. For the receiving subsystem: the MIKE series software or Delft3D model is used to simulate hydrodynamic and water quality diffusion, considering the water level fluctuation and water quality standard of the receiving water body, and taking the water level of the receiving water body and the area of the external catchment area as the downstream boundary conditions, taking the pump head-flow curve, inspection well overflow threshold, sewage treatment plant treatment capacity, etc. as internal constraints, and taking the rainfall event duration, intensity change, and pipe network response time as time constraints, so as to determine the corresponding physical simulation model of the target area.

[0045] Based on the above technical solution, determining the training data set according to the physical simulation model and the graph structure model includes: simulating according to a preset simulation condition and the physical simulation model to determine simulation data corresponding to the preset simulation condition; mapping the simulation data into the graph structure model to obtain the training data set.

[0046] Among them, the preset simulation condition can be understood as a preset simulation condition, which can include situations such as normal rainfall, extreme rainstorm, changes in storage strategies, equipment failures, etc. The simulation data can be the results obtained by simulating each simulation condition through the physical simulation model.

[0047] Specifically, different simulation scenarios are preset, such as urban drainage systems and basin hydrological processes, and a set of scenarios covering historical extreme events and potential over-limit conditions is designed. For example, for the drainage system, rainfall events of different intensities and durations can be set, combined with parameter perturbations such as land use changes and pipeline aging, to generate multi-dimensional simulation scenarios. Through hydraulic / hydrodynamic models, such as SWMM and MIKE 21, the responses under each scenario are calculated, and the time series data of key state variables, such as pipeline flow, node water level, and overflow volume, are extracted. Then, the generated physical simulation data is mapped to the graph structure model. Exemplarily, the spatial identifiers in the simulation data are aligned with the graph nodes, and the time series data is discretized at a fixed time step to dynamically update the attributes of the graph nodes, such as state parameters like water level and flow. Multiple physical variables, such as flow and pollutant concentration, are combined into multi-attribute vectors of nodes / edges. At the same time, the connection strength of the edges, such as the flow capacity or transmission delay, is dynamically adjusted according to the water flow direction, thereby realizing the mapping between the simulation data and the graph structure model.

[0048] S130. Train the first graph structure neural network model based on the training data set to obtain a second graph neural network model, and determine the drainage system prediction data corresponding to the target area according to the second graph neural network model.

[0049] Among them, the first graph neural network model is a graph neural network constructed according to the graph structure model. The first graph structure neural network model consists of a spatial structure modeling module and a time dynamics modeling module, and the spatial structure modeling module uses graph convolution or graph attention mechanism.

[0050] Specifically, the first graph structure neural network model is iteratively trained using the training data set, and the model parameters are adjusted through the backpropagation algorithm to minimize the prediction error. It should be noted that techniques such as cross-validation and early stopping strategies can be used during the training process to improve the generalization ability and training efficiency of the model. And during the training process, the model is regularly evaluated using the validation data set. The evaluation metrics can include accuracy, recall, F1 score, etc., as well as specific evaluation metrics for the characteristics of the drainage system. Based on the evaluation results during the training process, the model with the optimal performance is selected as the second graph neural network model, and the drainage system prediction data corresponding to the target area is determined according to the second graph neural network model.

[0051] Based on the above technical solution, the determining the drainage system prediction data corresponding to the target area according to the second graph neural network model includes: obtaining the real-time drainage system state and weather state corresponding to the target area; inputting the real-time drainage system state and the weather state into the second graph neural network model to obtain the drainage system prediction data corresponding to the target area output by the second graph neural network model.

[0052] Among them, the real-time drainage system state can be the factual operation state data of the drainage system collected by sensors. The weather state can be understood as the weather information corresponding to the target area.

[0053] Specifically, sensors installed in the drainage system, such as flow sensors, water level sensors, etc., are used to collect data in real time, such as the water flow velocity in the pipeline, the water level height, the working state of the pumping station, etc., and obtain real-time weather data corresponding to the target area, including rainfall, temperature, wind speed, humidity, etc. Then, the real-time drainage system state and weather state data are integrated to form a format suitable for input into the second graph neural network model, and the integrated data is input into the second graph neural network model. The model performs reasoning and calculations based on the input data and the spatial and temporal characteristics learned internally. Furthermore, the second graph neural network model outputs drainage system prediction data corresponding to the target area, including the drainage volume, water level change, pumping station load, etc. within a future period of time in the target area.

[0054] The technical solution of the embodiment of the present invention obtains the drainage system design data of the target area, determines the graph structure model corresponding to the target area according to the drainage system design data, establishes the physical simulation model corresponding to the target area according to the drainage system design data, determines the training data set according to the physical simulation model and the graph structure model, trains the first graph structure neural network model based on the training data set to obtain the second graph neural network model, and determines the drainage system prediction data corresponding to the target area according to the second graph neural network model. Based on the above technical solution, taking the graph structure as the basic data expression form, a graph neural network model that can cover all process elements is constructed, and the overall proxy and replacement of the traditional multi-physical model combination system are realized through this model, improving the determination efficiency and prediction accuracy of the prediction data.

[0055] Embodiment 2

[0056] Figure 2 It is a flowchart of a method for determining drainage system prediction data provided by an embodiment of the present invention. This embodiment further optimizes the technical solution of the method for determining drainage system prediction data on the basis of the above technical solution. As Figure 2 shown, the method includes:

[0057] Construct a graph structure model: By abstracting the urban drainage system into a graph structure to carry the physical connection relationship and functional coupling mechanism of the entire "source-network-plant-river" system. It should be noted that the urban drainage system essentially consists of a series of spatial units with clear hydrodynamic behaviors, and these units form a dynamically coupled transmission network through water flow paths. To uniformly represent the attribute states of various units and the conduction relationships between systems, a multi-type and multi-attribute directed graph structure is introduced as the basis of the model input. Among them, the nodes of the graph are used to represent the functional units in the urban drainage system, and the edges of the graph are used to express the physical or logical connection relationships between nodes, and a comprehensive description of static structure parameters and state variables is achieved through attribute attachment. In this graph structure, the nodes include both surface water collection units, representing the rainfall runoff generation process in different catchment areas; and physical nodes of the drainage pipe network, such as inspection wells, stormwater inlets, overflow outlets, etc., used to simulate the water quantity and water quality transfer behaviors in the underground pipeline system; also include treatment or regulation facility nodes such as water plants, pump stations, storage tanks, etc., used to describe the sewage or rainwater treatment process and its scheduling and control capabilities; at the same time, key points in the river system are also incorporated, such as confluence nodes, intersection points, hydrological stations, etc., to express the hydrological response and pollution conduction mechanism of the receiving water body. Each node not only contains static structure parameters, such as area, slope, treatment capacity, etc., but also carries state variables that change over time, such as liquid level, flow rate, water quality concentration, etc. The edges in the graph are used to represent the physical transfer channels between nodes, including pipe connections, river sections, transmission paths from the outlet to the water plant, connection relationships between the plant effluent and the river, etc., and can also be used to express the flow regulation paths under control relationships or scheduling strategies. Each edge can be attached with parameters such as directionality, distance, hydraulic transmission capacity, operation rules, etc., and multi-edge modeling and edge attribute extension are carried out according to specific simulation requirements. Through the above abstraction, the present invention realizes the unified mapping of the "source-network-plant-river" system, which traditionally requires multiple heterogeneous models to be constructed and coupled separately, into a graph structure with structural expression integrity and behavioral mapping continuity, laying a unified data foundation and model structure support for subsequent overall simulation and prediction based on graph neural networks.

[0058] Constructing training data: After the unified graph structure is completed, in order to train a graph neural network model to have the ability to simulate and predict the whole process behavior of the urban drainage system, for the data generation method of this graph structure, the core source of training data is the multi-model integrated simulation results after calibrating the monitoring data. Since the operation of the urban drainage system is jointly affected by multiple factors such as rainfall, regulation, terrain, and pollution load, it is difficult to cover all the behavior boundaries and response types of the system only relying on historical monitoring data. Therefore, based on physical mechanisms, this invention establishes refined models of each subsystem through existing hydrodynamic and water quality simulation software. After calibration and verification, it is used as a highly reliable response generator to systematically output the data sets required for training. In the source area (surface runoff part), models such as MIKE URBAN and SWMM can be used to simulate the runoff generation process under different land use conditions and rainfall intensities; in the pipe network system, a model with complete structural parameters and scheduling mechanisms is constructed based on SWMM, InfoWorks or MIKE+; in the sewage treatment plant part, models such as GPS-X or BioWin are used to dynamically simulate the water quality treatment process; in the river and receiving water body part, models such as HEC-RAS and MIKE 11 can be called to model processes such as water level, water quality, and transport and diffusion. It should be noted that the above models are calibrated (calibrated) with measured flow, liquid level, and water quality data as constraints to ensure the physical consistency of the simulation results and the adaptability of boundary conditions. By setting multiple working condition combinations, including scenarios such as regular rainfall, extreme rainstorms, changes in storage strategies, and equipment failures, the completed models will output the full-system response data of the urban drainage system at each time step, covering information such as the runoff generation intensity in the source area, the liquid level and flow rate of each node in the pipe network, the characteristic parameters of the water inlet and outlet of the water plant, and the water level and pollutant concentration of each cross-section in the river. All data are remapped with the graph structure as the carrier, that is, the simulation results are corresponding to various nodes and edges in the graph, forming a tensor expression of node / edge attributes for the time series state, and constructing a training sample set. The finally generated training data not only has wide coverage and controllability, but also can fully describe the response laws of the system under various boundaries and operating states, providing high-quality and structure-consistent training data support for the subsequent learning process of the graph neural network model, and at the same time ensuring that the training results can maintain a good balance between simulation accuracy and actual application requirements.

[0059] Training of the Graph Neural Network Model: The graph neural network model constructed in this application can directly receive the current state of the system and external driving variables (such as rainfall and boundary flow) as inputs. Through the information propagation and time evolution mechanisms within the network, it outputs the state changes of each node or edge at future time steps, thereby realizing the surrogate modeling of the entire urban drainage system. The model structure is constructed based on the aforementioned unified graph structure. The input end is the attribute tensors of all nodes and edges in the graph at multiple consecutive time steps, including but not limited to historical liquid levels, flows, water quality concentrations, rainfall intensities, plant loads, etc. The output end is the state values under the same graph structure within the target prediction period. To adapt to the heterogeneity in functions and behaviors of different types of nodes in the drainage system, the present invention adopts a node type embedding mechanism to encode different categories of nodes such as "sources, networks, plants, and rivers", and combines the heterogeneous message passing method in the graph neural network to make the interactions between different nodes follow the propagation rules under physical constraints. At the network structure level, the main body of the model consists of two types of modules: one is the spatial structure modeling module, which uses graph convolution or graph attention mechanisms to aggregate and fuse the neighborhood states of each node based on the node adjacency relationship, capturing the propagation effects of water or pollutants in the spatial structure; the other is the time dynamics modeling module, which uses structures such as recurrent neural networks (RNNs), gated recurrent units (GRUs), and temporal convolutional networks (TCNs) to learn the evolution patterns of each node in the time dimension.

[0060] The two types of modules can be integrated in series, alternately, or in parallel to form a joint spatio-temporal modeling architecture to achieve the overall prediction ability of the system state. During the model training process, the physical model simulation data generated in the above steps is used as the supervision signal, and a multi-objective loss function is defined, such as liquid level prediction error, flow prediction error, water quality index deviation, etc. Through backpropagation and optimizer iteration training, the convergence of model parameters is achieved. After the model training is completed, it can run independently of the original physical model. By inputting the current state and driving variables, it can output the system prediction results at several future time steps, covering the key indicators of the whole process from source runoff to the response of the receiving water body.

[0061] Determination of drainage system prediction data: After completing the construction of the structure and parameter training of the graph neural network model, by inputting the system state information and external driving variables at the current moment, directly output the state evolution results of various nodes and edges in the drainage system within the future time period, realizing the rapid simulation and multi-dimensional prediction of the whole process of "source-network-plant-river". In the specific reasoning process, the user can input the latest state monitoring values, such as rainfall, liquid level, water quality concentration, etc., to the model at any starting moment. The model will perform iterative propagation and rolling prediction on the states of each node based on the spatio-temporal coupling mapping relationship learned during the training stage, and output system indicators including surface water collection intensity, pipe network flow pattern, water treatment plant processing load, river water level change, water quality change trend, etc. Since multiple physical processes have been embedded in the unified structure of the model itself, the reasoning process no longer needs to rely on the call of traditional software models or the data transfer of external interfaces, significantly reducing the system operation coupling, calculation burden and deployment complexity.

[0062] In the technical solution of the embodiment of the present invention, by obtaining the drainage system design data of the target area, determining the graph structure model corresponding to the target area according to the drainage system design data, establishing the physical simulation model corresponding to the target area according to the drainage system design data, determining the training data set according to the physical simulation model and the graph structure model, training the first graph structure neural network model based on the training data set to obtain the second graph neural network model, and determining the drainage system prediction data corresponding to the target area according to the second graph neural network model. Based on the above technical solution, using the graph structure as the basic data expression form, constructing a graph neural network model that can cover all process elements, and realizing the overall agency and substitution of the traditional multi-physical model combination system through this model, improving the determination efficiency and prediction accuracy of the prediction data.

[0063] Embodiment III

[0064] Figure 3 It is a structural schematic diagram of a device for determining drainage system prediction data provided by an embodiment of the present invention. As Figure 3 shown, the device includes:

[0065] A graph structure model determination module 310, configured to obtain the drainage system design data of the target area, and determine the graph structure model corresponding to the target area according to the drainage system design data;

[0066] A data set determination module 320, configured to establish a physical simulation model corresponding to the target area according to the drainage system design data, and determine a training data set according to the physical simulation model and the graph structure model;

[0067] A prediction data determination module 330, configured to train a first graph structure neural network model based on the training data set to obtain a second graph neural network model, and determine drainage system prediction data corresponding to the target area according to the second graph neural network model, where the first graph neural network model is a graph neural network constructed according to the graph structure model.

[0068] Based on the above technical solution, the graph structure model determination module is configured to determine the nodes of the graph structure model based on the drainage system design data and the node types, where the node types include source water area nodes, drainage pipe network nodes, treatment facility nodes, and receiving water body nodes; determine the connection relationship between each node according to the drainage system design data, and determine the edges of the graph structure model according to the connection relationship; establish the graph structure model based on the nodes of the graph structure model and the edges of the graph structure model.

[0069] Based on the above technical solution, after establishing the graph structure model based on the nodes of the graph structure model and the edges of the graph structure model, the graph structure model determination module is configured to determine node structure parameters and node state parameters corresponding to the nodes of the graph structure model according to the drainage system design data; determine edge structure attributes corresponding to the edges of the graph structure model according to the drainage system design data; add the node structure parameters, the node state parameters, and the edge structure attributes to the graph structure model.

[0070] Based on the above technical solution, the data set determination module is configured to determine the drainage subsystem of the target area according to the drainage system design data, and obtain historical drainage system data corresponding to the target area; establish a subsystem physical simulation model corresponding to the drainage subsystem with the historical drainage system data as a constraint condition; use the subsystem physical simulation model as the physical simulation model corresponding to the target area.

[0071] Based on the above technical solution, the data set determination module is configured to perform simulation according to a preset simulation condition and the physical simulation model, and determine simulation data corresponding to the preset simulation condition; map the simulation data into the graph structure model to obtain the training data set.

[0072] Based on the above technical solution, the prediction data determination module is configured to obtain the real-time drainage system state and weather state corresponding to the target area; input the real-time drainage system state and the weather state into the second graph neural network model to obtain drainage system prediction data corresponding to the target area output by the second graph neural network model.

[0073] Based on the above technical solution, the first graph-structured neural network model is composed of a spatial structure modeling module and a temporal dynamics modeling module, and the spatial structure modeling module adopts graph convolution or graph attention mechanism.

[0074] In the technical solution of the embodiment of the present invention, by obtaining the drainage system design data of the target area, determining the graph structure model corresponding to the target area according to the drainage system design data, establishing the physical simulation model corresponding to the target area according to the drainage system design data, determining the training data set according to the physical simulation model and the graph structure model, training the first graph-structured neural network model based on the training data set to obtain the second graph neural network model, and determining the drainage system prediction data corresponding to the target area according to the second graph neural network model. Based on the above technical solution, taking the graph structure as the basic data expression form, constructing a graph neural network model that can cover all process elements, and realizing the overall agency and substitution of the traditional multi-physical model combination system through this model, improving the determination efficiency and prediction accuracy of the prediction data.

[0075] The device for determining the drainage system prediction data provided by the embodiment of the present invention can execute the method for determining the drainage system prediction data provided by any embodiment of the present invention, and has the corresponding functional modules and beneficial effects for executing the method.

[0076] Embodiment Four

[0077] Figure 4 FIG. shows a schematic structural diagram of an electronic device 10 that can be used to implement the embodiments of the present invention. The electronic device is intended to represent various forms of digital computers, such as, laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as, personal digital processing, cellular phones, smart phones, wearable devices (such as helmets, glasses, watches, etc.) and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely illustrative and are not intended to limit the implementation of the present invention described and / or claimed herein.

[0078] As Figure 4As shown, the electronic device 10 includes at least one processor 11 and a memory communicatively connected to the at least one processor 11, such as a read-only memory (ROM) 12, a random access memory (RAM) 13, etc. Among them, the memory stores a computer program executable by the at least one processor. The processor 11 can execute various appropriate actions and processes according to the computer program stored in the read-only memory (ROM) 12 or the computer program loaded from the storage unit 18 into the random access memory (RAM) 13. In the RAM 13, various programs and data required for the operation of the electronic device 10 can also be stored. The processor 11, the ROM 12, and the RAM 13 are connected to each other via a bus 14. An input / output (I / O) interface 15 is also connected to the bus 14.

[0079] Multiple components in the electronic device 10 are connected to the I / O interface 15, including: an input unit 16, such as a keyboard, a mouse, etc.; an output unit 17, such as various types of displays, speakers, etc.; a storage unit 18, such as a disk, an optical disc, etc.; and a communication unit 19, such as a network card, a modem, a wireless communication transceiver, etc. The communication unit 19 allows the electronic device 10 to exchange information / data with other devices through a computer network such as the Internet and / or various telecommunication networks.

[0080] The processor 11 can be various general and / or special processing components with processing and computing capabilities. Some examples of the processor 11 include but are not limited to a central processing unit (CPU), a graphics processing unit (GPU), various dedicated artificial intelligence (AI) computing chips, various processors running machine learning model algorithms, a digital signal processor (DSP), and any appropriate processor, controller, microcontroller, etc. The processor 11 executes the various methods and processes described above, such as the method for determining drainage system prediction data.

[0081] In some embodiments, the method for determining drainage system prediction data can be implemented as a computer program tangibly contained in a computer-readable storage medium, such as the storage unit 18. In some embodiments, part or all of the computer program can be loaded and / or installed onto the electronic device 10 via the ROM 12 and / or the communication unit 19. When the computer program is loaded into the RAM 13 and executed by the processor 11, one or more steps of the method for determining drainage system prediction data described above can be executed. Alternatively, in other embodiments, the processor 11 can be configured to execute the method for determining drainage system prediction data in any other appropriate manner (e.g., by means of firmware).

[0082] The various embodiments of the systems and techniques described above in this document can be implemented in digital electronic circuitry, integrated circuit systems, field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), systems on a chip (SOCs), complex programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments can include: being implemented in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which can be a special-purpose or general-purpose programmable processor that receives data and instructions from a storage system, at least one input device, and at least one output device, and transmits the data and instructions to the storage system, the at least one input device, and the at least one output device.

[0083] The computer programs for implementing the methods of the present invention can be written in any combination of one or more programming languages. These computer programs can be provided to a processor of a general purpose computer, a special purpose computer, or other programmable data processing apparatus, such that the computer programs, when executed by the processor, cause the functions / operations specified in the flowchart and / or block diagram to be implemented. The computer programs can be executed entirely on the machine, partly on the machine, as a stand-alone software package partly on the machine and partly on a remote machine or entirely on the remote machine or server.

[0084] In the context of the present invention, a computer-readable storage medium can be a tangible medium that can contain or store a computer program for use by or in connection with an instruction execution system, apparatus, or device. The computer-readable storage medium can include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination of the foregoing. Alternatively, the computer-readable storage medium can be a machine-readable signal medium. More specific examples of the machine-readable storage medium would include an electrical connection based on one or more wires, a portable computer diskette, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or Flash memory), an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0085] To provide interaction with a user, the systems and techniques described herein can be implemented on an electronic device having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and a pointing device (e.g., a mouse or a trackball) through which the user can provide input to the electronic device. Other kinds of devices can also be used to provide interaction with the user; for example, the feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including acoustic input, voice input, or tactile input).

[0086] The systems and techniques described herein can be implemented in a computing system including backend components (e.g., as a data server), or a computing system including middleware components (e.g., an application server), or a computing system including frontend components (e.g., a user computer having a graphical user interface or a web browser through which the user can interact with an implementation of the systems and techniques described herein), or a computing system including any combination of such backend components, middleware components, or frontend components. The components of the system can be interconnected to each other by digital data communication in any form or medium (e.g., a communication network). Examples of communication networks include: local area network (LAN), wide area network (WAN), blockchain network, and the Internet.

[0087] The computing system can include a client and a server. The client and the server are generally far from each other and usually interact through a communication network. The client-server relationship is created by computer programs running on the respective computers and having a client-server relationship with each other. The server can be a cloud server, also known as a cloud computing server or a cloud host, which is a host product in the cloud computing service system and solves the defects of difficult management and weak business scalability existing in traditional physical hosts and VPS services.

[0088] It should be understood that various forms of the processes shown above can be used, with steps reordered, added, or deleted. For example, the steps recited in the present invention can be executed in parallel, sequentially, or in a different order, as long as the desired results of the technical solution of the present invention can be achieved, and no limitation is imposed herein.

[0089] The above specific embodiments do not constitute a limitation on the protection scope of the present invention. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention shall be included within the protection scope of the present invention.

Claims

1. A method for determining prediction data of a drainage system, characterized in that, Including: Obtain the drainage system design data of the target area, and determine a graph structure model corresponding to the target area according to the drainage system design data; Establish a physical simulation model corresponding to the target area according to the drainage system design data, and determine a training data set according to the physical simulation model and the graph structure model; Train a first graph structure neural network model based on the training data set to obtain a second graph neural network model, and determine drainage system prediction data corresponding to the target area according to the second graph neural network model, wherein the first graph neural network model is a graph neural network constructed according to the graph structure model.

2. The method according to claim 1, wherein The determining the graph structure model corresponding to the target area according to the drainage system design data includes: Determine the nodes of the graph structure model based on the drainage system design data and node types, wherein the node types include source water area nodes, drainage pipe network nodes, treatment facility nodes and receiving water body nodes; Determine the connection relationship between each node according to the drainage system design data, and determine the edges of the graph structure model according to the connection relationship; Establish the graph structure model based on the nodes of the graph structure model and the edges of the graph structure model.

3. The method according to claim 2, wherein After establishing the graph structure model based on the nodes of the graph structure model and the edges of the graph structure model, it further includes: Determine node structure parameters and node state parameters corresponding to the nodes of the graph structure model according to the drainage system design data; Determine edge structure attributes corresponding to the edges of the graph structure model according to the drainage system design data; Add the node structure parameters, the node state parameters and the edge structure attributes to the graph structure model.

4. The method according to claim 1, wherein Establishing a physical simulation model corresponding to the target area according to the drainage system design data includes: Determine the drainage subsystem of the target area according to the drainage system design data, and obtain historical drainage system data corresponding to the target area; Establish a subsystem physical simulation model corresponding to the drainage subsystem with the historical drainage system data as a constraint condition; Use the subsystem physical simulation model as the physical simulation model corresponding to the target area.

5. The method according to claim 1, wherein The determining the training data set according to the physical simulation model and the graph structure model includes: Perform simulation according to a preset simulation condition and the physical simulation model to determine simulation data corresponding to the preset simulation condition; Map the simulation data into the graph structure model to obtain the training data set.

6. The method according to claim 1, wherein The determining the drainage system prediction data corresponding to the target area according to the second graph neural network model includes: Obtain the real-time drainage system state and weather state corresponding to the target area; Input the real-time drainage system state and the weather state into the second graph neural network model to obtain drainage system prediction data corresponding to the target area output by the second graph neural network model.

7. The method according to claim 1, characterized in that The first graph-structured neural network model consists of a spatial structure modeling module and a temporal dynamics modeling module, and the spatial structure modeling module adopts graph convolution or graph attention mechanism.

8. A device for determining prediction data of a drainage system, characterized in that, Including: A graph structure model determination module, configured to obtain the drainage system design data of the target area, and determine a graph structure model corresponding to the target area according to the drainage system design data; A data set determination module, configured to establish a physical simulation model corresponding to the target area according to the drainage system design data, and determine a training data set according to the physical simulation model and the graph structure model; A prediction data determination module, configured to train a first graph-structured neural network model based on the training data set to obtain a second graph neural network model, and determine drainage system prediction data corresponding to the target area according to the second graph neural network model, wherein the first graph neural network model is a graph neural network constructed according to the graph structure model.

9. An electronic device, characterized in that, The electronic device includes: At least one processor; and A memory communicatively connected to the at least one processor; wherein, The memory stores a computer program executable by the at least one processor, and the computer program is executed by the at least one processor so that the at least one processor can execute the method for determining the drainage system prediction data according to any one of claims 1-7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer instructions for causing a processor to implement the method for determining the drainage system prediction data according to any one of claims 1-7 when executed.

Images