Urban water runoff control method and system

Abstract

The invention provides an urban water runoff control method and system, and relates to the technical field of urban water runoff control, and the method comprises the steps: obtaining historical rainfall data, real-time data, modeling parameters and urban drainage system parameters; carrying out rainfall frequency analysis processing on the historical rainfall data based on a preset Bayesian inference mathematical model to obtain feature information; carrying out regional division processing according to the feature information and the rainwater pipe network layout map to obtain a segmentation result; a rainfall prediction model is constructed according to the segmentation result, and a prediction change result is obtained through calculation; performing simulation processing according to the prediction result and the rainwater pipe network layout map to obtain a simulation result; and performing optimization processing on the initial control scheme based on a simulation result to obtain an urban water runoff dynamic control scheme. The urban drainage system is comprehensively analyzed and optimized by comprehensively considering various factors such as terrain, buildings and soil permeability characteristics and adopting a graph embedding algorithm and a multi-objective optimization method, and the overall efficiency of the system is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of urban water body runoff control, and in particular to an urban water body runoff control method and system. Background Art

[0002] With the continuous acceleration of urbanization, the design and management of urban drainage systems have become increasingly important. Urban runoff control plays a key role in preventing floods, reducing water pollution, and protecting urban infrastructure. However, there are many problems with the current urban drainage system management technology. In the prior art, the common method is to design and evaluate the drainage system based on empirical formulas and simple statistical analysis. For example, the conventional rainfall frequency analysis method is used to predict the rainfall intensity and duration, and then simulated by the drainage network model. However, these methods have significant defects: first, they cannot effectively use real-time data to dynamically adjust the operation of the drainage system; second, traditional models cannot accurately reflect the changes in complex urban terrain, building density and soil permeability characteristics, resulting in a large deviation between the simulation results and the actual situation.

[0003] Based on the above shortcomings of the prior art, there is an urgent need for a method and system for controlling urban water runoff. Summary of the invention

[0004] The purpose of the present invention is to provide a method and system for controlling urban water body runoff to improve the above problems. In order to achieve the above purpose, the technical solution adopted by the present invention is as follows:

[0005] In a first aspect, the present application provides a method for controlling water runoff in an urban area, comprising:

[0006] Acquire historical rainfall data, real-time data, modeling parameters and urban drainage system parameters, wherein the real-time data includes real-time rainfall data and flow monitoring data, the modeling parameters include urban topographic maps, building models and soil property distribution maps, and the urban drainage system includes a rainwater pipe network layout map and an initial control plan;

[0007] Based on a preset Bayesian inference mathematical model, the historical rainfall data is subjected to rainfall frequency analysis to obtain characteristic information, wherein the characteristic information includes urban rainfall characteristics and trends;

[0008] Performing regional division processing according to the characteristic information and the rainwater pipe network layout diagram, dividing the city into at least one drainage partition based on drainage characteristics to obtain a segmentation result;

[0009] A rainfall prediction model is constructed according to the segmentation results, and the real-time data is used as an input value of the rainfall prediction model to calculate the predicted change result, wherein the predicted result includes the rainfall and flow change in a future preset time period;

[0010] Performing simulation processing according to the prediction results and the rainwater pipe network layout diagram, modeling the urban drainage system as a directed graph structure, and analyzing the node pressure, pipe flow and overflow risk under different scenarios to obtain simulation results;

[0011] The initial control scheme is optimized based on the simulation results to obtain a dynamic control scheme for urban water body runoff.

[0012] In the second aspect, the present application also provides a water runoff control system for an urban area, comprising:

[0013] An acquisition module, used to acquire historical rainfall data, real-time data, modeling parameters and urban drainage system parameters, wherein the real-time data includes real-time rainfall data and flow monitoring data, the modeling parameters include urban topographic maps, building models and soil property distribution maps, and the urban drainage system includes a rainwater pipe network layout map and an initial control plan;

[0014] An analysis module, which performs rainfall frequency analysis on the historical rainfall data based on a preset Bayesian inference mathematical model to obtain characteristic information, wherein the characteristic information includes urban rainfall characteristics and trends;

[0015] A division module, used for performing regional division processing according to the characteristic information and the rainwater pipe network layout diagram, and dividing the city into at least one drainage zone based on drainage characteristics to obtain a segmentation result;

[0016] A prediction module, used to construct a rainfall prediction model according to the segmentation results, and use the real-time data as an input value of the rainfall prediction model to calculate the predicted change results, wherein the predicted results include rainfall and flow changes within a preset time period in the future;

[0017] A simulation module, used for performing simulation processing according to the prediction results and the rainwater pipe network layout diagram, by modeling the urban drainage system as a directed graph structure and analyzing the node pressure, pipe flow and overflow risk under different scenarios to obtain simulation results;

[0018] The optimization module optimizes the initial control scheme based on the simulation results to obtain a dynamic control scheme for urban water body runoff.

[0019] The beneficial effects of the present invention are:

[0020] The present invention comprehensively analyzes and optimizes the urban drainage system by comprehensively considering multiple factors such as terrain, buildings and soil permeability characteristics, and adopts a graph embedding algorithm and a multi-objective optimization method, thereby improving the overall efficiency of the system.

[0021] Other features and advantages of the present invention will be described in the following description, and partly become apparent from the description, or be understood by implementing the embodiments of the present invention. The purpose and other advantages of the present invention can be realized and obtained by the structures particularly pointed out in the written description, claims, and drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings required for use in the embodiments are briefly introduced below. It should be understood that the following drawings only show certain embodiments of the present invention and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other related drawings can be obtained based on these drawings without creative work.

[0023] Figure 1 It is a schematic flow chart of the urban water body runoff control method described in an embodiment of the present invention;

[0024] Figure 2 Schematic diagram of the hierarchical clustering process. DETAILED DESCRIPTION

[0025] In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, rather than all of the embodiments. The components of the embodiments of the present invention generally described and shown in the drawings here can be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the present invention provided in the drawings is not intended to limit the scope of the claimed invention, but merely represents selected embodiments of the present invention. Based on the embodiments in the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0026] It should be noted that similar reference numerals and letters represent similar items in the following drawings, so once an item is defined in one drawing, it does not need to be further defined and explained in the subsequent drawings. At the same time, in the description of the present invention, the terms "first", "second", etc. are only used to distinguish the description and cannot be understood as indicating or implying relative importance.

[0027] Embodiment 1:

[0028] This embodiment provides a method for controlling water runoff in an urban area.

[0029] See also Figure 1 , the figure shows that the method includes steps S100 to S600.

[0030] Step S100, obtaining historical rainfall data, real-time data, modeling parameters and urban drainage system parameters, the real-time data including real-time rainfall data and flow monitoring data, the modeling parameters including urban topographic map, building model and soil property distribution map, the urban drainage system including rainwater pipe network layout map and initial control plan;

[0031] It is understandable that historical rainfall data can reveal the rainfall patterns of cities under different seasons and climate conditions, providing a basis for subsequent rainfall frequency analysis. Real-time data includes real-time rainfall data and flow monitoring data, which can dynamically reflect the current rainfall situation and the operating status of the drainage system. It is collected in real time through automated monitoring equipment such as rain gauges and flow meters, and transmitted to the central control system in real time through the Internet of Things technology to ensure the timeliness and accuracy of the data. The urban topographic map and building model in the modeling parameters are constructed through remote sensing technology and geographic information system (GIS). These high-precision spatial data can describe the topographic characteristics and building layout of the city in detail, reflecting the impact of the urban surface on rainwater runoff. The soil property distribution map is obtained through field sampling and laboratory analysis, focusing on the characteristics of soil permeability, saturation and water content, which directly affect the infiltration and runoff rate of rainwater. The rainwater pipe network layout map and the initial control plan come from the design drawings of the urban planning department and the drainage system. The rainwater pipe network layout map records the path, diameter and material of the drainage pipe in detail, and the initial control plan includes the operation strategy and scheduling rules of the existing drainage facilities. This information is integrated into a comprehensive database, providing complete basic data for subsequent model construction and simulation.

[0032] Step S200: Perform rainfall frequency analysis on historical rainfall data based on a preset Bayesian inference mathematical model to obtain characteristic information, where the characteristic information includes urban rainfall characteristics and trends;

[0033] Understandably, Bayesian inference provides a systematic method to combine prior knowledge with actual data to obtain more accurate and reliable rainfall characteristic information. Characteristic information includes rainfall distribution (such as average rainfall, standard deviation and extreme value of rainfall), rainfall event frequency (such as the frequency and time interval distribution of rainfall events), rainfall duration (such as average rainfall duration and its fluctuation), rainfall intensity (such as the distribution of rainfall intensity and extreme rainfall intensity), seasonal changes (such as rainfall patterns in different seasons and monthly or quarterly rainfall), and long-term trends (such as the trend of rainfall over time and the impact of climate change). Rainfall characteristic information can fully reveal the laws and characteristics of urban rainfall, and provide a scientific basis for the design of urban drainage systems, extreme weather response and flood control measures, thereby improving urban flood control capabilities and infrastructure safety.

[0034] Step S300: performing regional division processing according to the characteristic information and the rainwater pipe network layout diagram, and dividing the city into at least one drainage zone based on the drainage characteristics to obtain a segmentation result;

[0035] It should be noted that the characteristic information includes the frequency, intensity and duration of urban rainfall, which can reflect the response of different regions to rainfall events. With this information, areas in the city that are prone to waterlogging and flooding can be identified. The rainwater pipe network layout map provides detailed information about the urban drainage system, including the path, diameter, material, etc. of the drainage pipes. Combined with this information, the drainage capacity of each area and the distribution of existing drainage facilities can be understood in detail. By analyzing this data, the drainage characteristics of each area can be determined. For example, some areas have low drainage efficiency due to aging or unreasonable design of the pipe network, while other areas may be prone to waterlogging due to low terrain. When dividing the area, the topography, building distribution and soil permeability characteristics of the city must be considered. Topographic data can help determine the natural flow direction and collection point of rainwater, building models can reflect the obstruction of buildings to the flow of rainwater, and soil property distribution maps provide information on the permeability of soil in different areas. These factors work together to determine the drainage characteristics of each area. By comprehensively analyzing this information, the city can be divided into reasonable areas. For example, high-density building areas and low-density residential areas can be divided into different drainage zones due to different drainage needs and drainage characteristics. At the same time, low-lying areas prone to waterlogging need special attention. By strengthening the drainage facilities and management of these areas, the risk of waterlogging can be effectively reduced. This step can achieve accurate management of the urban drainage system through regional division. The divided drainage zones can better reflect the specific drainage needs and drainage capacity of each area, thereby formulating more effective drainage management strategies.

[0036] Step S400: construct a rainfall prediction model based on the segmentation results, and use the real-time data as the input value of the rainfall prediction model to calculate the predicted change results, which include the rainfall and flow changes in the future preset time period;

[0037] It can be understood that this step can achieve high-precision prediction of future rainfall and flow changes by building and running the rainfall prediction model, providing a scientific basis for the dynamic management of urban drainage systems. The introduction of real-time data improves the response speed and accuracy of the model, allowing cities to take preventive measures earlier and reduce the risk of waterlogging and flood disasters. At the same time, the prediction results can also help optimize the operation strategy of the drainage system, improve the overall drainage efficiency, and ensure the safety of urban residents and the stable operation of urban infrastructure.

[0038] Step S500: Perform simulation processing according to the prediction results and the rainwater pipe network layout diagram, and obtain simulation results by modeling the urban drainage system as a directed graph structure and analyzing the node pressure, pipe flow and overflow risk under different scenarios;

[0039] It can be understood that first, based on the rainfall forecast results and the rainwater pipe network layout diagram, a directed graph model of the urban drainage system is constructed. The nodes in the directed graph structure represent key points in the drainage system, such as drainage wells, rainwater inlets, and pipe intersections, and the edges represent the drainage pipes connecting these nodes. By assigning corresponding weights to nodes and edges, where the weight of the node represents the flow demand or pressure it bears, and the weight of the edge represents the flow capacity and pressure bearing capacity of the pipe, it can more realistically reflect the characteristics of the actual drainage system. In scenarios simulating different rainfall intensities and durations, potential overflow risk areas are identified by calculating the pressure changes of each node under these scenarios and the flow distribution of the pipe.

[0040] Step S600: Optimize the initial control scheme based on the simulation results to obtain a dynamic control scheme for urban water body runoff.

[0041] Understandably, the optimization process first requires an evaluation of the initial control scheme to identify its deficiencies. For example, some drainage pipes are prone to overload in high rainfall scenarios, or some nodes are at risk of overflow under certain conditions. By identifying these problems, targeted improvement measures can be proposed. The key to the optimization process is to use the data from the simulation results to find the best combination of control measures through multi-objective optimization algorithms and machine learning techniques. These measures include adjusting the flow distribution of drainage pipes, optimizing drainage paths, increasing the capacity of reservoirs, or deploying temporary drainage equipment in high-risk areas. The optimized dynamic control scheme needs to be able to respond to changes in actual conditions in real time. Therefore, it is also necessary to combine real-time monitoring data and continuously adjust and optimize the control strategy through a feedback control mechanism. In this way, the urban drainage system can flexibly respond to different rainfall intensities and durations to ensure the efficient operation of the drainage system.

[0042] Further, step S200 includes steps S210 to S240:

[0043] Step S210, performing data sorting and processing according to historical rainfall data, and obtaining a rainfall data set by filling missing values ​​and removing abnormal values;

[0044] Step S220, performing parameter estimation processing according to the rainfall data set, calculating the rainfall amount, rainfall duration and rainfall interval of each rainfall event by maximum likelihood estimation method, and obtaining a preliminary rainfall event characteristic parameter set;

[0045] Specifically, the basic elements of rainfall events are first defined, where rainfall refers to the total amount of precipitation measured in a specific time period, rainfall duration is the length of time from the start to the end of the event, and rainfall interval refers to the time interval between two rainfall events. Then, probability distribution assumptions are made for rainfall, duration, and interval. It is assumed that rainfall follows a gamma distribution with parameter θ1, duration follows a normal distribution with parameter θ2, and interval follows an exponential distribution with parameter θ3. Based on these assumptions, a likelihood function L(θ) is constructed to represent the joint probability of observing these rainfall events under a given parameter θ. For independent and identically distributed samples, the likelihood function is expressed as the product of the probability density functions of all observed data. The formula is as follows:

[0046]

[0047] Where L(θ1,θ2,θ3) represents the likelihood function; θ1, θ2 and θ3 represent the distribution parameters of rainfall, rainfall duration and rainfall interval respectively; i represents the sequence number of the rainfall event; z i represents the total number of rainfall events; x i represents the rainfall of the ith rainfall event; y i represents the duration of the i-th rainfall event; z i represents the interval time of the i-th rainfall event; P(x i |θ1)、P(y i |θ2) and P(z i |θ3) represent the probability density functions of rainfall amount, rainfall duration and rainfall interval time under given parameters θ1, θ2 and θ3 respectively.

[0048] Next, the optimal parameter estimate is found by solving the maximum value of the logarithmic form of the likelihood function (log-likelihood function), preferably by a numerical optimization algorithm, such as gradient descent or Newton-Raphson method. The calculation formula is as follows:

[0049]

[0050] in, and They represent the optimal parameter estimates of rainfall amount, rainfall duration and rainfall interval respectively.

[0051] Through the maximum likelihood estimation method, the most likely parameter values ​​of rainfall, duration and interval time of each rainfall event can be estimated respectively. Finally, a preliminary set of rainfall event characteristic parameters is generated, which contains the statistical characteristics of all rainfall events and provides basic data for further Bayesian inference and rainfall frequency analysis.

[0052] Step S230, performing Bayesian inference processing based on the preliminary rainfall event characteristic parameter set, by setting the gamma distribution as the prior distribution and using the historical rainfall data to update the posterior distribution, to obtain the posterior distribution result of each rainfall event parameter;

[0053] Specifically, the basic idea of ​​Bayesian inference is to update the knowledge of parameters by combining prior information and observed data. In this embodiment, the gamma distribution is selected as the prior distribution of the rainfall parameter because the gamma distribution can well describe the positive skewed distribution characteristics of rainfall. The probability density function of the gamma distribution is defined as:

[0054]

[0055] Among them, α and β are the shape parameter and scale parameter of the gamma distribution; θ represents the parameter of the rainfall event; P(θ|α,β) is the probability density function of the parameter θ, which describes the prior knowledge of the parameter θ before there is any observed data; Γ(α) represents the gamma function; and e represents a natural constant.

[0056] Next, we use historical rainfall data to perform likelihood estimation. The likelihood function is defined as follows:

[0057]

[0058] Where D = {x1, x2, ..., x n} represents the observed historical rainfall data set; i represents the sequence number of the rainfall event; x i represents the i-th rainfall event; n represents the total number of rainfall events; P(x i |θ) indicates that a rainfall event x is observed under parameter θ i The probability density function of .

[0059] In Bayesian inference, we need to calculate the posterior distribution P(x i |θ), which is defined according to Bayes’ theorem as:

[0060]

[0061] The calculation of the posterior distribution can be expressed as:

[0062] P(θ|D)∝P(D|θ)P(θ);

[0063] Among them, P(θ|D) represents the likelihood function; P(θ) represents the prior distribution; and P(D) represents the marginal likelihood of the observed data.

[0064] Furthermore, the gamma distribution is used as a prior distribution. After updating with historical rainfall data, the resulting posterior distribution is still in the form of a gamma distribution, but its parameters α and β will be updated according to the observed data. The updated parameters α' and β' can be calculated using the following formula:

[0065] α′=α+n;

[0066]

[0067] Where i represents the sequence number of the rainfall event; x i represents the i-th rainfall event; n represents the total number of rainfall events; α and β are the shape parameter and scale parameter of the gamma distribution; α' and β' represent the updated shape parameter and scale parameter.

[0068] Through Bayesian inference processing, the parameters of rainfall events can be estimated more accurately, improving the accuracy of rainfall prediction models. The posterior distribution provides uncertainty information of parameters, which helps to improve the reliability and scientificity of predictions and provide a more solid data foundation for the design and management of urban drainage systems.

[0069] Step S240: Perform rainfall frequency analysis processing according to the posterior distribution result, generate rainfall events through simulation, and obtain characteristic information by statistically analyzing the frequency distribution of rainfall amount, rainfall duration and rainfall interval time.

[0070] It is understandable that this step uses the Monte Carlo method to randomly extract samples from the posterior distribution to generate rainfall events, and simulates possible rainfall scenarios through a large number of repeated random sampling. Specifically, a certain number of samples are extracted from each posterior distribution, and the generated sample data are combined to form a complete set of rainfall events. Then, the simulated rainfall event data is statistically analyzed to calculate the frequency distribution of rainfall, rainfall duration, and rainfall interval time. Further, by drawing a histogram or cumulative distribution function, the distribution of these parameters is displayed to obtain the characteristic information of the rainfall event, including the common range of rainfall, the typical length of rainfall duration, and the interval time between rainfall events. Through this method, the characteristics and changing laws of urban rainfall events can be understood in detail, providing reliable data support for subsequent rainfall trend prediction and urban drainage system optimization. This not only improves the understanding of urban rainfall patterns, but also provides a scientific basis for flood prevention and disaster reduction and drainage system optimization, and improves the ability of cities to cope with extreme weather events.

[0071] Example 2

[0072] The difference between this embodiment and embodiment 1 is that step S300 includes steps S310 to S330:

[0073] Step S310: Perform runoff simulation processing according to the characteristic information and modeling parameters, and analyze the convergence, diversion and water accumulation areas of the runoff to obtain the urban drainage characteristic vector by simulating rainfall events of different rainfall intensities and durations;

[0074] Further, step S310 includes steps S311 to S314:

[0075] Step S311, performing terrain modeling processing according to the city terrain map and the building model, and generating static terrain modeling data including terrain features and building distribution through a multi-scale terrain modeling algorithm;

[0076] It is understandable that the multi-scale terrain modeling algorithm (MSTM) is able to process terrain data of different scales and generate a high-precision three-dimensional terrain model. The multi-scale terrain modeling algorithm combines large-scale overall terrain data with small-scale local detail data to generate a comprehensive model that contains both macro-terrain features and reflects micro details. This method not only improves the accuracy of terrain modeling, but also effectively reduces the consumption of computing resources. Specifically, the urban terrain map is first preprocessed to extract terrain elevation data and important terrain features such as rivers, hills and low-lying areas. Next, the building model data is integrated into the terrain map to reflect the specific location, shape and height of the building. Then, through the multi-scale terrain modeling algorithm, these terrain and building data are fused to generate static terrain modeling data containing terrain features and building distribution. The static terrain modeling data generated in this step describes the terrain features and building distribution of the city in detail, and can accurately reflect the surface morphology and structure of the city.

[0077] Step S312: Perform dynamic change process simulation processing according to the characteristic information and the soil property distribution map, and obtain dynamic infiltration data of different areas by performing time series analysis on the saturation change of the soil in rainfall events with different rainfall intensities and durations;

[0078] In the simulation of dynamic change process, it is necessary to consider the infiltration and saturation characteristics of soil under different rainfall conditions. Different types of soils will show different infiltration capacity and saturation characteristics during rainfall. For example, sandy soil has a faster infiltration rate, while clay has a slower infiltration rate. In order to accurately simulate this process, it is necessary to use the time series analysis method. The specific steps are as follows:

[0079] Step S3121: Based on the soil property distribution map, using the existing soil sample data, construct a preliminary soil permeability model for each area;

[0080] Specifically, soil samples are first collected from different areas of the city and laboratory analyzed to obtain physical and chemical property data such as particle size distribution, water content, permeability coefficient, and saturation. These laboratory analysis results are then integrated with the soil property distribution map to ensure that the soil properties of each area are fully described. Next, a preliminary soil permeability model for each area is constructed using statistical methods and numerical modeling methods, preferably, such as multiple regression analysis, geostatistical methods, or machine learning algorithms, including permeability coefficient and saturation characteristics. Finally, the accuracy of the model is verified and corrected by comparing it with actual observation data and known soil properties to ensure that it can accurately reflect the permeability characteristics of soil in different areas. By constructing a preliminary soil permeability model, the permeability capacity and saturation characteristics of soil in different areas of the city can be understood in detail, providing the necessary basic parameters for subsequent dynamic simulation of rainfall events and runoff analysis.

[0081] Step S3122, simulating various rainfall events according to the data of different rainfall intensities and durations in the characteristic information;

[0082] It should be noted that this step uses the extracted rainfall characteristic information to determine the rainfall and rainfall time curve for each simulation event, and then sets the initial conditions based on the preliminary soil infiltration model, including the initial soil moisture content and saturation. Next, a hydrological model or soil water balance model is used for numerical simulation to simulate the infiltration, retention and runoff process of water in the soil during rainfall, focusing on analyzing the saturation changes and infiltration rate of the soil during rainfall events.

[0083] Step S3123: For each simulated rainfall event, a time series analysis is performed to track the infiltration rate and saturation change of the soil, and the saturation change of the soil during and after the rainfall is calculated through numerical simulation and the infiltration equation;

[0084] It can be understood that this step uses time series analysis methods to identify trends and periodic changes in the data, focusing on tracking the changes in soil at each stage of rainfall. Next, numerical simulation technology is applied to calculate the infiltration process of soil under different rainfall intensities and durations through the Richards equation, and the saturation change of soil from the initial state to the end of rainfall is analyzed. Through these analyses and calculations, the infiltration characteristics and saturation change laws of soil under different rainfall conditions can be fully understood.

[0085] The one-dimensional form of the Richards equation is:

[0086]

[0087] Where θ represents volumetric water content; t represents time; z represents soil depth; K(θ) represents soil water permeability coefficient; and ψ represents soil matrix potential.

[0088] In practical applications, it is necessary to perform numerical solution in combination with the characteristic curve of a specific soil. In this embodiment, it is assumed that the soil characteristic curve is:

[0089]

[0090] where ψ(θ) represents the soil matrix potential as a function of volumetric water content θ; K(θ) represents the soil water permeability coefficient; ψ s represents the matrix potential of soil under saturated state; θ s Indicates the saturated water content of the soil; K s represents the permeability coefficient of soil under saturated state; λ and η represent empirical parameters that reflect soil characteristics. Substituting these relationships into the Richards equation, numerical simulation can be performed to solve the distribution and change of soil moisture content under rainfall conditions.

[0091] Step S3124: Arrange the results of the time series analysis to obtain dynamic penetration data.

[0092] It can be understood that this step converts the collated data into a data set that reflects the changes in soil infiltration rate and saturation over time under different rainfall intensities and durations. Each data set contains data at multiple time points, describing the response of the soil throughout the rainfall event.

[0093] Step S313, constructing a dynamic heterogeneous graph model according to the static terrain modeling data and the dynamic infiltration data;

[0094] It should be noted that this step combines static terrain data with dynamic infiltration data to form a comprehensive model that can dynamically reflect the water flow and soil infiltration characteristics during rainfall events. The nodes in the model represent important terrain feature points, building location points, and soil sample points, and the edges represent the mutual influence relationship between these nodes, including water flow paths and infiltration paths. Among them, the dynamicity is reflected in the fact that the model can reflect the hydrological changes during and after rainfall events in real time. Static terrain modeling data provides the terrain characteristics and building distribution of the city, which is a relatively fixed basic data. The dynamic infiltration data comes from time series analysis, reflecting the changes in the infiltration rate and saturation of the soil under different rainfall intensities and durations. These dynamic data enable the model to update the moisture content and infiltration of the soil in real time, and simulate the dynamic hydrological changes during and after rainfall. Secondly, heterogeneity is reflected in the multi-level and multi-factor integration of the model. The hydrological process in the urban environment is complex and changeable, and the terrain, buildings, and soil characteristics of different regions are different. The nodes in the model represent important terrain feature points, building location points, and soil sample points, and these nodes themselves are heterogeneous. For example, the water flow and infiltration characteristics of high-density built-up areas and open green spaces are significantly different, and the model needs to handle these heterogeneous characteristics separately. The edges represent the mutual influence relationship between these nodes, including water flow paths and infiltration paths. Through this construction method, the dynamic heterogeneous graph model can not only reflect the hydrological response characteristics of different areas of the city under various rainfall conditions in real time, but also accurately describe the mutual influence and comprehensive effects between different areas.

[0095] Step S314: According to the dynamic heterogeneous graph model, the node feature vectors are convolved and aggregated using a graph embedding algorithm to generate a node embedding vector, and the node embedding vector is mapped to a high-dimensional feature space to obtain an urban drainage feature vector.

[0096] Specifically, first, the feature data of the nodes in the dynamic heterogeneous graph model are integrated to form the initial node feature vector. Then, the graph embedding algorithm is used to convolve and aggregate the initial node feature vector. The feature information of the node and its neighboring nodes is captured through the convolution operation, and the features of the neighboring nodes are integrated through the aggregation operation to generate the embedding vector of each node. These embedding vectors contain not only the original feature information of the node, but also the relationship information with other nodes, so that they can more comprehensively represent the node characteristics in the high-dimensional space. Then the node embedding vector is mapped to the high-dimensional feature space. Through the mapping process, the node embedding vector can be represented in a higher-dimensional space to capture complex features and relationships. Finally, the high-dimensional embedding vectors of all nodes are integrated to obtain the urban drainage feature vectors, which can comprehensively describe the performance of the urban drainage system under different rainfall conditions, including drainage capacity, waterlogging risk, and drainage efficiency.

[0097] Step S320: performing clustering processing according to the urban drainage model and the rainwater pipe network layout diagram to obtain a clustering result;

[0098] Further, step S320 includes steps S321 to S324:

[0099] Step S321: perform similarity calculation processing based on the city drainage feature vectors, calculate the similarity between each city drainage feature vector by using the cosine similarity method, and obtain a similarity matrix;

[0100] It can be understood that when calculating similarity, the cosine value of the angle between two vectors is determined by the dot product and norm of the vectors. The similarity value is between -1 and 1. The closer the value is to 1, the more similar the two vectors are. The obtained similarity matrix is ​​a symmetric matrix, and each element in the matrix represents the similarity between the corresponding two eigenvectors.

[0101] Step S322: Perform hierarchical clustering processing based on the similarity matrix and the rainwater pipe network layout diagram, and obtain preliminary clustering results by combining the geographical proximity and pipe network connectivity in the pipe network;

[0102] It should be noted that the focus of this step is to make full use of the geographical proximity and pipe network connectivity of the urban drainage system to ensure that the clustering results can reflect both the similarity of drainage characteristics and the actual drainage network structure. Specifically, the similarity matrix and the rainwater pipe network layout map are first used as input data. The former describes in detail the similarity between urban drainage feature vectors, and the latter shows the geographical location and connection relationship of the drainage system. Next, weights are defined for the proximity and connectivity of each region, where the geographical proximity weight reflects the geographical proximity of the region, and the pipe network connectivity weight reflects the actual connection of the drainage pipe. By introducing these weights, the hierarchical clustering algorithm is improved to ensure that the actual geographical and pipe network structure are given priority in the clustering process. That is, each time the clusters are merged, not only the cosine similarity of the regional feature vectors is considered, but also the geographical proximity weight and pipe network connectivity weight are introduced, and the merging criterion is modified to weighted similarity. Then, the hierarchical clustering algorithm gradually merges the clusters with the highest similarity to generate a hierarchical tree. The merging process is as follows: Figure 2 As shown in the figure, nodes A to E represent various areas in the urban drainage system. The final clusters 1 and 2 show that the city is divided into two main drainage areas. In the merging process, areas with high geographical proximity and pipe network connectivity are preferentially merged to form a clustering result that is more in line with the actual drainage system structure. Finally, the preliminary clustering results of each area are determined through the hierarchical tree structure.

[0103] Step S323: using a density-based spatial constraint clustering algorithm to refine and optimize the preliminary clustering results, by combining the flow direction in the pipe network and the density of regional water accumulation points, processing boundary points and noise points to obtain optimized clustering results;

[0104] It should be noted that, in this embodiment, the boundary point refers to a point located at the junction of different clusters, with unclear features and easily affected by the surrounding area. These points show uncertain drainage behavior in actual rainfall events. Noise points are points that are scattered and isolated in space. These points are caused by abnormal rainfall or special terrain conditions. Specifically, the boundary points and noise points in the preliminary clustering results are first identified. The boundary points are often located at the turning point of the drainage system or the junction of different terrain features, and the noise points are some small-scale water accumulation points or areas with poor drainage. Then, the density of the points is calculated using a density-based spatial constraint clustering algorithm to identify core points, boundary points and noise points. In this process, the core point is a high-density point with more neighboring points around it, while the boundary point and the noise point are low-density points and isolated points, respectively. The advantage of the density-based spatial constraint clustering algorithm is that it can handle clusters of any shape and effectively identify noise points. In the process of refinement and optimization, the flow direction reflects the actual path of the water flow in the urban drainage system. For example, if a boundary point located upstream shows a trend of moving downstream during rainfall, it should be reclassified into the downstream cluster instead of remaining in the original cluster. Such adjustments can more accurately reflect the actual drainage situation. High-density waterlogging areas are usually places with high drainage pressure. These areas show significant waterlogging characteristics during rainfall events and need to be clustered separately for better management. For example, a city square continues to have waterlogging problems during multiple rainfall events. This square should be identified as a high-density waterlogging point and treated separately during the clustering process so that targeted drainage measures can be taken. This step ensures that each point can be reasonably attributed to a cluster by redistributing boundary points and noise points, thereby improving the accuracy and effectiveness of the clustering results.

[0105] Step S324: Perform feature fusion processing according to the optimized clustering result, and fuse the drainage feature vectors in each cluster by weighted average method to obtain the final clustering result.

[0106] The weighted average calculation formula is:

[0107]

[0108] Among them, v f represents the fused drainage feature vector; i represents the sequence number of the drainage feature vector; n represents the total number of drainage feature vectors in the cluster; v i represents the i-th drainage feature vector in each cluster; w iRepresents the weight of the i-th drainage eigenvector.

[0109] Step S330: Divide the city into at least one drainage partition according to the clustering result to obtain a segmentation result.

[0110] Further, step S330 includes steps S331 to S333:

[0111] Step S331, performing region recognition processing according to the optimized clustering result to obtain a region recognition result;

[0112] It can be understood that this step is based on optimizing the clustering results to identify the various drainage areas in the city. These areas are divided according to similarity, density and actual pipe network connection relationships to ensure that the drainage characteristics of each area are relatively consistent. During the identification process, it is necessary to combine the city's geographic information system (GIS) data, map the clustering results onto the actual city map, and determine the geographical boundaries of each cluster.

[0113] Step S332: Determine the drainage outlet and key nodes of each drainage partition according to the area recognition result, and obtain a preliminary drainage partition result;

[0114] It should be noted that, based on the identification, the drainage outlets and key nodes of each drainage zone are further determined. The drainage outlet refers to the main outlet where rainwater is discharged through the drainage system, and the key nodes include important connection points, flow regulation points and nodes prone to water accumulation in the drainage system. The determination of these nodes and outlets helps to optimize the management and control of the entire drainage system.

[0115] Step S333: Evaluate and adjust the boundaries of the drainage partitions according to the preliminary drainage partition results to obtain the final segmentation results.

[0116] Furthermore, the preliminary drainage zoning results are evaluated and adjusted as necessary based on the actual geographical conditions and the operation of the drainage system. The evaluation criteria include drainage efficiency, waterlogging risk, and ease of maintenance. If necessary, the zoning boundaries are adjusted to avoid placing areas with large differences in drainage characteristics into the same zoning to improve the overall coordination of the drainage system within the zoning. After the evaluation and adjustment are completed, the final segmentation result, i.e., the complete drainage zoning map, is generated. The map shows the boundaries, drainage outlets, and key nodes of each drainage zoning in detail.

[0117] Through the above specific steps, the optimized clustering results are transformed into actual drainage zones, ensuring that each zone has reasonable boundaries and clear drainage nodes. This not only helps to improve the efficiency and reliability of the urban drainage system, but also provides a scientific zoning management basis for cities to cope with extreme rainfall events.

[0118] Example 3

[0119] The difference between this embodiment and embodiment 1 or 2 is that step S400 includes steps S410 to S430:

[0120] Step S410: performing rainfall pattern recognition processing according to the segmentation result, and obtaining rainfall pattern feature data by identifying and extracting the typical rainfall pattern of each partition;

[0121] Step S420: Perform time series modeling processing according to rainfall pattern characteristics, establish a prediction model through a long short-term memory network, and perform model training and optimization processing in combination with historical rainfall data to obtain a rainfall prediction model;

[0122] Step S430: perform prediction processing on the real-time data based on the rainfall prediction model to obtain a predicted change result.

[0123] Specifically, rainfall pattern recognition is first performed based on the results of urban segmentation. For each partition, by collecting and analyzing historical rainfall data, combined with the terrain characteristics and drainage system characteristics within the partition, the typical rainfall pattern of each partition is identified and extracted to generate rainfall pattern feature data. By deeply analyzing these features, pattern data reflecting the unique rainfall characteristics of each partition can be extracted. Then, time series modeling is performed based on the rainfall pattern characteristics. Long short-term memory networks are used to train and optimize the model in combination with historical rainfall data. Long short-term memory networks are good at processing time series data and can capture data dependencies over a long span. By introducing features related to urban drainage, such as drainage system pressure, historical waterlogging data, soil permeability, etc., the model can more accurately predict future rainfall trends. For example, the model can predict the intensity and duration of future rainfall events by learning the changing patterns of rainfall in historical data and combining the current soil saturation and drainage system pressure. During the optimization training process, special attention is paid to the adaptability of the model in different partitions so that it can reflect the unique rainfall and drainage characteristics of each partition. Finally, based on the optimized rainfall prediction model, real-time data is predicted. By inputting real-time rainfall data and drainage system flow monitoring data into the long short-term memory network model, the model can predict rainfall and flow changes in a preset time period in the future in real time. By combining real-time monitoring data, the model can not only predict the overall trend of future rainfall events, but also provide accurate flow change forecasts to help identify potential waterlogging risk areas.

[0124] Example 4

[0125] The difference between this embodiment and embodiment 1, 2 or 3 is that step S500 includes steps S510 to S540:

[0126] Step S510: construct a directed graph structure according to the predicted change result, and convert the key nodes and connection relationships in the urban drainage system into a directed graph structure through a topological sorting algorithm to obtain a preliminary directed graph model;

[0127] It should be noted that, first of all, according to the predicted change results, the key nodes in the urban drainage system are determined, including the main drainage outlets, flow regulation nodes and nodes in areas prone to waterlogging. The determination of key nodes depends on the flow change and waterlogging risk assessment in the predicted change results, and the points with the greatest impact on the drainage system under different rainfall scenarios are identified. Then, the topological sorting algorithm is used to process these key nodes and their connection relationships to construct a directed graph structure. The topological sorting algorithm can effectively process a directional relationship network to ensure that the logical order and flow path between nodes are maintained during the construction process. Specifically, according to the direction of water flow from upstream to downstream, the nodes are arranged in a topological order to ensure the correctness and coherence of the water flow path. Topological sorting forms a loop-free directed graph structure by sorting the nodes of the entire drainage system according to the dependency relationship. Furthermore, in the process of constructing the directed graph, special attention is paid to the connection relationship between the nodes, which reflects the water flow path and mutual influence between different nodes in the drainage system. By combining the predicted change results, the connection strength and direction between the nodes are dynamically adjusted to ensure that the directed graph can truly reflect the operation of the actual drainage system under different rainfall conditions. The preliminary directed graph model not only shows the overall structure of the urban drainage system, but also reveals the dependencies and flow paths between key nodes, providing a clear framework for further detailed analysis and system optimization.

[0128] Step S520: performing node weight and edge weight assignment processing according to the preliminary directed graph model and the rainwater pipe network layout diagram, and setting the node weight to the node flow demand and the edge weight to the flow capacity and pressure bearing capacity of the pipe by a weight calculation method based on flow and pressure, thereby obtaining a weighted directed graph model;

[0129] It is understandable that in the node weight assignment process, the weight of each node is mainly based on its flow demand. The flow demand reflects the amount of water that the node needs to handle under different rainfall scenarios. By analyzing the predicted change results, the maximum flow demand of each node can be determined. For the edge weight assignment process, it is mainly based on the flow capacity and pressure bearing capacity of the pipeline. The flow capacity of the pipeline refers to the maximum flow that the pipeline can carry when it is running at full load, while the pressure bearing capacity reflects the stability and safety of the pipeline under different water pressure conditions. By combining these physical properties, a reasonable weight can be set for each edge (i.e. pipeline). The weighted directed graph model not only shows the structure of the urban drainage system, but also reflects the workload and capacity of each node and pipeline under different rainfall scenarios.

[0130] Step S530: Perform scenario analysis processing according to the weighted directed graph model, by simulating rainfall events of different rainfall intensities and durations, combining rainfall and flow changes in the predicted change results, and using a dynamic shortest path algorithm to calculate the pressure and flow of each node under different scenarios to obtain scenario simulation results;

[0131] Specifically, rainfall scenario parameters of different intensities and durations are first defined to ensure the comprehensiveness of the scenario analysis. Different rainfall events are then simulated according to the rainfall scenario parameters, and the rainfall data for each scenario is generated by combining the rainfall and flow data in the predicted change results. Among them, the dynamic shortest path algorithm is used because the algorithm can effectively handle path optimization problems in complex networks, and is particularly suitable for urban drainage systems under real-time changing rainfall and flow conditions. Through this algorithm, the pressure and flow distribution of each node under different scenarios can be calculated, the optimal path from the source node to the target node can be found, and possible bottlenecks and high-risk areas can be identified. In this embodiment, combined with a weighted directed graph model, the algorithm can dynamically adjust the path selection, taking into account the flow capacity and pressure bearing capacity of the pipeline, and providing more accurate simulation results. Calculation formula:

[0132] d(s,t)=min∑ (u,v)∈E d(u,v);

[0133]

[0134] Where s represents the source node; t represents the target node; u and v represent any two nodes; E represents the edge set; d(s,t) represents the shortest path distance from the source node s to the target node t; w(u,v) represents the weight of the edge (u,v), which represents the pressure bearing capacity; d(u,v) represents the flow demand from node u to node v; d(s,v) represents the shortest path distance from the source node s to the target node v; P v represents the pressure at node v; F v Represents the traffic of node v.

[0135] Step S540: perform overflow risk analysis based on the scenario simulation results, integrate node pressure, pipeline flow and regional water accumulation through a multi-objective optimization algorithm, identify high-risk areas and potential overflow points, and obtain the final simulation results.

[0136] It should be noted that the multi-objective optimization algorithm can consider multiple objectives at the same time. In this embodiment, these objectives include node pressure, pipeline flow, and regional water accumulation. Node pressure represents the load level of the system under different scenarios, pipeline flow reflects the ability of water to flow through each pipeline, and regional water accumulation reveals the possible risk of flooding. By considering these objectives comprehensively, the optimization algorithm can balance the relationship between different factors and find the optimal system operation state.

[0137] Example 5

[0138] The difference between this embodiment and embodiment 1, 2, 3 or 4 is that step S600 includes steps S610 to S630:

[0139] Step S610, evaluating the effectiveness of the initial control scheme according to the simulation results to obtain an evaluation result;

[0140] It can be understood that this step identifies the effectiveness and shortcomings of each measure in the plan by examining its performance under different rainfall scenarios. Focusing on the response effects of key nodes and high-risk areas, the performance of the drainage system in actual operation is evaluated. The evaluation results include the overall effectiveness analysis of the initial control plan, the specific performance under different rainfall scenarios, the main problems identified and improvement suggestions.

[0141] Step S620: Identify optimization requirements based on the evaluation results;

[0142] It should be noted that optimization needs include increasing drainage capacity, adjusting pipeline layout, optimizing node pressure distribution, etc. Specific optimization directions and goals are determined based on the weak links and potential problems exposed in different scenarios.

[0143] Step S630: formulate and implement an optimization plan based on optimization requirements, and make dynamic adjustments through real-time monitoring and feedback mechanisms.

[0144] It is understandable that dynamic adjustments include opening or closing storage reservoir gates, adjusting drainage network flows, optimizing overflow settings, etc., to adapt to real-time changes in rainfall and flow conditions.

[0145] Embodiment 6:

[0146] This embodiment provides a water body runoff control system for an urban area, including:

[0147] An acquisition module is used to acquire historical rainfall data, real-time data, modeling parameters and urban drainage system parameters. The real-time data includes real-time rainfall data and flow monitoring data. The modeling parameters include urban topographic maps, building models and soil property distribution maps. The urban drainage system includes a rainwater pipe network layout map and an initial control plan.

[0148] The analysis module performs rainfall frequency analysis on historical rainfall data based on a preset Bayesian inference mathematical model to obtain characteristic information, which includes urban rainfall characteristics and trends;

[0149] A partitioning module is used to perform regional partitioning according to the characteristic information and the rainwater pipe network layout diagram, and to divide the city into at least one drainage partition based on the drainage characteristics to obtain a segmentation result;

[0150] The prediction module is used to construct a rainfall prediction model based on the segmentation results, and use the real-time data as the input value of the rainfall prediction model to calculate the predicted change results, which include the rainfall and flow changes in the future preset time period;

[0151] The simulation module is used to perform simulation processing based on the prediction results and the rainwater pipe network layout diagram. The simulation results are obtained by modeling the urban drainage system as a directed graph structure and analyzing the node pressure, pipe flow and overflow risk under different scenarios.

[0152] The optimization module optimizes the initial control scheme based on the simulation results to obtain the dynamic control scheme of urban water runoff.

[0153] In the embodiments disclosed in the present application, the analysis module includes:

[0154] The first processing unit is used to perform data sorting and processing based on historical rainfall data, and obtain a rainfall data set by filling missing values ​​and removing abnormal values;

[0155] The first calculation unit is used to perform parameter estimation processing according to the rainfall data set, calculate the rainfall amount, rainfall duration and rainfall interval of each rainfall event by maximum likelihood estimation method, and obtain a preliminary rainfall event characteristic parameter set;

[0156] The second processing unit is used to perform Bayesian inference processing based on the preliminary rainfall event characteristic parameter set, by setting the gamma distribution as the prior distribution and using the historical rainfall data to update the posterior distribution, so as to obtain the posterior distribution result of each rainfall event parameter;

[0157] The third processing unit is used to perform rainfall frequency analysis processing according to the posterior distribution result, generate rainfall events through simulation, and obtain characteristic information by statistically analyzing the frequency distribution of rainfall amount, rainfall duration and rainfall interval time.

[0158] In the implementation mode disclosed in the present application, the division module includes:

[0159] The first simulation unit is used to perform runoff simulation processing according to the characteristic information and modeling parameters, and analyze the convergence, diversion and water accumulation areas of the runoff to obtain the urban drainage characteristic vector by simulating rainfall events of different rainfall intensities and durations;

[0160] The first clustering unit is used to perform clustering processing according to the urban drainage model and the rainwater pipe network layout diagram to obtain a clustering result;

[0161] The first division unit is used to divide the city into at least one drainage partition according to the clustering result to obtain a segmentation result.

[0162] The above is only a specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art can easily think of changes or substitutions within the technical scope disclosed by the present invention, which should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention should be based on the protection scope of the claims.

Claims

1. A method for controlling water runoff in an urban area, characterized in that: include: Acquire historical rainfall data, real-time data, modeling parameters and urban drainage system parameters, wherein the real-time data includes real-time rainfall data and flow monitoring data, the modeling parameters include urban topographic maps, building models and soil property distribution maps, and the urban drainage system includes a rainwater pipe network layout map and an initial control plan; Based on a preset Bayesian inference mathematical model, the historical rainfall data is subjected to rainfall frequency analysis to obtain characteristic information, wherein the characteristic information includes urban rainfall characteristics and trends; Performing regional division processing according to the characteristic information and the rainwater pipe network layout diagram, dividing the city into at least one drainage partition based on drainage characteristics to obtain a segmentation result; A rainfall prediction model is constructed according to the segmentation results, and the real-time data is used as an input value of the rainfall prediction model to calculate the predicted change result, wherein the predicted result includes the rainfall and flow change in a future preset time period; Performing simulation processing according to the prediction results and the rainwater pipe network layout diagram, modeling the urban drainage system as a directed graph structure, and analyzing the node pressure, pipe flow and overflow risk under different scenarios to obtain simulation results; The initial control scheme is optimized based on the simulation results to obtain a dynamic control scheme for urban water body runoff.

2. The urban water body runoff control method according to claim 1, characterized in that: Based on the preset Bayesian inference mathematical model, the historical rainfall data is subjected to rainfall frequency analysis to obtain characteristic information, including: Performing data sorting and processing according to the historical rainfall data, and obtaining a rainfall data set by filling missing values ​​and removing abnormal values; Parameter estimation is performed according to the rainfall data set, and the rainfall amount, rainfall duration and rainfall interval of each rainfall event are calculated by maximum likelihood estimation method to obtain a preliminary rainfall event characteristic parameter set; Performing Bayesian inference processing according to the preliminary rainfall event characteristic parameter set, setting the gamma distribution as the prior distribution, and using the historical rainfall data to update the posterior distribution, to obtain the posterior distribution result of each rainfall event parameter; The rainfall frequency analysis is performed according to the posterior distribution result, rainfall events are generated by simulation, and the frequency distribution of rainfall amount, rainfall duration and rainfall interval is statistically analyzed to obtain characteristic information.

3. The urban water body runoff control method according to claim 1, characterized in that: Performing regional division processing according to the characteristic information and the rainwater pipe network layout diagram, dividing the city into at least one drainage zone based on drainage characteristics to obtain a segmentation result, including: Performing runoff simulation processing according to the characteristic information and the modeling parameters, analyzing the convergence, diversion and water accumulation areas of the runoff to obtain the urban drainage characteristic vector by simulating rainfall events of different rainfall intensities and durations; Performing clustering processing according to the urban drainage model and the rainwater pipe network layout diagram to obtain a clustering result; The city is divided into at least one drainage partition according to the clustering result to obtain a segmentation result.

4. The urban water body runoff control method according to claim 3 is characterized in that: Runoff simulation processing is performed according to the characteristic information and the modeling parameters, and by simulating rainfall events of different rainfall intensities and durations, the convergence, diversion and waterlogging areas of the runoff are analyzed to obtain the urban drainage characteristic vector, including: Performing terrain modeling processing according to the urban terrain map and the building model, and generating static terrain modeling data including terrain features and building distribution through a multi-scale terrain modeling algorithm; Performing dynamic change process simulation processing according to the characteristic information and the soil property distribution map, and obtaining dynamic infiltration data of different regions by performing time series analysis on the saturation change of soil in rainfall events with different rainfall intensities and durations; Constructing a dynamic heterogeneous graph model according to the static terrain modeling data and the dynamic infiltration data; According to the dynamic heterogeneous graph model, a graph embedding algorithm is used to convolve and aggregate node feature vectors to generate node embedding vectors, and the node embedding vectors are mapped to a high-dimensional feature space to obtain an urban drainage feature vector.

5. The urban water runoff control method according to claim 3, characterized in that: Clustering results are obtained by performing clustering processing according to the urban drainage model and the rainwater pipe network layout diagram, including: Performing similarity calculation processing according to the city drainage feature vector, calculating the similarity between each city drainage feature vector by a cosine similarity method, and obtaining a similarity matrix; Performing hierarchical clustering processing based on the similarity matrix and the rainwater pipe network layout diagram, and obtaining preliminary clustering results by combining geographical adjacency and pipe network connectivity in the pipe network; The preliminary clustering results are refined and optimized using a density-based spatial constraint clustering algorithm, and the optimized clustering results are obtained by processing boundary points and noise points in combination with the flow direction in the pipe network and the density of regional water accumulation points; The feature fusion processing is performed according to the optimized clustering result, and the drainage feature vectors in each cluster are fused by a weighted average method to obtain the final clustering result.

6. The urban water body runoff control method according to claim 1, characterized in that: A rainfall prediction model is constructed according to the segmentation result, and the real-time data is used as an input value of the rainfall prediction model to calculate the predicted change result, including: Performing rainfall pattern recognition processing according to the segmentation results, and obtaining rainfall pattern feature data by identifying and extracting a typical rainfall pattern of each partition; Performing time series modeling processing according to the rainfall pattern characteristics, establishing a prediction model through a long short-term memory network, and performing model training and optimization processing in combination with the historical rainfall data to obtain a rainfall prediction model; The real-time data is predicted and processed based on the rainfall prediction model to obtain a predicted change result.

7. The urban water body runoff control method according to claim 1, characterized in that: According to the prediction results and the rainwater pipe network layout diagram, simulation processing is performed, and the simulation results are obtained by modeling the urban drainage system as a directed graph structure and analyzing the node pressure, pipe flow and overflow risk under different scenarios, including: According to the predicted change results, a directed graph structure is constructed, and key nodes and connection relationships in the urban drainage system are converted into a directed graph structure through a topological sorting algorithm to obtain a preliminary directed graph model; According to the preliminary directed graph model and the rainwater pipe network layout diagram, node weights and edge weights are assigned, and the weights of the nodes are set to the flow demand of the nodes and the weights of the edges are set to the flow capacity and pressure bearing capacity of the pipes through a weight calculation method based on flow and pressure, so as to obtain a weighted directed graph model; Scenario analysis processing is performed according to the weighted directed graph model, by simulating rainfall events of different rainfall intensities and durations, combining rainfall and flow changes in the predicted change results, and using a dynamic shortest path algorithm to calculate the pressure and flow of each node under different scenarios to obtain scenario simulation results; The overflow risk analysis is carried out based on the scenario simulation results. The node pressure, pipeline flow and regional water accumulation are comprehensively considered through a multi-objective optimization algorithm, and high-risk areas and potential overflow points are identified to obtain the final simulation results.

8. A water runoff control system for an urban area, characterized in that: include: An acquisition module, used to acquire historical rainfall data, real-time data, modeling parameters and urban drainage system parameters, wherein the real-time data includes real-time rainfall data and flow monitoring data, the modeling parameters include urban topographic maps, building models and soil property distribution maps, and the urban drainage system includes a rainwater pipe network layout map and an initial control plan; An analysis module, which performs rainfall frequency analysis on the historical rainfall data based on a preset Bayesian inference mathematical model to obtain characteristic information, wherein the characteristic information includes urban rainfall characteristics and trends; A division module, used to perform regional division processing according to the characteristic information and the rainwater pipe network layout diagram, and divide the city into at least one drainage zone based on drainage characteristics to obtain a segmentation result; A prediction module, used to construct a rainfall prediction model according to the segmentation results, and use the real-time data as an input value of the rainfall prediction model to calculate the predicted change results, wherein the predicted results include rainfall and flow changes within a preset time period in the future; A simulation module, used for performing simulation processing according to the prediction results and the rainwater pipe network layout diagram, by modeling the urban drainage system as a directed graph structure and analyzing the node pressure, pipe flow and overflow risk under different scenarios to obtain simulation results; The optimization module optimizes the initial control scheme based on the simulation results to obtain a dynamic control scheme for urban water body runoff.

9. The urban water runoff control system according to claim 8, characterized in that: The analysis module comprises: A first processing unit is used to perform data sorting and processing according to the historical rainfall data, and obtain a rainfall data set by filling missing values ​​and removing abnormal values; A first calculation unit is used to perform parameter estimation processing according to the rainfall data set, calculate the rainfall amount, rainfall duration and rainfall interval of each rainfall event by maximum likelihood estimation method, and obtain a preliminary rainfall event characteristic parameter set; A second processing unit is used to perform Bayesian inference processing according to the preliminary rainfall event characteristic parameter set, by setting the gamma distribution as the prior distribution and using the historical rainfall data to update the posterior distribution, so as to obtain the posterior distribution result of each rainfall event parameter; The third processing unit is used to perform rainfall frequency analysis processing according to the posterior distribution result, generate rainfall events through simulation, and obtain characteristic information by statistically analyzing the frequency distribution of rainfall amount, rainfall duration and rainfall interval time.

10. The urban water body runoff control system according to claim 8, characterized in that: The division module comprises: A first simulation unit is used to perform runoff simulation processing according to the characteristic information and the modeling parameters, and to obtain an urban drainage characteristic vector by analyzing the convergence, diversion and water accumulation areas of the runoff by simulating rainfall events of different rainfall intensities and durations; A first clustering unit, used for performing clustering processing according to the urban drainage model and the rainwater pipe network layout diagram to obtain a clustering result; The first division unit is used to divide the city into at least one drainage partition according to the clustering result to obtain a segmentation result.

Images