A method and device for rapid flood prediction based on Kriging model

Abstract

The invention discloses a method and device for rapid flood prediction based on a Kriging model, wherein the method comprises: determining a modeling scope according to a research area, and collecting relevant information and data; constructing a two-dimensional hydrodynamic model based on a group of hydrodynamic control equations and using an unstructured triangular unit grid; combining historical monitoring data and design data of a hydrological water level station, generating a large number of flow processes according to different upstream flood peak flows, and constructing an input sample library; simulating and calculating the evolution and change of floods in the research area based on the data content of the input sample library, and obtaining corresponding flood indicators through statistical calculation to form an output sample library; establishing a mapping relationship between upstream floods and flood indicators of key sections by using the Kriging method according to the input sample library and the output sample library, and constructing a rapid flood prediction model; and predicting the flood indicators of key sections of upstream floods by using the rapid flood prediction model. The invention improves the efficiency and accuracy of flood prediction.

Description

Technical Field

[0001] The present invention relates to the technical field of water conservancy digital twin and flood prediction, and in particular to a method and device for rapid flood prediction based on a Kriging model. Background Art

[0002] In response to the need for rapid rehearsal and risk assessment of flood disasters, experts and scholars in the industry have proposed a variety of different solutions, such as numerical solution and calculation of flood processes through Muskingum, one-dimensional and two-dimensional hydrodynamic models, and then obtaining flood process deduction, and further analyzing and judging flood risks.

[0003] The existing flood calculation or prediction models mainly use Muskingum method or one-dimensional hydrodynamic model for rapid calculation, and a small number use two-dimensional hydrodynamic model for flood prediction and calculation.

[0004] The inventors of this application found that the prior art has at least the following technical problems during the implementation of the present invention:

[0005] Muskingum method can quickly calculate the relationship between upstream and downstream flood flow and water level processes, but it is difficult to know the flood process and changes along the way between upstream and downstream;

[0006] Based on the one-dimensional hydrodynamic model, it is possible to calculate each section along the main stream of the river, obtain the flow and water level relationship at each section under the flood scenario, and then calculate the peak flow and water level and time. However, it is difficult to grasp the disaster situation of key villages or areas along the downstream. In addition, due to too many assumptions required for model calculation, it cannot adapt well to the complex and changeable different basin environments, and cannot guarantee the matching degree between the model calculation results and the current situation of the basin.

[0007] Based on the two-dimensional hydrodynamic model, the flow and water level of each node under the flood scenario can be calculated by integrating the river section, river topography and terrain data along the way, and the peak flow and water level and time of any grid point can be obtained. The peak flow and water level and time of the key sections as well as the disaster situation can be obtained through statistical calculation. However, due to the limitations of computer performance and the iteration speed of its own algorithm, it is impossible to quickly complete the calculation of the entire flood process within a few seconds or more than ten seconds, and it cannot be widely used in assisting quick decision-making and rapid response. Summary of the invention

[0008] In order to solve the technical problem that the existing technical solutions cannot provide good guidance for the assessment of downstream flood risks due to poor calculation accuracy or slow calculation efficiency, the present invention constructs input and output based on a two-dimensional hydrodynamic flood evolution model to form a sample library, and based on the sample library, innovatively uses the Kriging method to construct a rapid flood prediction model, establishes a mapping relationship between upstream floods and key section floods, so that the model can perform rapid calculations based on flood prediction data provided by basin managers.

[0009] The technical solution adopted by the present invention is as follows:

[0010] The first aspect provides a rapid flood prediction method based on the Kriging model, including:

[0011] Determine the modeling scope based on the study area, collect digital elevation model data, historical monitoring data of hydrological and water level stations, and design information within the modeling scope;

[0012] Based on the hydrodynamic control equations, a two-dimensional hydrodynamic model is constructed using unstructured triangular element grids.

[0013] The two-dimensional hydrodynamic model was calibrated and verified using historical monitoring data from hydrological and water level stations and real-time monitoring data from the study area;

[0014] Combining historical monitoring data from hydrological and water level stations with design information, a large number of flow processes are generated according to different upstream flood peak flows to construct an input sample library;

[0015] Based on the data content of the input sample library, the hydrodynamic control equations of the two-dimensional hydrodynamic model after verification are used to simulate and calculate the flood evolution changes in the study area, and the corresponding flood indicators are obtained based on the statistical calculation of the flood evolution changes in the study area to form an output sample library;

[0016] Based on the input sample library and the output sample library, the Kriging method is used to establish the mapping relationship between upstream floods and key section flood indicators, and to build a rapid flood prediction model;

[0017] The rapid flood prediction model is used to predict flood indicators of key sections based on upstream floods.

[0018] In one implementation, the modeling scope is determined according to the study area, and digital elevation model data, historical monitoring data of hydrological and water level stations, and design information within the modeling scope are collected, including:

[0019] The terrain within the research area is determined based on the vector map and satellite image of the research area. The modeling scope is divided around the river and the areas on both sides of the river. The boundary sections of the upstream and downstream of the river and the hillsides on both sides or the areas where the terrain meets the preset conditions are used as the boundaries of the modeling scope.

[0020] Collect digital elevation model data within the modeling range;

[0021] Collect historical hydrological monitoring data and design information within the modeling scope.

[0022] In one embodiment, based on the hydrodynamic control equations, an unstructured triangular unit grid is used to construct a two-dimensional hydrodynamic model, including:

[0023] Based on the collected digital elevation model data, historical monitoring data of hydrological and water level stations and design information within the modeling range, combined with satellite images, river bank and shoreline mapping results, sections or villages along the way and actual needs, determine the areas in the study area that need to be meshed, and gradually reduce the mesh density from the river bank to the hillside or areas that meet the preset conditions;

[0024] Based on the unstructured grid of irregular triangles, a two-dimensional finite element grid is constructed in the study area, and the terrain elevation variable information is assigned at each grid node to carry out flood evolution calculation;

[0025] Based on the historical monitoring data of the hydrological water level station or the real-time monitoring data of the study area, the flow process at the upstream boundary section and the water level process at the downstream boundary section in the typical period are selected as the boundary input data for model construction;

[0026] Set up the hydrodynamic control equations, where the hydrodynamic control equations are in the form of:

[0027]

[0028] In the formula, represents the conservation vector, t represents the time, and denote the convective flux in the x and y directions, respectively, and denote the diffusion flux in the x and y directions respectively; S denotes the source term.

[0029] In one implementation, historical monitoring data of hydrological and water level stations are combined with design data, a large number of flow processes are generated according to different upstream flood peak flows, and an input sample library is constructed, including:

[0030] Based on the historical monitoring data of hydrological and water level stations, considering the peak sizes of different natural floods, and combining the design data, the data of a peak size is interpolated to generate several flood process data sets;

[0031] Considering the reservoir regulation process within the study area, the upstream and downstream flood datasets are generated by combining the generated flood process datasets and reservoir operation experience rules;

[0032] Clean and format the generated upstream and downstream flood data sets to obtain the time series data of the upstream and downstream flood data sets, and match the upstream and downstream floods to obtain the matched time series data;

[0033] The matched data are integrated to form an input sample library.

[0034] In one embodiment, based on the data content of the input sample library, the hydrodynamic control equations of the two-dimensional hydrodynamic model after verification are used to simulate and calculate the flood evolution changes in the study area, and the corresponding flood indicators are obtained based on the statistical calculation of the flood evolution changes in the study area to form an output sample library, including:

[0035] Extract the time series data of upstream and downstream flood data sets of the same flood process from the input sample library, use them to construct the input conditions of the two-dimensional hydrodynamic model, and use the hydrodynamic control equations to perform flood simulation calculations based on the input conditions;

[0036] Combined with the objects of interest within the research scope, the adjacent river sections are taken as key sections;

[0037] Based on the results of flood simulation calculation, the time series change process of flow rate and water depth of key sections is extracted, and the corresponding flood indicators are calculated statistically, including peak flow rate, peak arrival time and peak water level;

[0038] Based on the calculated flood index, the peak time interval and regional disaster area corresponding to the key sections of concern are calculated under the input conditions of the constructed two-dimensional hydrodynamic model, and the peak time interval and regional disaster area corresponding to the key sections are associated with the peak flow;

[0039] The associated data is organized into a structured data format to form an output sample library.

[0040] In one embodiment, a flood simulation calculation is performed using a group of hydrodynamic control equations according to the input conditions, including:

[0041] Based on the input conditions of the constructed two-dimensional hydrodynamic model, the water depth and velocity data of each grid node are calculated using the hydrodynamic control equations to obtain the river flood evolution changes as the result of the flood simulation calculation.

[0042] In one implementation, based on the input sample library and the output sample library, the Kriging method is used to establish a mapping relationship between upstream floods and flood indicators of key sections, and a rapid flood prediction model is constructed, including:

[0043] Matching the input-output relationship of the obtained input sample library and output sample library;

[0044] Sampling and selecting several input-output relationships to form a calibration data set;

[0045] The Gaussian process regression of the input-output relationship is calculated using the Kriging method to obtain the response surface mapping relationship between different flood scenarios and hydraulic elements of key sections, and to build a rapid flood prediction model;

[0046] After matching, data outside the calibration data set is sampled and selected from the input sample library and output sample library to form a verification data set. The accuracy of the rapid flood prediction model is verified based on the data set. If the accuracy does not meet the requirements, re-sampling and Gaussian process regression operations of the input-output relationship are calculated using the Kriging method, and the model parameters are adjusted until the accuracy meets the requirements.

[0047] Based on the same inventive concept, the second aspect of the present invention provides a rapid flood prediction device based on a Kriging model, comprising:

[0048] The data collection module is used to determine the modeling scope according to the study area, and collect digital elevation model data, historical monitoring data of hydrological and water level stations, and design information within the modeling scope;

[0049] A two-dimensional hydrodynamic model building module is used to build a two-dimensional hydrodynamic model based on the hydrodynamic control equations using an unstructured triangular unit grid;

[0050] Model calibration module, used to calibrate and verify the two-dimensional hydrodynamic model using historical monitoring data of hydrological and water level stations and real-time monitoring data of the study area;

[0051] The input sample library construction module is used to combine the historical monitoring data of the hydrological and water level stations with the design data, generate a large number of flow processes according to different upstream flood peak flows, and construct the input sample library;

[0052] The output sample library construction module is used to simulate and calculate the flood evolution changes in the study area based on the data content of the input sample library using the hydrodynamic control equations of the two-dimensional hydrodynamic model after the calibration verification, and obtain the corresponding flood indicators based on the statistical calculation of the flood evolution changes in the study area to form the output sample library;

[0053] The flood rapid prediction model building module is used to establish the mapping relationship between upstream floods and key section flood indicators based on the input sample library and the output sample library using the Kriging method, and to build a flood rapid prediction model;

[0054] The flood prediction module is used to predict flood indicators of key sections based on upstream floods using a rapid flood prediction model.

[0055] Based on the same inventive concept, the third aspect of the present invention provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the Kriging model-based rapid flood prediction method described in the first aspect.

[0056] Based on the same inventive concept, the fourth aspect of the present invention provides a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, the method for rapid flood prediction based on the Kriging model described in the first aspect is implemented.

[0057] Compared with the prior art, the advantages and beneficial technical effects of the present invention are as follows:

[0058] The present invention provides a method for rapid flood prediction based on the Kriging model. After collecting relevant data, based on the hydrodynamic control equations, an unstructured triangular unit grid is used to construct a two-dimensional hydrodynamic model, and an input sample library and an output sample library are constructed based on the two-dimensional hydrodynamic flood evolution model. According to the input sample library and the output sample library, the Kriging method is used to establish a mapping relationship between upstream floods and flood indicators of key sections, and a rapid flood prediction model is constructed, so that the model can be quickly calculated according to the flood prediction data provided by the basin management personnel. Specifically, it is possible to obtain calculation response results at the second level for the future upstream flood inflow process, and the flood situation of key sections, such as the peak flow water level and peak time data, and the impact of the flood on key villages, can be calculated in a short time, which can assist in consultation and decision-making and risk assessment and rapid response in flood scenarios, and help realize the digital management of the basin. BRIEF DESCRIPTION OF THE DRAWINGS

[0059] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0060] Figure 1 A flowchart of a method for rapid flood prediction based on a Kriging model provided by an embodiment of the present invention;

[0061] Figure 2 A triangular mesh schematic diagram of a two-dimensional hydrodynamic model in an embodiment of the present invention;

[0062] Figure 3 A schematic diagram of a triangular mesh encryption area constructed in an embodiment of the present invention;

[0063] Figure 4A schematic diagram of the response data of the Kriging model constructed in an embodiment of the present invention;

[0064] Figure 5 A structural diagram of a rapid flood prediction device based on a Kriging model provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0065] The present invention provides a method and device for rapid flood prediction based on a Kriging model, which solves the technical problem that the prior art cannot provide good guidance for the assessment of downstream flood risks due to poor calculation accuracy or low calculation efficiency.

[0066] In order to solve the above technical problems, the key technical points adopted by the present invention include:

[0067] 1. Based on the two-dimensional shallow water equations, an unstructured triangular unit grid is used to construct a two-dimensional flood evolution model and complete the calibration and verification of the model;

[0068] 2. Construct an input sample library and calculate the output sample library in combination with the two-dimensional flood evolution model;

[0069] 3. Develop and construct a rapid flood prediction model based on the Kriging method and the two-dimensional flood evolution model to achieve the mapping relationship between the flood process and the hydraulic elements of key sections. The model can calculate the peak flow, peak water level, peak time and the corresponding regional disaster area based on the predicted flood.

[0070] Based on the above technical solution, the present invention can obtain calculation response results at the second level for future upstream flood inflow processes. The flood conditions of key sections, such as peak flow water level and peak time data, as well as the impact of the flood on key villages, can be calculated in a short time. It can assist in consultation and decision-making, risk assessment and rapid response in flood scenarios, and help realize digital management of the river basin.

[0071] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0072] Embodiment 1

[0073] The present invention discloses a method for rapid flood prediction based on Kriging model. Figure 1 , the method comprising:

[0074] S1: Determine the modeling scope according to the study area, collect digital elevation model data, historical monitoring data of hydrological water level stations and design information within the modeling scope;

[0075] S2: Based on the hydrodynamic control equations, a two-dimensional hydrodynamic model is constructed using an unstructured triangular element grid;

[0076] S3: Calibrate and verify the two-dimensional hydrodynamic model using historical monitoring data of hydrological water level stations and real-time monitoring data of the study area;

[0077] S4: Combine the historical monitoring data of the hydrological and water level stations with the design data, generate a large number of flow processes according to different upstream flood peak flows, and build an input sample library;

[0078] S5: Based on the data content of the input sample library, the hydrodynamic control equations of the two-dimensional hydrodynamic model after the verification are used to simulate and calculate the flood evolution changes in the study area, and the corresponding flood indicators are obtained based on the statistical calculation of the flood evolution changes in the study area to form an output sample library;

[0079] S6: Based on the input sample library and the output sample library, the Kriging method is used to establish the mapping relationship between the upstream flood and the flood indicators of the key sections, and a rapid flood prediction model is constructed;

[0080] S7: Use the rapid flood prediction model to predict flood indicators of key sections based on upstream floods.

[0081] Specifically, S1 is the collection of data. During the specific implementation process, after data collection, the data can be cleaned and formatted and organized into a structured input data set.

[0082] S2 is the construction of a two-dimensional hydrodynamic model. A finite element mesh based on unstructured triangular unit topological data is established, and a two-dimensional hydrodynamic model is constructed in combination with a group of hydrodynamic control equations. The model is subsequently used to calculate the evolution of river floods under a given flood process scenario.

[0083] S3 is the calibration of the two-dimensional hydrodynamic model. Specifically, some data from a typical flood period can be selected as the calibration period to calibrate the model parameters, and then the calibrated parameters can be used to verify in other flood periods until the NSE (Nash efficiency coefficient) reaches the expectation.

[0084] S4 is the construction of the input sample library.

[0085] S5 is based on the data content of the input sample library and uses the calibrated model to simulate and calculate the flood evolution changes in the study area, thereby constructing an output sample library.

[0086] S6 is based on the input sample library and output sample library of the previous steps, and uses the Kriging method to establish the mapping relationship between upstream floods and key section floods, and then constructs a rapid flood prediction model. In addition, the rapid flood prediction model can also be calibrated and verified in combination with the input and output sample libraries.

[0087] S7 uses the rapid flood prediction model established in S6, combined with the predicted upstream floods, to quickly calculate data such as the peak flow, peak water level and regional disaster-stricken area of ​​key sections to assist in flood risk assessment and analysis.

[0088] In one implementation, the modeling scope is determined according to the study area, and digital elevation model data, historical monitoring data of hydrological and water level stations, and design information within the modeling scope are collected, including:

[0089] The terrain within the research area is determined based on the vector map and satellite image of the research area. The modeling scope is divided around the river and the areas on both sides of the river. The boundary sections of the upstream and downstream of the river and the hillsides on both sides or the areas where the terrain meets the preset conditions are used as the boundaries of the modeling scope.

[0090] Collect digital elevation model data within the modeling range;

[0091] Collect historical hydrological monitoring data and design information within the modeling scope.

[0092] During the specific implementation process, when the collected digital elevation model data is longitude and latitude coordinate data, it is converted into CGCS (China Geodetic Coordinate System) or UTM (UNIVERSAL TRANSVERSEMERCARTOR GRID SYSTEM) coordinate data to ensure that the terrain data element unit within the modeling range is meter and is based on a plane rectangular coordinate system to facilitate model construction and calculation.

[0093] Collect historical hydrological monitoring data and design information within the modeling scope, including section elevation, monitoring water level, monitoring flow, etc. In addition, the data format can be unified to facilitate subsequent model calibration and verification and other related work.

[0094] In one embodiment, based on the hydrodynamic control equations, an unstructured triangular unit grid is used to construct a two-dimensional hydrodynamic model, including:

[0095] Based on the collected digital elevation model data, historical monitoring data of hydrological and water level stations and design information within the modeling range, combined with satellite images, river bank and shoreline mapping results, sections or villages along the way and actual needs, determine the areas in the study area that need to be meshed, and gradually reduce the mesh density from the river bank to the hillside or areas that meet the preset conditions;

[0096] Based on the unstructured grid of irregular triangles, a two-dimensional finite element grid is constructed in the study area, and the terrain elevation variable information is assigned at each grid node to carry out flood evolution calculation;

[0097] Based on the historical monitoring data of the hydrological water level station or the real-time monitoring data of the study area, the flow process at the upstream boundary section and the water level process at the downstream boundary section in the typical period are selected as the boundary input data for model construction;

[0098] Set up the hydrodynamic control equations, where the hydrodynamic control equations are in the form of:

[0099]

[0100] In the formula, represents the conservation vector, t represents the time, and denote the convective flux in the x and y directions, respectively, and denote the diffusion flux in the x and y directions respectively; S denotes the source term.

[0101] Specifically, mesh refinement refers to increasing the density of the mesh. Figure 2 and Figure 3 , Figure 2 Schematic diagram of a triangular mesh of a two-dimensional hydrodynamic model in an embodiment of the present invention. Figure 3 Schematic diagram of the triangular mesh encryption area constructed in an embodiment of the present invention.

[0102] In the specific implementation process, the expressions of the conservation quantities in the hydrodynamic control equations are as follows:

[0103]

[0104]

[0105]

[0106]

[0107]

[0108] In the formula, the physical quantities are: h represents water depth; u represents the average vertical velocity component in the x direction; v represents the average vertical velocity component in the y direction; b represents the elevation of the riverbed bottom; g represents the gravitational acceleration; It represents the turbulent viscosity coefficient in the horizontal direction and is related to the shear velocity of the bed surface.

[0109] As a preference, in order to comprehensively consider river hydraulics and meteorological driving factors, additional source terms and dissipation terms are introduced into the model equations, which can be expressed as:

[0110]

[0111] In the formula, as well as Represent the riverbed slopes in the x and y directions respectively, and their expressions are: as well as ; and They represent the bottom friction loss in the x and y directions, respectively, which are determined by the roughness, water depth and flow velocity; f represents the Coriolis force coefficient, which is determined according to the latitude of each node; and It represents the wind stress in the x and y directions, which is obtained through the wind speed and correlation coefficient of the atmospheric wind field at a height of ten meters; p represents the increase in static pressure on the river surface caused by the change in air pressure (a negative number represents a decrease).

[0112] In one implementation, historical monitoring data of hydrological and water level stations are combined with design data, a large number of flow processes are generated according to different upstream flood peak flows, and an input sample library is constructed, including:

[0113] Based on the historical monitoring data of hydrological and water level stations, considering the peak sizes of different natural floods, and combining the design data, the data of a peak size is interpolated to generate several flood process data sets;

[0114] Considering the reservoir regulation process within the study area, the upstream and downstream flood datasets are generated by combining the generated flood process datasets and reservoir operation experience rules;

[0115] Clean and format the generated upstream and downstream flood data sets to obtain the time series data of the upstream and downstream flood data sets, and match the upstream and downstream floods to obtain the matched time series data;

[0116] The matched data are integrated to form an input sample library.

[0117] Specifically, based on the collected historical monitoring data of the hydrological and water level stations and any flood peak size (a value), flood process data (a series of values) are obtained through interpolation. Among them, several flood process data sets include flood flow and flood water level data at each moment; the generated upstream and downstream flood data sets include flood flow data at each moment in the upstream and flood water level data at each moment in the downstream.

[0118] In one embodiment, based on the data content of the input sample library, the hydrodynamic control equations of the two-dimensional hydrodynamic model after verification are used to simulate and calculate the flood evolution changes in the study area, and the corresponding flood indicators are obtained based on the statistical calculation of the flood evolution changes in the study area to form an output sample library, including:

[0119] a. Extract the time series data of upstream and downstream flood data sets of the same flood process from the input sample library, use it to construct the input conditions of the two-dimensional hydrodynamic model, and use the hydrodynamic control equations to perform flood simulation calculations based on the input conditions;

[0120] b. Combined with the objects of interest within the research scope, the adjacent river sections are taken as key sections;

[0121] c. Based on the results of flood simulation calculations, i.e., the data of time-series water depth, flow velocity, flow flux and physical quantities (such as Froude number, etc.) in the calculation process at each two-dimensional grid node, the time-series change process of flow rate and water depth at key sections is extracted, and the corresponding flood indicators are statistically calculated, including peak flow, peak arrival time and peak water level;

[0122] d. Based on the calculated flood index, the peak-to-appearance time interval and regional disaster-affected area corresponding to the key sections of concern are calculated under the input conditions of the constructed two-dimensional hydrodynamic model. The peak-to-appearance time interval can be obtained by the time difference between the arrival time of the flood peak at the key section and the time of the upstream flood peak. The regional disaster-affected area can be obtained by counting the area occupied by the nodes where the flood peak water level appears in the region. The peak-to-appearance time interval and regional disaster-affected area corresponding to the key sections are associated with the flood peak flow;

[0123] e. Organize the associated data into a structured data format to form an output sample library.

[0124] Specifically, the objects of interest within the scope of the study include historical flood-prone areas, key villages or streets, and key downstream facilities and equipment. It should be noted that steps ad are executed cyclically, and each time a piece of data is extracted from the input sample library to perform the operation of steps ad until the calculation of all input sample libraries is completed, and the calculated key section flow and water level process, flood peak flow and water level, and the associated disaster-stricken area are organized into a structured data format to form an output sample library.

[0125] In one embodiment, a flood simulation calculation is performed using a group of hydrodynamic control equations according to the input conditions, including:

[0126] Based on the input conditions of the constructed two-dimensional hydrodynamic model, the water depth and velocity data of each grid node are calculated using the hydrodynamic control equations to obtain the river flood evolution changes as the result of the flood simulation calculation.

[0127] In one implementation, based on the input sample library and the output sample library, the Kriging method is used to establish a mapping relationship between upstream floods and flood indicators of key sections, and a rapid flood prediction model is constructed, including:

[0128] a. Match the input-output relationship of the input sample library and the output sample library;

[0129] b. Sampling and selecting a number of input-output relationships to form a calibration data set;

[0130] c. Use Kriging method to calculate Gaussian process regression of input-output relationship, obtain the response surface mapping relationship between different flood scenarios and hydraulic elements of key sections, and build a rapid flood prediction model;

[0131] d. After matching, sample and select data outside the calibration data set to form a validation data set. Verify the accuracy of the rapid flood prediction model based on this data set. If the accuracy does not meet the requirements, re-sample and use the Kriging method to calculate the Gaussian process regression of the input-output relationship, and adjust the model parameters until the accuracy meets the requirements.

[0132] In the specific implementation process, if the accuracy does not meet the requirements, re-sampling and the Gaussian process regression operation of calculating the input-output relationship using the Kriging method are performed, which means that when the accuracy does not meet the requirements, steps b and c are re-executed.

[0133] The Kriging model used in step c can be expressed as:

[0134] Assume that the sample points X={x1,x2,...,x n}, where x1 to x n represents n different upstream flood peak values; the corresponding actual response value Y={y1,y2,...,y n}, where y1 to y n It represents the set of flood peak flow, flood peak water level, peak time interval and the corresponding disaster area under the flow conditions of the key section under different flood scenarios. The relationship between can be expressed as:

[0135]

[0136] In the formula, (x) is the predicted response value of variable x; is the regression model after interpolation calculation; is the unknown coefficient; Z(x) means the mean is 0 and the variance is A random Gaussian process with zero covariance. The covariance matrix of Z(x) can be expressed as:

[0137]

[0138] In the formula, and Represents two different sample points in the sample space (i.e., constructing the calibration data set), i and j represent spatial location information, and have the relationship 1≤i,j≤n; the expression Represents the spatial correlation between two sample points, using the Gaussian model, that is, Calculate, n in the relationship represents the total number of sample space points, are the parameters of the model to be determined.

[0139] Among them, the maximum likelihood estimation that maximizes the response value is used to calculate the unknown coefficients in the model and variance , through the following maximum likelihood function:

[0140]

[0141] Take its about and The derivative of , we get the mean that maximizes the function and variance :

[0142]

[0143]

[0144] Where, F=[f(x1),f(x2),...,f(x n )]T, which means that each sample point The regression function f(x) after interpolation calculation on (1≤i≤n, n represents the total number of sample points n) i ); Y represents the actual response value y(x i ); R is the correlation matrix between sample points and corresponding response values, which can be described as:

[0145]

[0146] Finally, the response value calculated by Kriging method for any upstream flood peak is:

[0147]

[0148] See also Figure 4 , is a schematic diagram of the Kriging model response data constructed in the embodiment of the present invention. Different solid line trajectories represent the influence of different θ values ​​on the prediction results.

[0149] In the specific implementation process, the flood rapid prediction model is used to predict the flood indicators of key sections based on the upstream floods. Based on the results calculated by the flood rapid prediction model, the flood risk within the research scope can be analyzed and judged. The specific process input is:

[0150] a. Based on the predicted flood process or reservoir discharge conditions, transmit them as input conditions to the flood rapid prediction model after calibration and verification, and calculate important data such as peak flow, peak water level, peak time and disaster-affected area of ​​key sections downstream;

[0151] b. Combine the flood peak flow, water level, peak time and affected area data with the design flood and historical flood data of the study area, and give corresponding early warning reminders for areas that may be in danger.

[0152] The method proposed in the present invention is described below through specific examples.

[0153] In the central my country region, a tributary of the Yangtze River has an overall basin area of ​​about 1,500 square kilometers, which is susceptible to heavy rainstorms and produces large floods, affecting the flood discharge scheduling and decision-making of reservoirs along the way, and thus causing flooding and disaster risks to villages and towns downstream. Therefore, in response to this flood-prone phenomenon, it is necessary to develop and construct a targeted rapid flood prediction model to assist relevant personnel of reservoirs and basin management bureaus in rapid response and decision-making to heavy rains and floods.

[0154] Based on the basin water system map and design flood data obtained from the local basin management bureau, combined with public terrain data, satellite images, vector maps and other resources, the terrain within the scope is determined and the modeling scope within the study area is determined. The boundaries of the modeling scope are the main stream of the river and the part that flows into the tributaries, the areas on both sides of the river, the upstream and downstream boundary sections of the river, and the hillsides or high-lying areas on both sides. DEM data is collected within the modeling scope, including mountain terrain and river terrain, and the data is stored in the UTM coordinate system. The data element unit of DEM data is meter, with an accuracy of about 1m, which can be used to construct a two-dimensional flood evolution model within the modeling scope. In addition, after the modeling scope is determined, the local meteorological bureau, basin management bureau and water conservancy bureau are applied for and obtained to obtain the historical hydrological element data within the scope (including historical monitoring data of hydrological and water level stations and design data) for subsequent model calibration and verification related work.

[0155] According to the collected information and basic data, combined with satellite images, river bank surveying and mapping results, sections along the way and the actual needs of the local watershed management bureau, the areas in the modeling area that need to be meshed are determined, mainly the main stream and tributary mouths of the river, several key sections along the way, key villages downstream and flood-prone areas, etc. An unstructured triangular unit grid is constructed within the modeling range. The horizontal and vertical XY coordinates of each node of the grid use the UTM coordinate system that matches the DEM data. The mesh accuracy is encrypted for the encrypted area. The overall grid is gradually expanded from 1~2m accuracy to about 100m accuracy at the boundary of the mountain in the modeling range. The number of overall grid nodes is about 100,000.

[0156] Based on the collected DEM data, the node coordinates of the grid are matched and interpolated, and the terrain elevation data is assigned to each node of the grid for flood evolution calculation. According to the collected hydrological element data, the data of a flood period is selected as the input conditions of the upper and lower boundaries, and the initial water depth of the river is set to a uniform 1m to ensure the normal flow of the river. Based on the two-dimensional shallow water equations (hydrodynamic control equations), the water depth and the average vertical velocity at each node are calculated to obtain the evolution of the river flood. In this embodiment, the wind speed is not considered for the time being and the value is 0; p represents the increase in the static pressure on the river surface caused by the change in air pressure (negative number means decrease). In this embodiment, the normal atmospheric pressure is considered, so the value of p is 0.

[0157] Based on the collected historical monitoring data of hydrological and water level stations and design information, the data are sorted out to obtain data from several typical flood periods, and 50% of the data are selected for model parameter calibration. By continuously adjusting the correlation coefficient and roughness of each physical quantity in the model, and repeatedly using the two-dimensional shallow water equations to calculate the model results, the remaining 50% of the flood period data are used to verify the model until the Nash efficiency coefficient obtained by the model calibration verification reaches the expectation. In this embodiment, the NSE coefficient of the constructed and practical two-dimensional flood evolution model can reach an average of more than 0.85.

[0158] The historical flood process was analyzed using the collected and collated hydrological element data. In combination with local flood control requirements and project characteristics, several representative large floods that were unfavorable to flood control were selected from the measured data as typical examples. The design flood volume of the design section was used as control, and the corresponding flood volume of each area was calculated according to the proportion of the flood volume of each area in a typical year to the flood volume of the same flood in the design section. The natural flood process under the conditions of each typical year was obtained, and several flood process data sets were generated based on this process by interpolation.

[0159] In this embodiment, more than 1,000 flood process data were generated considering the accuracy of subsequent model establishment and the efficiency of model construction. In addition, since there is a reservoir regulation and storage process in the basin of the study area, the corresponding upstream and downstream flood data sets were generated by combining the obtained flood process data set and the local reservoir regulation experience rules, considering that the initial reservoir water level is at the daily flood control high water level.

[0160] The upstream and downstream data sets are cleaned to ensure that there are no missing data, and then the format is regularized and converted into the time-series data required for calculation. After matching the upstream floods with the downstream water level changes, the relevant data are integrated to form an input sample library.

[0161] Based on the content of the input sample library and combined with the grid data of unstructured triangular units, the hydrodynamic control equations are used to calculate the flood evolution changes in the study area. Based on the calculation results, statistical analysis of flood indicators is performed to obtain the peak flow, peak time, peak water level of the section under various flood conditions, and the affected area data along the way and villages. The data are matched with the input conditions and then integrated into the output sample library.

[0162] Based on the sample library of input and output, some data are sampled and selected for the construction and calibration of the rapid flood prediction model to form a calibration data set. The Gaussian process regression between the input and output data is calculated using the Kriging method to obtain the response surface mapping relationship between the flow process and the hydraulic elements of the key sections, and then a rapid flood prediction model is built.

[0163] Assume that the sample points X={x1,x2,...,x100}, where x1 to x 100 represents 100 different upstream flood peak values; the corresponding response value Y={y1,y2,...,y 100}, where y1 to y 100 It represents the peak flow, peak water level, peak time interval of key sections under different flood scenarios and the corresponding disaster-affected area under the flow conditions of the section. In this embodiment, the most appropriate Finally, the response value calculated by the Kriging method under any upstream flood peak is obtained.

[0164] The remaining data in the input and output sample library are selected for the verification of the rapid flood prediction model to form a verification data set; the accuracy of the rapid flood prediction model is verified based on the data set. In this embodiment, the required accuracy is to meet the average NSE coefficient of 0.7 or above. If the accuracy does not meet the requirements, the calibration process of the rapid flood prediction model is repeated, and the model parameters are adjusted before verification.

[0165] Combined with the predicted flood process provided by the local management bureau personnel, it is passed as an input condition to the rapid flood prediction model and calculated to obtain important data such as the peak flow, peak water level, peak time interval and disaster-stricken area of ​​the downstream key sections; after multiple tests by personnel, the calculation response time of the model established in this embodiment is within 30s, which can be used to assist local management personnel in making rapid decisions and responses. By comparing with the design flood and historical flood data in the research scope, it can assist personnel in giving corresponding early warning reminders for areas that may be dangerous, helping local residents and management units to safely pass through possible flood disasters during the flood season.

[0166] Embodiment 2

[0167] Based on the same inventive concept, this embodiment discloses a rapid flood prediction device based on a Kriging model, see Figure 5 ,include:

[0168] The data collection module 501 is used to determine the modeling scope according to the research area, and collect digital elevation model data, historical monitoring data of hydrological and water level stations, and design information within the modeling scope;

[0169] A two-dimensional hydrodynamic model construction module 502 is used to construct a two-dimensional hydrodynamic model based on a group of hydrodynamic control equations using an unstructured triangular unit grid;

[0170] Model calibration module 503, used to calibrate and verify the two-dimensional hydrodynamic model using historical monitoring data of hydrological and water level stations and real-time monitoring data of the study area;

[0171] An input sample library construction module 504 is used to combine the historical monitoring data of the hydrological and water level stations with the design data, generate a large number of flow processes according to different upstream flood peak flows, and construct an input sample library;

[0172] The output sample library construction module 505 is used to simulate and calculate the flood evolution changes in the study area based on the data content of the input sample library using the hydrodynamic control equations of the two-dimensional hydrodynamic model after the verification, and obtain the corresponding flood indicators based on the statistical calculation of the flood evolution changes in the study area to form an output sample library;

[0173] The flood rapid prediction model building module 506 is used to establish a mapping relationship between upstream floods and key section flood indicators based on the input sample library and the output sample library by using the Kriging method, and to build a flood rapid prediction model;

[0174] The flood prediction module 507 is used to predict the flood index of the key section based on the upstream flood using the rapid flood prediction model.

[0175] Since the device introduced in the second embodiment of the present invention is a device used to implement the flood rapid prediction method based on the Kriging model in the first embodiment of the present invention, based on the method introduced in the first embodiment of the present invention, the person skilled in the art can understand the specific structure and deformation of the device, so it is not repeated here. All devices used in the method in the first embodiment of the present invention belong to the scope of protection of the present invention.

[0176] Embodiment 3

[0177] Based on the same inventive concept, the present invention further provides a computer-readable storage medium on which a computer program is stored. When the program is executed by a processor, the method described in the first embodiment is implemented.

[0178] Since the computer-readable storage medium introduced in the third embodiment of the present invention is a computer-readable storage medium used to implement the flood rapid prediction method based on the Kriging model in the first embodiment of the present invention, based on the method introduced in the first embodiment of the present invention, those skilled in the art can understand the specific structure and deformation of the computer-readable storage medium, so it is not repeated here. All computer-readable storage media used in the method of the first embodiment of the present invention belong to the scope of protection of the present invention.

[0179] Embodiment 4

[0180] The present invention also provides a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the method described in Embodiment 1 when executing the program.

[0181] Since the computer device introduced in the fourth embodiment of the present invention is a computer device used to implement the method for rapid flood prediction based on the Kriging model in the first embodiment of the present invention, based on the method introduced in the first embodiment of the present invention, the person skilled in the art can understand the specific structure and deformation of the computer device, so it is not repeated here. All computer devices used in the method of the first embodiment of the present invention belong to the scope of protection of the present invention.

[0182] It will be appreciated by those skilled in the art that embodiments of the present invention may be provided as methods, systems, or computer program products. Therefore, the present invention may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Furthermore, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0183] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0184] Although preferred embodiments of the present invention have been described, additional changes and modifications may be made to these embodiments by those skilled in the art once the basic creative concepts are known. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments and all changes and modifications that fall within the scope of the present invention. Obviously, those skilled in the art may make various changes and modifications to the embodiments of the present invention without departing from the spirit and scope of the embodiments of the present invention. Thus, if these modifications and variations of the embodiments of the present invention fall within the scope of the claims of the present invention and their equivalents, the present invention is also intended to include these modifications and variations.

Claims

1. A rapid flood prediction method based on Kriging model, characterized in that: include: Determine the modeling scope based on the study area, collect digital elevation model data, historical monitoring data of hydrological and water level stations, and design information within the modeling scope; Based on the hydrodynamic control equations, a two-dimensional hydrodynamic model is constructed using unstructured triangular element grids. The two-dimensional hydrodynamic model was calibrated and verified using historical monitoring data from hydrological and water level stations and real-time monitoring data from the study area; Combining historical monitoring data from hydrological and water level stations with design information, a large number of flow processes are generated according to different upstream flood peak flows to construct an input sample library; Based on the data content of the input sample library, the hydrodynamic control equations of the two-dimensional hydrodynamic model after verification are used to simulate and calculate the flood evolution changes in the study area, and the corresponding flood indicators are obtained based on the statistical calculation of the flood evolution changes in the study area to form an output sample library; Based on the input sample library and the output sample library, the Kriging method is used to establish the mapping relationship between upstream floods and key section flood indicators, and to build a rapid flood prediction model; Use the rapid flood prediction model to predict flood indicators of key sections of upstream floods; Among them, combining the historical monitoring data of the hydrological and water level stations with the design data, a large number of flow processes are generated according to different upstream flood peak flows, and an input sample library is constructed, including: Based on the historical monitoring data of hydrological and water level stations, considering the peak sizes of different natural floods, and combining the design data, the data of a peak size is interpolated to generate several flood process data sets; Considering the reservoir regulation process within the study area, the upstream and downstream flood datasets are generated by combining the generated flood process datasets and reservoir operation experience rules; Clean and format the generated upstream and downstream flood data sets to obtain the time series data of the upstream and downstream flood data sets, and match the upstream and downstream floods to obtain the matched time series data; Integrate the matched data to form an input sample library; Based on the data content of the input sample library, the hydrodynamic control equations of the two-dimensional hydrodynamic model after verification are used to simulate and calculate the flood evolution changes in the study area, and the corresponding flood indicators are obtained based on the statistical calculation of the flood evolution changes in the study area to form an output sample library, including: Extract the time series data of upstream and downstream flood data sets of the same flood process from the input sample library, use them to construct the input conditions of the two-dimensional hydrodynamic model, and use the hydrodynamic control equations to perform flood simulation calculations based on the input conditions; Combined with the objects of interest within the research scope, the adjacent river sections are taken as key sections; Based on the results of flood simulation calculation, the time series change process of flow rate and water depth of key sections is extracted, and the corresponding flood indicators are calculated statistically, including peak flow rate, peak arrival time and peak water level; Based on the calculated flood index, the peak time interval and regional disaster area corresponding to the key sections of concern are calculated under the input conditions of the constructed two-dimensional hydrodynamic model, and the peak time interval and regional disaster area corresponding to the key sections are associated with the peak flow; Organize the associated data into a structured data format to form an output sample library; Based on the input sample library and the output sample library, the Kriging method is used to establish the mapping relationship between upstream floods and key section flood indicators, and to build a rapid flood prediction model, including: Matching the input-output relationship of the obtained input sample library and output sample library; Sampling and selecting several input-output relationships to form a calibration data set; The Gaussian process regression of the input-output relationship is calculated using the Kriging method to obtain the response surface mapping relationship between different flood scenarios and hydraulic elements of key sections, and to build a rapid flood prediction model; After matching, data outside the calibration data set is sampled and selected from the input sample library and output sample library to form a verification data set. The accuracy of the rapid flood prediction model is verified based on the data set. If the accuracy does not meet the requirements, re-sampling and Gaussian process regression operations of the input-output relationship are calculated using the Kriging method, and the model parameters are adjusted until the accuracy meets the requirements.

2. The method for rapid flood prediction based on the Kriging model according to claim 1, characterized in that: Determine the modeling scope based on the study area, collect digital elevation model data, historical monitoring data of hydrological and water level stations, and design information within the modeling scope, including: The terrain within the research area is determined based on the vector map and satellite image of the research area. The modeling scope is divided around the river and the areas on both sides of the river. The boundary sections of the upstream and downstream of the river and the hillsides on both sides or the areas where the terrain meets the preset conditions are used as the boundaries of the modeling scope. Collect digital elevation model data within the modeling range; Collect historical hydrological monitoring data and design information within the modeling scope.

3. The method for rapid flood prediction based on the Kriging model according to claim 1, characterized in that: Based on the hydrodynamic control equations, a two-dimensional hydrodynamic model is constructed using an unstructured triangular unit grid, including: Based on the collected digital elevation model data, historical monitoring data of hydrological and water level stations and design information within the modeling range, combined with satellite images, river bank and shoreline mapping results, sections or villages along the way and actual needs, determine the areas in the study area that need to be meshed, and gradually reduce the mesh density from the river bank to the hillside or areas that meet the preset conditions; Based on the unstructured grid of irregular triangles, a two-dimensional finite element grid is constructed in the study area, and the terrain elevation variable information is assigned at each grid node to carry out flood evolution calculation; Based on the historical monitoring data of the hydrological water level station or the real-time monitoring data of the study area, the flow process at the upstream boundary section and the water level process at the downstream boundary section in the typical period are selected as the boundary input data for model construction; Set up the hydrodynamic control equations, where the hydrodynamic control equations are in the form of: In the formula, represents the conservation vector, t represents the time, and denote the convective flux in the x and y directions, respectively, and denote the diffusion flux in the x and y directions respectively; S denotes the source term.

4. The method for rapid flood prediction based on the Kriging model according to claim 1, characterized in that: Based on the input conditions, the flood simulation calculation is carried out using the hydrodynamic control equations, including: Based on the input conditions of the constructed two-dimensional hydrodynamic model, the water depth and velocity data of each grid node are calculated using the hydrodynamic control equations to obtain the river flood evolution changes as the result of the flood simulation calculation.

5. A rapid flood prediction device based on Kriging model, characterized in that: include: The data collection module is used to determine the modeling scope according to the study area, and collect digital elevation model data, historical monitoring data of hydrological and water level stations, and design information within the modeling scope; A two-dimensional hydrodynamic model building module is used to build a two-dimensional hydrodynamic model based on the hydrodynamic control equations using an unstructured triangular unit grid; Model calibration module, used to calibrate and verify the two-dimensional hydrodynamic model using historical monitoring data of hydrological and water level stations and real-time monitoring data of the study area; The input sample library construction module is used to combine the historical monitoring data of the hydrological and water level stations with the design data, generate a large number of flow processes according to different upstream flood peak flows, and construct the input sample library; The output sample library construction module is used to simulate and calculate the flood evolution changes in the study area based on the data content of the input sample library using the hydrodynamic control equations of the two-dimensional hydrodynamic model after the calibration verification, and obtain the corresponding flood indicators based on the statistical calculation of the flood evolution changes in the study area to form the output sample library; The flood rapid prediction model building module is used to establish the mapping relationship between upstream floods and key section flood indicators based on the input sample library and the output sample library using the Kriging method, and to build a flood rapid prediction model; The flood prediction module is used to predict flood indicators of key sections based on upstream floods using a rapid flood prediction model; The input sample library construction module is specifically used for: Based on the historical monitoring data of hydrological and water level stations, considering the peak sizes of different natural floods, and combining the design data, the data of a peak size is interpolated to generate several flood process data sets; Considering the reservoir regulation process within the study area, the upstream and downstream flood datasets are generated by combining the generated flood process datasets and reservoir operation experience rules; Clean and format the generated upstream and downstream flood data sets to obtain the time series data of the upstream and downstream flood data sets, and match the upstream and downstream floods to obtain the matched time series data; Integrate the matched data to form an input sample library; The output sample library construction module is specifically used for: Extract the time series data of upstream and downstream flood data sets of the same flood process from the input sample library, use them to construct the input conditions of the two-dimensional hydrodynamic model, and use the hydrodynamic control equations to perform flood simulation calculations based on the input conditions; Combined with the objects of interest within the research scope, the adjacent river sections are taken as key sections; Based on the results of flood simulation calculation, the time series change process of flow rate and water depth of key sections is extracted, and the corresponding flood indicators are calculated statistically, including peak flow rate, peak arrival time and peak water level; Based on the calculated flood index, the peak time interval and regional disaster area corresponding to the key sections of concern are calculated under the input conditions of the constructed two-dimensional hydrodynamic model, and the peak time interval and regional disaster area corresponding to the key sections are associated with the peak flow; Organize the associated data into a structured data format to form an output sample library; The flood rapid prediction model construction module is specifically used for: Matching the input-output relationship of the obtained input sample library and output sample library; Sampling and selecting several input-output relationships to form a calibration data set; The Gaussian process regression of the input-output relationship is calculated using the Kriging method to obtain the response surface mapping relationship between different flood scenarios and hydraulic elements of key sections, and to build a rapid flood prediction model; After matching, data outside the calibration data set is sampled and selected from the input sample library and output sample library to form a verification data set. The accuracy of the rapid flood prediction model is verified based on the data set. If the accuracy does not meet the requirements, re-sampling and Gaussian process regression operations of the input-output relationship are calculated using the Kriging method, and the model parameters are adjusted until the accuracy meets the requirements.

6. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the method for rapid flood prediction based on the Kriging model as described in any one of claims 1 to 4 is implemented.

7. A computer device comprising a memory, a processor and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the program, the method for rapid flood prediction based on the Kriging model as described in any one of claims 1 to 4 is implemented.

Images