Flood prediction method and device, electronic equipment and computer readable storage medium

Abstract

The invention provides a flood prediction method and device, electronic equipment and a computer readable storage medium, and relates to the technical field of urban drainage pipe networks. The method comprises the following steps: dividing a target point location in a rainwater pipe network into a plurality of clustering blocks; performing classification according to a plurality of preset waterlogging risk levels based on the ponding depths in the clustering blocks; training a basic flood prediction model based on the rainwater pipe network data in the clustering blocks until the prediction precision is reached, and obtaining a flood prediction model; according to the method, nodes in a rainwater pipe network are clustered, the rainwater pipe network is divided into multiple types of blocks according to different types, historical data in different blocks are utilized to train a prediction model comprising a double-layer long-short-term memory neural network, and the trained model has intelligent and automatic data capability and can meet the requirement of quick response; especially under extreme conditions, when large-scale data is processed, the calculation time is shortened, and the response speed is increased.

Description

Technical Field

[0001] The present invention relates to the technical field of urban drainage pipe networks, and in particular to a flood prediction method, device, electronic equipment and computer-readable storage medium. Background Art

[0002] Urban flooding has become a significant challenge to modern urban development. Its main causes include increased rainfall intensity, increased surface runoff coefficients due to urbanization, aging drainage systems, and inadequate management measures. The application of early warning systems also provides important support for urban waterlogging prevention. These systems, through weather forecasts and real-time monitoring data, issue warnings and guide relevant departments and residents in implementing emergency measures.

[0003] However, existing technologies have exposed numerous limitations in practical applications. First, existing technologies are inefficient when processing large-scale stormwater network data, making it difficult to meet the demand for rapid response. Second, traditional statistical methods and models lack sufficient prediction accuracy under extreme conditions, making it impossible to accurately assess the risk of urban flooding. Third, existing technologies lack intelligent and automated data analysis capabilities, making it difficult to dynamically analyze and predict complex drainage systems. Finally, existing technologies require long computation times when processing large-scale data, which affects response speed. Summary of the Invention

[0004] The purpose of the present invention is to provide a flood prediction method, device, electronic device and computer-readable storage medium. First, the nodes in the rainwater pipe network are clustered and divided into multiple types of blocks according to different categories. The historical data in different blocks are used to train a prediction model including a double-layer long short-term memory neural network. The trained model has intelligent and automated data capabilities and can meet the needs of rapid response. Especially in extreme cases, when processing large-scale data, it reduces calculation time and improves response speed.

[0005] In a first aspect, the present invention provides a flood prediction method, comprising: Obtain data on the rainwater pipe network, including basic pipe network data, catchment area data, and historical precipitation data. Cluster analysis is performed based on basic pipe network data to divide target points within the stormwater pipe network into multiple cluster blocks. Target points include inspection wells and stormwater pipes. Cluster blocks correspond one-to-one to cluster categories, and cluster categories represent the flood risk level of cluster blocks. Based on the waterlogging depth within the cluster blocks, the cluster blocks are classified according to multiple preset waterlogging risk levels to determine the prediction accuracy corresponding to the cluster blocks; different waterlogging risk levels correspond to different waterlogging depths; A basic flood prediction model is trained based on the rainwater pipe network data within the clustered blocks until prediction accuracy is achieved, thereby obtaining a flood prediction model; wherein the basic flood prediction model includes: a first long short-term memory neural network and a second long short-term memory neural network connected in sequence; the first long short-term memory neural network has input data of the rainwater pipe network, and output data of the liquid level height in the inspection well and the fullness of the rainwater pipe; the second long short-term memory neural network has input data of the rainwater pipe network, the liquid level height in the inspection well and the fullness of the rainwater pipe, and output data of the water accumulation depth; The rainfall event data is input into the flood prediction model, and the predicted water depth is output.

[0006] In some preferred embodiments of the present invention, the step of performing cluster analysis based on the basic data of the pipe network to divide the target points in the rainwater pipe network into multiple cluster blocks includes: Determine the preset proportion of target points in the stormwater pipe network based on the preset screening rules; Determine the elevation difference of the inspection well and the fullness of the stormwater pipe at the target point; The elevation difference of the inspection well and the fullness of the rainwater pipe are normalized and used as the feature vector; The Gaussian clustering model is trained based on the feature vector until the preset conditions are met to obtain the clustering results.

[0007] In some preferred embodiments of the present invention, the step of training a Gaussian clustering model based on the feature vector until a preset condition is met to obtain a clustering result includes: Initialization parameters; the parameters include: mixing weights, mean and covariance matrix; Calculate the probability of Gaussian distribution based on eigenvectors; The posterior probability is determined by the expectation maximization algorithm based on the probability of the Gaussian distribution; Based on the posterior probability, the parameters are updated and the probability of the Gaussian distribution of the eigenvector is continuously recalculated until the change in the log-likelihood function of the Gaussian clustering model is less than a preset threshold or the preset number of iterations is reached; Perform cluster analysis on the feature vectors to obtain clustering results.

[0008] In some preferred embodiments of the present invention, the log-likelihood function is constrained by the following formula: ; in, is the log-likelihood function, is the kth Gaussian distribution in x i The probability density function at , K is the number of Gaussian distributions, is the eigenvector of the i-th data point, are the parameters of the Gaussian clustering model.

[0009] In some preferred embodiments of the present invention, the method further comprises: determining a flood prediction model by the following steps: After standardizing the data of the rainwater pipe network, the first training data is constructed based on the time series; Training a first long short-term memory neural network based on the first training data, and updating parameters of the first long short-term memory neural network based on the optimizer until a preset training condition is met; Based on the trained first long short-term memory neural network, the liquid level in the inspection well and the fullness of the rainwater pipe are output; After standardizing the data of the rainwater pipe network, the liquid level height in the inspection well, and the fullness of the rainwater pipe, the second training data is constructed based on the time series; Training a second long short-term memory neural network based on the second training data, and updating parameters of the second long short-term memory neural network based on the optimizer until a preset training condition is met; Based on the trained second long short-term memory neural network, the water depth is output.

[0010] In some preferred embodiments of the present invention, the parameters of the first long short-term memory neural network and the parameters of the second long short-term memory neural network both include: 1 long short-term memory layer, 10 hidden units and a learning rate of 0.001.

[0011] In some preferred embodiments of the present invention, the method further comprises: Sort multiple risk levels from low to high; Sort multiple prediction accuracies from small to large; The sorted multiple risk levels are matched with the sorted multiple prediction accuracies.

[0012] In a second aspect, the present invention provides a flood prediction device, comprising: The data acquisition module is used to obtain the data of the rainwater pipe network; wherein the data of the rainwater pipe network includes: pipe network basic data, catchment area data and historical precipitation data; The cluster analysis module is used to perform cluster analysis based on the basic data of the pipe network, dividing the target points in the stormwater pipe network into multiple cluster blocks. The target points include: inspection wells and stormwater pipes. The cluster blocks correspond to cluster categories one by one. The cluster category represents the flood risk level of the cluster block. A prediction accuracy determination module is used to classify the waterlogging depths within the cluster blocks according to a plurality of preset waterlogging risk levels, and determine the prediction accuracy corresponding to the cluster blocks; wherein different waterlogging risk levels correspond to different waterlogging depths; A flood prediction model determination module is used to train a basic flood prediction model based on the rainwater pipe network data in the cluster block until the prediction accuracy is achieved, thereby obtaining a flood prediction model; wherein the basic flood prediction model includes: a first long short-term memory neural network and a second long short-term memory neural network connected in sequence; the input data of the first long short-term memory neural network is the data of the rainwater pipe network, and the output data is the liquid level height in the inspection well and the fullness of the rainwater pipe; the input data of the second long short-term memory neural network is the data of the rainwater pipe network, the liquid level height in the inspection well and the fullness of the rainwater pipe, and the output data is the depth of accumulated water; The waterlogging depth prediction module is used to input rainfall event data into the flood prediction model and output the predicted waterlogging depth.

[0013] In a third aspect, the present invention provides an electronic device comprising a processor and a memory, wherein the memory stores computer-executable instructions that can be executed by the processor, and the processor executes the computer-executable instructions to implement the flood prediction method provided in the first aspect.

[0014] In a fourth aspect, the present invention provides a computer-readable storage medium storing computer-executable instructions. When the computer-executable instructions are called and executed by a processor, the computer-executable instructions prompt the processor to implement the flood prediction method provided in the first aspect.

[0015] The present invention brings the following beneficial effects: The present invention provides a flood prediction method, device, electronic device and computer-readable storage medium, the method comprising: obtaining data of a rainwater pipe network; wherein the data of the rainwater pipe network comprises: pipe network basic data, catchment area data and historical precipitation data; performing cluster analysis based on the pipe network basic data, dividing target points in the rainwater pipe network into a plurality of cluster blocks; wherein the target points comprise: inspection wells and rainwater pipes; cluster blocks correspond to cluster categories one by one; cluster categories characterize flood risk levels of cluster blocks; classifying according to a plurality of preset waterlogging risk levels based on the depth of accumulated water in the cluster blocks, and determining the prediction accuracy corresponding to the cluster blocks; wherein different waterlogging risk levels correspond to different waterlogging depths; training a basic flood prediction model based on the rainwater pipe network data in the cluster blocks until the prediction accuracy is reached, thereby obtaining a flood prediction model; wherein the basic flood prediction model comprises: sequentially connecting The first long short-term memory neural network and the second long short-term memory neural network are connected; the input data of the first long short-term memory neural network is the data of the rainwater pipe network, and the output data is the liquid level height in the inspection well and the fullness of the rainwater pipe; the input data of the second long short-term memory neural network is the data of the rainwater pipe network, the liquid level height in the inspection well and the fullness of the rainwater pipe, and the output data is the depth of accumulated water; the rainfall event data is input into the flood prediction model, and the predicted depth of accumulated water is output; first, the nodes in the rainwater pipe network are clustered, and the rainwater pipe network is divided into multiple blocks according to different categories, and the historical data in different blocks are used to train the prediction model including the double-layer long short-term memory neural network. The trained model has intelligent and automated data capabilities, which can meet the needs of rapid response, especially in extreme cases, when processing large-scale data, reducing calculation time and improving response speed. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] In order to more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the specific embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0017] Figure 1 A flow chart of a flood prediction method provided by an embodiment of the present invention; Figure 2 A schematic structural diagram of a flood prediction device provided by an embodiment of the present invention; Figure 3 A schematic structural diagram of an electronic device provided by an embodiment of the present invention.

[0018] Icons: 310 - data acquisition module; 320 - cluster analysis module; 330 - prediction accuracy determination module; 340 - flood prediction model determination module; 350 - water depth prediction module; 400 - memory; 401 - processor; 402 - bus; 403 - communication interface. DETAILED DESCRIPTION

[0019] To make the objectives, technical solutions, and advantages of the embodiments of the present invention more clear, the technical solutions of the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Generally, the components of the embodiments of the present invention described and shown in the drawings herein can be arranged and designed in various different configurations.

[0020] Therefore, the following detailed description of the embodiments of the present invention provided in the accompanying drawings is not intended to limit the scope of the invention as claimed, but rather merely represents selected embodiments of the present invention. All other embodiments derived by persons of ordinary skill in the art based on the embodiments of the present invention without creative effort shall fall within the scope of protection of the present invention.

[0021] It should be noted that similar reference numerals and letters denote similar items in the following drawings, and therefore, once an item is defined in one drawing, it does not need to be further defined or explained in subsequent drawings.

[0022] In the description of the present invention, it should be noted that the terms "center," "upper," "lower," "left," "right," "vertical," "horizontal," "inner," and "outer" and the like indicate positions or locations based on the positions shown in the accompanying drawings, or the positions or locations in which the inventive product is typically placed when in use. These terms are intended solely to facilitate the description of the present invention and to simplify the description, and are not intended to indicate or imply that the devices or components referred to must have a specific orientation, be constructed, or operate in a specific orientation. Therefore, they should not be construed as limitations on the present invention. Furthermore, the terms "first," "second," and "third," etc., are used solely to distinguish descriptions and should not be construed as indicating or implying relative importance.

[0023] Furthermore, terms such as "horizontal," "vertical," and "overhanging" do not necessarily imply that a component must be absolutely horizontal or overhanging, but rather that it can be slightly tilted. For example, "horizontal" simply means that its direction is more horizontal than "vertical," and does not mean that the structure must be completely horizontal, but rather that it can be slightly tilted.

[0024] In the description of the present invention, it should also be noted that, unless otherwise expressly specified or limited, the terms "disposed," "installed," "connected," and "connected" should be understood in a broad sense. For example, they may refer to fixed connections, detachable connections, or integral connections; they may refer to mechanical connections or electrical connections; they may refer to direct connections or indirect connections through an intermediate medium; and they may refer to internal communication between two components. Those skilled in the art will understand the specific meanings of the above terms in the present invention based on the specific circumstances.

[0025] Urban flooding has become a significant challenge hindering the development of modern cities. Its main drivers include increasing rainfall intensity, increased surface runoff coefficients due to urbanization, aging drainage systems, and inadequate management measures. To address this challenge, current flood prevention and mitigation strategies rely primarily on the development of hardware infrastructure, such as stormwater pumping stations and drainage pipelines, combined with software technologies, including geographic information systems (GIS), hydraulic models (such as SWMM), and sensor monitoring, to model drainage systems and enable real-time monitoring of drainage conditions. Furthermore, the application of early warning systems provides crucial support for preventing urban waterlogging. These systems, through weather forecasts and real-time monitoring data, issue warnings and guide relevant departments and residents in taking emergency measures.

[0026] However, existing technologies have exposed numerous limitations in practical applications. First, GIS and hydraulic models suffer from low computational efficiency when processing large-scale stormwater network data, making it difficult to meet the demand for rapid response. Second, traditional statistical methods and SWMM simulations lack sufficient prediction accuracy under extreme rainfall conditions, making it impossible to accurately assess the risk of urban flooding. Furthermore, existing technologies lack intelligent and automated data analysis capabilities, making it difficult to dynamically analyze and predict complex drainage systems. Finally, existing monitoring and early warning systems require long computation times when processing large-scale data and complex models, which affects response speed.

[0027] Currently, a large number of physical models and machine learning algorithms are used for early warning and prediction of urban waterlogging. Common models include SWMM (Storm Water Management Model), InfoWorks ICM (Integrated Catchment Management), and Mike Urban (software for modeling and simulating urban drainage systems). These models solve hydrodynamic equations to simulate water flow states, providing highly accurate numerical simulation results. These models, such as SWMM and InfoWorks ICM (Integrated Catchment Management), have been widely used. Similar models exist in China, such as the Urban Stormwater Forecasting Model proposed by Zhang Min et al. Machine learning methods include random forests, support vector machines, and XGBoost (an optimized distributed gradient boosting library), as well as deep learning methods such as recurrent neural networks and artificial neural networks. These methods achieve rapid predictions by learning characteristic factors (such as rainfall intensity, slope, and drainage network density).

[0028] However, in large-scale scenarios and complex underlying surfaces, hydrodynamic models are complex and time-consuming to run and solve, making them difficult to meet the demands of rapid forecasting. These models require extensive input data and complex parameter calibration, placing high demands on data quality and computing power. Machine learning models require extensive historical data for training, and insufficient or low-quality data can compromise forecast accuracy. The models also have limited generalization capabilities, making them difficult to adapt to extreme rainfall or unseen scenarios.

[0029] The purpose of the present invention is to provide a flood prediction method, device, electronic device and computer-readable storage medium. First, the nodes in the rainwater pipe network are clustered and divided into multiple types of blocks according to different categories. The historical data in different blocks are used to train a prediction model including a double-layer long short-term memory neural network. The trained model has intelligent and automated data capabilities and can meet the needs of rapid response. Especially in extreme cases, when processing large-scale data, it reduces calculation time and improves response speed.

[0030] The following embodiments of the present invention are described in detail with reference to the accompanying drawings. In the absence of conflict, the following embodiments and features in the embodiments may be combined with each other.

[0031] Example 1 The embodiment of the present invention provides a flood prediction method, see Figure 1 The flowchart of a flood prediction method provided by an embodiment of the present invention is shown, and the method includes: Step S102, obtaining data of the rainwater pipe network; wherein the data of the rainwater pipe network includes: pipe network basic data, catchment area data and historical precipitation data.

[0032] Specifically, the stormwater network system primarily consists of stormwater wells, stormwater pipes, stormwater pumping stations, and stormwater outlets. Basic data for the stormwater network system is collected and organized, including: wellhead elevation, wellhead elevation, pipe bottom elevation, pipe level elevation H, pipe slope i, and stormwater outlet elevation. Regional data on catchment areas, catchment areas, and runoff coefficients is collected. Historical rainfall intensity within the region is collected and organized, and rainfall types are categorized.

[0033] Historical data of monitoring points are set up in the rainwater pipe network system. Under different rainfall conditions, the data such as the liquid level and flow in the rainwater pipe and the depth of ground water near the rainwater well are sorted out.

[0034] Step S104: Perform cluster analysis based on the basic data of the pipe network to divide the target points in the rainwater pipe network into multiple cluster blocks; the target points include: inspection wells and rainwater pipes; the cluster blocks correspond to cluster categories one by one; the cluster categories represent the flood risk level of the cluster blocks.

[0035] Specifically, since the rainwater pipe network is a very large system, if all rainwater inspection wells and rainwater pipes are regarded as nodes, there will be tens of thousands or even hundreds of thousands of such nodes in a large city, which is not conducive to data processing.

[0036] Furthermore, in some preferred embodiments of the present invention, cluster analysis is performed based on the basic data of the pipe network, and the target points in the rainwater pipe network are divided into multiple cluster blocks, including: determining a preset proportion of target points in the rainwater pipe network based on preset screening rules; determining the elevation difference of the inspection wells and the fullness of the rainwater pipes in the target points; normalizing the elevation difference of the inspection wells and the fullness of the rainwater pipes and using them as feature vectors; training a Gaussian clustering model based on the feature vectors until the preset conditions are met to obtain clustering results.

[0037] Specifically, several key points are selected as target points in the huge rainwater pipe network system for real-time monitoring. For example, the monitoring points are generally about 1% to 5% of the total number of rainwater nodes.

[0038] Target points are generally selected at the nodes of main pipes and branch pipes, and are arranged at key nodes such as the intersection of main pipes in the pipeline network, branch pipe access points, and upstream and downstream of pump stations. These locations can reflect the operating status of the entire pipeline network; low-lying areas can also be selected, which are generally high-risk areas for urban flooding, such as downstream of the catchment area and depressions in the terrain; hydraulic sensitive points can also be selected, which are generally locations in the pipeline network where hydraulic anomalies are prone to occur, such as sudden changes in water flow velocity, changes in pipe diameter, and slope turning.

[0039] The elevation difference data of each stormwater inspection well in the study area were collected under various rainfall intensities within the statistical analysis area. The elevation difference was constrained by the following formula: ΔE = E1-E2; where E1 is the ground elevation of the stormwater inspection well; E2 is the water level elevation of the stormwater inspection well; and ΔE is the elevation difference.

[0040] Collect the liquid level of the rainwater pipe with a liquid level monitoring point installed, record it as h, collect the pipe diameter D, and calculate the fullness a=h / D.

[0041] In some preferred embodiments of the present invention, parameters such as the slope, catchment area, precipitation curve, and runoff coefficient of the rainwater pipe are obtained, and a simulation model is run to obtain the predicted ΔE and a in the drainage network. Error analysis is then performed between the predicted ΔE and a and the actual ΔE and a, and the model is adjusted until the error is less than 10%. The simulation model can then be used to fit ΔE and a.

[0042] For example, SWMM software is used to fit ΔE and a in the entire drainage network, and then the fitting results of the SWMM software are exported to an INP file. Specifically, the data of the entire rainwater pipe network system, E1, E2, h, D and other data are read, and the analytical data ΔE and a of the entire rainwater pipe network are calculated.

[0043] Furthermore, ΔE and a are normalized so that the values ​​of each feature are on the same scale. Generally, each data is obtained by subtracting the mean and dividing by the standard deviation. The normalization formula is as follows: Where z is the normalized data, x is the original data, μ is the mean, and σ is the standard deviation. ΔE and a are used as the input of the feature vector.

[0044] Furthermore, in some preferred embodiments of the present invention, the step of training a Gaussian clustering model based on eigenvectors until a preset condition is met and obtaining a clustering result includes: initializing parameters; wherein the parameters include: mixing weights, mean and covariance matrix; calculating the probability of the Gaussian distribution based on the eigenvectors; determining the posterior probability through the expectation-maximization algorithm based on the probability of the Gaussian distribution; updating the parameters based on the posterior probability, and continuing to recalculate the probability of the Gaussian distribution of the eigenvectors until the change in the log-likelihood function of the Gaussian clustering model is less than a preset threshold or reaches a preset number of iterations; performing cluster analysis on the eigenvectors to obtain clustering results.

[0045] Specifically, the first step is to initialize the parameters of the Gaussian clustering model, where the mixing weight π: initializes the weight of each Gaussian distribution, usually set to a uniform distribution; mean μ: initializes the mean of each Gaussian distribution, which can be pre-initialized by randomly selecting data points or using K-Means (an iterative clustering analysis algorithm); covariance matrix Σ: initializes the covariance matrix of each Gaussian distribution, usually set to the unit matrix or the covariance matrix of the data.

[0046] The second step is to calculate the probability that each data point belongs to each Gaussian distribution: ; Among them, x i is the eigenvector of the i-th data point, π k is the updated mixing weight of the kth Gaussian distribution, k is the total number of data points, μ k is the mean vector of the k-th Gaussian distribution, and Σk is the covariance matrix of the k-th Gaussian distribution.

[0047] The third step is to calculate the expectation maximization algorithm, which calculates the posterior probability and updates the mixing weights, mean and covariance matrix. See the following calculation formula for details: ; ; ; ; Among them, π k is the updated mixing weight of the kth Gaussian distribution, N is the total number of data points, r ik is the data point x i The posterior probability of belonging to the kth Gaussian distribution, x i is the i-th data point, μ k is the mean of the k-th Gaussian distribution after the update, Σk is the covariance matrix of the k-th Gaussian distribution after the update, is the data point x i The probability density under the k-th Gaussian distribution, μ k is the mean vector of the Gaussian distribution, Σk is the covariance matrix of the Gaussian distribution, and the summation part in the denominator is The sum of the weighted probability densities of all Gaussian distributions is used to normalize the posterior probability to ensure that the sum of all posterior probabilities is 1. is the data point x i and mean μ k The outer product of .

[0048] Step 4: Repeat steps 2 and 3 until the change in the log-likelihood function is less than a certain threshold (e.g., e -6 ) or reaches the maximum number of iterations, when |L (t+1) -L (t) ∣<ϵ (e.g. ϵ=10 -6 ) terminates the iteration.

[0049] Furthermore, in some preferred embodiments of the present invention, the log-likelihood function is constrained by the following formula: ;in, is the log-likelihood function, is the kth Gaussian distribution in x i The probability density function at , K is the number of Gaussian distributions, is the eigenvector of the i-th data point, are the parameters of the Gaussian clustering model.

[0050] Gaussian mixture clustering (GMM) was used to perform cluster analysis on the stormwater network and identify risk areas, significantly improving data processing efficiency. GMM can effectively handle large amounts of data, reduce computational complexity, and ensure the system's high efficiency and rapid response when dealing with complex urban drainage systems.

[0051] Step S106 , classifying the cluster blocks according to a plurality of preset waterlogging risk levels based on the waterlogging depths within the cluster blocks, and determining the prediction accuracy corresponding to the cluster blocks; wherein different waterlogging risk levels correspond to different waterlogging depths.

[0052] Specifically, by performing cluster analysis under different rainfall intensities, the cluster analysis results of the entire rainwater pipe network system can be obtained. For example, the categories of the rainwater pipe network system obtained by the cluster analysis results can be divided into: I, II, III, V, VI, VII, VIII, etc.

[0053] It should be emphasized that the level of urban flooding risk can be determined by experience. In the embodiment of the present invention, seven levels, namely I, II, III, V, VI, VII, and VIII, are adopted.

[0054] The data corresponding to the waterlogging risk levels correspond to different prediction accuracies when training the waterlogging model.

[0055] In some preferred embodiments of the present invention, the method further includes: sorting multiple risk levels from low to high; sorting multiple prediction accuracies from small to large; and matching the sorted multiple risk levels with the sorted multiple prediction accuracies.

[0056] Generally, in areas with lower risk levels, the elevation difference of inspection wells in these areas is larger, the filling degree is lower, and the flood risk is lower. The prediction accuracy required in the prediction model can be appropriately lower; in areas with higher risk levels, the elevation difference of these areas is lower, the filling degree is higher, and the flood risk is higher. The prediction accuracy required in the prediction model needs to be higher. The two can be sorted separately, and the sorted multiple risk levels can be matched with the sorted multiple prediction accuracies; one risk level can correspond to one prediction accuracy, or several adjacent risk levels can correspond to one prediction accuracy.

[0057] For example, take 5 inspection wells as an example, see the raw data list shown in Table 1: Table 1

[0058] Perform standardization on the above original numbers and calculate the mean and standard deviation: μ ΔE =(2.5+1.8+0.9+0.5+1.2) / 5=1.38;μ a =(0.3+0.5+0.7+0.8+0.4) / 5=0.54;σ ΔE ≈0.72;σ a ≈0.18; the data were normalized using the following formula: ; ; The standardized values ​​for each inspection well are shown in Table 2: Table 2

[0059] Take 3 Gaussian distributions as an example, that is, K=3.

[0060] First, initialize the parameters: the mixing weight π: , mean μ: Randomly select the eigenvectors of three inspection wells as the initial mean, for example, inspection wells 1, 3, and 4: , , ; Covariance matrix (Σ): identity matrix.

[0061] Compute Gaussian distribution probabilities: For each data point x i , calculate its probability density under each Gaussian distribution: .

[0062] Take inspection well 1 (eigenvector [1.55, -1.33]) as an example: For the first Gaussian distribution ( ): ; Similarly, calculate the probability under other Gaussian distributions: For the second Gaussian distribution ( ), , , .

[0063] For the third Gaussian distribution ( ): , , .

[0064] Expectation-maximization algorithm: Calculate the posterior probability: ; get: r 11 ≈0.9925; r 12 ≈0.00741; r 13 ≈0.000496; similarly, calculate the posterior probabilities of other inspection wells.

[0065] Update the blending weights: ; Update the mean: ; Update the covariance matrix: .

[0066] Perform convergence check: Repeat the calculation steps until the change in the log-likelihood function is less than a certain threshold (such as e -6 ) or the maximum number of iterations is reached.

[0067] Clustering the results: After multiple iterations, the cluster categories of each inspection well are finally obtained as shown in Table 3 below: Table 3

[0068] Category I: Low-risk areas. Manholes 1, 2, and 5 fall into this category. These manholes have large elevation differences, low fill levels, and low flood risk. Category II: Medium-risk areas. Manhole 3 falls into this category. Its elevation difference and fill levels are moderate. Category III: High-risk areas. Manhole 4 falls into this category. Its elevation difference is low, its fill levels are high, and its flood risk is high.

[0069] For example, the prediction accuracy for low-risk areas is ≥85%, for medium-risk areas ≥90%, and for high-risk areas ≥95%. Given the tens of thousands or even hundreds of thousands of nodes in the entire stormwater network, achieving consistent overall accuracy during training is difficult due to the large model computational complexity, numerous iterations, long computational times, and inability to converge. Therefore, a step-by-step approach to prediction accuracy is established based on the probability of occurrence to improve overall prediction accuracy while achieving rapid convergence and rapid response to early warnings.

[0070] Step S108, training a basic flood prediction model based on the rainwater pipe network data in the clustering block until the prediction accuracy is achieved to obtain a flood prediction model; wherein, the basic flood prediction model includes: a first long short-term memory neural network and a second long short-term memory neural network connected in sequence; the input data of the first long short-term memory neural network is the data of the rainwater pipe network, and the output data is the liquid level height in the inspection well and the fullness of the rainwater pipe; the input data of the second long short-term memory neural network is the data of the rainwater pipe network, the liquid level height in the inspection well and the fullness of the rainwater pipe, and the output data is the depth of accumulated water.

[0071] Specifically, training data is prepared, mainly including: basic values ​​such as the ground elevation of the rainwater well mouth, the ground elevation of the rainwater well, the elevation of the top of the rainwater pipe, the elevation of the bottom of the rainwater pipe, the pipe slope i, the elevation of the rainwater outlet, etc.; the watershed block attributes, watershed area, runoff coefficient, etc. of the collection area; The historical monitoring data (mainly precipitation data) are divided into 80% as a training group and 20% as a test group. The historical monitoring data of the training group and the basic data are input into the long short-term memory neural network.

[0072] Furthermore, in some preferred embodiments of the present invention, a flood prediction model is determined by the following steps: after standardizing the data of the rainwater pipe network, first training data is constructed based on the time series; a first long short-term memory neural network is trained based on the first training data, and the parameters of the first long short-term memory neural network are updated based on the optimizer until the preset training conditions are reached; based on the trained first long short-term memory neural network, the liquid level height in the inspection well and the fullness of the rainwater pipe are output; after standardizing the data of the rainwater pipe network, the liquid level height in the inspection well and the fullness of the rainwater pipe, second training data is constructed based on the time series; based on the second training data, a second long short-term memory neural network is trained, and the parameters of the second long short-term memory neural network are updated based on the optimizer until the preset training conditions are reached; based on the trained second long short-term memory neural network, the depth of accumulated water is output.

[0073] Specifically, the training process mainly includes two stages: the first stage outputs the elevation difference ΔE and the rainwater pipe fullness a, and the second stage outputs the surface water depth H.

[0074] For the first stage: Step 1, enter basic data: 1. GIS basic data of the rainwater pipe network system: These basic data represent the basic properties of the rainwater pipe network and are invariants: they include pipe length, pipe diameter, pipe length, pipe slope, pipe top elevation, pipe bottom elevation, inspection well ground elevation, inspection well bottom elevation, and the elevation of rainwater discharge to natural rivers.

[0075] 2. There is no basic data, but there are guidance or recommended data and the parameters need to be adjusted and verified: 1) Total catchment area S1: the surface area of ​​water flowing into the inspection well; 2) Surface runoff area S2: the surface runoff area along the pipeline connected to the inspection well; 3) Slope of surface runoff area i: average slope of surface runoff area; 4) Effective runoff coefficient u: the effective runoff coefficient allocated to the surface runoff area.

[0076] 3. External dynamic change parameters: 1) Rainfall intensity (q): the rainfall intensity in the past T hours, including the rainfall intensity at the current time; 2) Liquid level in rainwater inspection well (GH): The liquid level in the rainwater inspection well under the current rainfall intensity.

[0077] 3) Stormwater pipe fullness (a): The fullness of the stormwater pipe under the current rainfall conditions.

[0078] The second step is to train the first long short-term memory neural network model: 1. Data preprocessing: Normalize the underlying data so that the values ​​of each feature are on the same scale.

[0079] Construct time series data, where each sample contains data from the past few time steps, to predict the liquid level in the rainwater inspection well and the fullness of the rainwater pipe at the next time step.

[0080] 2. Model construction: Define the first long short-term memory neural network model, including LSTM layer and fully connected layer.

[0081] The LSTM layer is used to process time series data, and the fully connected layer is used to output prediction results.

[0082] 3. Training process Set hyperparameters such as the number of input features, number of hidden units, number of LSTM layers, learning rate, and training epochs.

[0083] In each training cycle, training data is fed into the model and the loss between the predicted values ​​and the target values ​​is calculated.

[0084] Use an optimizer (such as Adam) to update the model parameters to minimize the loss.

[0085] 4. Prediction results: The historical rainfall intensity data of 20% of the test group was input to predict the liquid level in the inspection well of the stormwater pipe system and the fullness of the stormwater pipe.

[0086] For the second stage: predict the water depth of the waterlogged points at the nodes on the street (JH): The first step is to prepare the input data: 1. The first stage predicts the liquid level of nodes in the stormwater pipe system and the fullness of the stormwater pipe.

[0087] 2. GIS basic data of the rainwater pipe network system: These basic data represent the basic properties of the rainwater pipe network and are invariants: they include pipe length, pipe diameter, pipe length, pipe slope, pipe top elevation, pipe bottom elevation, inspection well ground elevation, inspection well bottom elevation, and the elevation of rainwater discharge to natural rivers.

[0088] 3. There is no basic data, but there are guidance or recommended data and the parameters need to be adjusted and verified: 1) Total catchment area: the total drainage area contributed to the inspection well.

[0089] 2) Surface runoff area: the surface runoff area along the pipeline connected to the inspection well.

[0090] 3) Slope of surface runoff area: the average slope of the surface runoff area.

[0091] 4) Effective runoff coefficient: the effective runoff coefficient allocated to the surface runoff area.

[0092] 4. External dynamic change parameters: 1) Rainfall intensity (q): Rainfall intensity in the past T hours, including the rainfall intensity at the current time.

[0093] 2) Waterlogging depth at waterlogging points (JH): The depth of waterlogging on the ground surface under certain rainfall intensity conditions.

[0094] The second step is to train the second long short-term memory neural network model: 1. Data preprocessing: Normalize the data so that the values ​​of each feature are on the same scale.

[0095] Construct time series data, where each sample contains data from the past few time steps, and is used to predict the water depth at the next time step.

[0096] 2. Model construction: Define the second long short-term memory neural network model, including LSTM layer and fully connected layer.

[0097] The LSTM layer is used to process time series data, and the fully connected layer is used to output prediction results.

[0098] 3. Training process: Set hyperparameters such as the number of input features, number of hidden units, number of LSTM layers, learning rate, and training epochs.

[0099] In each training cycle, training data is fed into the model and the loss between the predicted values ​​and the target values ​​is calculated.

[0100] Use an optimizer (such as Adam) to update the model parameters to minimize the loss.

[0101] 4. Prediction: Input new rainfall event data and predict the waterlogging depth (JH) of nodes on the street.

[0102] Furthermore, in some preferred embodiments of the present invention, the LSTM model is constrained by the following formula: 1. Input Gate: .

[0103] Where t is the activation vector of the input gate, W i and R i are the weight matrices for input and recurrent connections, b i is the bias vector, x t is the input vector of the current time step, h t-1 is the hidden state at the previous time step.

[0104] 2. Forget Gate: .

[0105] Among them, f t is the activation vector of the forget gate, W f and R f are the weight matrices of input and recurrent connections, b f is the bias vector.

[0106] 3. Output Gate: .

[0107] Among them, t is the activation vector of the output gate, W o and R o are the weight matrices of input and recurrent connections, b o is the bias vector.

[0108] 4. Cell State .

[0109] Among them, c t is the cell state at the current time step, is the candidate unit state, is the element-wise product.

[0110] 5. Hidden State .

[0111] Among them, h t is the hidden state at the current time step.

[0112] Using a long short-term memory (LSTM) network to process time series data can accurately predict the liquid level and fullness of stormwater pipe networks, as well as the depth of waterlogging. By learning the time series characteristics of historical data, the LSTM model can effectively capture the dynamic relationship between rainfall intensity and waterlogging depth, thereby improving prediction accuracy.

[0113] Furthermore, in some preferred embodiments of the present invention, the parameters of the first long short-term memory neural network and the parameters of the second long short-term memory neural network both include: 1 long short-term memory layer, 10 hidden units and a learning rate of 0.001.

[0114] Specifically, take the simulation data shown in Table 4 as an example: Table 4

[0115] 1. Data preprocessing: After standardizing the data, construct the time series input: input features: rainfall intensity, manhole liquid level, pipeline fullness; output target: water depth.

[0116] 2. LSTM model training: Assume we use one LSTM layer with 10 hidden units and a learning rate of 0.001. The model parameters are obtained after training.

[0117] Step S110: inputting rainfall event data into a flood prediction model and outputting a predicted waterlogging depth.

[0118] Specifically, using the data shown in Table 4 above as an example, new rainfall event data (for example, rainfall intensity of 45 mm / h) is input to predict the water depth. The predicted result: the water depth is 50 cm (which is a severe waterlogging risk level).

[0119] Combining GMM cluster analysis and LSTM prediction results, the system dynamically assesses and provides early warnings for urban flooding risks. The system can set different prediction accuracies based on risk levels, ensuring more accurate predictions for high-risk areas. This provides a scientific basis for city managers to take timely countermeasures.

[0120] The present invention provides a flood prediction method, comprising: obtaining data of a rainwater pipe network; wherein the data of the rainwater pipe network includes: pipe network basic data, catchment area data and historical precipitation data; performing cluster analysis based on the pipe network basic data, and dividing target points in the rainwater pipe network into multiple cluster blocks; wherein the target points include: inspection wells and rainwater pipes; cluster blocks correspond to cluster categories one by one; cluster categories represent flood risk levels of cluster blocks; based on the depth of accumulated water in the cluster blocks, classifying them according to multiple preset waterlogging risk levels, and determining the prediction accuracy corresponding to the cluster blocks; wherein different waterlogging risk levels correspond to different waterlogging depths; training a basic flood prediction model based on the rainwater pipe network data in the cluster blocks until the prediction accuracy is reached, thereby obtaining a flood prediction model; wherein the basic flood prediction model includes: a first long short-term memory neural network connected in sequence network and the second long short-term memory neural network; the input data of the first long short-term memory neural network is the data of the rainwater pipe network, and the output data is the liquid level height in the inspection well and the fullness of the rainwater pipe; the input data of the second long short-term memory neural network is the data of the rainwater pipe network, the liquid level height in the inspection well and the fullness of the rainwater pipe, and the output data is the depth of accumulated water; the rainfall event data is input into the flood prediction model, and the predicted depth of accumulated water is output; first, the nodes in the rainwater pipe network are clustered, and the rainwater pipe network is divided into multiple blocks according to different categories, and the historical data in different blocks are used to train the prediction model including the double-layer long short-term memory neural network. The trained model has intelligent and automated data capabilities, which can meet the needs of rapid response, especially in extreme cases, when processing large-scale data, reducing calculation time and improving response speed.

[0121] Example 2 Based on the above embodiment, the present invention provides a flood prediction device. Figure 2 The schematic diagram of a flood prediction device provided by an embodiment of the present invention is shown, and the device includes: The data acquisition module 310 is used to acquire data of the rainwater pipe network. The data of the rainwater pipe network includes: basic pipe network data, catchment area data and historical precipitation data.

[0122] Cluster analysis module 320 is used to perform cluster analysis based on the basic pipe network data, dividing target points in the stormwater pipe network into multiple cluster blocks; wherein the target points include: inspection wells and stormwater pipes; cluster blocks correspond to cluster categories one by one; cluster categories represent the flood risk level of cluster blocks; The prediction accuracy determination module 330 is used to classify the waterlogging depths within the cluster blocks according to a plurality of preset waterlogging risk levels, and determine the prediction accuracy corresponding to the cluster blocks; wherein different waterlogging risk levels correspond to different waterlogging depths.

[0123] The flood prediction model determination module 340 is used to train a basic flood prediction model based on the rainwater pipe network data in the cluster block until the prediction accuracy is achieved to obtain a flood prediction model; wherein, the basic flood prediction model includes: a first long short-term memory neural network and a second long short-term memory neural network connected in sequence; the input data of the first long short-term memory neural network is the data of the rainwater pipe network, and the output data is the liquid level height in the inspection well and the fullness of the rainwater pipe; the input data of the second long short-term memory neural network is the data of the rainwater pipe network, the liquid level height in the inspection well and the fullness of the rainwater pipe, and the output data is the depth of accumulated water.

[0124] The water depth prediction module 350 is used to input rainfall event data into the flood prediction model and output the predicted water depth.

[0125] Furthermore, in some preferred embodiments of the present invention, the cluster analysis module 320 is used to determine a preset proportion of target points in the rainwater pipe network based on preset screening rules; determine the elevation difference of the inspection wells and the fullness of the rainwater pipes in the target points; normalize the elevation difference of the inspection wells and the fullness of the rainwater pipes and use them as feature vectors; train a Gaussian clustering model based on the feature vectors until the preset conditions are met to obtain clustering results.

[0126] Furthermore, in some preferred embodiments of the present invention, the cluster analysis module 320 is used to initialize parameters; wherein the parameters include: mixing weights, mean and covariance matrix; calculating the probability of Gaussian distribution based on the eigenvector; determining the posterior probability through the expectation-maximization algorithm based on the probability of Gaussian distribution; updating the parameters based on the posterior probability, and continuing to recalculate the probability of Gaussian distribution of the eigenvector until the change in the log-likelihood function of the Gaussian clustering model is less than a preset threshold or reaches a preset number of iterations; performing cluster analysis on the eigenvector to obtain a clustering result.

[0127] Furthermore, in some preferred embodiments of the present invention, the log-likelihood function is constrained by the following formula: ;in, is the log-likelihood function, is the kth Gaussian distribution in x i The probability density function at , K is the number of Gaussian distributions, is the eigenvector of the i-th data point, are the parameters of the Gaussian clustering model.

[0128] Furthermore, in some preferred embodiments of the present invention, the device also includes: a model training module, which is used to determine the flood prediction model through the following steps: after standardizing the data of the rainwater pipe network, constructing the first training data based on the time series; training the first long short-term memory neural network based on the first training data, and updating the parameters of the first long short-term memory neural network based on the optimizer until the preset training conditions are reached; outputting the liquid level height in the inspection well and the fullness of the rainwater pipe based on the trained first long short-term memory neural network; after standardizing the data of the rainwater pipe network, the liquid level height in the inspection well and the fullness of the rainwater pipe, constructing the second training data based on the time series; training the second long short-term memory neural network based on the second training data, and updating the parameters of the second long short-term memory neural network based on the optimizer until the preset training conditions are reached; outputting the depth of accumulated water based on the trained second long short-term memory neural network.

[0129] Furthermore, in some preferred embodiments of the present invention, the parameters of the first long short-term memory neural network and the parameters of the second long short-term memory neural network both include: 1 long short-term memory layer, 10 hidden units and a learning rate of 0.001.

[0130] Furthermore, in some preferred embodiments of the present invention, the device also includes: a prediction accuracy determination module, which is used to sort multiple risk levels from low to high; sort multiple prediction accuracies from small to large; and correspond the sorted multiple risk levels to the sorted multiple prediction accuracies.

[0131] Those skilled in the art will clearly understand that, for the convenience and brevity of description, the specific working process of the flood prediction device described above can refer to the corresponding process in the aforementioned embodiment of the flood prediction method, and will not be repeated here.

[0132] Example 3 The embodiment of the present invention further provides an electronic device for executing a flood prediction method; see Figure 3 The embodiment of the present invention shown is a structural schematic diagram of an electronic device provided, which includes a memory 400 and a processor 401, wherein the memory 400 is used to store one or more computer instructions, and the one or more computer instructions are executed by the processor 401 to implement the above-mentioned flood prediction method.

[0133] Furthermore, Figure 3 The electronic device shown further includes a bus 402 and a communication interface 403 , and the processor 401 , the communication interface 403 and the memory 400 are connected via the bus 402 .

[0134] The memory 400 may include high-speed random access memory (RAM) and may also include non-volatile memory, such as at least one disk storage. The communication connection between the system network element and at least one other network element is achieved through at least one communication interface 403 (which may be wired or wireless), and the Internet, wide area network, local area network, metropolitan area network, etc. may be used. The bus 402 may be an ISA bus, a PCI bus, or an EISA bus. The bus can be divided into an address bus, a data bus, a control bus, etc. For ease of representation, Figure 3 Only one bidirectional arrow is used in the diagram, but this does not mean that there is only one bus or one type of bus.

[0135] The processor 401 may be an integrated circuit chip with signal processing capabilities. During implementation, each step of the above method can be completed by hardware integrated logic circuits in the processor 401 or by software instructions. The above processor 401 may be a general-purpose processor, including a central processing unit (CPU), a network processor (NP), etc.; it may also be a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components. It can implement or execute the various methods, steps, and logic block diagrams disclosed in the embodiments of the present invention. The general-purpose processor may be a microprocessor or any conventional processor. The steps of the method disclosed in conjunction with the embodiments of the present invention can be directly implemented and executed by a hardware decoding processor, or by a combination of hardware and software modules in the decoding processor. The software module can be located in a storage medium well-known in the art, such as random access memory, flash memory, read-only memory, programmable read-only memory, electrically erasable programmable memory, registers, etc. The storage medium is located in memory 400, and processor 401 reads the information in memory 400 and, in conjunction with its hardware, completes the steps of the method of the aforementioned embodiment.

[0136] An embodiment of the present invention also provides a computer-readable storage medium, which stores computer-executable instructions. When the computer-executable instructions are called and executed by a processor, the computer-executable instructions prompt the processor to implement the above-mentioned flood prediction method. The specific implementation can be found in the method embodiment, which will not be repeated here.

[0137] The computer program products of the flood prediction method, device, and electronic device provided in the embodiments of the present invention include a computer-readable storage medium storing program code. The instructions included in the program code can be used to execute the methods in the previous method embodiments. For specific implementation, please refer to the method embodiments and will not be repeated here.

[0138] Those skilled in the art will clearly understand that, for the convenience and brevity of description, the specific working process of the system and / or device described above can refer to the corresponding process in the aforementioned method embodiment and will not be repeated here.

[0139] In addition, in the description of the embodiments of the present invention, unless otherwise expressly specified or limited, the terms "mounted," "connected," and "connected" should be understood in a broad sense. For example, they may refer to fixed connections, detachable connections, or integral connections; they may refer to mechanical connections or electrical connections; they may refer to direct connections or indirect connections through an intermediate medium; and they may refer to internal communication between two components. Those skilled in the art will understand the specific meanings of the above terms in the present invention based on the specific circumstances.

[0140] If the functions are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the portion that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions for causing a computer device (which can be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in various embodiments of the present invention. The aforementioned storage media include various media capable of storing program code, such as USB flash drives, mobile hard drives, read-only memories (ROMs), random access memories (RAMs), magnetic disks, or optical disks.

[0141] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the above embodiments, or replace some or all of the technical features therein with equivalents. However, these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A flood prediction method, characterized in that: include: Acquire data of the rainwater pipe network; wherein the data of the rainwater pipe network includes: pipe network basic data, catchment area data and historical precipitation data; Cluster analysis is performed based on the basic data of the pipe network to divide target points in the rainwater pipe network into multiple cluster blocks; wherein the target points include: inspection wells and rainwater pipes; the cluster blocks correspond to cluster categories one by one; the cluster categories represent the flood risk level of the cluster blocks; Classifying the waterlogging depths within the cluster blocks according to a plurality of preset waterlogging risk levels, and determining prediction accuracies corresponding to the cluster blocks; Training a basic flood prediction model based on the rainwater pipe network data in the cluster block until the prediction accuracy is achieved to obtain a flood prediction model; The rainfall event data is input into the flood prediction model, and the predicted water depth is output.

2. The flood prediction method according to claim 1, characterized in that: The step of performing cluster analysis based on the basic data of the pipe network to divide the target points in the rainwater pipe network into multiple cluster blocks includes: Determining a preset proportion of the target points in the stormwater pipe network based on a preset screening rule; Determining the elevation difference of the inspection well and the fullness of the rainwater pipe at the target point; Normalizing the elevation difference of the inspection well and the fullness of the rainwater pipe to obtain a characteristic vector; A Gaussian clustering model is trained based on the feature vector until a preset condition is met to obtain a clustering result.

3. The flood prediction method according to claim 2, characterized in that: The step of training a Gaussian clustering model based on the feature vector until a preset condition is met and obtaining a clustering result includes: Initialization parameters; wherein the parameters include: mixing weights, mean and covariance matrix; Calculating a probability of a Gaussian distribution based on the eigenvector; Determining the posterior probability by an expectation-maximization algorithm based on the probability of the Gaussian distribution; Updating the parameters based on the posterior probability, and continuing to recalculate the probability of the Gaussian distribution of the feature vector until the change in the log-likelihood function of the Gaussian clustering model is less than a preset threshold or a preset number of iterations is reached; Perform cluster analysis on the feature vectors to obtain clustering results.

4. The flood prediction method according to claim 3, characterized in that: The log-likelihood function is constrained by the following formula: ； in, is the log-likelihood function, is the kth Gaussian distribution in x i The probability density function at , K is the number of Gaussian distributions, is the eigenvector of the i-th data point, are the parameters of the Gaussian clustering model.

5. The flood prediction method according to claim 1, characterized in that: The method further includes: determining the flood prediction model by the following steps: After standardizing the data of the rainwater pipe network, constructing first training data based on time series; Training the first long short-term memory neural network based on the first training data, and updating parameters of the first long short-term memory neural network based on an optimizer until a preset training condition is met; Based on the trained first long short-term memory neural network, output the liquid level height in the inspection well and the fullness of the rainwater pipe; After standardizing the data of the rainwater pipe network, the liquid level height in the inspection well, and the fullness of the rainwater pipe, second training data is constructed based on the time series; Training the second long short-term memory neural network based on the second training data, and updating parameters of the second long short-term memory neural network based on an optimizer until a preset training condition is met; Based on the trained second long short-term memory neural network, the water depth is output.

6. The flood prediction method according to claim 5, characterized in that: The parameters of the first long short-term memory neural network and the parameters of the second long short-term memory neural network both include: 1 long short-term memory layer, 10 hidden units, and a learning rate of 0.

001.

7. The flood prediction method according to claim 1, characterized in that: The method further comprises: sorting the plurality of risk levels from low to high; sorting the plurality of prediction accuracies in ascending order; The sorted risk levels are matched to the sorted prediction accuracies.

8. A flood prediction device, characterized in that: include: A data acquisition module is used to acquire data of the rainwater pipe network; wherein the data of the rainwater pipe network includes: pipe network basic data, catchment area data and historical precipitation data; A cluster analysis module is configured to perform cluster analysis based on the basic pipe network data, dividing target points within the rainwater pipe network into a plurality of cluster blocks; wherein the target points include inspection wells and rainwater pipes; the cluster blocks correspond to cluster categories one by one; and the cluster categories represent the flood risk level of the cluster blocks; a prediction accuracy determination module, configured to classify the waterlogging depths within the cluster blocks according to a plurality of preset waterlogging risk levels, and determine the prediction accuracy corresponding to the cluster blocks; wherein different waterlogging risk levels correspond to different waterlogging depths; A flood prediction model determination module is configured to train a basic flood prediction model based on the rainwater pipe network data within the cluster block until the prediction accuracy is achieved, thereby obtaining a flood prediction model; wherein the basic flood prediction model comprises: a first long short-term memory neural network and a second long short-term memory neural network connected in sequence; the first long short-term memory neural network receives input data of the rainwater pipe network, and outputs the liquid level in the inspection well and the fullness of the rainwater pipe; the second long short-term memory neural network receives input data of the rainwater pipe network, the liquid level in the inspection well and the fullness of the rainwater pipe, and outputs the water depth; The waterlogging depth prediction module is used to input rainfall event data into the flood prediction model and output the predicted waterlogging depth.

9. An electronic device, characterized in that: The method comprises a processor and a memory, wherein the memory stores computer-executable instructions that can be executed by the processor, and the processor executes the computer-executable instructions to implement the flood prediction method according to any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that The computer-readable storage medium stores computer-executable instructions. When the computer-executable instructions are called and executed by the processor, the computer-executable instructions prompt the processor to implement the flood prediction method according to any one of claims 1 to 7.

Images