Flood disaster situation prediction method and device, computer equipment, storage medium and product

Abstract

The invention relates to the technical field of disaster situation prediction, and discloses a flood disaster situation prediction method and device, computer equipment, a storage medium and a product. The historical data set comprises historical disaster data of the target area and historical factor values of flood disaster situation influence factors of the target area; constructing a generated data set based on the historical data set, wherein the generated data set comprises generated disaster data and generated factor values; training a flood disaster situation prediction model based on the historical data set and the generated data set; obtaining a current factor value of the flood disaster situation influence factor of the target area, and determining rainfall prediction data of the target area based on the current factor value; and taking the rainfall prediction data and the current factor value as input features of the flood disaster situation prediction model to obtain flood disaster situation prediction data. According to the method, the disaster situation estimation efficiency and precision can be improved.

Description

Technical Field

[0001] The present invention relates to the technical field of disaster prediction, and in particular to a flood disaster prediction method, device, computer equipment, storage medium and product. Background Art

[0003] Traditional disaster statistics methods still rely mainly on post-disaster reporting by governments at all levels, which consumes manpower and material resources and has the risk of missing the best time for rescue. At the same time, disaster warning systems often rely on different hydrological and meteorological observations and geographic information systems to estimate the affected areas. The hydrological simulation of this method for natural disaster events mostly stops at variables such as water level and flow of the event itself, and cannot be directly linked to disaster losses. At the same time, hydrological simulation uses the watershed as the calculation unit, which does not match the target and spatial scope of the disaster loss estimation of the administrative area in actual emergency decision-making, and cannot achieve effective flood disaster prediction. Summary of the invention

[0004] In view of this, the present invention provides a flood disaster prediction method, device, computer equipment, storage medium and product, which can improve the efficiency and accuracy of disaster prediction.

[0005] In a first aspect, the present invention provides a flood disaster prediction method, which includes: obtaining a historical data set of a target area, the historical data set including historical disaster-affected data of the target area and historical factor values ​​of flood disaster influencing factors of the target area; constructing a generated data set based on the historical data set, the generated data set including generated disaster-affected data and generated factor values; training a flood disaster prediction model based on the historical data set and the generated data set; obtaining current factor values ​​of flood disaster influencing factors in the target area, and determining precipitation prediction data for the target area based on the current factor values; using the precipitation prediction data and the current factor values ​​as input features of the flood disaster prediction model to obtain flood disaster prediction data.

[0006] In this implementation, by constructing a generated data set based on the historical data set, the problem of insufficient sample size of disaster data in traditional statistics, which leads to overfitting of the model and inaccurate prediction, can be solved. At the same time, by jointly training the flood disaster prediction model based on the constructed historical data and the generated data set, the accuracy of the flood disaster prediction model can be improved with the real seat labels of the affected data. Furthermore, in the actual application process, the precipitation forecast data is used as the input data of the flood disaster prediction model, and the flood disaster losses in the target area can be quickly predicted through hydrological and meteorological observation data and related flood disaster influencing factors, thereby improving the efficiency and accuracy of disaster prediction and providing technical support for flood emergency management and resource allocation.

[0007] In an optional implementation, the target area includes a target administrative area and a target watershed area, the historical factor values ​​include historical geographical factor values ​​and historical precipitation values, and obtaining a historical data set of the target area includes: determining the target administrative area to be predicted; obtaining historical disaster data and historical geographical factor values ​​of the target administrative area; expanding the scope based on the target administrative area to obtain a target watershed area; and obtaining historical precipitation values ​​of the target watershed area.

[0008] In this implementation, by dividing the administrative regions and watershed areas, the watershed scope of the rainfall runoff process can be considered, and the historical precipitation factor values ​​and historical precipitation values ​​can be obtained in the administrative regions and watershed areas respectively, which can further improve the comprehensiveness of historical data collection and improve the accuracy of subsequent model predictions.

[0009] In an optional implementation, a generated data set is constructed based on a historical data set, and the generated data set includes generated disaster data and generated factor values, including: inputting the historical data set into a conditional generative adversarial model, the conditional generative adversarial model includes a generator and a discriminator, the generator is used to obtain generated factor values ​​based on historical disaster data, and the discriminator is used to compare and discriminate the generated factor values ​​and historical factor values; iteratively training the generator and the discriminator to output the generated factor values, and using the historical disaster data corresponding to the generated factor values ​​as the generated disaster data.

[0010] In this implementation, a conditional generative adversarial model is used with real disaster data as supervision to generate a generated data set that conforms to the flood disaster characteristics of the target area. This can improve the quality of the generated data set, expand the data set, and solve the problem of insufficient sample size of disaster data in traditional statistics, resulting in model overfitting and inaccurate predictions.

[0011] In an optional embodiment, a flood disaster prediction model includes a convolutional neural layer, a long short-term memory layer, and a fully connected layer. The flood disaster prediction model is trained based on a historical data set and a generated data set, including: extracting features of historical factor values ​​and generated factor values ​​based on the convolutional neural layer; performing time series analysis on historical factor values ​​and generated factor values ​​based on the long short-term memory layer; connecting the convolutional neural layer and the long short-term memory layer based on the fully connected layer, and performing flood disaster prediction to obtain training disaster prediction data; comparing the training disaster prediction data with the corresponding disaster data to adjust the parameters of the flood disaster prediction model to obtain a trained flood disaster prediction model.

[0012] In this implementation, by constructing a flood disaster prediction model and training the flood disaster prediction model based on real data sets and generated data sets, the spatial and temporal characteristics of the data can be effectively extracted, thereby improving the robustness of the flood disaster prediction model.

[0013] In an optional implementation, precipitation forecast data for a target area is determined based on current factor values, including: inputting current factor values ​​into a meteorological research and forecasting model; simulating meteorological change processes based on a variety of meteorological equations in the meteorological research and forecasting model to obtain precipitation forecast data.

[0014] In this implementation, the meteorological research and forecasting model is used to predict precipitation data, which can efficiently and accurately simulate complex phenomena such as atmospheric circulation, the development and evolution of weather systems, and the formation of clouds and precipitation, extract refined precipitation data related to flood disasters, solve the uncertainty of the output caused by precipitation forecasts, and effectively improve the accuracy of disaster forecast simulations.

[0015] In an optional implementation, precipitation prediction data and current factor values ​​are used as input features of a flood disaster prediction model to obtain flood disaster prediction data, including: determining the corresponding precipitation prediction data and current precipitation values ​​based on the target prediction days, the precipitation prediction data and current precipitation values ​​being data within a continuous time period; using the current geographical factor values, current precipitation values ​​and precipitation prediction data as input features of the flood disaster prediction model, using the flood disaster prediction model to perform predictions, and obtaining flood disaster prediction data for the target prediction days.

[0016] In this implementation, based on the future precipitation forecast and combined with the current geographical factor values, it is possible to quickly make quantitative estimates of different disaster assessment indicators by administrative region, providing effective support for flood disaster management and resource allocation.

[0017] In a second aspect, the present invention provides a flood disaster prediction device, which includes: a first acquisition module, used to acquire a historical data set of a target area, the historical data set including historical disaster-affected data of the target area and historical factor values ​​of flood disaster influencing factors of the target area; a construction module, used to construct a generated data set based on the historical data set, the generated data set including generated disaster-affected data and generated factor values; a training module, used to train a flood disaster prediction model based on the historical data set and the generated data set; a second acquisition module, used to acquire current factor values ​​of flood disaster influencing factors in the target area, and determine precipitation prediction data of the target area based on the current factor values; a prediction module, used to use the precipitation prediction data and the current factor values ​​as input features of the flood disaster prediction model to obtain flood disaster prediction data.

[0018] In a third aspect, the present invention provides a computer device, comprising: a memory and a processor, the memory and the processor are communicatively connected to each other, the memory stores computer instructions, and the processor executes the flood disaster prediction method of the first aspect or any corresponding embodiment thereof by executing the computer instructions.

[0019] In a fourth aspect, the present invention provides a computer-readable storage medium having computer instructions stored thereon, the computer instructions being used to enable a computer to execute the flood disaster prediction method of the first aspect or any corresponding embodiment thereof.

[0020] In a fifth aspect, the present invention provides a computer program product, comprising computer instructions for causing a computer to execute the flood disaster prediction method of the first aspect or any corresponding embodiment thereof. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] In order to more clearly illustrate the specific implementation methods of the present invention or the technical solutions in the prior art, the drawings required for use in the specific implementation methods or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are some implementation methods of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0022] Figure 1 is a flow chart of a flood disaster prediction method according to an embodiment of the present invention;

[0023] Figure 2 is a flow chart of another flood disaster prediction method according to an embodiment of the present invention;

[0024] Figure 3 is a schematic diagram of a conditional generative adversarial model according to an embodiment of the present invention;

[0025] Figure 4 is a structural block diagram of a flood disaster prediction device according to an embodiment of the present invention;

[0026] Figure 5 It is a schematic diagram of the hardware structure of a computer device according to an embodiment of the present invention. DETAILED DESCRIPTION

[0027] 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 those skilled in the art without creative work are within the scope of protection of the present invention.

[0028] The frequent occurrence of extreme weather events has led to heavy rain and flood disasters, which have caused serious damage to human society, economy and ecological environment. The current disaster statistics method still mainly relies on the post-disaster reporting of governments at all levels, which consumes manpower and material resources, and there is a risk of missing the best time for rescue. At the same time, the disaster warning system often relies on different hydrological and meteorological observations and geographic information systems to estimate the affected areas, but this method has certain limitations when dealing with complex and changeable rainfall processes. On the one hand, due to the uncertainty of precipitation, there is uncertainty in hydrological forecasts; on the other hand, the forecast of natural disaster events by hydrological simulation mostly stops at variables such as water level and flow of the event itself, which cannot be directly related to disaster losses; in addition, hydrological simulation uses the watershed as the calculation unit, which has a mismatch between the target and spatial scope of the disaster loss estimation of the administrative area required in actual emergency decision-making.

[0029] Therefore, this application proposes a flood disaster prediction method, which can solve the problem of inaccurate disaster prediction by determining the spatial scope of actual emergency decision-making, associating the factors affecting disasters with actual disaster information, and accurately predicting the disaster situation.

[0030] According to an embodiment of the present invention, a flood disaster prediction method embodiment is provided. It should be noted that the steps shown in the flowchart of the accompanying drawings can be executed in a computer system such as a set of computer executable instructions, and although a logical order is shown in the flowchart, in some cases, the steps shown or described can be executed in an order different from that shown here.

[0031] In this embodiment, a flood disaster prediction method is provided. Figure 1 is a flow chart of a flood disaster prediction method according to an embodiment of the present invention. It should be noted that if there are substantially the same results, this embodiment does not use Figure 1 The process sequence shown is limited. Figure 1 As shown, the process includes the following steps:

[0032] Step S101, obtaining a historical data set of a target area.

[0033] The historical data set includes the historical disaster data of the target area and the historical factor values ​​of the factors affecting the flood disaster in the target area.

[0034] Among them, the disaster data include disaster assessment indicators such as the affected population, affected houses, affected agriculture, and affected economy.

[0035] Among them, the factors affecting flood disasters include elevation, slope, slope direction, population density, river distance, land use type, administrative area coverage, daily precipitation data, runoff, river water level, etc.

[0036] Specifically, the target area to be predicted is determined, and the disaster-affected time and affected area of ​​historical flood disasters that occurred in the target area are obtained, and the corresponding historical disaster data and historical factor values ​​are obtained according to the corresponding disaster-affected time and affected area.

[0037] In one implementation, historical disaster data of the target area is obtained based on disaster statistical data such as meteorological disaster yearbooks and documents of the Ministry of Water Resources.

[0038] In one implementation, historical elevation data is obtained using a public geospatial data cloud website, slope data and aspect data are extracted using the ArcToolbox toolbox in ArcGIS, the distance between each grid center point in the target area and the river control station is calculated using ArcGIS, the density of the affected population is calculated using the ratio of the number of affected people in the target area to the area of ​​the target area, and the land use type is obtained using a public geographic information resource directory service system.

[0039] Step S102, constructing a generated data set based on the historical data set, where the generated data set includes generated disaster data and generated factor values.

[0040] According to the statistical laws of the historical data set, a generated data set that is different from the historical data set is reconstructed.

[0041] Specifically, historical disaster data and historical factor values ​​corresponding to each historical flood disaster are obtained. For each set of historical disaster data, generating factor values ​​different from the historical factor values ​​are constructed according to the data distribution characteristics of the historical factor values, and the historical disaster data are used as the generated disaster data corresponding to the generating factor values.

[0042] In one implementation, the data distribution characteristic indicators include probability density function (PDF), cumulative density function (CDF), auto-correlation function (ACF), and by analyzing the data distribution characteristic indicators of the historical factor values ​​and the data distribution characteristic indicators of the constructed generation factor values, the data distribution of the generation factor values ​​is adjusted so that the historical factor values ​​and the generation factor values ​​have a high similarity. Disaster data and generation factor values ​​are further generated.

[0043] Step S103: training based on the historical data set and the generated data set.

[0044] Specifically, the historical data set and the generated data set are combined to construct a training data set. The factor values ​​are used as input data and the disaster-affected data are used as labels to train the disaster warning model.

[0045] Exemplarily, the disaster warning model includes a multilayer perceptron model (Multilayer Perceptron, MLP), convolutional neural networks (Convolutional Neural Networks, CNN), recurrent neural networks (Recurrent Neural Networks, RNN), long short-term memory networks (Long Short-Term Memory, LSTM), gated recurrent units (Gated Recurrent Unit, GRU) and other models or multiple combined models.

[0046] Step S104, obtaining current factor values ​​of flood disaster influencing factors in the target area, and determining precipitation prediction data for the target area based on the current factor values.

[0047] Specifically, when conducting real-time flood disaster analysis, the current factor value of each flood disaster influencing factor in the target area under the current situation is obtained, and the impact of the current factor value on precipitation is analyzed to obtain precipitation forecast data for the target area within a certain period of time.

[0048] In one implementation, historical factor values ​​and corresponding historical precipitation data are obtained to analyze the influence of factors on precipitation, and precipitation data is predicted based on current factor values. Exemplarily, a precipitation prediction model is constructed, historical factor values ​​are used as input data, historical precipitation data are used as labels, and the precipitation prediction model is trained. The current factor values ​​are used as input data, and precipitation prediction data is predicted using the precipitation prediction model.

[0049] In another implementation, a meteorological precipitation analysis equation is constructed, and the current factor values ​​are used to analyze meteorological changes and predict precipitation data.

[0050] Step S105, using the precipitation prediction data and the current factor value as input features to obtain flood disaster prediction data.

[0051] Specifically, precipitation forecast data within the first time is obtained, and the precipitation forecast data within the first time and the current factor value are used together as input data after training to predict the flood disaster data of the target area within the first time.

[0052] Exemplarily, the flood disaster data include the flood disaster loss values ​​of population, housing, agriculture, and economy in the target area caused by precipitation.

[0053] The flood disaster prediction method provided in this embodiment can solve the problem of overfitting of the model and inaccurate prediction caused by insufficient sample size of disaster data in traditional statistics by constructing a generated data set based on historical data sets. At the same time, by jointly training the flood disaster prediction model based on the constructed historical data and the generated data set, the accuracy of the flood disaster prediction model can be improved with real seat labels of disaster-affected data. Furthermore, in actual application, precipitation forecast data is used as input data of the flood disaster prediction model, and the flood disaster losses in the target area can be quickly predicted through hydrological and meteorological observation data and related flood disaster influencing factors, thereby improving the efficiency and accuracy of disaster prediction and providing technical support for flood emergency management and resource allocation.

[0054] In this embodiment, a flood disaster prediction method is provided. Figure 2 is a flow chart of another flood disaster prediction method according to an embodiment of the present invention. It should be noted that if there are substantially the same results, this embodiment does not use Figure 2 The process sequence shown is limited. Figure 2 As shown, the process includes the following steps:

[0055] Step S201, obtaining a historical data set of a target area.

[0056] The historical data set includes the historical disaster data of the target area and the historical factor values ​​of the flood disaster influencing factors in the target area. The historical factor values ​​include historical geographical factor values ​​and historical precipitation values.

[0057] Specifically, the above step S201 includes:

[0058] Step S2011, determine the target administrative area to be predicted, and obtain the historical disaster data and historical geographical factor values ​​of the target administrative area.

[0059] Specifically, administrative divisions are performed on target areas where flood disasters occur to obtain corresponding target administrative areas, and historical disaster data and historical geographical factor values ​​within the spatial scope of the target administrative areas are acquired.

[0060] Exemplarily, all grid records covered by the target administrative area are D1, the longitude east boundary of the target administrative area is D1lon_max, the longitude west boundary is D1lon_min, the latitude south boundary is D1lat_min, and the latitude north boundary is D1lat_max. Obtain disaster data such as the affected population, affected houses, affected agriculture, and affected economy in the target administrative area, and obtain historical geographical factor values ​​such as elevation, slope, slope direction, population density, river distance, land use type, and administrative area coverage.

[0061] Furthermore, the time lengths of historical and geographical factor values ​​are unified and normalized.

[0062] Step S2012, expanding the scope based on the target administrative area, obtaining the target watershed area, and acquiring the historical precipitation value of the target watershed area.

[0063] Specifically, the precipitation runoff process is expanded outward within the target administrative area to obtain the target watershed area, and the historical precipitation values ​​within the target watershed area are obtained.

[0064] For example, all grid records covered by the target watershed area are P1, and the longitude east boundary of the target watershed area is determined as P1_e, the longitude west boundary is P1_w, the latitude south boundary is P1_s, and the latitude north boundary is P1_n according to the area range of the target administrative area. The calculation formula is:

[0065] P1_e=D1lon_max+|D1lon_max-D1lon_min|.

[0066] P1_w=D1lon_min-|D1lon_max-D1lon_min|.

[0067] P1_n=D1lat_max+|D1lat_max-D1lat_min|.

[0068] P1_s=D1lat_max+|D1lat_max-D1lat_min|.

[0069] Obtain historical precipitation values ​​such as cumulative precipitation, precipitation duration, and precipitation intensity for each grid in the target watershed area.

[0070] In this implementation, by dividing the administrative regions and watershed areas, the watershed scope of the rainfall runoff process can be considered, and the historical precipitation factor values ​​and historical precipitation values ​​can be obtained in the administrative regions and watershed areas respectively, which can further improve the comprehensiveness of historical data collection and improve the accuracy of subsequent model predictions.

[0071] Step S202: constructing a generated data set based on the historical data set, wherein the generated data set includes generating disaster affected data and generating factor values.

[0072] According to the statistical laws of the historical data set, a generated data set that is different from the historical data set is reconstructed.

[0073] In one implementation, a generative adversarial module is used to construct a generated dataset to expand the sample data.

[0074] Specifically, the above step S202 includes:

[0075] Step S2021, input the historical data set into the conditional generation adversarial model.

[0076] Among them, the Conditional Generative Adversarial Network (CGAN) includes a generator and a discriminator. The generator model maps the random variable z sampled from the latent space with a prior noise distribution Pz to the data space. The discriminator model estimates the probability that the sample comes from the real training data rather than from the generated model sample.

[0077] Step S2022: Iteratively train the generator and the discriminator to output the generated factor values, and use the historical disaster data corresponding to the generated factor values ​​as the generated disaster data.

[0078] In one implementation, see Figure 3 , Figure 3 is a schematic diagram of a conditional generative adversarial model according to an embodiment of the present invention, wherein the generator is used to obtain a generated factor value based on historical disaster data, and the discriminator is used to compare and discriminate the generated factor value and the historical factor value.

[0079] Specifically, the model parameters of the conditional generative adversarial model, such as the number of neurons, dimension, learning rate, batch size, etc., are set. The historical disaster data in the constructed historical data set is used as the condition y, and the historical factor values ​​such as precipitation, runoff, and surface parameters are used as the feature x to combine and obtain {historical factor value|historical disaster data}.

[0080] This application adds historical disaster data as condition y in both the generator and the discriminator.

[0081] In the generator part, random noise z is input and condition y is superimposed, that is, {random noise|historical disaster data}, and the generated factor value xf is output to obtain the corresponding {generated factor value}. The generated factor value is used as the feature xf, and the historical disaster data is used as the condition y to obtain {generated factor value|historical disaster data}.

[0082] In the discriminator part, the generated factor value xf of the superimposed condition y, that is, {generated factor value|historical disaster data} and the historical factor value xr of the superimposed condition y, that is, {historical factor value|historical disaster data}, are input to calculate the probability that the generated factor value is close to the historical factor value under the same historical disaster data.

[0083] Furthermore, the distribution characteristics of the generated factor values ​​and historical factor values ​​are analyzed using probability density functions, cumulative density functions, autocorrelation functions, etc., the generator and discriminator are adjusted through the loss function, and the generator and discriminator are further used to discriminate and generate generated factor values ​​that have similar statistical laws to the historical factor values. Among them, the loss function is:

[0084]

[0085] in, is the objective function of the conditional adversarial neural network, Indicates that x follows the sample distribution p data expect; Indicates that z follows the sample distribution p z expect.

[0086] The generated generation factor values ​​and the corresponding historical disaster data are constructed to obtain a generation data set.

[0087] In this implementation, a conditional generative adversarial model is used with real disaster data as supervision to generate a generated data set that conforms to the flood disaster characteristics of the target area. This can improve the quality of the generated data set, expand the data set, and solve the problem of insufficient sample size of disaster data in traditional statistics, resulting in model overfitting and inaccurate predictions.

[0088] Step S203: training based on the historical data set and the generated data set.

[0089] In one implementation, a CNN-LSTM prediction model is trained using a historical data set and a generated data set.

[0090] Among them, the CNN-LSTM prediction model includes convolutional neural layer, long short-term memory layer and fully connected layer.

[0091] Specifically, the convolutional neural layer includes a convolutional layer and a pooling layer. The convolutional neural layer performs static feature extraction on the spatial information of the historical factor values ​​and the generating factor values, wherein the spatial features can be effectively extracted from the disaster-causing factors and the effective information at different locations can be captured. The long short-term memory layer includes multiple long short-term memory units. The long short-term memory layer performs time series analysis on the historical factor values ​​and the generating factor values, extracts time information, and the output of each long short-term memory unit is used as the input of the next unit. The output of the long short-term memory unit corresponding to the last time node is used as the output data of the long short-term memory layer. The fully connected layer connects the convolutional neural layer and the long short-term memory layer, and performs flood disaster prediction to obtain training disaster prediction data.

[0092] The training disaster prediction data and the corresponding disaster-affected data are compared and adjusted to obtain the trained data.

[0093] Specifically, parameter training is performed. This application is used to predict flood disaster data covering multiple prediction tasks such as population, housing, agriculture, and economy. A loss function is constructed for each disaster task, and the weighted sum of the loss functions of each task is taken to obtain the total loss function.

[0094] In one implementation, the Pareto optimality principle is used to set the weight of each prediction task, and the weighted sum of the loss function of each task is taken to obtain the total loss function.

[0095] Among them, the method of setting weights can balance the training of each task so that each task can achieve beneficial improvements.

[0096] Furthermore, an evaluation model is set up, and the relative error between the estimated disaster situation and the actual disaster situation is calculated by using the evaluation model. When the relative error is less than or equal to the preset value, the model is used for subsequent disaster prediction, and when the relative error is greater than the preset value, the model is further adjusted.

[0097] In this implementation, by constructing a flood disaster prediction model and training the flood disaster prediction model based on real data sets and generated data sets, the spatial and temporal characteristics of the data can be effectively extracted, thereby improving the robustness of the flood disaster prediction model.

[0098] Step S204, obtaining current factor values ​​of flood disaster influencing factors in the target area, and determining precipitation prediction data for the target area based on the current factor values.

[0099] Specifically, when conducting real-time flood disaster analysis, the current factor value of each flood disaster influencing factor in the target area under the current situation is obtained, and the impact of the current factor value on precipitation is analyzed to obtain precipitation forecast data for the target area within a certain period of time.

[0100] In another implementation, a meteorological precipitation analysis equation is constructed, and the meteorological change situation is analyzed using the current factor values ​​to predict precipitation prediction data. Specifically, the above step S204 includes:

[0101] Step S2041, input the current factor value into the meteorological research and forecasting model.

[0102] Among them, the meteorological research and forecasting model includes a variety of atmospheric dynamics equations, such as the Navier-Stokes equations that describe fluid motion, the energy equations that control heat transfer and temperature changes in the atmosphere, and the water vapor equations that describe water vapor circulation, evaporation, condensation, etc. The atmospheric dynamics equations can describe physical processes such as the flow pattern of wind in the atmosphere, the temporal and spatial evolution of temperature, and the phase change and circulation of water vapor.

[0103] Step S2042, simulating the meteorological change process based on various meteorological equations in the meteorological research and forecasting model to obtain precipitation prediction data.

[0104] Specifically, the Weather Research and Forecasting Model (WRF) includes a WPS preprocessing system, an ARW dynamic solver, and a post-processing module.

[0105] Among them, the WPS preprocessing system includes geogrid program, ungrib program, metgrid program, and real program.

[0106] The geogrid program is responsible for determining the specific geographic location of the target area, the fineness of the grid, that is, the resolution, and the map projection method used to ensure that the simulation results can accurately reflect the geographical characteristics of the target area. Geogrid will interpolate the current factor values ​​provided, such as terrain height, land use type, soil type and its parameters, to adapt them to the set simulation grid, providing the necessary geographical background information for subsequent simulations.

[0107] The ungrib program is responsible for extracting the required boundary conditions and initial field data from the standard meteorological data format (such as GRIB) of global meteorological models such as GFS or ECMWF, and passing the extracted boundary conditions and initial field data to the metgrid program.

[0108] The metgrid program is responsible for horizontally interpolating data including wind speed, temperature, humidity, and air pressure onto the grid of the WRF simulation domain.

[0109] Furthermore, the real program is responsible for vertically interpolating data including wind speed, temperature, humidity, and air pressure onto the grid of the WRF simulation domain. It further uses observational data and model background fields, and through data analysis and assimilation techniques, accurately interpolates meteorological data onto the eta layer used by the WRF model to generate high-quality initial fields.

[0110] The ARW dynamic solver is responsible for performing numerical integration operations based on the current factor values. Specifically, the integration length, output frequency, simulation area size, time step, and parameterization scheme of physical processes (such as microphysical processes, cumulus convection parameterization, boundary layer processing, etc.) can be adjusted according to forecast requirements. Through these highly configurable options, the ARW solver can efficiently and accurately simulate complex phenomena such as atmospheric circulation, the development and evolution of weather systems, and the formation of clouds and precipitation.

[0111] The post-processing module includes a variety of NCL, Python scripts and other tools to conduct in-depth analysis and visualization of the simulation results after the ARW dynamic solver simulation is completed. Users can extract and display precipitation forecast data such as cumulative precipitation, precipitation rate, wind field, temperature field, etc. according to the needs of research or forecasting.

[0112] In this implementation, the meteorological research and forecasting model is used to predict precipitation data, which can efficiently and accurately simulate complex phenomena such as atmospheric circulation, the development and evolution of weather systems, and the formation of clouds and precipitation, extract refined precipitation data related to flood disasters, solve the uncertainty of the output caused by precipitation forecasts, and effectively improve the accuracy of disaster forecast simulations.

[0113] Step S205, using the precipitation prediction data and the current factor value as input features to obtain flood disaster prediction data.

[0114] Specifically, the above step S205 includes:

[0115] Step S2051, determining corresponding precipitation forecast data and current precipitation value based on the target forecast days.

[0116] Specifically, the corresponding precipitation forecast data and current precipitation value are determined according to the required precipitation data days and the target forecast days, wherein the precipitation forecast data and current precipitation value are data within a continuous time period.

[0117] For example, if precipitation data with a time length of 7 days is required and the target prediction days are 3 days, then 4 days of current precipitation values ​​and 3 days of precipitation prediction data are obtained.

[0118] Step S2052, using the current geographical factor value, the current precipitation value and the precipitation forecast data as input features, and using them to make predictions to obtain flood disaster forecast data for the target prediction days.

[0119] Specifically, the precipitation forecast data of the target forecast days and the current factor values ​​are used as the input data of the CNN-LSTM model to predict the flood disaster data of the target area in the first time.

[0120] Exemplarily, the input features of the CNN-LSTM model are elevation, slope, slope aspect, population density, river distance, land use type, administrative area coverage, 4-day current precipitation value, 3-day precipitation forecast data, and runoff data from the previous seven days; finally, the final prediction result of the CNN-LSTM model is output, namely the multi-dimensional flood disaster situation.

[0121] Among them, sliding input can estimate the dynamic multi-dimensional loss values ​​of the target area affected by flood disasters over time, and provide disaster warning information for decision makers.

[0122] The flood disaster prediction method provided in this embodiment solves the problem of overfitting of forecasts caused by insufficient flood disaster samples, the problem of decision-making timeliness caused by statistical delays, and the problem of mismatch between disaster event monitoring and estimation and the statistical spatial scope of disaster conditions. The flood disaster prediction is made more real-time, fast, accurate, and comprehensive. The conditional generative adversarial model is used to integrate data and generate more flood disaster scene samples to solve the problem of model overfitting caused by insufficient disaster data samples in traditional statistics. WRF is used to perform refined precipitation forecasts to solve the uncertainty of the output caused by precipitation forecasts, effectively improving the accuracy of disaster forecasts. At the same time, the method combines disaster-causing factors and basic socioeconomic assessment indicators of each administrative region, including population, housing, agriculture, and economic elements, to achieve rapid quantitative estimation of different disaster assessment indicators by administrative region, providing effective support for flood disaster management and resource allocation.

[0123] In this embodiment, a flood disaster prediction device is also provided, which is used to implement the above-mentioned embodiments and preferred implementation modes, and the descriptions that have been made will not be repeated. As used below, the term "module" can implement a combination of software and / or hardware of a predetermined function. Although the devices described in the following embodiments are preferably implemented in software, the implementation of hardware, or a combination of software and hardware, is also possible and conceivable.

[0124] This embodiment provides a flood disaster prediction device, such as Figure 4 As shown, including:

[0125] The first acquisition module 401 is used to acquire a historical data set of a target area, where the historical data set includes historical disaster data of the target area and historical factor values ​​of flood disaster influencing factors of the target area.

[0126] The construction module 402 is used to construct a generated data set based on the historical data set, and the generated data set includes generated disaster data and generated factor values.

[0127] The training module 403 is used for training based on the historical data set and the generated data set.

[0128] The second acquisition module 404 is used to obtain the current factor values ​​of the flood disaster influencing factors in the target area, and determine the precipitation prediction data of the target area based on the current factor values.

[0129] The prediction module 405 is used to use the precipitation prediction data and the current factor value as input features to obtain flood disaster prediction data.

[0130] In some optional implementations, the first acquisition module 401 includes:

[0131] The determination unit is used to determine the target administrative area to be predicted.

[0132] The first acquisition unit is used to acquire historical disaster data and historical geographical factor values ​​of the target administrative area.

[0133] The expansion unit is used to expand the scope based on the target administrative area to obtain the target watershed area.

[0134] The second acquisition unit is used to obtain historical precipitation values ​​of the target watershed area.

[0135] In some optional implementations, the construction module 402 includes:

[0136] The first input unit is used to input the historical data set into the conditional generative adversarial model. The conditional generative adversarial model includes a generator and a discriminator. The generator is used to obtain the generated factor value based on the historical disaster data, and the discriminator is used to compare and discriminate the generated factor value and the historical factor value.

[0137] The generating unit is used to iteratively train and output the generated factor value by using the generator and the discriminator, and use the historical disaster data corresponding to the generated factor value as the generated disaster data.

[0138] In some optional implementations, the training module 403 includes:

[0139] The feature extraction unit is used to extract features of the historical factor values ​​and the generated factor values ​​based on the convolutional neural layer.

[0140] The time series analysis unit is used to perform time series analysis on the historical factor values ​​and the generated factor values ​​based on the long short-term memory layer.

[0141] The connection unit is used to connect the convolutional neural layer and the long short-term memory layer based on the fully connected layer, and perform flood disaster prediction to obtain training disaster prediction data.

[0142] The training unit is used to compare and adjust the parameters of the training disaster prediction data and the corresponding disaster-affected data to obtain the trained data.

[0143] In some optional implementations, the prediction module 405 includes:

[0144] The second input unit is used to input the current factor value into the meteorological research and forecasting model.

[0145] The precipitation prediction unit is used to simulate the meteorological change process based on various meteorological equations in the meteorological research and forecasting model to obtain precipitation prediction data.

[0146] In some optional implementations, the second acquisition module 404 includes:

[0147] The determination unit is used to determine the corresponding precipitation prediction data and current precipitation value based on the target prediction days, and the precipitation prediction data and current precipitation value are data within a continuous time period.

[0148] The prediction unit is used to use the current geographical factor value, the current precipitation value and the precipitation prediction data as input features, and use them to make predictions to obtain the flood disaster prediction data for the target prediction days.

[0149] The further functional description of each of the above modules and units is the same as that of the above corresponding embodiments and will not be repeated here.

[0150] The flood disaster prediction device in this embodiment is presented in the form of a functional unit, where the unit refers to an ASIC (Application Specific Integrated Circuit) circuit, a processor and memory that executes one or more software or fixed programs, and / or other devices that can provide the above functions.

[0151] The embodiment of the present invention also provides a computer device having the above Figure 4 The flood disaster prediction device shown.

[0152] See also Figure 5 , Figure 5 is a schematic diagram of the structure of a computer device provided by an optional embodiment of the present invention, such as Figure 5 As shown, the computer device includes: one or more processors 10, a memory 20, and interfaces for connecting various components, including high-speed interfaces and low-speed interfaces. Various components are connected to each other using different buses for communication, and can be installed on a common mainboard or installed in other ways as needed. The processor can process the instructions executed in the computer device, including instructions stored in or on the memory to display the graphical information of the GUI on an external input / output device (such as, a display device coupled to the interface). In some optional embodiments, if necessary, multiple processors and / or multiple buses can be used together with multiple memories and multiple memories. Similarly, multiple computer devices can be connected, and each device provides some necessary operations (for example, as a server array, a group of blade servers, or a multi-processor system). Figure 5 A processor 10 is taken as an example.

[0153] The processor 10 may be a central processing unit, a network processor or a combination thereof. The processor 10 may further include a hardware chip. The hardware chip may be a dedicated integrated circuit, a programmable logic device or a combination thereof. The programmable logic device may be a complex programmable logic device, a field programmable gate array, a general purpose array logic or any combination thereof.

[0154] The memory 20 stores instructions executable by at least one processor 10, so that the at least one processor 10 executes the method shown in the above embodiment.

[0155] The memory 20 may include a program storage area and a data storage area, wherein the program storage area may store an operating system, an application required for at least one function; the data storage area may store data created according to the use of the computer device, etc. In addition, the memory 20 may include a high-speed random access memory, and may also include a non-transient memory, such as at least one disk storage device, a flash memory device, or other non-transient solid-state storage device. In some optional embodiments, the memory 20 may optionally include a memory remotely arranged relative to the processor 10, and these remote memories may be connected to the computer device via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0156] The memory 20 may include a volatile memory, such as a random access memory; the memory may also include a non-volatile memory, such as a flash memory, a hard disk or a solid state drive; the memory 20 may also include a combination of the above types of memory.

[0157] The computer device also includes an input device 30 and an output device 40. The processor 10, the memory 20, the input device 30 and the output device 40 may be connected via a bus or other means. Figure 5 The example of connecting through bus is taken in the following.

[0158] The input device 30 can receive input digital or character information, and generate key signal input related to the user settings and function control of the computer device, such as a touch screen, a keypad, a mouse, a track pad, a touch pad, an indicator bar, one or more mouse buttons, a trackball, a joystick, etc. The output device 40 may include a display device, an auxiliary lighting device (e.g., an LED) and a tactile feedback device (e.g., a vibration motor), etc. The above-mentioned display device includes but is not limited to a liquid crystal display, a light emitting diode, a display and a plasma display. In some optional embodiments, the display device can be a touch screen.

[0159] The embodiment of the present invention also provides a computer-readable storage medium. The method according to the embodiment of the present invention can be implemented in hardware, firmware, or can be implemented as a computer code that can be recorded in a storage medium, or can be implemented as a computer code that is originally stored in a remote storage medium or a non-temporary machine-readable storage medium and will be stored in a local storage medium through a network download, so that the method described herein can be stored in such software processing on a storage medium using a general-purpose computer, a dedicated processor, or programmable or dedicated hardware. Among them, the storage medium can be a magnetic disk, an optical disk, a read-only storage memory, a random access memory, a flash memory, a hard disk or a solid-state hard disk, etc.; further, the storage medium can also include a combination of the above types of memories. It can be understood that a computer, a processor, a microprocessor controller, or programmable hardware includes a storage component that can store or receive software or computer code. When the software or computer code is accessed and executed by a computer, a processor, or hardware, the method shown in the above embodiment is implemented.

[0160] A part of the present invention may be applied as a computer program product, such as a computer program instruction, which, when executed by a computer, can call or provide the method and / or technical solution according to the present invention through the operation of the computer. Those skilled in the art should understand that the existence of the computer program instruction in a computer-readable medium includes, but is not limited to, a source file, an executable file, an installation package file, etc., and accordingly, the way in which the computer program instruction is executed by the computer includes, but is not limited to: the computer directly executes the instruction, or the computer compiles the instruction and then executes the corresponding compiled program, or the computer reads and executes the instruction, or the computer reads and installs the instruction and then executes the corresponding installed program. Here, the computer-readable medium may be any available computer-readable storage medium or communication medium accessible to the computer.

[0161] Although the embodiments of the present invention have been described in conjunction with the accompanying drawings, those skilled in the art may make various modifications and variations without departing from the spirit and scope of the present invention, and such modifications and variations are all within the scope defined by the appended claims.

Claims

1. A flood disaster prediction method, characterized in that: The method comprises: Acquire a historical data set of a target area, wherein the historical data set includes historical disaster data of the target area and historical factor values ​​of flood disaster influencing factors of the target area; Constructing a generated data set based on the historical data set, wherein the generated data set includes generated disaster data and generated factor values; Training a flood disaster prediction model based on the historical data set and the generated data set; Acquire current factor values ​​of factors affecting flood disasters in the target area, and determine precipitation forecast data for the target area based on the current factor values; The precipitation prediction data and the current factor value are used as input features of the flood disaster prediction model to obtain flood disaster prediction data.

2. The flood disaster prediction method according to claim 1, characterized in that: The target area includes a target administrative area and a target watershed area, the historical factor value includes a historical geographical factor value and a historical precipitation value, and the historical data set of the target area is obtained, including: Determining the target administrative area to be predicted; Acquire the historical disaster data and the historical geographical factor values ​​of the target administrative area; Expanding the scope based on the target administrative area to obtain the target watershed area; Obtain the historical precipitation value of the target watershed area.

3. The flood disaster prediction method according to claim 1, characterized in that: The generating data set is constructed based on the historical data set, wherein the generating data set includes generating disaster affected data and generating factor values, including: Inputting the historical data set into a conditional generative adversarial model, the conditional generative adversarial model comprising a generator and a discriminator, the generator being used to obtain a generated factor value based on the historical disaster data, and the discriminator being used to compare and discriminate the generated factor value with the historical factor value; The generator and the discriminator are used to iteratively train and output the generated factor value, and the historical disaster data corresponding to the generated factor value is used as the generated disaster data.

4. The flood disaster prediction method according to claim 3, characterized in that: The flood disaster prediction model includes a convolutional neural layer, a long short-term memory layer and a fully connected layer. The flood disaster prediction model is trained based on the historical data set and the generated data set, including: Performing feature extraction on the historical factor value and the generated factor value based on the convolutional neural layer; Performing time series analysis on the historical factor values ​​and the generated factor values ​​based on the long short-term memory layer; Connecting the convolutional neural layer and the long short-term memory layer based on the fully connected layer, and performing flood disaster prediction to obtain training disaster prediction data; The training disaster prediction data and the corresponding disaster-affected data are compared to adjust the parameters of the flood disaster prediction model to obtain the trained flood disaster prediction model.

5. The flood disaster prediction method according to claim 2, characterized in that: The step of determining the precipitation prediction data for the target area based on the current factor value comprises: inputting the current factor values ​​into a meteorological research and forecasting model; The meteorological change process is simulated based on the various meteorological equations in the meteorological research and forecasting model to obtain the precipitation prediction data.

6. The flood disaster prediction method according to claim 5, characterized in that: The method of using the precipitation prediction data and the current factor value as input features of the flood disaster prediction model to obtain flood disaster prediction data includes: Determine the corresponding precipitation forecast data and current precipitation value based on the target forecast days, wherein the precipitation forecast data and current precipitation value are data within a continuous time period; The current geographical factor value, the current precipitation value and the precipitation forecast data are used as input features of the flood disaster prediction model, and the flood disaster prediction model is used to perform prediction to obtain the flood disaster prediction data for the target prediction days.

7. A flood disaster prediction device, characterized in that: The device comprises: A first acquisition module is used to acquire a historical data set of a target area, wherein the historical data set includes historical disaster data of the target area and historical factor values ​​of flood disaster influencing factors of the target area; A construction module, used for constructing a generated data set based on the historical data set, wherein the generated data set includes generated disaster data and generated factor values; A training module, used for training a flood disaster prediction model based on the historical data set and the generated data set; A second acquisition module is used to acquire current factor values ​​of flood disaster influencing factors in the target area, and determine precipitation forecast data for the target area based on the current factor values; The prediction module is used to use the precipitation prediction data and the current factor value as input features of the flood disaster prediction model to obtain flood disaster prediction data.

8. A computer device, characterized in that: include: A memory and a processor, wherein the memory and the processor are communicatively connected to each other, the memory stores computer instructions, and the processor executes the flood disaster prediction method according to any one of claims 1 to 6 by executing the computer instructions.

9. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer instructions, and the computer instructions are used to enable a computer to execute the flood disaster prediction method according to any one of claims 1 to 6.

10. A computer program product, characterized in that The method comprises computer instructions, wherein the computer instructions are used to enable a computer to execute the flood disaster prediction method according to any one of claims 1 to 6.

Images