A real-time river water level monitoring and early warning method and system based on deep learning

Abstract

The present application relates to a real-time river water level monitoring and early warning method and system based on deep learning. The method includes the following steps: obtaining remote sensing images of the target watershed and real-time hydrological data of multiple monitoring points; performing water flow direction topological analysis on the remote sensing images of the target watershed and constructing a water flow topological network; performing dynamic change characteristics analysis on the remote sensing images of the target watershed and generating dynamic change data of the watershed structure; performing topological situation evolution fitting on the water flow topological network based on the dynamic change data of the watershed structure to obtain a dynamic topological evolution network; performing multi-scale time series trend evolution analysis on the real-time hydrological data of multiple monitoring points to generate water flow time series evolution characteristics; performing evolution feature spatial position mapping on the dynamic topological evolution network based on the water flow time series evolution characteristics to construct a three-dimensional hydrological evolution model. The invention realizes timely and accurate river water level monitoring and early warning.

Description

Technical Field

[0001] The present invention relates to the field of water level monitoring and early warning, and in particular to a real-time river water level monitoring and early warning method and system based on deep learning. Background Art

[0002] With the acceleration of urbanization, floods and waterlogging have posed a serious threat to people's lives and property safety. In recent years, frequent rainstorms and floods have also highlighted the importance of real-time monitoring and early warning systems. Timely and accurate river water level monitoring and early warning are essential to reducing the damage caused by floods and waterlogging. Traditional river water level monitoring methods usually rely on manual observation and mechanical measuring equipment, which often have problems such as untimely and inaccurate early warning capabilities, and are unable to prevent and respond in advance. Therefore, an intelligent river water level monitoring and early warning method is needed. Summary of the invention

[0003] In order to solve the above technical problems, the present invention proposes a real-time river water level monitoring and early warning method and system based on deep learning to solve at least one of the above technical problems.

[0004] To achieve the above object, the present invention provides a real-time river water level monitoring and early warning method based on deep learning, comprising the following steps:

[0005] Step S1: Acquire remote sensing images of the target watershed and real-time hydrological data of multiple monitoring points; perform water flow topology analysis on the remote sensing images of the target watershed and construct a water flow topology network;

[0006] Step S2: Analyze the dynamic change characteristics of the watershed structure of the target watershed remote sensing image to generate watershed structure dynamic change data; perform topological situation evolution fitting on the water flow topology network based on the watershed structure dynamic change data to obtain a dynamic topological evolution network;

[0007] Step S3: Perform multi-scale time series trend evolution analysis on real-time hydrological data from multiple monitoring points to generate water flow time series evolution characteristics;

[0008] Step S4: mapping the spatial position of the evolution characteristics of the dynamic topological evolution network based on the time series evolution characteristics of the water flow, and constructing a three-dimensional hydrological evolution model; generating basin hydrodynamic simulation data based on the three-dimensional hydrological evolution model;

[0009] Step S5: Predicting the water level trend of the basin hydrological dynamic simulation data to generate a water level trend prediction value; performing smooth fitting on the water level trend prediction value to construct a water level trend prediction curve;

[0010] Step S6: Perform dynamic water level risk analysis on the water level trend prediction curve based on the preset water level risk assessment standard, generate a water level risk decision strategy, and execute real-time dynamic water level warning operations.

[0011] The present invention performs a topological analysis of the water flow direction on the remote sensing image of the target watershed, determines the flow direction and flow path of the water flow according to the terrain and water flow distribution in the image, constructs a water flow topological network, represents the water flow distribution as a topological structure, provides a visual representation of the water flow path and the relationship between them, performs a dynamic change characteristic analysis of the watershed structure on the remote sensing image of the target watershed, analyzes the impact of changes in factors such as terrain and land use in the watershed on the water flow, and provides a quantitative description of the changes in the watershed structure. Based on the dynamic change data of the watershed structure, the water flow topological network is topologically constructed. The water flow evolution fitting is used to simulate the dynamic changes of water flow in the basin, reflect the temporal and spatial distribution of water flow, conduct multi-scale time series trend evolution analysis on real-time hydrological data of multiple monitoring points, analyze the changing trend of hydrological data, including long-term trends and short-term fluctuations, generate water flow time series evolution characteristics, extract key features of hydrological data, such as change amplitude and periodicity, provide input parameters for subsequent hydrological dynamic simulation, and map the spatial position of evolution characteristics of dynamic topological evolution network based on water flow time series evolution characteristics. The changes of hydrological data correspond to the evolution of topological network, and establish hydrological evolution. The model is constructed to comprehensively represent the hydrological data, topological network and geographic spatial location, providing a comprehensive understanding of the evolution of the hydrological system. Based on the three-dimensional hydrological evolution model, the basin hydrodynamic simulation data is generated to simulate the spatiotemporal distribution and changes of water flow in the basin, providing a basis for water level trend prediction. The water level trend is predicted for the basin hydrodynamic simulation data. Through the model prediction method, the trend of water level changes in the future is obtained, and the prediction results of future water level changes are provided. The water level trend prediction value is smoothed and fitted to the smooth prediction curve to reduce the prediction error and fluctuation, and improve the reliability and stability of the prediction results. Based on the preset water level risk assessment standard, the water level trend prediction curve is subjected to dynamic water level risk analysis. According to the preset risk assessment standard, the water level trend is risk assessed and classified, and a water level risk decision-making strategy is generated. According to the risk assessment results, corresponding water level warning and response strategies are formulated, and measures are taken in advance to deal with the flood risks. Real-time dynamic water level warning operations are performed. According to the water level risk assessment results, water level warning information is released in a timely manner to remind relevant departments and the public to take necessary precautions and response measures to reduce the losses caused by floods.

[0012] Preferably, step S1 comprises the following steps:

[0013] Step S11: obtaining remote sensing images of the target watershed through a drone; obtaining real-time hydrological data of multiple monitoring points based on the hydrological monitoring points;

[0014] Step S12: performing global brightness average calculation on the target watershed remote sensing image to obtain the image brightness mean;

[0015] Step S13: performing brightness balance optimization on the target watershed remote sensing image based on the image brightness mean value to obtain a brightness balanced remote sensing image;

[0016] Step S14: performing river texture feature enhancement processing on the brightness balanced remote sensing image to obtain a texture enhanced remote sensing image;

[0017] Step S15: Perform water flow direction topology analysis on the texture enhanced remote sensing image to construct a water flow topology network.

[0018] The present invention uses a drone to perform remote sensing monitoring of the Yangtze River Basin to obtain high-resolution, wide-coverage remote sensing images of the target basin, provide detailed surface information, obtain real-time hydrological data of multiple monitoring points based on hydrological monitoring points, obtain hydrological parameters at different locations in the basin, including water level, flow, etc., provide real-time hydrological information, perform global brightness average calculation on the remote sensing image of the target basin to obtain the brightness mean of the image, which is used for subsequent brightness balance optimization. The image brightness mean provides a reference value for the overall brightness of the image, which is used to adjust the brightness distribution of the image to make the image clearer. Based on the image brightness mean, the brightness balance optimization of the remote sensing image of the target basin is performed to adjust the brightness distribution of the image, enhance the contrast and details of the image, and obtain a brightness balanced remote sensing image. The remote sensing image improves the visual quality of the image, making the surface features more clearly visible, which is conducive to the subsequent image analysis and water flow topology analysis. The river texture feature enhancement processing is performed on the brightness balanced remote sensing image to highlight the texture features of the river, making the river more conspicuous in the image, and the texture enhanced remote sensing image is obtained to improve the recognizability and distinction of the river. For subsequent water flow topology analysis and water level monitoring, the water flow direction topology analysis is performed on the texture enhanced remote sensing image. The water flow direction is judged and analyzed according to the texture features of the river, the flow direction and flow path of the water flow are determined, and the construction of the water flow topology network provides a visual representation of the water flow path and the relationship between them, understands the distribution and evolution of the water flow, and provides a basis for real-time river water level monitoring and early warning.

[0019] Preferably, the specific steps of step S15 are:

[0020] Perform water flow path edge contour recognition on texture enhanced remote sensing images and extract water flow path edge contour lines;

[0021] Based on the edge contours of the water flow path, the water flow path analysis of the whole basin is carried out to obtain multiple water flow paths;

[0022] Remote sensing images are set to fine-grained grid cells based on texture enhancement;

[0023] Based on the fine-grained grid unit, the texture enhanced remote sensing image is subjected to regional rasterization processing to obtain a regional grid;

[0024] Performing grid cell connection analysis on the regional grid to obtain grid cell connection data;

[0025] Performing logical analysis of water flow directions on multiple water flow paths according to grid unit connection data to generate logical data of water flow directions;

[0026] Perform spatial topological distribution analysis on multiple water flow paths to obtain spatial topological distribution data of water flow paths;

[0027] Based on the logical data of water flow path direction, the global path topology connectivity of the spatial topological distribution data of the water flow path is visualized to construct a water flow topology network.

[0028] The present invention extracts the contour line of the edge of the water flow path by performing water flow path edge contour recognition on the texture enhanced remote sensing image, separates the water flow path from the image, extracts the water flow path edge contour line to accurately outline the shape and boundary of the water flow, and provides accurate data for subsequent water flow path analysis. Based on the water flow path edge contour line, the water flow path analysis of the entire river basin is performed to identify multiple water flow paths, determine the distribution and path of the water flow in the entire river basin, and obtain multiple water flow paths, which provide a specific understanding of the direction of the water flow and provide more comprehensive basic data for water level monitoring and early warning. Based on the texture enhanced remote sensing image, the image is divided into smaller grid units at the fine granularity of the grid unit, which increases the precision and accuracy of the image analysis. The fine granularity division of the grid unit better captures the detailed information in the image, which plays an important role in the recognition and analysis of the water flow path. Based on the fine granularity of the grid unit, the texture enhanced remote sensing image is subjected to regional rasterization processing to convert the image into a regional grid, which is convenient for subsequent analysis and processing. The obtained regional grid provides a structured representation of the image, which facilitates the connection and analysis of the water flow path. The regional grid is connected to the grid unit The connection relationship between grid cells is determined by direct analysis, and the topological structure between grid cells is established. The grid cell connection data provides spatial relationship information between grid cells, which provides a basis for the logical analysis of the flow direction of water flow paths. According to the grid cell connection data, the flow direction logic of multiple water flow paths is analyzed to determine the flow direction of water flow in each grid cell, and the flow direction logic of water flow paths is established. The flow path flow direction logic data provides the flow direction information of water flow in the entire basin, which provides an important reference for water level monitoring and early warning. The spatial topological distribution analysis of multiple water flow paths is performed to understand the relationship and distribution characteristics between water flow paths, revealing the overall structure of the water flow network. The spatial topological distribution data of water flow paths provides spatial correlation information between water flow paths, analyzes the convergence and dispersion of water flows, and visualizes the global path topology connectivity of the spatial topological distribution data of water flow paths based on the flow path flow direction logic data to visualize the relationship between water flow paths and form a water flow topological network. The water flow topological network provides a comprehensive representation of the connection relationship and flow direction between water flow paths, providing an intuitive visualization tool for real-time river water level monitoring and early warning.

[0029] Preferably, the specific steps of step S2 are:

[0030] Step S21: performing multi-time point time series analysis on the remote sensing image of the target watershed to obtain a remote sensing image time series sequence;

[0031] Step S22: Analyze the dynamic change characteristics of the watershed structure on the remote sensing image time series to generate watershed structure dynamic change data;

[0032] Step S23: performing deep dynamic representation learning on the dynamic change data of the watershed structure to obtain the dynamic evolution law of the watershed structure;

[0033] Step S24: fitting the topological situation evolution of the water flow topological network based on the dynamic evolution law of the watershed structure to obtain a dynamic topological evolution network.

[0034] The present invention obtains a series of time series by performing multi-time point time series analysis on remote sensing images of the target watershed, that is, data of remote sensing images at different time points. The remote sensing image time series sequence provides dynamic change information of the watershed, which is used to analyze and monitor the evolution process of the watershed. The remote sensing image time series sequence is subjected to a dynamic change feature analysis of the watershed structure to detect structural changes within the watershed, such as expansion and contraction of the river channel, changes in the confluence area, etc. The generated watershed structure dynamic change data provides a quantitative description of the evolution of the internal structure of the watershed, providing a basis for subsequent analysis and modeling. The watershed structure dynamic change data is subjected to deep dynamic characterization learning to learn the dynamic evolution law of the watershed structure from the data. The deep dynamic characterization learning extracts the potential characteristics of the watershed structure evolution, reveals the structural relationship and change trend between different moments, and performs topological situation evolution fitting on the water flow topology network based on the dynamic evolution law of the watershed structure. The dynamic change of the watershed structure is combined with the water flow topology network to infer the topological situation evolution of the water flow network. The obtained dynamic topological evolution network provides a simulation result of the change of the water flow topology structure over time, which can predict and display the change trend of the water flow path and the change of the topological relationship.

[0035] Preferably, the specific steps of step S3 are:

[0036] Step S31: performing abnormal missing value detection on the real-time hydrological data of multiple monitoring points to obtain hydrological missing values;

[0037] Step S32: interpolating and filling the real-time hydrological data of multiple monitoring points based on the missing hydrological values ​​to obtain optimized real-time hydrological data;

[0038] Step S33: performing stage flow calculation on the optimized real-time hydrological data to obtain stage water flow parameters at multiple monitoring points;

[0039] Step S34: Perform multi-scale time series trend evolution analysis on the water flow parameters at multiple monitoring points to generate water flow time series evolution characteristics.

[0040] The present invention detects abnormal missing values ​​on real-time hydrological data of multiple monitoring points to identify abnormalities and missing values ​​in the data, obtains hydrological missing values ​​to provide information on data quality, helps to judge the reliability and integrity of the data, interpolates and fills the real-time hydrological data of multiple monitoring points based on the hydrological missing values, fills the missing values ​​in the data, and generates complete optimized real-time hydrological data. The optimized real-time hydrological data provides more accurate and continuous hydrological observation data, and provides a reliable basis for subsequent analysis and model establishment. The optimized real-time hydrological data is subjected to stage flow calculation, and the corresponding stage water flow parameters, such as average flow, maximum flow, etc., are calculated according to the water level data. The stage water flow parameters of multiple monitoring points provide a quantitative description of the water flow, and reflect the change of water flow and the distribution of water volume in the basin. The multi-scale time series trend evolution analysis of the stage water flow parameters of multiple monitoring points is performed to reveal the evolution law of water flow at different time scales. The generated water flow time series evolution characteristics include information such as the change trend of water flow and periodic change, and provide a basis for understanding the law and trend of water level change.

[0041] Preferably, the specific steps of step S34 are:

[0042] The multi-scale power spectrum of water flow is obtained by multi-time-frequency decomposition of water flow parameters at multiple monitoring points.

[0043] The periodic fluctuation characteristics are extracted from the multi-scale power spectrum of water flow to obtain the periodic fluctuation characteristics;

[0044] Identify non-stationary variables of periodic fluctuation characteristics to obtain non-stationary variable values ​​of power spectrum;

[0045] According to the non-stationary variable value of the power spectrum, the trend turning point of the multi-scale power spectrum of water flow is marked to obtain the trend turning point;

[0046] The time series trend evolution analysis is performed on the trend turning points to generate the time series evolution characteristics of water flow.

[0047] The present invention decomposes water flow data into components of different scales by performing multi-time-frequency decomposition on water flow parameters at multiple monitoring points, and obtains a multi-scale power spectrum of water flow. The multi-scale power spectrum of water flow provides energy distribution at different frequencies and time scales, reveals the periodic fluctuation characteristics of water flow, extracts periodic fluctuation characteristics of the multi-scale power spectrum of water flow, identifies periodic components and fluctuation characteristics in the power spectrum, and obtains periodic fluctuation characteristics including information such as peaks, troughs, and fluctuation amplitudes, which are used to describe the periodic changes and fluctuations of water flow. Non-stationary variable identification is performed on the periodic fluctuation characteristics to detect non-stationary components in the power spectrum. The stationary variable values ​​and non-stationary variable values ​​provide non-stationary information on water flow fluctuations, understand and analyze the changing characteristics of water flow, mark the trend turning points of the multi-scale power spectrum of water flow according to the non-stationary variable values ​​of the power spectrum, determine the turning points of the trend in the power spectrum, and the trend turning points reveal the turning points of the changing trend of water flow, mark important fluctuation turning positions, perform time series trend evolution analysis on the trend turning points, analyze the trend changes and evolution laws in the water flow time series data, and the generated water flow time series evolution characteristics may include trend growth, decline, periodic changes and other information, providing a time series characteristic description of water flow changes.

[0048] Preferably, the specific steps of step S4 are:

[0049] Step S41: performing spatial correlation analysis on the real-time hydrological data of multiple monitoring points to generate spatial correlation data of the monitoring points;

[0050] Step S42: performing spatial position registration on the time series evolution characteristics of water flow according to the spatial correlation data of the monitoring points to obtain the spatial position of the evolution characteristics;

[0051] Step S43: mapping the time series evolution characteristics of water flow to the dynamic topological evolution network based on the evolution characteristic spatial position, and performing dynamic rendering modeling to construct a three-dimensional hydrological evolution model;

[0052] Step S44: Perform hydrological dynamic simulation calculations for future time periods on the three-dimensional hydrological evolution model to generate basin hydrological dynamic simulation data.

[0053] The present invention performs spatial correlation analysis on real-time hydrological data of multiple monitoring points, analyzes the spatial relationship between different monitoring points, and generates spatial correlation data of monitoring points, which provides the relative position and spatial correlation degree between monitoring points, understands the mutual influence and spatial variation characteristics between different monitoring points, and performs spatial position registration on the time series evolution characteristics of water flow according to the spatial correlation data of monitoring points, aligns the evolution characteristics of different monitoring points to the corresponding spatial positions, and the obtained spatial position of the evolution characteristics provides the spatial positioning of the time series evolution characteristics of water flow, so that the evolution characteristics of different monitoring points can be intuitively compared and analyzed. The feature spatial position maps the time-series evolution characteristics of water flow to the dynamic topological evolution network. The evolution characteristics are combined with the dynamic topological evolution network. Dynamic rendering modeling maps the evolution feature spatial position to the three-dimensional hydrological evolution model to show the spatial distribution and topological changes of hydrological evolution, provide intuitive visualization effects, and perform hydrological dynamic simulation calculations on the three-dimensional hydrological evolution model to simulate the hydrological changes in the future period. The generated basin hydrological dynamic simulation data provides prediction results of hydrological elements such as water level and water flow in the future period, and provides an estimate of future hydrological status for the real-time river water level monitoring and early warning system.

[0054] Preferably, the specific steps of step S5 are:

[0055] Step S51: Analyze the water level distribution at different time points on the basin hydrological dynamic simulation data to obtain water level distribution data at multiple time points;

[0056] Step S52: performing water level trend prediction on the water level distribution data at multiple time points to generate a water level trend prediction value;

[0057] Step S53: Optimizing the water level trend prediction value by hyperparameters to obtain trend prediction hyperparameters;

[0058] Step S54: performing iterative sliding time series prediction based on trend prediction hyperparameters, thereby obtaining a water level trend prediction sequence;

[0059] Step S55: Perform smooth fitting on the water level trend prediction sequence to construct a water level trend prediction curve.

[0060] The present invention analyzes the water level distribution at different time points of the basin hydrological dynamic simulation data to understand the water level distribution at different time points, generates water level distribution data at multiple time points to provide water level information at different times, analyzes the spatiotemporal variation characteristics of the water level, performs water level trend prediction on the water level distribution data at multiple time points, predicts the variation trend of the water level in the future, generates water level trend prediction values ​​to provide the prediction result of the future water level, provides the trend information of the water level variation for the real-time river water level monitoring and early warning system, performs hyperparameter optimization on the water level trend prediction values ​​to adjust the parameters of the prediction model, improves the accuracy and stability of the prediction, and obtains The obtained trend prediction hyperparameters improve the performance of the water level trend prediction model and make the prediction results more reliable and accurate. Based on the trend prediction hyperparameters, iterative sliding time series prediction is performed according to historical data to obtain a prediction sequence of water level trends. The generated water level trend prediction sequence provides detailed prediction results of water level changes in the future, and provides more specific water level trend information for real-time river water level monitoring and early warning systems. The water level trend prediction sequence is smoothed and fitted to smooth the prediction results, eliminating noise and instability in the prediction. The constructed water level trend prediction curve provides a smooth trend of water level changes and more intuitively displays the changing trend and fluctuation characteristics of the water level.

[0061] Preferably, the specific steps of step S6 are:

[0062] Step S61: Calculate the water level change rate of the water level trend prediction curve to obtain the water level trend change rate;

[0063] Step S62: performing a water level peak quantitative analysis based on the water level trend change rate to obtain water level peak trend data;

[0064] Step S63: performing dynamic water level risk analysis on the water level peak trend data based on a preset water level risk assessment standard to obtain water level risk assessment data;

[0065] Step S64: performing risk decision analysis on the water level risk assessment data to generate a water level risk decision strategy;

[0066] Step S65: Execute real-time dynamic water level warning operation through water level risk decision strategy.

[0067] The present invention quantifies the rate of change of the water level by calculating the water level change rate on the water level trend prediction curve. The generated water level trend change rate provides the rate information of the water level change, evaluates the speed of the water level change, performs a water level peak quantitative analysis based on the water level trend change rate to identify the peak characteristics of the water level, and generates water level peak trend data to provide a prediction result of the water level peak, understands the peak period of the water level and the high water level change trend, performs a dynamic water level risk analysis on the water level peak trend data based on a preset water level risk assessment standard to evaluate the risk degree of the water level, and generates water level risk assessment data to provide an assessment result of the water level risk, judges whether the water level reaches a preset risk threshold, performs a risk decision analysis on the water level risk assessment data to formulate a decision strategy for different water level risk situations, and generates a water level risk decision strategy to provide decision suggestions for different water level risk levels, takes corresponding early warning and emergency measures, and performs a real-time dynamic water level early warning operation through a water level risk decision strategy. The real-time dynamic water level early warning is performed according to a preset decision strategy, and the implemented real-time dynamic water level early warning operation can timely issue a water level early warning signal, provide decision support and early warning prompts, and help reduce flood risks.

[0068] In this specification, a real-time river water level monitoring and early warning system based on deep learning is provided, which is used to execute the real-time river water level monitoring and early warning method based on deep learning as described above, including:

[0069] The flow direction topology module is used to obtain remote sensing images of the target watershed and real-time hydrological data of multiple monitoring points; conduct water flow direction topology analysis on the remote sensing images of the target watershed and construct a water flow topology network;

[0070] The situation evolution module is used to analyze the dynamic change characteristics of the watershed structure of the target watershed remote sensing image and generate the dynamic change data of the watershed structure; based on the dynamic change data of the watershed structure, the topological situation evolution of the water flow topology network is fitted to obtain the dynamic topological evolution network;

[0071] The time series trend evolution module is used to perform multi-scale time series trend evolution analysis on real-time hydrological data from multiple monitoring points and generate water flow time series evolution characteristics;

[0072] The hydrological evolution model module is used to map the spatial position of the evolution characteristics of the dynamic topological evolution network based on the time-series evolution characteristics of water flow, and to construct a three-dimensional hydrological evolution model; based on the three-dimensional hydrological evolution model, it generates basin hydrological dynamic simulation data;

[0073] The water level trend prediction module is used to predict the water level trend of the basin hydrological dynamic simulation data and generate water level trend prediction values; perform smooth fitting on the water level trend prediction values ​​and construct a water level trend prediction curve;

[0074] The water level risk warning module is used to perform dynamic water level risk analysis on the water level trend prediction curve based on the preset water level risk assessment standard, generate a water level risk decision strategy, and perform real-time dynamic water level warning operations.

[0075] The present invention provides spatial information and hydrological data of the basin by acquiring remote sensing images of the target basin and real-time hydrological data of multiple monitoring points, performs water flow direction topological analysis to reveal the transmission path and flow direction relationship of the water flow in the basin, and constructs a water flow topological network to form the topological structure of the water flow in the basin, providing a basis for subsequent hydrological analysis. The dynamic change characteristics of the basin structure are analyzed for the remote sensing images of the target basin to understand the changes in factors such as topography and land use in the basin, and the dynamic change data of the basin structure are generated to provide information on the evolution of the basin structure. The changing trends and characteristics in the basin are analyzed, and the topological situation evolution fitting of the water flow topological network is performed to combine the changes in the basin structure with the water flow topological relationship to obtain a dynamic topological evolution network, which provides simulation results of the changes in the water flow topology over time, understands the dynamic changes in the water flow path in the basin, performs multi-scale time series trend evolution analysis on the real-time hydrological data of multiple monitoring points to reveal the changing laws of water flow, generates water flow time series evolution characteristics to provide time series evolution information of hydrological data, understands the trend and characteristics of water flow Periodic changes, based on the time series evolution characteristics of water flow, the dynamic topological evolution network is mapped to the spatial position of the evolution characteristics, combining hydrological data with the topological network, constructing a three-dimensional hydrological evolution model, providing simulation results of hydrological processes in the basin, comprehensively analyzing the changes and interactions of hydrological elements, and predicting water level trends for the basin hydrological dynamic simulation data to understand the trend and peak of water level changes in advance, generating water level trend prediction values, providing prediction results of water levels in the future, formulating corresponding early warning strategies and response measures, smoothing and fitting water level trend prediction values ​​to eliminate noise and mutations, and obtaining a smoother water level prediction curve. Based on the preset water level risk assessment standards, the water level trend prediction curve is subjected to dynamic water level risk analysis to assess the risk level of the water level, generating water level risk decision strategies, providing decision suggestions for different water level risk levels, formulating water level warning and emergency response strategies, and executing real-time dynamic water level warning operations to issue water level warning signals in a timely manner, provide decision support and warning prompts, and help reduce flood risks. BRIEF DESCRIPTION OF THE DRAWINGS

[0076] Figure 1 A schematic diagram of the steps of a real-time river water level monitoring and early warning method based on deep learning of the present invention;

[0077] Figure 2 Detailed implementation flow chart of step S1;

[0078] Figure 3 Detailed implementation flow chart of step S2;

[0079] Figure 4 Detailed implementation flow chart of step S3. DETAILED DESCRIPTION

[0080] It should be understood that the specific embodiments described herein are only used to explain the present invention, and are not used to limit the present invention.

[0081] The present application example provides a real-time river water level monitoring and early warning method and system based on deep learning. The execution subject of the real-time river water level monitoring and early warning method and system based on deep learning includes but is not limited to: mechanical equipment, data processing platform, cloud server node, network upload device, etc. equipped with the system can be regarded as the general computing node of the present application, and the data processing platform includes but is not limited to: at least one of an audio image management system, an information management system, and a cloud data management system.

[0082] See also Figures 1 to 4 The present invention provides a real-time river water level monitoring and early warning method based on deep learning, comprising the following steps:

[0083] Step S1: Acquire remote sensing images of the target watershed and real-time hydrological data of multiple monitoring points; perform water flow topology analysis on the remote sensing images of the target watershed and construct a water flow topology network;

[0084] Step S2: Analyze the dynamic change characteristics of the watershed structure of the target watershed remote sensing image to generate watershed structure dynamic change data; perform topological situation evolution fitting on the water flow topology network based on the watershed structure dynamic change data to obtain a dynamic topological evolution network;

[0085] Step S3: Perform multi-scale time series trend evolution analysis on real-time hydrological data from multiple monitoring points to generate water flow time series evolution characteristics;

[0086] Step S4: mapping the spatial position of the evolution characteristics of the dynamic topological evolution network based on the time series evolution characteristics of the water flow, and constructing a three-dimensional hydrological evolution model; generating basin hydrodynamic simulation data based on the three-dimensional hydrological evolution model;

[0087] Step S5: Predicting the water level trend of the basin hydrological dynamic simulation data to generate a water level trend prediction value; performing smooth fitting on the water level trend prediction value to construct a water level trend prediction curve;

[0088] Step S6: Perform dynamic water level risk analysis on the water level trend prediction curve based on the preset water level risk assessment standard, generate a water level risk decision strategy, and execute real-time dynamic water level warning operations.

[0089] The present invention performs a topological analysis of the water flow direction on the remote sensing image of the target watershed, determines the flow direction and flow path of the water flow according to the terrain and water flow distribution in the image, constructs a water flow topological network, represents the water flow distribution as a topological structure, provides a visual representation of the water flow path and the relationship between them, performs a dynamic change characteristic analysis of the watershed structure on the remote sensing image of the target watershed, analyzes the impact of changes in factors such as terrain and land use in the watershed on the water flow, and provides a quantitative description of the changes in the watershed structure. Based on the dynamic change data of the watershed structure, the water flow topological network is topologically constructed. The water flow evolution fitting is used to simulate the dynamic changes of water flow in the basin, reflect the temporal and spatial distribution of water flow, conduct multi-scale time series trend evolution analysis on real-time hydrological data of multiple monitoring points, analyze the changing trend of hydrological data, including long-term trends and short-term fluctuations, generate water flow time series evolution characteristics, extract key features of hydrological data, such as change amplitude and periodicity, provide input parameters for subsequent hydrological dynamic simulation, and map the spatial position of evolution characteristics of dynamic topological evolution network based on water flow time series evolution characteristics. The changes of hydrological data correspond to the evolution of topological network, and establish hydrological evolution. The model is constructed to comprehensively represent the hydrological data, topological network and geographic spatial location, providing a comprehensive understanding of the evolution of the hydrological system. Based on the three-dimensional hydrological evolution model, the basin hydrodynamic simulation data is generated to simulate the spatiotemporal distribution and changes of water flow in the basin, providing a basis for water level trend prediction. The water level trend is predicted for the basin hydrodynamic simulation data. Through the model prediction method, the trend of water level changes in the future is obtained, and the prediction results of future water level changes are provided. The water level trend prediction value is smoothed and fitted to the smooth prediction curve to reduce the prediction error and fluctuation, and improve the reliability and stability of the prediction results. Based on the preset water level risk assessment standard, the water level trend prediction curve is subjected to dynamic water level risk analysis. According to the preset risk assessment standard, the water level trend is risk assessed and classified, and a water level risk decision-making strategy is generated. According to the risk assessment results, corresponding water level warning and response strategies are formulated, and measures are taken in advance to deal with the flood risks. Real-time dynamic water level warning operations are performed. According to the water level risk assessment results, water level warning information is released in a timely manner to remind relevant departments and the public to take necessary precautions and response measures to reduce the losses caused by floods.

[0090] In the embodiment of the present invention, refer to Figure 1 , is a schematic diagram of the steps of a real-time river water level monitoring and early warning method based on deep learning of the present invention. In this example, the steps of the real-time river water level monitoring and early warning method based on deep learning include:

[0091] Step S1: Acquire remote sensing images of the target watershed and real-time hydrological data of multiple monitoring points; perform water flow topology analysis on the remote sensing images of the target watershed and construct a water flow topology network;

[0092] In this embodiment, high-resolution remote sensing image data of the river basin section to be analyzed in the Yangtze River Basin, such as satellite images, drone images, etc., are collected, and real-time water level, flow and other hydrological observation data of multiple hydrological monitoring stations in the Yangtze River Basin are obtained at the same time. The remote sensing images are finely segmented and extracted, and the extracted water body area is divided into a series of regular grids, each grid unit represents a flow unit, and combined with terrain data such as digital elevation models (DEMs), the flow direction of each grid unit is calculated, and the direction and flow of the water flow are determined according to the elevation difference between adjacent grid units. All grid units and their flow direction relationships are constructed into a complete water flow topology network. The nodes in the network represent grid units, and the edges represent flow direction relationships. According to the results of the water flow direction topological analysis, a water flow topology network is constructed. The water flow topology network is a network structure composed of water flow paths and connection relationships, which reflects the flow of water in the target basin.

[0093] Step S2: Analyze the dynamic change characteristics of the watershed structure of the target watershed remote sensing image to generate watershed structure dynamic change data; perform topological situation evolution fitting on the water flow topology network based on the watershed structure dynamic change data to obtain a dynamic topological evolution network;

[0094] In this embodiment, semantic segmentation or target detection technology based on deep learning is used to identify and extract geographical features in the watershed, such as rivers, lakes, vegetation, etc. By comparing the segmentation results of different time nodes, the dynamic change characteristics of the geographical features, including area, position, morphology, etc., are analyzed to generate time series data describing the dynamic changes of the watershed structure, such as indicators such as river channel extension and lake area reduction. The dynamic change data of the watershed structure are used to update the attributes and connection relationships of each node (grid unit) in the network. According to dynamic changes such as river channel changes and lake growth and decline, the connectivity and weight of the edges in the network are adjusted to simulate the evolution process of the water flow topology structure. Through time series fitting, a complete dynamic topology evolution network is generated to describe the evolution trajectory of the water flow topology on a time scale.

[0095] Step S3: Perform multi-scale time series trend evolution analysis on real-time hydrological data from multiple monitoring points to generate water flow time series evolution characteristics;

[0096] In this embodiment, real-time water level, flow and other hydrological observation data of multiple hydrological monitoring stations in the Yangtze River Basin are collected, and the collected raw data are preprocessed by quality inspection, missing value filling and other preprocessing to ensure the integrity and reliability of the data. For the hydrological data series of each monitoring station, different time scales (such as day, week, month, quarter, year) are used for trend analysis, and various statistical analysis methods, such as linear regression, Mann-Kendall trend test, etc., are used to identify linear or nonlinear trends in the data series, and analyze the changing trends of hydrological elements at different time scales, such as increase or decrease of water level / flow, seasonal fluctuations, etc. For each monitoring station, the trend analysis results at different time scales are integrated into a time series evolution feature vector, which contains the changing trend information of hydrological elements at multiple time scales, and can more comprehensively describe the evolution state of the hydrological time series. The time series evolution feature vectors of all monitoring stations are aggregated to form a complete hydrological time series evolution feature database.

[0097] Step S4: mapping the spatial position of the evolution characteristics of the dynamic topological evolution network based on the time series evolution characteristics of the water flow, and constructing a three-dimensional hydrological evolution model; generating basin hydrodynamic simulation data based on the three-dimensional hydrological evolution model;

[0098] In this embodiment, the temporal evolution characteristic data of the hydrological monitoring sites are used to associate with the corresponding geographic spatial locations (monitoring site coordinates), and spatial interpolation or other geographic information processing methods are used to expand the discrete monitoring point data to the grid space of the entire basin. The dynamic topological evolution network is used as the basic framework, and the dynamic changes in the basin structure and the temporal evolution characteristics of the water flow are mapped to the network nodes and edges. Using 3D visualization technology, a three-dimensional model that can display the spatiotemporal evolution of the basin hydrology is constructed. The model can dynamically display the changes in hydrological elements (such as rivers, lakes, etc.) within the basin and the spatiotemporal distribution of hydrological parameters (such as water level, flow). Based on the three-dimensional hydrological evolution model, numerical simulation or machine learning methods are used to deduce the hydrological dynamic process within the basin. The simulation results include the spatiotemporal changes in hydrological parameters such as water level, flow, water velocity, and water flow direction, and generate high spatiotemporal resolution hydrological dynamic simulation data covering the entire basin.

[0099] Step S5: Predicting the water level trend of the basin hydrological dynamic simulation data to generate a water level trend prediction value; performing smooth fitting on the water level trend prediction value to construct a water level trend prediction curve;

[0100] In this embodiment, based on the basin hydrological and dynamic simulation data, the water level time series data of each monitoring point are extracted, and time series analysis is performed on these water level time series data to identify the trend components therein. The water level trend is predicted by time series prediction methods such as linear regression, ARIMA, and exponential smoothing, and the water level trend is generated to generate water level trend prediction values ​​for a period of time in the future. The predicted water level trend values ​​are smoothed to remove the noise components. Commonly used smoothing fitting methods include moving average smoothing, cubic spline smoothing, local weighted regression smoothing, etc. Through smooth fitting, a smooth water level trend prediction curve is constructed.

[0101] Step S6: Perform dynamic water level risk analysis on the water level trend prediction curve based on the preset water level risk assessment standard, generate a water level risk decision strategy, and execute real-time dynamic water level warning operations.

[0102] In this embodiment, water level risk assessment standards applicable to the current basin are formulated based on the geographical characteristics of the basin, historical water conditions data and flood control standards. These standards include thresholds such as water level warning lines, flood warning lines, and super-warning water levels. These thresholds are reasonably set to ensure that different levels of water level risks can be identified and responded to in a timely and effective manner. The water level trend prediction curve is compared and analyzed with the preset water level risk assessment standards to identify the intersection of the prediction curve and each risk threshold, determine the water level risk level that will appear in the future, and promptly discover potential water level super-warning risks, so as to reserve sufficient response time for subsequent emergency plans. According to the results of dynamic risk analysis, a corresponding emergency response strategy is formulated, which includes specific measures such as warning release, emergency material preparation, and evacuation under different risk levels, so as to ensure that effective response actions can be taken in a timely manner when facing different degrees of water level risks. The formulated water level risk decision-making strategy is converted into an operational warning release and emergency response process, and a linkage mechanism for information monitoring, warning release, emergency response and other links is established to monitor the changes in the water level trend prediction curve in real time. Once the warning standard is reached, the corresponding emergency plan is immediately initiated.

[0103] In this embodiment, refer to Figure 2 , is a flowchart of detailed implementation steps of step S1. In this embodiment, the detailed implementation steps of step S1 include:

[0104] Step S11: obtaining remote sensing images of the target watershed through a drone; obtaining real-time hydrological data of multiple monitoring points based on the hydrological monitoring points;

[0105] Step S12: performing global brightness average calculation on the target watershed remote sensing image to obtain the image brightness mean;

[0106] Step S13: performing brightness balance optimization on the target watershed remote sensing image based on the image brightness mean value to obtain a brightness balanced remote sensing image;

[0107] Step S14: performing river texture feature enhancement processing on the brightness balanced remote sensing image to obtain a texture enhanced remote sensing image;

[0108] Step S15: Perform water flow direction topology analysis on the texture enhanced remote sensing image to construct a water flow topology network.

[0109] In this embodiment, an unmanned aerial vehicle platform is used to carry out remote sensing monitoring of the Yangtze River Basin. The unmanned aerial vehicle is equipped with remote sensing equipment, such as a camera or a multispectral sensor, to perform aerial photography or remote sensing image acquisition of the target basin. The maneuverability and flexibility of the unmanned aerial vehicle are used to obtain high-resolution, multi-angle remote sensing image data. Hydrological monitoring points, which are water level measuring stations, flow measuring stations or other hydrological observation equipment, are set up in the target basin. These monitoring points are used to monitor hydrological data in real time, such as water level, flow, rainfall, etc. The real-time data obtained through the monitoring points are used for subsequent hydrological analysis and decision support. The global brightness calculation is performed on the loaded remote sensing image, that is, the average brightness value of all pixels in the image is calculated. This is achieved by summing the brightness value of each pixel and dividing it by the total number of pixels. The brightness value of each pixel in the image is adjusted to be close to the overall Brightness mean, through this brightness balance optimization, the problem of excessive local brightness differences in the image is reduced. For the river part in the brightness balanced remote sensing image, texture feature enhancement processing is performed. By applying image enhancement algorithms, such as filtering and sharpening, the texture features of the river are highlighted and enhanced to make it clearer and more recognizable. The texture enhanced remote sensing image is used to perform topological analysis of the water flow direction. This is achieved through image processing and analysis algorithms, such as edge detection and connectivity analysis. The goal is to extract the topological structure of the water flow, including the main river channels, tributaries and confluence areas. According to the results of the topological analysis of the water flow direction, the extracted water flow topological structure is connected and organized to construct a water flow topological network. This network is represented in the form of a graph, in which the relationship between the river channel and the water flow is represented as the edge of the graph, and the node represents the intersection or branch point of the river channel.

[0110] In this embodiment, the specific steps of step S15 are:

[0111] Perform water flow path edge contour recognition on texture enhanced remote sensing images and extract water flow path edge contour lines;

[0112] Based on the edge contours of the water flow path, the water flow path analysis of the whole basin is carried out to obtain multiple water flow paths;

[0113] Remote sensing images are set to fine-grained grid cells based on texture enhancement;

[0114] Based on the fine-grained grid unit, the texture enhanced remote sensing image is subjected to regional rasterization processing to obtain a regional grid;

[0115] Performing grid cell connection analysis on the regional grid to obtain grid cell connection data;

[0116] Performing logical analysis of water flow directions on multiple water flow paths according to grid unit connection data to generate logical data of water flow directions;

[0117] Perform spatial topological distribution analysis on multiple water flow paths to obtain spatial topological distribution data of water flow paths;

[0118] Based on the logical data of water flow path direction, the global path topology connectivity of the spatial topological distribution data of the water flow path is visualized to construct a water flow topology network.

[0119] In this embodiment, edge detection and contour extraction are performed on the texture-enhanced remote sensing image to identify the clear edge contour line of the river, extract the edge boundary information of the water flow path, and perform water flow path analysis in the entire basin based on the extracted water flow path edge contour line. Through topological analysis, connectivity analysis and other methods, multiple complete water flow paths are identified, and the texture-enhanced remote sensing image is divided into fine-grained grid units. The scale of the grid unit is appropriately selected according to the actual situation to achieve a suitable resolution. The texture-enhanced remote sensing image is subjected to regional gridding processing, and the pixel attributes inside each grid unit are unified to obtain regional grid data, and the adjacent grid units are analyzed. The spatial connection relationship between grid units is obtained to obtain complete grid unit connection data, which provides a basis for subsequent flow direction analysis. Combined with the grid unit connection data, the flow direction logic analysis of the multiple flow paths identified previously is performed to determine the direction of each grid unit connection, determine the flow direction of each grid unit on the flow path, determine the specific flow direction of each flow path, form the flow path flow direction logic data, analyze the spatial distribution topological relationship of multiple flow paths, obtain the specific spatial distribution data of the flow path within the basin, merge the flow path flow direction logic data and the spatial topological distribution data, construct a complete water flow topology network model, and visualize the hydrological system of the entire basin.

[0120] In this embodiment, refer to Figure 3 , is a flowchart of detailed implementation steps of step S2. In this embodiment, the detailed implementation steps of step S2 include:

[0121] Step S21: performing multi-time point time series analysis on the remote sensing image of the target watershed to obtain a remote sensing image time series sequence;

[0122] Step S22: Analyze the dynamic change characteristics of the watershed structure on the remote sensing image time series to generate watershed structure dynamic change data;

[0123] Step S23: performing deep dynamic representation learning on the dynamic change data of the watershed structure to obtain the dynamic evolution law of the watershed structure;

[0124] Step S24: fitting the topological situation evolution of the water flow topological network based on the dynamic evolution law of the watershed structure to obtain a dynamic topological evolution network.

[0125] In this embodiment, remote sensing image data of multiple time points including the target watershed are obtained. These time points are remote sensing images within adjacent time periods to capture the dynamic changes of the watershed. The remote sensing images at each time point are preprocessed, including denoising, radiation correction, atmospheric correction, etc. The preprocessed remote sensing images are combined in chronological order to form a time series of remote sensing images. The image at each time point is represented as a matrix or a multi-band image, and the time series is composed of images at multiple time points. The dynamic changes of watershed landforms, river channels, vegetation and other elements in the time series remote sensing images are compared and analyzed, and the structural change characteristics of each element within the watershed in the time series are identified. A data set describing the dynamic changes of the watershed structure is generated, and a deep learning time series analysis method is used to model the dynamic change data of the watershed structure. Through deep learning training, the potential laws of the evolution of the watershed structure are extracted and characterized. The model or parameters that can describe the dynamic evolution mechanism of the basin structure are obtained, the water flow topology network is integrated with the dynamic evolution law of the basin structure, and the topological structure of the water flow topology network is predicted and simulated based on the dynamic evolution law. By applying the dynamic evolution law of the basin structure to the topological structure of the water flow topology network, the topological structure of the network changes over time. According to the fitted topological situation evolution, network science related theories, such as graph theory and complex network analysis, are used to fit and predict the evolution trend of the basin water system structure, track and analyze the changes in river channels and water system structures in the basin, identify the dynamic evolution processes such as river changes and water system remodeling in the basin, generate a dynamic topological evolution network, which is represented by the changes in nodes and edges in the water flow topology network at different time points, and use graph theory and network analysis methods to construct and analyze the dynamic topological evolution network.

[0126] In this embodiment, reference Figure 4 The above is a schematic flow chart of the detailed implementation steps of step S3. In this embodiment, the detailed implementation steps of step S3 include:

[0127] Step S31: performing abnormal missing value detection on the real-time hydrological data of multiple monitoring points to obtain hydrological missing values;

[0128] Step S32: interpolating and filling the real-time hydrological data of multiple monitoring points based on the missing hydrological values ​​to obtain optimized real-time hydrological data;

[0129] Step S33: performing stage flow calculation on the optimized real-time hydrological data to obtain stage water flow parameters at multiple monitoring points;

[0130] Step S34: Perform multi-scale time series trend evolution analysis on the water flow parameters at multiple monitoring points to generate water flow time series evolution characteristics.

[0131] In this embodiment, abnormal missing value detection is performed on real-time hydrological data. Statistical methods, time series analysis, machine learning and other technologies are used to detect abnormal values ​​and missing values ​​in the data. Abnormal values ​​are abnormal data caused by equipment failure or abnormal conditions, while missing values ​​are data missing due to data collection or transmission problems. According to the characteristics of hydrological data and the distribution of missing values, a suitable interpolation method is selected. Commonly used interpolation methods include linear interpolation, spline interpolation, Kriging interpolation, etc. The selected interpolation method should be able to accurately estimate missing values ​​and maintain data continuity. The selected interpolation method is used to fill in the missing values ​​in the hydrological data. According to the existing data points around, the missing values ​​are interpolated. Calculation is used to estimate missing values. The filled data will replace the missing values ​​in the original data to obtain optimized real-time hydrological data. The optimized real-time hydrological data sequence is segmented according to a certain time scale (such as day, week, month, etc.). For each time period, the flow calculation formula is used to calculate the corresponding water flow parameters, such as average flow, peak flow, etc., to form a segmented water flow parameter sequence containing multiple monitoring points. The time series analysis method is used for trend analysis to identify the evolution characteristics of water flow at each monitoring point on different time scales, such as seasonal changes, long-term trends, etc. The time series evolution characteristics of multi-point water flow are comprehensively analyzed to characterize the dynamic evolution law of the entire basin hydrological system.

[0132] In this embodiment, the specific steps of step S34 are:

[0133] The multi-scale power spectrum of water flow is obtained by multi-time-frequency decomposition of water flow parameters at multiple monitoring points.

[0134] The periodic fluctuation characteristics are extracted from the multi-scale power spectrum of water flow to obtain the periodic fluctuation characteristics;

[0135] Identify non-stationary variables of periodic fluctuation characteristics to obtain non-stationary variable values ​​of power spectrum;

[0136] According to the non-stationary variable value of the power spectrum, the trend turning point of the multi-scale power spectrum of water flow is marked to obtain the trend turning point;

[0137] The time series trend evolution analysis is performed on the trend turning points to generate the time series evolution characteristics of water flow.

[0138] In this embodiment,

[0139] In this embodiment, step S4 includes the following steps:

[0140] Step S41: performing spatial correlation analysis on the real-time hydrological data of multiple monitoring points to generate spatial correlation data of the monitoring points;

[0141] Step S42: performing spatial position registration on the time series evolution characteristics of water flow according to the spatial correlation data of the monitoring points to obtain the spatial position of the evolution characteristics;

[0142] Step S43: mapping the time series evolution characteristics of water flow to the dynamic topological evolution network based on the evolution characteristic spatial position, and performing dynamic rendering modeling to construct a three-dimensional hydrological evolution model;

[0143] Step S44: Perform hydrological dynamic simulation calculations for future time periods on the three-dimensional hydrological evolution model to generate basin hydrological dynamic simulation data.

[0144] In this embodiment, a time-frequency analysis method (such as wavelet transform, Hilbert-Huang transform, etc.) is used to perform multi-time-scale spectrum decomposition on these water flow data, and the water flow parameters are decomposed into components of different scales and frequencies. A multi-scale power spectrum describing the periodic characteristics of water flow at different time scales is obtained, and the multi-scale power spectrum of water flow is analyzed to identify the significant periodic fluctuation characteristics at each time scale, extract the relevant characteristics of these significant periodic fluctuations, such as the dominant period, power intensity, etc., evaluate the stability of the periodic fluctuation characteristics in time, quantify the degree of non-stationarity for the periodic fluctuation characteristics with non-stationarity, obtain the non-stationary variable value of the power spectrum, analyze the multi-scale power spectrum of water flow, identify the key time points that show trend turning at different time scales, mark these trend turning points, prepare for the subsequent time series evolution analysis, perform segmented analysis on the time series of water flow parameters at multiple monitoring points, characterize the trend change characteristics of each segment, comprehensively analyze the trend evolution characteristics of each monitoring point, and summarize the time series dynamic evolution law of water flow at the basin scale.

[0145] In this embodiment, step S5 includes the following steps:

[0146] Step S51: Analyze the water level distribution at different time points on the basin hydrological dynamic simulation data to obtain water level distribution data at multiple time points;

[0147] Step S52: performing water level trend prediction on the water level distribution data at multiple time points to generate a water level trend prediction value;

[0148] Step S53: Optimizing the water level trend prediction value by hyperparameters to obtain trend prediction hyperparameters;

[0149] Step S54: performing iterative sliding time series prediction based on trend prediction hyperparameters, thereby obtaining a water level trend prediction sequence;

[0150] Step S55: Perform smooth fitting on the water level trend prediction sequence to construct a water level trend prediction curve.

[0151] In this embodiment, a spatial statistical analysis method (such as Kriging interpolation, covariance function fitting, etc.) is used to perform spatial correlation analysis on the hydrological data of these monitoring points to obtain spatial correlation data of the monitoring points that describe the spatial relationship between the monitoring points, such as correlation coefficient, spatial distance, etc., and these time series evolution characteristics are spatially mapped and aligned to determine the specific position of each time series evolution feature in the basin space, generate evolution feature spatial position data, map the spatial position of the water flow time series evolution feature to the corresponding node in the dynamic topological evolution network, and map the evolution feature to the spatial position of the node in the dynamic topological evolution network by matching the spatial position information of the evolution feature with the node position of the dynamic topological evolution network. The dynamic topological evolution network and the evolution characteristics are corresponded to each other, and dynamic rendering modeling is performed by using the correspondence between the dynamic topological evolution network and the evolution characteristics. According to the changes in the evolution characteristics, the nodes in the network are dynamically changed accordingly to simulate the spatiotemporal evolution process of the hydrological system and construct a three-dimensional hydrological evolution model. The parameters of the constructed three-dimensional hydrological evolution model are set, including watershed characteristics, terrain data, initial conditions, etc. Based on the set model parameters, hydrological dynamic simulation calculations for future periods are performed. Using appropriate hydrological models, taking into account factors such as rainfall, evaporation, and groundwater flow, the hydrological process of the basin in the future period is simulated, and the corresponding hydrological dynamic simulation data are generated.

[0152] In this embodiment, the specific steps of step S6 are:

[0153] Step S61: Calculate the water level change rate of the water level trend prediction curve to obtain the water level trend change rate;

[0154] Step S62: performing a water level peak quantitative analysis based on the water level trend change rate to obtain water level peak trend data;

[0155] Step S63: performing dynamic water level risk analysis on the water level peak trend data based on a preset water level risk assessment standard to obtain water level risk assessment data;

[0156] Step S64: performing risk decision analysis on the water level risk assessment data to generate a water level risk decision strategy;

[0157] Step S65: Execute real-time dynamic water level warning operation through water level risk decision strategy.

[0158] In this embodiment, the prediction curve is differentiated to obtain the water level change rate at each moment, that is, the water level trend change rate. The water level change rate is obtained by calculating the difference between the water levels at adjacent time points and dividing it by the time interval. The water level trend change rate data is used to detect the occurrence of the water level peak. The position of the water level peak is determined by finding the local maximum value on the change rate curve. The time point of the water level peak and the corresponding water level value are extracted to form water level peak trend data. These data are used to analyze the change trend and characteristics of the water level peak. According to specific circumstances and needs, preset water level risk assessment standards are formulated. These standards include different water level levels, corresponding risk levels and thresholds, etc. According to the water level peak trend data and the preset water level risk assessment standards According to the water level risk assessment data and actual needs, indicators for risk decision-making are determined, such as risk level, early warning threshold, etc. Based on the risk decision indicators, water level risk assessment data are analyzed, and response strategies and decision-making actions under different risk levels are formulated, such as early warning measures and emergency response plans for different risk levels. According to the results of risk decision analysis, water level risk decision strategies are generated, including early warning levels, triggering conditions, response measures, etc.

[0159] In this embodiment, a real-time river water level monitoring and early warning system based on deep learning is provided, which is used to execute the real-time river water level monitoring and early warning method based on deep learning as described above, including:

[0160] The flow direction topology module is used to obtain remote sensing images of the target watershed and real-time hydrological data of multiple monitoring points; conduct water flow direction topology analysis on the remote sensing images of the target watershed and construct a water flow topology network;

[0161] The situation evolution module is used to analyze the dynamic change characteristics of the watershed structure of the target watershed remote sensing image and generate the dynamic change data of the watershed structure; based on the dynamic change data of the watershed structure, the topological situation evolution of the water flow topology network is fitted to obtain the dynamic topological evolution network;

[0162] The time series trend evolution module is used to perform multi-scale time series trend evolution analysis on real-time hydrological data from multiple monitoring points and generate water flow time series evolution characteristics;

[0163] The hydrological evolution model module is used to map the spatial position of the evolution characteristics of the dynamic topological evolution network based on the time-series evolution characteristics of water flow, and to construct a three-dimensional hydrological evolution model; based on the three-dimensional hydrological evolution model, it generates basin hydrological dynamic simulation data;

[0164] The water level trend prediction module is used to predict the water level trend of the basin hydrological dynamic simulation data and generate the water level trend prediction value; perform smooth fitting on the water level trend prediction value and construct the water level trend prediction curve;

[0165] The water level risk warning module is used to perform dynamic water level risk analysis on the water level trend prediction curve based on the preset water level risk assessment standard, generate a water level risk decision strategy, and perform real-time dynamic water level warning operations.

[0166] The present invention provides spatial information and hydrological data of the basin by acquiring remote sensing images of the target basin and real-time hydrological data of multiple monitoring points, performs water flow direction topological analysis to reveal the transmission path and flow direction relationship of the water flow in the basin, and constructs a water flow topological network to form the topological structure of the water flow in the basin, providing a basis for subsequent hydrological analysis. The dynamic change characteristics of the basin structure are analyzed for the remote sensing images of the target basin to understand the changes in factors such as topography and land use in the basin, and the dynamic change data of the basin structure are generated to provide information on the evolution of the basin structure. The changing trends and characteristics in the basin are analyzed, and the topological situation evolution fitting of the water flow topological network is performed to combine the changes in the basin structure with the water flow topological relationship to obtain a dynamic topological evolution network, which provides simulation results of the changes in the water flow topology over time, understands the dynamic changes in the water flow path in the basin, performs multi-scale time series trend evolution analysis on the real-time hydrological data of multiple monitoring points to reveal the changing laws of water flow, generates water flow time series evolution characteristics to provide time series evolution information of hydrological data, understands the trend and characteristics of water flow Periodic changes, based on the time series evolution characteristics of water flow, the dynamic topological evolution network is mapped to the spatial position of the evolution characteristics, combining hydrological data with the topological network, constructing a three-dimensional hydrological evolution model, providing simulation results of hydrological processes in the basin, comprehensively analyzing the changes and interactions of hydrological elements, and predicting water level trends for the basin hydrological dynamic simulation data to understand the trend and peak of water level changes in advance, generating water level trend prediction values, providing prediction results of water levels in the future, formulating corresponding early warning strategies and response measures, smoothing and fitting water level trend prediction values ​​to eliminate noise and mutations, and obtaining a smoother water level prediction curve. Based on the preset water level risk assessment standards, the water level trend prediction curve is subjected to dynamic water level risk analysis to assess the risk level of the water level, generating water level risk decision strategies, providing decision suggestions for different water level risk levels, formulating water level warning and emergency response strategies, and executing real-time dynamic water level warning operations to issue water level warning signals in a timely manner, provide decision support and warning prompts, and help reduce flood risks.

[0167] Therefore, the embodiments should be regarded as illustrative and non-restrictive from all points, and the scope of the present invention is limited by the appended claims rather than the above description, and it is therefore intended that all changes falling within the meaning and range of equivalent elements of the application documents are included in the present invention.

[0168] The above is only a specific embodiment of the present invention, so that those skilled in the art can understand or implement the present invention. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to the embodiments shown herein, but should conform to the widest scope consistent with the principles and novel features invented herein.

Claims

1. A real-time river water level monitoring and early warning method based on deep learning, characterized in that: The following steps are involved: Step S1: Obtain remote sensing images of the target watershed and real-time hydrological data of multiple monitoring points; Conduct water flow topology analysis on remote sensing images of the target watershed and construct a water flow topology network; Step S2: Analyze the dynamic change characteristics of the watershed structure of the target watershed remote sensing image to generate watershed structure dynamic change data; perform topological situation evolution fitting on the water flow topology network based on the watershed structure dynamic change data to obtain a dynamic topological evolution network; The specific steps of step S2 are: Step S21: performing multi-time point time series analysis on the remote sensing image of the target watershed to obtain a remote sensing image time series sequence; Step S22: Analyze the dynamic change characteristics of the watershed structure on the remote sensing image time series to generate watershed structure dynamic change data; Step S23: performing deep dynamic representation learning on the dynamic change data of the watershed structure to obtain the dynamic evolution law of the watershed structure; Step S24: fitting the topological situation evolution of the water flow topological network based on the dynamic evolution law of the watershed structure to obtain a dynamic topological evolution network; Step S3: Perform multi-scale time series trend evolution analysis on real-time hydrological data from multiple monitoring points to generate water flow time series evolution characteristics; Step S4: mapping the spatial position of the evolution characteristics of the dynamic topological evolution network based on the time series evolution characteristics of the water flow, and constructing a three-dimensional hydrological evolution model; generating basin hydrodynamic simulation data based on the three-dimensional hydrological evolution model; Step S5: Predicting the water level trend of the basin hydrological dynamic simulation data to generate a water level trend prediction value; performing smooth fitting on the water level trend prediction value to construct a water level trend prediction curve; Step S6: Perform dynamic water level risk analysis on the water level trend prediction curve based on the preset water level risk assessment standard, generate a water level risk decision strategy, and execute real-time dynamic water level warning operations.

2. The real-time river water level monitoring and early warning method based on deep learning according to claim 1 is characterized in that: The specific steps of step S1 are: Step S11: obtaining remote sensing images of the target watershed through a drone; obtaining real-time hydrological data of multiple monitoring points based on the hydrological monitoring points; Step S12: performing global brightness average calculation on the target watershed remote sensing image to obtain the image brightness mean; Step S13: performing brightness balance optimization on the target watershed remote sensing image based on the image brightness mean value to obtain a brightness balanced remote sensing image; Step S14: performing river texture feature enhancement processing on the brightness balanced remote sensing image to obtain a texture enhanced remote sensing image; Step S15: Perform water flow direction topology analysis on the texture enhanced remote sensing image to construct a water flow topology network.

3. The real-time river water level monitoring and early warning method based on deep learning according to claim 2 is characterized in that: The specific steps of step S15 are: Perform water flow path edge contour recognition on texture enhanced remote sensing images and extract water flow path edge contour lines; Based on the edge contours of the water flow path, the water flow path analysis of the whole basin is carried out to obtain multiple water flow paths; Remote sensing images are set to fine-grained grid cells based on texture enhancement; Based on the fine-grained grid unit, the texture enhanced remote sensing image is subjected to regional rasterization processing to obtain a regional grid; Performing grid cell connection analysis on the regional grid to obtain grid cell connection data; Performing logical analysis of water flow directions on multiple water flow paths according to grid unit connection data to generate logical data of water flow directions; Perform spatial topological distribution analysis on multiple water flow paths to obtain spatial topological distribution data of water flow paths; Based on the logical data of water flow path direction, the global path topology connectivity of the spatial topological distribution data of the water flow path is visualized to construct a water flow topology network.

4. The real-time river water level monitoring and early warning method based on deep learning according to claim 1 is characterized in that: The specific steps of step S3 are: Step S31: performing abnormal missing value detection on the real-time hydrological data of multiple monitoring points to obtain hydrological missing values; Step S32: interpolating and filling the real-time hydrological data of multiple monitoring points based on the missing hydrological values ​​to obtain optimized real-time hydrological data; Step S33: performing stage flow calculation on the optimized real-time hydrological data to obtain stage water flow parameters at multiple monitoring points; Step S34: Perform multi-scale time series trend evolution analysis on the water flow parameters at multiple monitoring points to generate water flow time series evolution characteristics.

5. The real-time river water level monitoring and early warning method based on deep learning according to claim 4 is characterized in that: The specific steps of step S34 are: The multi-scale power spectrum of water flow is obtained by multi-time-frequency decomposition of water flow parameters at multiple monitoring points. The periodic fluctuation characteristics are extracted from the multi-scale power spectrum of water flow to obtain the periodic fluctuation characteristics; Identify non-stationary variables of periodic fluctuation characteristics to obtain non-stationary variable values ​​of power spectrum; According to the non-stationary variable value of the power spectrum, the trend turning point of the multi-scale power spectrum of water flow is marked to obtain the trend turning point; The time series trend evolution analysis is performed on the trend turning points to generate the time series evolution characteristics of water flow.

6. The real-time river water level monitoring and early warning method based on deep learning according to claim 1 is characterized in that: The specific steps of step S4 are: Step S41: performing spatial correlation analysis on the real-time hydrological data of multiple monitoring points to generate spatial correlation data of the monitoring points; Step S42: performing spatial position registration on the time series evolution characteristics of water flow according to the spatial correlation data of the monitoring points to obtain the spatial position of the evolution characteristics; Step S43: mapping the time series evolution characteristics of water flow to the dynamic topological evolution network based on the evolution characteristic spatial position, and performing dynamic rendering modeling to construct a three-dimensional hydrological evolution model; Step S44: Perform hydrological dynamic simulation calculations for future time periods on the three-dimensional hydrological evolution model to generate basin hydrological dynamic simulation data.

7. The real-time river water level monitoring and early warning method based on deep learning according to claim 1 is characterized in that: The specific steps of step S5 are: Step S51: Analyze the water level distribution at different time points on the basin hydrological dynamic simulation data to obtain water level distribution data at multiple time points; Step S52: performing water level trend prediction on the water level distribution data at multiple time points to generate a water level trend prediction value; Step S53: Optimizing the water level trend prediction value by hyperparameters to obtain trend prediction hyperparameters; Step S54: performing iterative sliding time series prediction based on trend prediction hyperparameters, thereby obtaining a water level trend prediction sequence; Step S55: Perform smooth fitting on the water level trend prediction sequence to construct a water level trend prediction curve.

8. The real-time river water level monitoring and early warning method based on deep learning according to claim 1 is characterized in that: The specific steps of step S6 are: Step S61: Calculate the water level change rate of the water level trend prediction curve to obtain the water level trend change rate; Step S62: performing a water level peak quantitative analysis based on the water level trend change rate to obtain water level peak trend data; Step S63: performing dynamic water level risk analysis on the water level peak trend data based on a preset water level risk assessment standard to obtain water level risk assessment data; Step S64: performing risk decision analysis on the water level risk assessment data to generate a water level risk decision strategy; Step S65: Execute real-time dynamic water level warning operation through water level risk decision strategy.

9. A real-time river water level monitoring and early warning system based on deep learning, characterized in that: The method for real-time river water level monitoring and early warning based on deep learning as claimed in claim 1 comprises: The flow direction topology module is used to obtain remote sensing images of the target watershed and real-time hydrological data of multiple monitoring points; conduct water flow direction topology analysis on the remote sensing images of the target watershed and construct a water flow topology network; The situation evolution module is used to analyze the dynamic change characteristics of the watershed structure of the target watershed remote sensing image and generate the dynamic change data of the watershed structure; based on the dynamic change data of the watershed structure, the topological situation evolution of the water flow topology network is fitted to obtain the dynamic topological evolution network; The time series trend evolution module is used to perform multi-scale time series trend evolution analysis on real-time hydrological data from multiple monitoring points and generate water flow time series evolution characteristics; The hydrological evolution model module is used to map the spatial position of the evolution characteristics of the dynamic topological evolution network based on the time-series evolution characteristics of water flow, and to construct a three-dimensional hydrological evolution model; based on the three-dimensional hydrological evolution model, it generates basin hydrodynamic simulation data; The water level trend prediction module is used to predict the water level trend of the basin hydrological dynamic simulation data and generate water level trend prediction values; perform smooth fitting on the water level trend prediction values ​​and construct a water level trend prediction curve; The water level risk warning module is used to perform dynamic water level risk analysis on the water level trend prediction curve based on the preset water level risk assessment standard, generate a water level risk decision strategy, and perform real-time dynamic water level warning operations.

Images