A flood disaster forecasting and early warning method and system in plain river network areas based on hydrological and hydrodynamic model

By introducing deep neural network into the hydrological and hydrodynamic model for analysis of rainfall and spatial topology between sites, the accuracy and efficiency of flood disaster forecasting and early warning in the plain river network area in traditional methods is solved, and the rapid flood disasters are achieved , accurate and intelligent detection and warning, improving the accuracy and reliability of flood disaster warnings.

CN118114827BActive Publication Date: 2024-11-08JIANGSU SURVEYING & DESIGN INST OF WATER RESOURCES
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202410261487.2
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2024-03-07
Publication Date
2024-11-08
Estimated Expiration
2044-03-07

AI Technical Summary

Technical Problem

Traditional flood disaster forecasting and early warning methods have problems such as difficult to determine model parameters, large calculation volume, low calculation efficiency and high requirements for input data in the plain river network area, making it difficult to provide accurate and reliable flood disaster information.

Method used

Using a method based on hydrological hydrodynamic model, the rainfall and inter-site spatial topology structures of multiple sites are extracted from hydrological data, and the correlation analysis of rainfall and inter-site spatial topology is carried out through deep neural network to capture the semantics of hydrological dynamics. Features to determine the warning level of flood disasters.

Benefits of technology

It improves the accuracy and reliability of flood disaster warnings, achieves rapid, accurate and intelligent detection and early warning of flood disasters in plain river network areas, and reduces the harm of flood disasters.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118114827B_ABST
    Figure CN118114827B_ABST
Patent Text Reader

Abstract

The present invention discloses a flood disaster forecasting and warning method and system in a plain river network area based on a hydrological and hydrodynamic model. First, a set of rainfall of a set of monitored sites is extracted from hydrological data, then, a spatial topological matrix between sites of the set of monitored sites is constructed, then, the set of rainfall is time-series arranged and feature analyzed to obtain a global rainfall correlation feature vector, then, the spatial distribution feature of the spatial topological matrix between sites is extracted to obtain a spatial topological feature matrix between sites, then, a hydrological and hydrodynamic model based on a deep neural network is used to process the spatial topological feature matrix between sites and the global rainfall correlation feature vector to obtain hydrological and dynamic semantic features, and finally, based on the hydrological and dynamic semantic features, a flood disaster warning level label is determined. In this way, the accuracy and reliability of flood disaster warnings can be improved, and the harm of flood disasters can be reduced.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present application relates to the field of flood disaster forecasting and early warning, and more specifically, to a flood disaster forecasting and early warning method and system in plain river network areas based on a hydrological and hydrodynamic model. Background Art

[0002] Flood disasters refer to natural disasters that pose serious threats to human life, property, and social economy due to surface waterlogging and river overflow caused by excessive rainfall, rising river water levels, or other reasons. In plain river network areas, the risk of flood disasters is relatively high due to the flat terrain and limited drainage capacity. In order to effectively prevent and mitigate the impact of flood disasters, it is necessary to establish a scientific and reasonable flood disaster forecasting and early warning system to provide accurate and reliable flood disaster information in a timely manner and guide flood prevention and disaster relief work.

[0003] Traditional flood disaster forecasting and warning methods are mainly based on hydrological and hydrodynamic models, which simulate the hydrological and hydrodynamic processes in river network areas, predict hydrological elements such as river water level, flow, and flow velocity, and judge the possibility and severity of flood disasters. However, this method has some limitations, such as difficulty in determining model parameters, large amount of calculation, low calculation efficiency, and high requirements for input data.

[0004] Therefore, an optimized flood disaster forecasting and early warning scheme in plain river network areas based on hydrological and hydrodynamic models is desired. Summary of the invention

[0005] In order to solve the above technical problems, this application is proposed. This application provides a flood disaster forecasting and warning method and system in plain river network areas based on hydrological and hydrodynamic models, which can extract the rainfall of multiple stations and the spatial topological structure between stations from hydrological data, and introduce artificial intelligence-based data processing and analysis algorithms at the back end to perform correlation analysis between rainfall and spatial topology between stations, so as to capture the semantic characteristics of hydrological dynamics, so as to detect the level of flood disasters.

[0006] According to one aspect of the present application, a flood disaster forecasting and early warning method in a plain river network area based on a hydrological and hydrodynamic model is provided, which comprises:

[0007] Obtain hydrological data;

[0008] Extracting a set of rainfall amounts for a set of monitored stations from the hydrological data;

[0009] Constructing an inter-site spatial topology matrix of the monitored site set, wherein the value of each non-diagonal position in the inter-site spatial topology matrix is ​​used to represent the altitude relationship between the corresponding two monitored sites;

[0010] Performing time series arrangement and feature analysis on the rainfall set to obtain a global rainfall correlation feature vector;

[0011] Extracting spatial distribution features of the inter-site spatial topology matrix to obtain an inter-site spatial topology feature matrix;

[0012] Using a hydrological and hydrodynamic model based on a deep neural network, the inter-station spatial topological feature matrix and the global rainfall association feature vector are processed to obtain a hydrological and dynamic semantic feature vector as a hydrological and dynamic semantic feature; and

[0013] Based on the hydrological dynamics semantic features, a flood disaster warning level label is determined.

[0014] According to another aspect of the present application, a flood disaster forecasting and early warning system for plain river network areas based on a hydrological and hydrodynamic model is provided, which includes:

[0015] Data acquisition module, used to obtain hydrological data;

[0016] A rainfall extraction module, used to extract a set of rainfalls of a set of monitored sites from the hydrological data;

[0017] An inter-site spatial topology matrix construction module, used to construct an inter-site spatial topology matrix of the monitored site set, wherein the value of each non-diagonal position in the inter-site spatial topology matrix is ​​used to represent the altitude relationship between the corresponding two monitored sites;

[0018] An arrangement and analysis module, used for performing time series arrangement and feature analysis on the rainfall set to obtain a global rainfall correlation feature vector;

[0019] A spatial distribution feature extraction module, used for extracting spatial distribution features from the inter-site spatial topology matrix to obtain an inter-site spatial topology feature matrix;

[0020] A semantic processing module, for processing the inter-station spatial topological feature matrix and the global rainfall association feature vector using a hydrological and hydrodynamic model based on a deep neural network to obtain a hydrological and hydrodynamic semantic feature vector as a hydrological and hydrodynamic semantic feature; and

[0021] The level analysis module is used to determine the flood disaster warning level label based on the hydrological dynamics semantic characteristics.

[0022] Compared with the prior art, the flood disaster forecasting and warning method and system for plain river network areas based on the hydrological and hydrodynamic model provided by the present application first extracts a set of rainfall of a set of monitored sites from hydrological data, then constructs a spatial topological matrix between sites of the set of monitored sites, then performs time series arrangement and feature analysis on the set of rainfall to obtain a global rainfall correlation feature vector, then extracts spatial distribution features of the spatial topological matrix between sites to obtain a spatial topological feature matrix between sites, then uses a hydrological and hydrodynamic model based on a deep neural network to process the spatial topological feature matrix between sites and the global rainfall correlation feature vector to obtain hydrological and dynamic semantic features, and finally determines a flood disaster warning level label based on the hydrological and dynamic semantic features. In this way, the accuracy and reliability of flood disaster warnings can be improved, and the harm of flood disasters can be reduced. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the following briefly introduces the drawings required for use in the description of the embodiments. The following drawings are not intentionally scaled to the actual size, and the focus is on illustrating the main purpose of the present application.

[0024] Figure 1 It is a flow chart of a flood disaster forecasting and early warning method in a plain river network area based on a hydrological and hydrodynamic model according to an embodiment of the present application.

[0025] Figure 2 Schematic diagram of the architecture of a flood disaster forecasting and warning method in a plain river network area based on a hydrological and hydrodynamic model according to an embodiment of the present application.

[0026] Figure 3 It is a block diagram of a flood disaster forecasting and early warning system in a plain river network area based on a hydrological and hydrodynamic model according to an embodiment of the present application.

[0027] Figure 4 This is an application scenario diagram of the flood disaster forecasting and early warning method in plain river network areas based on the hydrological and hydrodynamic model according to an embodiment of the present application. DETAILED DESCRIPTION

[0028] The technical solutions in the embodiments of the present application will be described clearly and completely below in conjunction with the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present application, not all embodiments. Based on the embodiments of the present application, all other embodiments obtained by ordinary technicians in this field without creative work also fall within the scope of protection of the present application.

[0029] As shown in this application and claims, unless the context clearly indicates an exception, the words "a", "an", "an" and / or "the" do not refer to the singular and may also include the plural. Generally speaking, the terms "include" and "comprise" only indicate the inclusion of the steps and elements that have been clearly identified, and these steps and elements do not constitute an exclusive list. The method or device may also include other steps or elements.

[0030] Although the present application makes various references to certain modules in the system according to the embodiments of the present application, any number of different modules can be used and run on the user terminal and / or server. The modules are only illustrative, and different aspects of the system and method can use different modules.

[0031] Flowcharts are used in the present application to illustrate the operations performed by the system according to the embodiments of the present application. It should be understood that the preceding or following operations are not necessarily performed accurately in order. On the contrary, various steps may be processed in reverse order or simultaneously as required. Meanwhile, other operations may also be added to these processes, or a certain step or several steps of operations may be removed from these processes.

[0032] Below, the exemplary embodiments according to the present application will be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments of the present application, and it should be understood that the present application is not limited to the exemplary embodiments described here.

[0033] In response to the above technical problems, the technical concept of the present application is to extract the rainfall of multiple stations and the spatial topological structure between stations from the hydrological data, and introduce artificial intelligence-based data processing and analysis algorithms at the back end to perform correlation analysis between rainfall and spatial topology between stations, so as to capture the semantic characteristics of hydrological dynamics, so as to perform flood disaster level detection. In this way, rapid, accurate and intelligent detection and early warning of flood disasters in plain river network areas are achieved, and the accuracy and reliability of flood disaster early warnings are improved, which helps to take timely measures to deal with flood disasters, thereby reducing the harm caused by flood disasters.

[0034] Figure 1 It is a flow chart of a flood disaster forecasting and early warning method in a plain river network area based on a hydrological and hydrodynamic model according to an embodiment of the present application. Figure 2 FIG. 1 is a schematic diagram of the architecture of a flood disaster forecasting and warning method for a plain river network area based on a hydrological and hydrodynamic model according to an embodiment of the present application. Figure 1 and Figure 2As shown, according to the embodiment of the present application, the flood disaster forecasting and warning method for the plain river network area based on the hydrological and hydrodynamic model includes the following steps: S110, obtaining hydrological data; S120, extracting a set of rainfall of a set of monitored sites from the hydrological data; S130, constructing an inter-site spatial topological matrix of the set of monitored sites, wherein the values ​​of each position on the non-diagonal position in the inter-site spatial topological matrix are used to represent the altitude relationship between the corresponding two monitored sites; S140, performing time series arrangement and feature analysis on the set of rainfall to obtain a global rainfall correlation feature vector; S150, performing spatial distribution feature extraction on the inter-site spatial topological matrix to obtain an inter-site spatial topological feature matrix; S160, using a hydrological and hydrodynamic model based on a deep neural network to process the inter-site spatial topological feature matrix and the global rainfall correlation feature vector to obtain a hydrological dynamic semantic feature vector as a hydrological dynamic semantic feature; and, S170, determining a flood disaster warning level label based on the hydrological dynamic semantic feature.

[0035] Specifically, in the technical solution of the present application, first, hydrological data is obtained. Then, a set of rainfall of a set of monitored sites is extracted from the hydrological data. It should be understood that by extracting a set of rainfall of a set of monitored sites from the hydrological data, rainfall data of different sites can be obtained. These data can be used for analysis and modeling to understand the spatial distribution characteristics of rainfall, and then judge the possibility and severity of flood disasters.

[0036] Then, the rainfall set is arranged into a global rainfall input vector according to the sample dimension, and then feature mining is performed in the site-to-site rainfall correlation feature extractor based on a one-dimensional convolutional layer to extract the local correlation feature information of the global rainfall between each site in the monitored site set, thereby obtaining a global rainfall correlation feature vector. In this way, the local spatial distribution characteristics of the global rainfall of each site can be captured, so as to facilitate the subsequent more accurate analysis and characterization of the semantic characteristics of hydrological dynamics, thereby carrying out flood disaster warning.

[0037] Accordingly, in step S140, the rainfall set is arranged in time series and feature analyzed to obtain a global rainfall correlation feature vector, including: arranging the rainfall set into a global rainfall input vector according to the sample dimension and then passing it through an inter-site rainfall correlation feature extractor based on a one-dimensional convolutional layer to obtain the global rainfall correlation feature vector.

[0038] Then, considering that in flood disaster forecasting and warning, the altitude relationship is of great significance to the prediction and warning of flood disasters, this is because in the plain river network area, the terrain is flat, and the altitude change may affect the flow direction and flow rate of water flow, thereby affecting the diffusion and accumulation of floods. Therefore, incorporating the altitude relationship into the model helps to more accurately simulate and predict the occurrence and development process of flood disasters. Based on this, in order to capture the spatial relationship between sites, especially the altitude relationship, so as to more accurately perform the correlation analysis between the precipitation of the sites and the flood disaster warning, in the technical solution of the present application, the inter-site spatial topology matrix of the monitored site set is further constructed, wherein the values ​​of each position on the non-diagonal position in the inter-site spatial topology matrix are used to represent the altitude relationship between the corresponding two monitored sites. Specifically, by constructing a spatial topology matrix between multiple sites, the spatial relationship between the sites can be represented as a matrix, wherein the rows and columns of the matrix correspond to each site in the monitored site set. The values ​​on the non-diagonal positions of the matrix are used to represent the altitude relationship between the corresponding two sites. In particular, if the altitude difference between the two sites is large, the value of the corresponding position can be recorded as 1, indicating a site with a high altitude. On the contrary, if the site is at a low altitude, the value of the corresponding position is recorded as -1. In this way, the values ​​in the matrix can reflect the altitude changes between sites.

[0039] Furthermore, the inter-site spatial topology matrix is ​​subjected to feature mining in a spatial distribution feature extractor based on a convolutional neural network model to extract spatial topology association feature information between the various sites, thereby obtaining an inter-site spatial topology feature matrix.

[0040] Accordingly, in step S150, spatial distribution features are extracted from the inter-site spatial topology matrix to obtain the inter-site spatial topology feature matrix, including: passing the inter-site spatial topology matrix through a spatial distribution feature extractor based on a convolutional neural network model to obtain the inter-site spatial topology feature matrix.

[0041] Furthermore, since the global rainfall correlation feature vector reflects the local correlation characteristics of the global rainfall between each monitored site, and the inter-site spatial topological feature matrix reflects the spatial topological distribution characteristics between each site. Therefore, in order to simulate and predict hydrological elements such as changes in water flow and rise in river water levels, and thus predict the possibility and severity of flood disasters, it is necessary to extract the semantic features of hydrodynamics. Specifically, in the technical solution of the present application, a hydrological and hydrodynamic model based on a deep neural network is further used to process the inter-site spatial topological feature matrix and the global rainfall correlation feature vector to obtain a hydrodynamic semantic feature vector. In this way, through processing with a deep neural network, the nonlinear relationship and feature representation in the input data can be learned, thereby extracting the semantic features of hydrodynamics, so as to better represent the characteristics and laws of hydrodynamics. In this way, more accurate and useful information can be provided for subsequent flood disaster prediction and early warning.

[0042] Accordingly, in step S160, the inter-station spatial topological feature matrix and the global rainfall association feature vector are processed using a hydrological and hydrodynamic model based on a deep neural network to obtain a hydrological and dynamic semantic feature vector as a hydrological and dynamic semantic feature, including: using the hydrological and hydrodynamic model based on a deep neural network to process the inter-station spatial topological feature matrix and the global rainfall association feature vector using the following calculation formula to obtain the hydrological and dynamic semantic feature vector; wherein the calculation formula is:

[0043]

[0044] Where M is the spatial topological feature matrix between the sites, V a is the global rainfall associated feature vector, represents matrix multiplication, V b is the hydrodynamic semantic feature vector.

[0045] Next, the hydrodynamic semantic feature vector is passed through a flood disaster forecast and warning device based on a classifier to obtain a flood disaster warning level label. Specifically, in the technical solution of the present application, the label of the classifier is a flood disaster warning level label. Therefore, after obtaining the classification result, the level of flood disasters can be detected and warned based on the classification result. In this way, rapid, accurate and intelligent detection and warning of flood disasters in plain river network areas are achieved, and the accuracy and reliability of flood disaster warnings are improved.

[0046] Accordingly, in step S170, based on the hydrodynamic semantic feature, a flood disaster warning level label is determined, including: passing the hydrodynamic semantic feature vector through a classifier-based flood disaster forecasting and warning device to obtain a flood disaster warning level label.

[0047] It should be understood that the role of the classifier is to use the given categories and known training data to learn classification rules and classifiers, and then classify (or predict) unknown data. Logistic regression and SVM are often used to solve binary classification problems. For multi-class classification problems, logistic regression or SVM can also be used, but multiple binary classifications are required to form a multi-classification. However, this is prone to errors and is not efficient. Commonly used multi-classification methods include the Softmax classification function.

[0048] Furthermore, in the technical solution of the present application, the flood disaster forecasting and warning method for plain river network areas based on the hydrological and hydrodynamic model also includes a training step: for training the inter-site rainfall correlation feature extractor based on the one-dimensional convolutional layer, the spatial distribution feature extractor based on the convolutional neural network model, the hydrological and hydrodynamic model based on the deep neural network and the flood disaster forecasting and warning device based on the classifier.

[0049] In one example, the training step includes: acquiring training data, wherein the training data includes training hydrological data and the true value of the flood disaster warning level label; extracting a set of training rainfall for a set of training monitored sites from the training hydrological data; constructing a training site-to-site spatial topology matrix for the set of training monitored sites, wherein the values ​​of each non-diagonal position in the training site-to-site spatial topology matrix are used to represent the altitude relationship between the corresponding two training monitored sites; arranging the set of training rainfall into a training global rainfall input vector according to the sample dimension and then passing it through the site-to-site rainfall correlation feature extractor based on the one-dimensional convolutional layer to obtain a training global rainfall correlation feature vector; passing the training site-to-site spatial topology matrix through the spatial distribution feature extractor based on the convolutional neural network model extractor to obtain the spatial topological feature matrix between training sites; use the hydrological and hydrodynamic model based on deep neural network to process the spatial topological feature matrix between training sites and the training global rainfall association feature vector to obtain a training hydrological and dynamic semantic feature vector; pass the training hydrological and dynamic semantic feature vector through the flood disaster forecast and early warning device based on the classifier to obtain a classification loss function value; and, use the classification loss function value to train the rainfall association feature extractor between sites based on one-dimensional convolution layer, the spatial distribution feature extractor based on convolution neural network model, the hydrological and hydrodynamic model based on deep neural network and the flood disaster forecast and early warning device based on classifier, wherein, in each round of iteration of the training, the training hydrological and dynamic semantic feature vector is corrected.

[0050] The method of passing the training hydrological dynamics semantic feature vector through the classifier-based flood disaster forecast and early warning device to obtain a classification loss function value comprises: processing the training hydrological dynamics semantic feature vector through the classifier-based flood disaster forecast and early warning device with the following classification training formula to obtain a training classification result; wherein the classification training formula is:

[0051] softmax{(W n , B n ):...:(W1,B1)|X}

[0052] Among them, W1 to W n is the weight matrix, B1 to B n is a bias vector, X is the training hydrodynamic semantic feature vector; and a cross entropy value between the training classification result and the true value is calculated as the classification loss function value.

[0053] In the above technical solution, the training global rainfall association feature vector expresses the one-dimensional local association feature of the training global rainfall input vector formed by arranging the set of training rainfall according to the sample dimension (that is, the local association feature between the training rainfalls between the various training monitored sites), and the spatial topological feature matrix between the training sites expresses the spatial topological distribution feature of the spatial topological matrix between the training sites. Therefore, after processing the spatial topological feature matrix between the training sites and the training global rainfall association feature vector using a hydrological and hydrodynamic model based on a deep neural network, the training hydrological and hydrodynamic semantic feature vector can include the rainfall association features of each monitored site in the training based on its spatial topological distribution, thereby improving the expression effect of the sample value association feature of the training hydrological and hydrodynamic semantic feature vector for the specific spatial topological distribution of multiple samples.

[0054] However, when the correlation features of multiple sample spatial topological distribution features and their corresponding sample values ​​are integrated, the significance of the feature distribution information based on their specific feature distribution of the spatial topological feature matrix between training sites and the training global rainfall correlation feature vector will be affected. The training hydrological dynamics semantic feature vector has a more significant problem of insufficient posterior distribution aggregation, which affects the expression effect of the training hydrological dynamics semantic feature vector.

[0055] Therefore, the present application preferably performs feature aggregation optimization on the training hydrodynamic semantic feature vector, for example, denoted as V, each time the training hydrodynamic semantic feature vector is subjected to classification iterative training through a classifier.

[0056] In one example, each time the training hydrological dynamics semantic feature vector is subjected to classification iterative training by a classifier, feature aggregation optimization is performed on the training hydrological dynamics semantic feature vector, including: performing feature aggregation optimization on the training hydrological dynamics semantic feature vector according to the following formula to obtain an optimized training hydrological dynamics semantic feature vector; wherein the formula is:

[0057]

[0058] Where V is the training hydrodynamic semantic feature vector, v i is the eigenvalue of the i-th position of the training hydrodynamic semantic feature vector, is the square of the 1-norm of the training hydrodynamic semantic feature vector V, ||V||2 -1 / 2 is the inverse of the square root of the 2-norm of the training hydrodynamic semantic feature vector V, L is the length of the training hydrodynamic semantic feature vector V, and ε is a scaling hyperparameter, log represents the logarithmic function value with base 2, v iis the eigenvalue of the ith position of the optimized training hydrodynamic semantic feature vector.

[0059] Here, the low-rank structure based on the norm of the training hydrodynamic semantic feature vector V is represented as a voting cluster of the feature value set aggregation of the training hydrodynamic semantic feature vector V, and each feature value of the training hydrodynamic semantic feature vector V is subjected to canonical voting relative to the information perception framework, so as to map the feature values ​​belonging to the same regression class to a similar local canonical coordinate set through the regression representation of the direction and scale of the aggregated feature distribution, thereby improving the aggregation effect of the feature set of the training hydrodynamic semantic feature vector V, improving the expression effect of the training hydrodynamic semantic feature vector V, and improving the accuracy of the classification result obtained by the classifier. In this way, a hydrodynamic semantic model can be constructed based on the rainfall and topological distribution of multiple stations to perform level detection and early warning of flood disasters. In this way, rapid, accurate and intelligent detection and early warning of flood disasters in plain river network areas are achieved, and the accuracy and reliability of flood disaster early warnings are improved, which helps to take timely measures to deal with flood disasters, thereby reducing the harm of flood disasters.

[0060] In summary, the flood disaster forecasting and warning method in the plain river network area based on the hydrological and hydrodynamic model according to the embodiment of the present application is clarified, which can improve the accuracy and reliability of flood disaster warning and reduce the harm of flood disasters.

[0061] Figure 3 FIG. 1 is a block diagram of a flood disaster forecasting and warning system 100 for a plain river network area based on a hydrological and hydrodynamic model according to an embodiment of the present application. Figure 3As shown, according to the embodiment of the present application, the flood disaster forecasting and early warning system 100 for plain river network areas based on the hydrological and hydrodynamic model includes: a data acquisition module 110, which is used to acquire hydrological data; a rainfall extraction module 120, which is used to extract a set of rainfalls of a monitored site set from the hydrological data; an inter-site spatial topology matrix construction module 130, which is used to construct an inter-site spatial topology matrix of the monitored site set, wherein the values ​​of each position on the non-diagonal position in the inter-site spatial topology matrix are used to represent the altitude relationship between the corresponding two monitored sites; an arrangement analysis module 140, which is used to analyze the rainfall The collection is arranged in time series and feature analyzed to obtain a global rainfall associated feature vector; a spatial distribution feature extraction module 150 is used to extract spatial distribution features of the inter-site spatial topology matrix to obtain an inter-site spatial topology feature matrix; a semantic processing module 160 is used to use a hydrological and hydrodynamic model based on a deep neural network to process the inter-site spatial topology feature matrix and the global rainfall associated feature vector to obtain a hydrological dynamics semantic feature vector as a hydrological dynamics semantic feature; and a level analysis module 170 is used to determine the flood disaster warning level label based on the hydrological dynamics semantic feature.

[0062] In one example, in the above-mentioned flood disaster forecasting and warning system 100 for plain river network areas based on the hydrological and hydrodynamic model, the arrangement analysis module 140 is used to: arrange the rainfall set into a global rainfall input vector according to the sample dimension and then obtain the global rainfall association feature vector through an inter-site rainfall association feature extractor based on a one-dimensional convolutional layer.

[0063] Here, those skilled in the art can understand that the specific functions and operations of each module in the above-mentioned plain river network area flood disaster forecasting and early warning system 100 based on the hydrological and hydrodynamic model have been referred to above. Figure 1 to Figure 2 It has been introduced in detail in the description of flood disaster forecasting and early warning method in plain river network areas based on hydrological and hydrodynamic model, and therefore, its repeated description will be omitted.

[0064] As described above, the flood disaster forecasting and warning system 100 for plain river network areas based on the hydrological and hydrodynamic model according to the embodiment of the present application can be implemented in various wireless terminals, such as a server having a flood disaster forecasting and warning algorithm for plain river network areas based on the hydrological and hydrodynamic model. In one example, the flood disaster forecasting and warning system 100 for plain river network areas based on the hydrological and hydrodynamic model according to the embodiment of the present application can be integrated into the wireless terminal as a software module and / or a hardware module. For example, the flood disaster forecasting and warning system 100 for plain river network areas based on the hydrological and hydrodynamic model can be a software module in the operating system of the wireless terminal, or can be an application developed for the wireless terminal; of course, the flood disaster forecasting and warning system 100 for plain river network areas based on the hydrological and hydrodynamic model can also be one of the many hardware modules of the wireless terminal.

[0065] Alternatively, in another example, the flood disaster forecasting and warning system 100 for plain river network areas based on hydrological and hydrodynamic models and the wireless terminal may also be separate devices, and the flood disaster forecasting and warning system 100 for plain river network areas based on hydrological and hydrodynamic models may be connected to the wireless terminal via a wired and / or wireless network, and transmit interactive information in accordance with an agreed data format.

[0066] Figure 4 FIG. 1 is an application scenario diagram of a flood disaster forecasting and warning method for a plain river network area based on a hydrological and hydrodynamic model according to an embodiment of the present application. Figure 4 As shown, in this application scenario, first, hydrological data (for example, Figure 4 Then, the hydrological data is input to a server (for example, Figure 4 In S) shown in , the server can use the flood disaster forecasting and warning algorithm for plain river network areas based on the hydrological and hydrodynamic model to process the hydrological data to obtain a flood disaster warning level label.

[0067] Furthermore, in another example of the present application, the system adopts a B / S architecture with multiple levels of users. In order to reduce the difficulty of system deployment, save operation and maintenance costs, increase the speed of project implementation, and facilitate unified maintenance and management, the system B / S view framework adopts OOP (object-oriented) to build the basic operation framework, and modular development is used as the form of functional development. The framework follows the principles of security, stability, flexibility, and power, is easy to expand functions, maintain codes, and has excellent secondary development capabilities. It can meet all application requirements of comprehensive scheduling. The system B / S view framework has the following characteristics: 1) Multi-terminal adaptation: The system adapts to PC computers, flat computers, mobile phones, WeChat and other devices; 2) System theme settings: strong theme setting function, you can set the head and bottom style and content; 3) System page settings: real-time editing, what you see is what you get, powerful module functions and css animations.

[0068] The system can be developed and deployed using a microservice technology framework. The storage system uses a combined architecture of large relational databases, semi-structured and unstructured databases. The server side is supported by a Web Server and corresponding middleware. The development follows the front-end and back-end separation architecture model, making the front-end and back-end interaction interfaces clearer and facilitating later system maintenance and optimization.

[0069] Furthermore, the system provides developers with convenient function and service publishing capabilities, which can quickly publish business functions, generate service specification contracts, and then generate default implementation code for services based on the contracts. Based on the service contracts, the input and output specifications of the service interfaces are determined, and the business service processes are orchestrated in a simple graphical way. The system simplifies the complexity of distributed service calls through orchestration, such as synchronous, asynchronous, asynchronous simulated synchronous, timeout retry, transaction compensation, etc., which are all completed by the service orchestration engine and no longer rely entirely on the coding ability of developers.

[0070] Since the system carries a lot of flood disaster warning and dispatch related information, once the system fails, it will cause immeasurable losses. The entire system should support operation based on autonomous and controllable equipment, operating systems, middleware and databases. Therefore, according to the information content stored in this system, the losses and impacts caused by the system failure, the system security is built in accordance with the third level security standard.

[0071] Regarding the data flow, forecasting is mainly based on basin basic data, water and rain monitoring data and geographic spatial data; warning is based on forecast result data and monitored water and rain conditions and video data to issue warning information; rehearsal is a three-dimensional simulation of basin floods through basin visualization model, flood control-related water conservancy model, mathematical simulation engine and various monitoring data and flood forecast result data collected by the front end, combined with flood control dispatch plan, and the simulation process will be displayed in the form of video, image and text; the rehearsal result data is used to support the dispatch consultation work, and the flood control dispatch plan is optimized in combination with basin basic data, monitoring data and socio-economic data; the flood plan will form a dispatch instruction based on the results of the consultation and display it, and the dispatch order data will be pushed to relevant units.

[0072] According to another aspect of the present application, a non-volatile computer-readable storage medium is provided, on which computer-readable instructions are stored. When the instructions are executed by a computer, the above-mentioned method can be executed.

[0073] The program part of the technology can be considered as a "product" or "manufactured product" in the form of executable code and / or related data, which is participated in or realized by computer-readable media. Tangible and permanent storage media can include any memory or storage used by any computer, processor, or similar device or related module. For example, various semiconductor memories, tape drives, disk drives or any similar devices that can provide storage functions for software.

[0074] In addition, it will be appreciated by those skilled in the art that various aspects of the present application may be illustrated and described by a number of patentable categories or situations, including any new and useful process, machine, product or combination of substances, or any new and useful improvements thereto. Accordingly, various aspects of the present application may be performed entirely by hardware, entirely by software (including firmware, resident software, microcode, etc.), or by a combination of hardware and software. The above hardware or software may all be referred to as "data blocks", "modules", "engines", "units", "components" or "systems". In addition, various aspects of the present application may be represented as a computer product located in one or more computer-readable media, which includes computer-readable program code.

[0075] Unless otherwise defined, all terms (including technical and scientific terms) used herein have the same meaning as commonly understood by those of ordinary skill in the art to which this application belongs. It should also be understood that terms such as those defined in common dictionaries should be interpreted as having a meaning consistent with their meaning in the context of the relevant technology, and should not be interpreted in an idealized or extremely formal sense, unless explicitly defined as such herein.

[0076] The above is an explanation of the present application and should not be considered as a limitation thereof. Although several exemplary embodiments of the present application are described, it will be readily understood by those skilled in the art that many modifications may be made to the exemplary embodiments without departing from the novel teachings and advantages of the present application. Therefore, all of these modifications are intended to be included within the scope of the present application as defined by the claims. It should be understood that the above is an explanation of the present application and should not be considered to be limited to the specific embodiments disclosed, and modifications to the disclosed embodiments and other embodiments are intended to be included within the scope of the appended claims. The present application is defined by the claims and their equivalents.

Claims

1. A flood disaster forecasting and early warning method in plain river network areas based on hydrological and hydrodynamic models, characterized in that: include: Obtain hydrological data; Extracting a set of rainfall amounts for a set of monitored stations from the hydrological data; Constructing an inter-site spatial topology matrix of the monitored site set, wherein the value of each non-diagonal position in the inter-site spatial topology matrix is ​​used to represent the altitude relationship between the corresponding two monitored sites; Performing time series arrangement and feature analysis on the rainfall set to obtain a global rainfall correlation feature vector; Extracting spatial distribution features of the inter-site spatial topology matrix to obtain an inter-site spatial topology feature matrix; Using a hydrological and hydrodynamic model based on a deep neural network, the inter-station spatial topological feature matrix and the global rainfall association feature vector are processed to obtain a hydrological and dynamic semantic feature vector as a hydrological and dynamic semantic feature; and Based on the hydrological dynamics semantic features, determining flood disaster warning level labels; The method further comprises a training step: for training an inter-station rainfall correlation feature extractor based on a one-dimensional convolutional layer, a spatial distribution feature extractor based on a convolutional neural network model, a hydrological and hydrodynamic model based on a deep neural network, and a flood disaster forecasting and warning device based on a classifier; The training step comprises: Acquiring training data, wherein the training data includes training hydrological data and the true value of the flood disaster warning level label; Extracting a set of training rainfall for a set of training monitored sites from the training hydrological data; Constructing a spatial topological matrix between training sites of the set of training monitored sites, wherein the value of each non-diagonal position in the spatial topological matrix between training sites is used to represent the altitude relationship between the corresponding two training monitored sites; Arranging the set of training rainfalls according to the sample dimension into a training global rainfall input vector and then passing it through the inter-site rainfall correlation feature extractor based on the one-dimensional convolutional layer to obtain a training global rainfall correlation feature vector; Passing the spatial topology matrix between training sites through the spatial distribution feature extractor based on the convolutional neural network model to obtain the spatial topology feature matrix between training sites; Using the hydrological and hydrodynamic model based on the deep neural network, the spatial topological feature matrix between the training sites and the training global rainfall association feature vector are processed to obtain a training hydrological and hydrodynamic semantic feature vector; Passing the training hydrological dynamics semantic feature vector through the classifier-based flood disaster forecasting and warning device to obtain a classification loss function value; and The one-dimensional convolutional layer-based site rainfall correlation feature extractor, the convolutional neural network model-based spatial distribution feature extractor, the deep neural network-based hydrological and hydrodynamic model, and the classifier-based flood disaster forecast and warning device are trained with the classification loss function value, wherein each time the training hydrological and hydrodynamic semantic feature vector is subjected to classification iterative training through the classifier, feature aggregation optimization is performed on the training hydrological and hydrodynamic semantic feature vector; Each time the training hydrological dynamics semantic feature vector is subjected to classification iterative training by the classifier, feature aggregation optimization is performed on the training hydrological dynamics semantic feature vector, including: feature aggregation optimization is performed on the training hydrological dynamics semantic feature vector according to the following formula to obtain an optimized training hydrological dynamics semantic feature vector; wherein the formula is: ; in, is the training hydrodynamic semantic feature vector, is the first The eigenvalues ​​at the positions, is the training hydrodynamic semantic feature vector The square of the 1-norm of is the training hydrodynamic semantic feature vector The reciprocal of the square root of the 2-norm of is the training hydrodynamic semantic feature vector The length of is a scaling hyperparameter, represents the logarithmic function value with base 2, is the first The feature value at each position.

2. The flood disaster forecasting and early warning method in plain river network area based on hydrological and hydrodynamic model according to claim 1 is characterized in that: The rainfall set is arranged in time series and analyzed for characteristics to obtain a global rainfall correlation feature vector, including: The rainfall set is arranged into a global rainfall input vector according to the sample dimension, and then passed through an inter-site rainfall correlation feature extractor based on a one-dimensional convolutional layer to obtain the global rainfall correlation feature vector.

3. The flood disaster forecasting and early warning method in plain river network area based on hydrological and hydrodynamic model according to claim 2 is characterized in that: Extracting spatial distribution features of the inter-site spatial topology matrix to obtain an inter-site spatial topology feature matrix includes: The inter-site spatial topology matrix is ​​passed through a spatial distribution feature extractor based on a convolutional neural network model to obtain the inter-site spatial topology feature matrix.

4. The flood disaster forecasting and early warning method in plain river network area based on hydrological and hydrodynamic model according to claim 3 is characterized in that: The inter-station spatial topological feature matrix and the global rainfall association feature vector are processed using a hydrological and hydrodynamic model based on a deep neural network to obtain a hydrological and dynamic semantic feature vector as a hydrological and dynamic semantic feature, including: The hydrological and hydrodynamic model based on a deep neural network is used to process the inter-station spatial topological feature matrix and the global rainfall association feature vector using the following calculation formula to obtain the hydrological and dynamic semantic feature vector; wherein the calculation formula is: ; in, is the spatial topological characteristic matrix between the sites, is the global rainfall associated feature vector, represents matrix multiplication, is the hydrodynamic semantic feature vector.

5. The flood disaster forecasting and early warning method in plain river network area based on hydrological and hydrodynamic model according to claim 4 is characterized in that: Based on the hydrological dynamics semantic features, the flood disaster warning level label is determined, including: The hydrological dynamics semantic feature vector is passed through a flood disaster forecasting and warning device based on a classifier to obtain a flood disaster warning level label.

6. The flood disaster forecasting and early warning method in plain river network area based on hydrological and hydrodynamic model according to claim 5 is characterized in that: The training hydrological dynamics semantic feature vector is passed through the classifier-based flood disaster forecasting and warning device to obtain a classification loss function value, including: The training hydrological dynamics semantic feature vector is processed by the classifier-based flood disaster forecasting and warning device using the following classification training formula to obtain a training classification result; wherein the classification training formula is: ; in, arrive is the weight matrix, arrive is the bias vector, for the training hydrological dynamics semantic feature vector; and A cross entropy value between the training classification result and the true value is calculated as the classification loss function value.

7. A flood disaster forecasting and early warning system for plain river network areas based on hydrological and hydrodynamic models, characterized in that: include: Data acquisition module, used to obtain hydrological data; A rainfall extraction module, used to extract a set of rainfalls of a set of monitored sites from the hydrological data; An inter-site spatial topology matrix construction module, used to construct an inter-site spatial topology matrix of the monitored site set, wherein the value of each non-diagonal position in the inter-site spatial topology matrix is ​​used to represent the altitude relationship between the corresponding two monitored sites; An arrangement and analysis module, used for performing time series arrangement and feature analysis on the rainfall set to obtain a global rainfall correlation feature vector; A spatial distribution feature extraction module, used for extracting spatial distribution features from the inter-site spatial topology matrix to obtain an inter-site spatial topology feature matrix; A semantic processing module, for processing the inter-station spatial topological feature matrix and the global rainfall association feature vector using a hydrological and hydrodynamic model based on a deep neural network to obtain a hydrological and hydrodynamic semantic feature vector as a hydrological and hydrodynamic semantic feature; and A level analysis module, used to determine a flood disaster warning level label based on the hydrological dynamics semantic features; The system also includes a training module: used to train an inter-site rainfall correlation feature extractor based on a one-dimensional convolutional layer, a spatial distribution feature extractor based on a convolutional neural network model, a hydrological and hydrodynamic model based on a deep neural network, and a flood disaster forecasting and warning device based on a classifier; The training module is used to: Acquiring training data, wherein the training data includes training hydrological data and the true value of the flood disaster warning level label; Extracting a set of training rainfall for a set of training monitored sites from the training hydrological data; Constructing a spatial topological matrix between training sites of the set of training monitored sites, wherein the value of each non-diagonal position in the spatial topological matrix between training sites is used to represent the altitude relationship between the corresponding two training monitored sites; Arranging the set of training rainfalls according to the sample dimension into a training global rainfall input vector and then passing it through the inter-site rainfall correlation feature extractor based on the one-dimensional convolutional layer to obtain a training global rainfall correlation feature vector; Passing the spatial topology matrix between training sites through the spatial distribution feature extractor based on the convolutional neural network model to obtain the spatial topology feature matrix between training sites; Using the hydrological and hydrodynamic model based on the deep neural network, the spatial topological feature matrix between the training sites and the training global rainfall association feature vector are processed to obtain a training hydrological and hydrodynamic semantic feature vector; Passing the training hydrological dynamics semantic feature vector through the classifier-based flood disaster forecasting and warning device to obtain a classification loss function value; and The one-dimensional convolutional layer-based site rainfall correlation feature extractor, the convolutional neural network model-based spatial distribution feature extractor, the deep neural network-based hydrological and hydrodynamic model, and the classifier-based flood disaster forecast and warning device are trained with the classification loss function value, wherein each time the training hydrological and hydrodynamic semantic feature vector is subjected to classification iterative training through the classifier, feature aggregation optimization is performed on the training hydrological and hydrodynamic semantic feature vector; Each time the training hydrological dynamics semantic feature vector is subjected to classification iterative training by the classifier, feature aggregation optimization is performed on the training hydrological dynamics semantic feature vector, including: feature aggregation optimization is performed on the training hydrological dynamics semantic feature vector according to the following formula to obtain an optimized training hydrological dynamics semantic feature vector; wherein the formula is: ; in, is the training hydrodynamic semantic feature vector, is the first The eigenvalues ​​at the positions, is the training hydrodynamic semantic feature vector The square of the 1-norm of is the training hydrodynamic semantic feature vector The reciprocal of the square root of the 2-norm of is the training hydrodynamic semantic feature vector The length of is a scaling hyperparameter, represents the logarithmic function value with base 2, is the first The feature value at each position.

8. The flood disaster forecasting and early warning system for plain river network areas based on the hydrological and hydrodynamic model according to claim 7 is characterized in that: The arrangement analysis module is used to: The rainfall set is arranged into a global rainfall input vector according to the sample dimension, and then passed through an inter-site rainfall correlation feature extractor based on a one-dimensional convolutional layer to obtain the global rainfall correlation feature vector.