A basin water level intelligent early warning method

Abstract

The present application relates to the technical field of water level early warning, in particular to a basin water level intelligent early warning method, comprising: obtaining a water level dynamic curve based on a section flow process line; performing periodic extension on a non-observed period water level sequence of the basin based on a time series extension algorithm to obtain an extended water level dynamic curve; verifying the extended water level dynamic curve based on a water level simulation method to obtain a 24-hour water level dynamic curve; setting a multi-level water level early warning threshold based on the 24-hour water level dynamic curve, and automatically sending a hierarchical early warning signal when an actual monitoring water level exceeds a corresponding level of the multi-level water level early warning threshold. The present application generates a 24-hour water level dynamic curve by using time series extension and water level simulation methods to interpolate and convert daily missing data, avoids the occurrence of missing water level maximum values in other periods, and makes the water level early warning threshold more reliable and the water level early warning effect better.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of water level early warning, and particularly relates to a basin water level intelligent early warning method. BACKGROUND

[0002] The traditional method adopts manual observation, which is generally observed once an hour from 8am to 17pm, and not observed at other times, resulting in unreliable water level early warning threshold and low water level early warning accuracy. The traditional method usually adopts a single water level threshold, lacks a multi-level early warning mechanism, and may rely on manual judgment, causing slow response speed. The traditional method may directly use original rainfall data or forecast data, lacks correction of rainfall data, and may rely on fixed data update frequency or long-delay real-time data, resulting in water level prediction lag. The traditional method usually relies on historical water level data and a single hydrological model, which may fail to fully consider the complex terrain and rainfall conditions within the basin, resulting in low hydrological analysis accuracy. SUMMARY

[0003] The present application solves the technical problems of the prior art, and provides a basin water level intelligent early warning method.

[0004] The technical scheme adopted to solve the above technical problems is as follows: a basin water level intelligent early warning method, comprising:

[0005] Obtain terrain data of a basin of a pumped storage power station, construct a model based on a digital elevation model (DEM) and the terrain data to obtain a three-dimensional terrain model of the basin;

[0006] Perform hydrological analysis on the basin based on the three-dimensional terrain model of the basin to obtain hydrological data corresponding to the basin;

[0007] Obtain actual rainfall data of the basin, obtain predicted rainfall data of the basin based on rainfall numerical prediction products, and correct the actual rainfall data and the predicted rainfall data based on a Bayesian probability correction algorithm to obtain corrected rainfall data;

[0008] Input the hydrological data and the corrected rainfall data into a motion wave hydrological model to obtain a cross-section flow process line of the basin, and obtain a water level dynamic curve based on the cross-section flow process line;

[0009] Perform period extension on a non-observation period water level sequence of the basin based on a time series extension algorithm to obtain an extended water level dynamic curve, and verify the extended water level dynamic curve based on a water level simulation method to obtain a 24-hour water level dynamic curve;

[0010] A plurality of water level warning thresholds are set based on the 24-hour water level dynamic curve, and when the actual monitoring water level exceeds the corresponding level of the plurality of water level warning thresholds, a hierarchical warning signal is automatically sent.

[0011] Preferably, a model is constructed based on a digital elevation model (DEM) and the terrain data to obtain a three-dimensional terrain model of the watershed, including:

[0012] The terrain data is topologically checked and corrected based on a terrain feature line to obtain DEM terrain data corresponding to the terrain data.

[0013] The DEM terrain data is three-dimensionally constructed based on GIS software to obtain the three-dimensional terrain model of the watershed.

[0014] Preferably, the watershed is hydrologically analyzed based on the three-dimensional terrain model of the watershed to obtain hydrological data corresponding to the watershed, including:

[0015] The watershed is gradient analyzed based on the three-dimensional terrain model of the watershed to obtain the gradient of the watershed.

[0016] River network and confluence area data of the watershed are obtained based on the three-dimensional terrain model of the watershed to obtain the hydrological data corresponding to the watershed.

[0017] Preferably, the actual rainfall data and the predicted rainfall data are corrected based on a Bayesian probability correction algorithm to obtain corrected rainfall data, including:

[0018] A probability function is constructed based on the Bayesian probability correction algorithm, the actual rainfall data and the predicted rainfall data to obtain a rainfall correction function, wherein the rainfall correction function has a mathematical expression as follows:

[0019] ;

[0020] wherein, represents a posterior distribution of the predicted rainfall data, represents the predicted rainfall data, represents the actual rainfall data, represents a prior distribution of the predicted rainfall data, represents a likelihood function, represents a marginal probability constant.

[0021] The posterior distribution of the predicted rainfall data is obtained based on the rainfall correction function, and a sample is extracted from the posterior distribution to correct the predicted rainfall data to obtain the corrected rainfall data.

[0022] Preferably, the hydrological data and the corrected rainfall data are input into a kinematic wave hydrological model to obtain the cross-sectional discharge process curve of the watershed, including:

[0023] Based on the hydrological data, the motion wave coefficients are set for the motion wave hydrological model to obtain the motion wave coefficients of the motion wave hydrological model. The motion wave coefficients include: the direct contribution weight of current rainfall to flow, the flow weight of historical rainfall, the maintenance weight of cross-sectional flow, and the comprehensive loss weight of cross-sectional flow.

[0024] Based on the kinematic wave hydrological model, the cross-sectional discharge of the corrected rainfall data is calculated to obtain the cross-sectional discharge process curve, wherein the cross-sectional discharge calculation formula is as follows:

[0025] ;

[0026] in, Indicates the current time The cross-sectional flow rate of the aforementioned watershed, Indicates the current time The corrected rainfall data, Indicates the time step. Indicates the previous moment Corrected rainfall, Indicates the previous moment The cross-sectional flow rate, This indicates the direct contribution of the current rainfall to the flow rate. This indicates the flow weight of the historical rainfall. This indicates the maintenance weight of the flow rate at the cross-section. This represents the overall loss weight of the flow rate at the cross-section.

[0027] Preferably, obtaining the water level dynamic curve based on the cross-sectional flow process curve includes:

[0028] Obtain the actual water level data of the cross section of the basin;

[0029] The simulated water level process line of the watershed is obtained based on the cross-sectional flow process line.

[0030] The simulated water level process line is dynamically corrected based on the actual water level data of the cross section and the extended Kalman filter algorithm to obtain the dynamic water level curve.

[0031] Preferably, the simulated water level process curve is dynamically corrected based on the actual water level data of the cross-section and the extended Kalman filter algorithm to obtain the dynamic water level curve, including:

[0032] Predicted water level data is obtained based on the simulated water level process curve;

[0033] correcting the predicted water level data based on an extended Kalman filtering algorithm and the cross-section actual water level data, to obtain the water level dynamic curve, wherein a correction formula of the predicted water level data is as follows:

[0034] ;

[0035] wherein, represents the correction water level data at time point , represents the predicted water level data at time point , represents a Kalman gain, represents the cross-section actual water level data at time point , represents the cross-section actual water level data at time point and an observation error of the predicted water level data.

[0036] Preferably, a non-observation period water level sequence of the river basin is periodically extended based on a time series extension algorithm, to obtain an extended water level dynamic curve, comprising:

[0037] a frequency domain water level sequence curve is obtained by performing frequency domain transformation on the water level dynamic curve based on Fourier transform;

[0038] an extended frequency domain water level sequence curve is obtained by performing frequency domain extension on the non-observation period water level sequence of the river basin based on the frequency domain water level sequence curve;

[0039] the extended frequency domain water level sequence curve is converted into a time domain water level sequence curve based on inverse Fourier transform, to obtain the extended water level dynamic curve.

[0040] Preferably, the extended water level dynamic curve is verified based on a water level simulation method, to obtain a 24-hour water level dynamic curve, comprising:

[0041] simulation water level data is obtained based on a preset extension period and the extended water level dynamic curve;

[0042] actual water level data of the preset extension period is obtained;

[0043] the simulation water level data and the actual water level data are compared, to obtain a verification result of the extended water level dynamic curve, and if the verification result has an error, the extended water level dynamic curve is adjusted and optimized, to obtain the 24-hour water level dynamic curve.

[0044] The beneficial effects of the present invention are as follows: (1) By setting multi-level water level warning thresholds, the present invention automatically sends a warning signal when the actual water level exceeds a certain level. The automated and graded response can respond to water level changes of different severity levels more quickly and accurately, thus improving the efficiency of disaster prevention and mitigation; (2) By using time-series extension and water level simulation methods, the present invention interpolates and transforms daily missing data to generate a 24-hour water level dynamic curve, avoiding the situation of missing the maximum water level value in other periods. The water level warning threshold is more reliable and the water level warning effect is better; (3) By combining hydrological data, corrected rainfall data and inference models such as kinematic wave hydrological models for hydrological analysis, the present invention can more accurately predict hydrological changes in the basin, thereby improving the accuracy of the water level dynamic curve; (4) By generating a 24-hour water level dynamic curve and setting warning thresholds, the present invention provides timely water level prediction and warning, which can better cope with sudden water level changes. Attached Figure Description

[0045] Figure 1 This is a schematic diagram of the overall method steps in one embodiment of the present invention; Detailed Implementation

[0046] Example 1, as Figure 1 As shown, the present invention proposes a method for intelligent early warning of water levels in a river basin, comprising:

[0047] S1. Obtain the topographic data of the watershed of the pumped storage power station, and construct a model based on the digital elevation model (DEM) and the topographic data to obtain a three-dimensional topographic model of the watershed.

[0048] S2. Perform hydrological analysis on the watershed based on the three-dimensional topographic model of the watershed to obtain the corresponding hydrological data of the watershed;

[0049] S3. Obtain actual rainfall data of the watershed, obtain predicted rainfall data of the watershed based on rainfall numerical forecast products, and correct the actual rainfall data and predicted rainfall data based on Bayesian probability correction algorithm to obtain corrected rainfall data.

[0050] S4. Input the hydrological data and corrected rainfall data into the kinematic wave hydrological model to obtain the cross-sectional flow process line of the watershed, and obtain the dynamic water level curve based on the cross-sectional flow process line.

[0051] S5. Based on the time-series extension algorithm, the water level sequence of the non-observation period in the basin is periodically extended to obtain the extended water level dynamic curve. The extended water level dynamic curve is verified based on the water level simulation method to obtain the 24-hour water level dynamic curve.

[0052] S6, setting multi-level water level warning thresholds based on the 24-hour water level dynamic curve, and automatically sending a hierarchical warning signal when the actual monitoring water level exceeds the corresponding level of the multi-level water level warning thresholds.

[0053] In the present application, the terrain data of the watershed refers to the digitized data of the terrain undulations of a certain area obtained through remote sensing technology or field survey; the digital elevation model refers to a digitized model representing the elevation information of the earth's surface, which is usually used to analyze the topographic features; the three-dimensional terrain model refers to a three-dimensional terrain model constructed based on DEM data, which is used to display the elevation changes and spatial structure of the earth's surface; the actual rainfall data refers to the recorded rainfall data; the predicted rainfall data refers to the prediction result of the rainfall in a future period of time through rainfall numerical prediction products; the Bayesian probability correction algorithm refers to a statistical method that adjusts the prediction result through historical data and probability theory to make it closer to the actual situation; the kinematic wave hydrological model refers to a hydrological model that simulates the interaction of water flow, rainwater and groundwater, which is suitable for hydrological analysis and prediction in the watershed; the cross-section flow process line refers to the curve of water flow change with time at a certain cross-section of the watershed, which is used to analyze the water flow dynamics; the water level dynamic curve refers to the curve describing the change of water level with time, which is usually used for water level change analysis of reservoirs, rivers and other water bodies; the time series extension algorithm refers to processing the existing time series data to predict future water level changes; the water level simulation method refers to simulating water level changes to verify and adjust the predicted water level dynamic curve; the multi-level water level warning threshold refers to setting different warning standards according to different water levels, which is used to trigger the corresponding warning level; the hierarchical warning signal refers to issuing different levels of alarms according to different water levels to remind relevant personnel or institutions to take appropriate measures.

[0054] In embodiment two, the watershed water level intelligent warning method proposed by the present application further includes:

[0055] Based on the digital elevation model DEM and the terrain data, a model is constructed to obtain a three-dimensional terrain model of the watershed, which includes:

[0056] A1, topological consistency check and correction of terrain data based on terrain feature lines to obtain DEM terrain data corresponding to the terrain data;

[0057] A2, three-dimensional construction of DEM terrain data based on GIS software to obtain a three-dimensional terrain model of the watershed.

[0058] In this embodiment, the GIS software refers to a tool for collecting, storing, managing, analyzing, and displaying geographic information. Common GIS software includes QGIS and the like, which can process geographic data, perform spatial analysis, and generate maps and models. Three-dimensional construction refers to generating a model with three-dimensional spatial characteristics by using DEM data and tools. The three-dimensional terrain model of the watershed refers to a three-dimensional visualization model generated based on DEM data and terrain data, which shows the changes in surface elevation and morphology of the watershed region.

[0059] In an optional embodiment, the watershed is hydrologically analyzed based on the three-dimensional terrain model of the watershed to obtain hydrological data corresponding to the watershed, including:

[0060] B1, slope analysis of the watershed is performed based on the three-dimensional terrain model of the watershed to obtain the slope of the watershed.

[0061] B2, river network and catchment area data of the watershed are obtained based on the three-dimensional terrain model of the watershed to obtain hydrological data corresponding to the watershed.

[0062] It should be noted that slope analysis refers to calculating DEM terrain data to obtain slope information of each point on the surface. Catchment area refers to the area of all water flows in the watershed that eventually converge to a point.

[0063] In an optional embodiment, the actual rainfall data and the predicted rainfall data are corrected based on a Bayesian probability correction algorithm to obtain corrected rainfall data, including:

[0064] C1, a probability function is constructed based on the Bayesian probability correction algorithm, the actual rainfall data, and the predicted rainfall data to obtain a rainfall correction function, wherein the mathematical expression of the rainfall correction function is as follows:

[0065] ;

[0066] wherein, represents the posterior distribution of the predicted rainfall data, represents the predicted rainfall data, represents the actual rainfall data, represents the prior distribution of the predicted rainfall data, represents the likelihood function, represents the marginal probability constant;

[0067] C2, the posterior distribution of the predicted rainfall data is obtained based on the rainfall correction function, and samples are extracted from the posterior distribution to correct the predicted rainfall data to obtain corrected rainfall data.

[0068] It should be noted that the rainfall correction function describes the probability of rainfall data and corrects it according to the actual situation; the posterior distribution refers to the probability distribution of rainfall data updated based on Bayes' theorem after observing the actual rainfall data; the prior distribution refers to the preliminary hypothetical distribution of rainfall data before observing the actual rainfall data; the likelihood function describes the probability of observing the actual rainfall data given the model and parameters; the marginal probability constant is the constant in Bayes' theorem used to ensure that the total probability of the posterior distribution is 1; the corrected rainfall data refers to the rainfall data corrected by the Bayesian algorithm, which takes into account the error of previous predictions and the actual observed data, thus making the final rainfall prediction more accurate.

[0069] In an optional embodiment, hydrological data and corrected rainfall data are input into a kinematic wave hydrological model to obtain the cross-sectional discharge process curve of the watershed, including:

[0070] D1. Based on hydrological data, set the motion wave coefficients for the motion wave hydrological model to obtain the motion wave coefficients of the motion wave hydrological model. The motion wave coefficients include: the direct contribution weight of current rainfall to flow, the flow weight of historical rainfall, the maintenance weight of cross-sectional flow, and the comprehensive loss weight of cross-sectional flow.

[0071] D2. Based on the kinematic wave hydrological model, cross-sectional discharge is calculated from the corrected rainfall data to obtain the cross-sectional discharge hydrograph. The formula for calculating cross-sectional discharge is as follows:

[0072] ;

[0073] in, Indicates the current time The cross-sectional flow of the basin, Indicates the current time Corrected rainfall data, Indicates the time step. Indicates the previous moment Corrected rainfall, Indicates the previous moment The cross-sectional flow rate, This indicates the direct contribution of current rainfall to the flow rate. The flow weight represents the historical rainfall. This indicates the maintenance weight of the cross-sectional flow. This represents the overall loss weight of the cross-sectional flow.

[0074] It should be noted that the motion wave coefficient refers to a parameter used to describe the relationship between rainfall and flow in the motion wave hydrological model, which includes several different weight values for describing the influence of different factors on flow; the time step refers to the time interval represented by each calculation in numerical simulation, and in the hydrological model, the time step determines the frequency of model updating and the speed of flow change.

[0075] In an optional embodiment, the water level dynamic curve is obtained based on the cross-section flow process line, comprising:

[0076] E1, obtaining cross-section actual water level data of a basin;

[0077] E2, obtaining a simulated water level process line of the basin based on the cross-section flow process line;

[0078] E3, dynamically correcting the simulated water level process line based on the cross-section actual water level data and an extended Kalman filtering algorithm to obtain the water level dynamic curve.

[0079] It should be noted that the extended Kalman filtering algorithm refers to an algorithm for state estimation of a nonlinear system, and the extended Kalman filtering can be used to correct the simulated water level process line and dynamically correct the prediction error.

[0080] In an optional embodiment, the water level dynamic curve is obtained based on the cross-section actual water level data and the extended Kalman filtering algorithm, comprising:

[0081] F1, obtaining predicted water level data based on the simulated water level process line;

[0082] F2, correcting the predicted water level data based on the extended Kalman filtering algorithm and the cross-section actual water level data to obtain the water level dynamic curve, wherein the correction formula of the predicted water level data is as follows:

[0083] ;

[0084] wherein, represents the corrected water level data at time , represents the predicted water level data at time , represents the Kalman gain, represents the cross-section actual water level data at time , and represents the observation error of the cross-section actual water level data and the predicted water level data at time .

[0085] ​​​It should be noted that the Kalman gain is a coefficient used to adjust the proportion of prediction error and observation error in the extended Kalman filter algorithm, which reflects the degree of trust in the current estimated value, and the Kalman gain determines how to weight the actual observation data and the predicted data.

[0086] In an optional embodiment, the non-observation period water level sequence of the basin is periodically extended based on the time series extension algorithm to obtain an extended water level dynamic curve, including:

[0087] G1, based on Fourier transform, the frequency domain water level sequence curve is obtained by transforming the water level dynamic curve in the frequency domain;

[0088] G2, based on the frequency domain water level sequence curve, the non-observation period water level sequence of the basin is extended in the frequency domain to obtain an extended frequency domain water level sequence curve;

[0089] G3, based on the inverse Fourier transform, the extended frequency domain water level sequence curve is converted into a time domain water level sequence curve to obtain an extended water level dynamic curve.

[0090] It should be noted that Fourier transform refers to a mathematical tool for converting a time domain signal (such as a water level dynamic curve) into a frequency domain signal. Through Fourier transform, a time domain signal can be represented as a combination of sine waves of different frequencies. Fourier transform helps to understand the frequency characteristics of water level changes. The frequency domain water level sequence curve is a curve obtained by Fourier transform from the time domain water level data, which represents the distribution of water level change signals at different frequencies. Frequency domain extension is a process of extending or interpolating the frequency domain water level sequence curve to predict or estimate the water level sequence in the non-observation period. The extension operation is usually based on the analysis of frequency domain signals to extend the frequency components of the signal to supplement the missing data. The extended frequency domain water level sequence curve is a curve obtained after frequency domain extension, which represents the frequency characteristics of the water level in the non-observation period. This curve, through frequency domain extension, adds new frequency components to the original frequency domain water level sequence, thus predicting the water level changes in the missing period. Inverse Fourier transform is a process of converting frequency domain signals back to time domain signals. For the frequency domain water level sequence curve, inverse Fourier transform can restore the frequency components to the original water level time sequence. This process converts the frequency domain data back to the time domain data that can be actually observed, thus obtaining the extended water level dynamic curve. The extended water level dynamic curve is a water level change curve obtained by converting the extended frequency domain water level sequence curve into a time domain sequence through inverse Fourier transform. This curve not only contains the actually observed water level data, but also includes the estimated data obtained through frequency domain extension, forming a complete and continuous water level dynamic change curve.

[0091] In an optional embodiment, the water level simulation method is used to verify the extended water level dynamic curve to obtain a 24-hour water level dynamic curve, including:

[0092] H1, obtaining simulation water level data based on the preset extended period and the extended water level dynamic curve;

[0093] H2, obtaining actual water level data of the preset extended period;

[0094] H3, comparing the simulation water level data and the actual water level data to obtain a verification result of the extended water level dynamic curve, and if there is an error in the verification result, adjusting and optimizing the extended water level dynamic curve to obtain a 24-hour water level dynamic curve.

[0095] It should be noted that the 24-hour water level dynamic curve refers to a curve representing the water level change in the next 24 hours, which is obtained by integrating simulation data, actual observation data and the optimization result of the extended water level dynamic curve, and shows the trend of water level change in the next day. This curve is very important for water level warning work.

[0096] The embodiments of the present application are described in detail above in combination with the drawings, but the present application is not limited thereto, and various changes can be made within the knowledge of those skilled in the art without departing from the purpose of the present application.

Claims

1. A basin water level intelligent early warning method, Comprising, characterized in that: Obtain topographic data of the basin of the pumped storage power station, construct a model based on a digital elevation model DEM and the topographic data to obtain a three-dimensional terrain model of the basin; Based on the three-dimensional terrain model of the basin, hydrological analysis is performed on the basin to obtain the corresponding hydrological data of the basin; Obtain actual rainfall data of the basin, obtain predicted rainfall data of the basin based on rainfall numerical prediction products, and correct the actual rainfall data and predicted rainfall data based on a Bayesian probability correction algorithm to obtain corrected rainfall data; Input the hydrological data and the corrected rainfall data into a kinematic wave hydrological model to obtain the cross-section flow process line of the basin, and obtain the water level dynamic curve based on the cross-section flow process line; Based on the time series extension algorithm, the unobserved period water level sequence of the basin is periodically expanded to obtain an expanded water level dynamic curve, and the expanded water level dynamic curve is verified based on a water level simulation method to obtain a 24-hour water level dynamic curve; Based on the 24-hour water level dynamic curve, set multi-level water level warning thresholds, and when the actual monitoring water level exceeds the corresponding level of the multi-level water level warning thresholds, automatically send a hierarchical warning signal. 2.The watershed water level intelligent early warning method according to claim 1, characterized in that, Based on the digital elevation model DEM and the topographic data, a model is constructed to obtain a three-dimensional terrain model of the basin, comprising: Based on the topographic feature line, the topographic data is topologically consistent and corrected to obtain DEM topographic data corresponding to the topographic data; Based on the GIS software, the DEM topographic data is three-dimensionally constructed to obtain the three-dimensional terrain model of the basin.

3. The intelligent basin water level early warning method according to claim 1, characterized in that, Based on the three-dimensional terrain model of the basin, hydrological analysis is performed on the basin to obtain the corresponding hydrological data of the basin, comprising: Based on the three-dimensional terrain model of the basin, slope analysis is performed on the basin to obtain the slope of the basin; Based on the three-dimensional terrain model of the basin, the river network and confluence area data of the basin are obtained to obtain the corresponding hydrological data of the basin.

4. The intelligent basin water level early warning method according to claim 1, characterized in that, Based on the Bayesian probability correction algorithm, the actual rainfall data and the predicted rainfall data are corrected to obtain corrected rainfall data, comprising: Based on the Bayesian probability correction algorithm, the actual rainfall data and the predicted rainfall data are constructed to obtain a probability function, wherein the rainfall correction function is mathematically expressed as follows: ； wherein, represents a posterior distribution of the predicted rainfall data, represents the predicted rainfall data, represents the actual rainfall data, represents a prior distribution of the predicted rainfall data, represents a likelihood function, represents a marginal probability constant; Based on the rainfall correction function, the posterior distribution of the predicted rainfall data is obtained, and samples are extracted from the posterior distribution to correct the predicted rainfall data to obtain the corrected rainfall data.

5. The intelligent basin water level early warning method according to claim 1, characterized in that, Input the hydrological data and the corrected rainfall data into a kinematic wave hydrological model to obtain the cross-section flow process line of the basin, comprising: Based on the hydrological data, the kinematic wave coefficient of the kinematic wave hydrological model is set to obtain the kinematic wave coefficient of the kinematic wave hydrological model, wherein the kinematic wave coefficient includes: the direct contribution weight of current rainfall to flow, the flow weight of historical rainfall, the maintenance weight of cross-section flow and the comprehensive loss weight of cross-section flow; The cross-section flow process line is obtained based on the correction rainfall data and the motion wave hydrological model, and a formula for calculating the cross-section flow is as follows: ； wherein, denotes the cross-sectional flow of the catchment at the current time denotes the correction rainfall data at the current time denotes the time step, denotes the correction rainfall at the previous time denotes the cross-sectional flow at the previous time denotes the direct contribution weight of the current rainfall to the flow, denotes the flow weight of the historical rainfall, denotes the maintenance weight of the cross-sectional flow, denotes the comprehensive loss weight of the cross-sectional flow.​​​​ 6. The intelligent basin water level early warning method according to claim 1, characterized in that, The water level dynamic curve is obtained based on the cross-section flow process line, including: Actual cross-section water level data of the basin is obtained; A simulated water level process line of the basin is obtained based on the cross-section flow process line; The simulated water level process line is dynamically corrected based on the actual cross-section water level data and an extended Kalman filtering algorithm to obtain the water level dynamic curve.

7. The intelligent basin water level early warning method according to claim 6, characterized in that, The simulated water level process line is dynamically corrected based on the actual cross-section water level data and an extended Kalman filtering algorithm to obtain the water level dynamic curve, including: Predicted water level data is obtained based on the simulated water level process line; The predicted water level data is corrected based on an extended Kalman filtering algorithm and the actual cross-section water level data to obtain the water level dynamic curve, and a formula for correcting the predicted water level data is as follows: ； wherein, represents the corrected water level data at the time point , represents the predicted water level data at the time point , represents the Kalman gain, represents the cross-section actual water level data at the time point , represents the observation error of the cross-section actual water level data and the predicted water level data at the time point . 8.The watershed water level intelligent early warning method according to claim 1, characterized in that, The water level sequence of the basin in a non-observed time period is periodically extended based on a time series extension algorithm to obtain an extended water level dynamic curve, including: The water level dynamic curve is frequency domain transformed based on a Fourier transform to obtain a frequency domain water level sequence curve; The water level sequence of the basin in a non-observed time period is frequency domain extended based on the frequency domain water level sequence curve to obtain an extended frequency domain water level sequence curve; The extended frequency domain water level sequence curve is converted into a time domain water level sequence curve based on an inverse Fourier transform to obtain the extended water level dynamic curve.

9. The intelligent basin water level early warning method according to claim 1, characterized in that, The extended water level dynamic curve is verified based on a water level simulation method to obtain a 24-hour water level dynamic curve, including: Simulated water level data is obtained based on a preset extension time period and the extended water level dynamic curve; Actual water level data of the preset extension time period is obtained; The simulated water level data and the actual water level data are compared to obtain a verification result of the extended water level dynamic curve, and if the verification result has an error, the extended water level dynamic curve is adjusted and optimized to obtain the 24-hour water level dynamic curve.

Images