An intelligent flood perception and decision-making method for multi-source rainfall data fusion

Abstract

The present invention belongs to the technical field of disaster decision-making and discloses an intelligent flood perception and decision-making method based on the fusion of multi-source rainfall data; the method comprises: constructing a spatially integrated rainfall monitoring network and collecting a multidimensional rainfall data set in real time; performing time alignment and spatial registration on the multidimensional rainfall data set to generate rainfall processing data; performing data fusion on the rainfall processing data to form fused rainfall data; obtaining hydrological environment data and constructing a hydrodynamic model; driving the hydrodynamic model to operate based on the fused rainfall data to obtain water flow characteristic data; combining the water flow characteristic data and the fused rainfall data to evaluate the flood risk level of different urban areas; and dynamically formulating differentiated emergency response plans based on a historical disaster database. The present invention realizes intelligent perception and response to urban flood disasters, can effectively support emergency response needs in a dynamic environment, and improve urban flood prevention and control capabilities and disaster resistance levels.

Description

Technical Field

[0001] The present invention relates to the technical field of disaster decision-making, and more specifically, to an intelligent flood perception decision-making method based on multi-source rainfall data fusion. Background Art

[0002] With the intensification of climate change and the frequent occurrence of extreme rainfall events, flood disasters have become a major challenge facing the world. Accurate rainfall data is the basis for flood warning and emergency decision-making. However, traditional methods rely on single-source data such as ground rain gauges, weather radars, and remote sensing satellites. They have problems such as limited coverage, insufficient accuracy, and data delays, making it difficult to meet the needs of rapid response and accurate decision-making. Therefore, an intelligent flood perception and decision-making method is urgently needed to integrate multi-source rainfall data to achieve early perception, dynamic monitoring, and scientific response to flood disasters, thereby effectively improving the scientific nature and reliability of flood disaster monitoring and response, and enhancing the efficiency and effectiveness of disaster response.

[0003] The patent with announcement number CN116070918B discloses a method for urban flood safety assessment and flood disaster prevention and control; it includes obtaining urban hydrological basic data; obtaining urban historical rainfall data and calculating based on the urban historical rainfall data and urban hydrological basic data to obtain designed rainfall process data; obtaining flood process data through calculation based on the designed rainfall process data; constructing an urban flood model based on the flood process data and urban hydrological basic data; evaluating the inundation results of the urban flood model to generate flood inundation result data; determining the flood risk level based on the flood inundation result data; obtaining flood safety adaptability analysis results based on the flood risk level; generating flood disaster prevention and control measures based on the flood safety adaptability analysis results; this invention proposes corresponding disaster prevention and control methods based on the flood inundation results, so that the flood safety risk level of the plot is adapted to the construction purpose.

[0004] However, although the above technologies have achieved flood perception and decision-making, they mainly rely on historical rainfall data and static hydrological basic data, lack the perception and response mechanism for real-time rainfall changes, and are unable to respond to sudden flood disasters in a timely manner; in addition, disaster prevention and control measures are generated based on static analysis results, mainly disaster prevention and mitigation measures, focusing on pre-disaster prevention and mitigation of disaster impacts, rather than emergency response plans after flood disasters occur. Therefore, it is difficult to effectively support emergency response needs in a dynamic environment, thereby limiting the actual effect of refined flood management in modern cities.

[0005] In view of this, the present invention proposes an intelligent flood perception decision-making method based on multi-source rainfall data fusion to solve the above problems. Summary of the Invention

[0006] In order to overcome the above-mentioned defects of the prior art and achieve the above-mentioned objectives, the present invention provides the following technical solution: an intelligent flood perception and decision-making method based on multi-source rainfall data fusion, comprising:

[0007] S1: Using multi-source data acquisition technology, we build a spatially integrated rainfall monitoring network to collect multi-dimensional rainfall datasets in real time;

[0008] S2: Temporal alignment and spatial registration of multidimensional rainfall datasets to generate rainfall processing data;

[0009] S3: Using data fusion algorithm, multi-dimensional data fusion is performed on rainfall processing data to form fused rainfall data;

[0010] S4: Acquire hydrological environment data and build a hydrological dynamics model based on the hydrological environment data;

[0011] S5: Drive the hydrological dynamics model based on the integrated rainfall data to simulate the water flow evolution process and obtain water flow characteristic data;

[0012] S6: Combine water flow characteristic data and integrated rainfall data to build a flood risk assessment system based on fuzzy comprehensive evaluation to intelligently assess the flood risk level of different urban areas;

[0013] S7: Based on the flood risk levels of different urban areas and combined with a pre-built historical disaster database, a multi-objective decision-making optimization algorithm is used to dynamically formulate differentiated emergency response plans.

[0014] Furthermore, the method for constructing a spatially integrated rainfall monitoring network includes:

[0015] set up monitoring stations, is an integer greater than 1; Randomly select monitoring sites monitoring sites, and as a set of sites, a total of Group site collections, , ; Obtain the city range and the monitoring range of each monitoring station, merge the monitoring range of each monitoring station in each group of station sets, and obtain the total monitoring range of each station set; superimpose the total monitoring range of each station set with the city range, mark the station sets whose total monitoring range is greater than or equal to the city range as candidate sets, and do not mark the station sets whose total monitoring range is less than the city range; count the number of monitoring stations in each candidate set and mark it as the number of stations, and mark the candidate set with the smallest number of stations as the screening set; set a rain gauge for each monitoring station in the screening set, and build a spatial integrated rainfall monitoring network by combining meteorological radar and remote sensing satellite;

[0016] Multidimensional rainfall datasets include The data consists of monitoring data, radar data and remote sensing data; the monitoring data is the rainfall collected by monitoring stations, the radar data is the rainfall collected by meteorological radars, and the remote sensing data is the rainfall collected by remote sensing satellites.

[0017] Furthermore, the method for temporal alignment of multidimensional rainfall datasets includes:

[0018] Get the acquisition time of monitoring data in the multidimensional rainfall dataset and mark it as the current time; get the acquisition time of radar data in the multidimensional rainfall dataset and mark it as radar time; get the acquisition time of remote sensing data in the multidimensional rainfall dataset and mark it as remote sensing time; subtract radar time from current time to get radar difference time; subtract remote sensing time from current time to get remote sensing difference time; get historical data, which includes Set of historical radar data and The historical remote sensing data is the historical radar data, the historical remote sensing data is the remote sensing data obtained at the historical moment. is an integer greater than 1;

[0019] Set the preset radar sampling period, radar difference time and The historical radar data of the group are input into the trained radar prediction model in turn to predict the corresponding real-time radar data. The real-time radar data is the radar data corresponding to the current time. The preset remote sensing sampling period, remote sensing difference time and The historical remote sensing data of the group are input into the trained remote sensing prediction model in turn to predict the corresponding real-time remote sensing data. The real-time remote sensing data is the remote sensing data corresponding to the current time. Both the radar prediction model and the remote sensing prediction model are deep neural network models.

[0020] Furthermore, the method of spatial registration of multidimensional rainfall datasets includes:

[0021] Obtain the geographic coordinates of the monitoring site corresponding to each monitoring data in the multidimensional rainfall dataset and mark them as site coordinates; combine the monitoring data and site coordinates of each monitoring site to generate monitoring distribution data; project the real-time radar data to the WGS84 coordinate system to obtain projected radar data; obtain the geographic coordinates corresponding to each rainfall amount in the real-time remote sensing data and mark them as remote sensing coordinates; obtain the geographic coordinates corresponding to each rainfall amount in the projected radar data and mark them as radar coordinates; mark the site coordinates that are different from the remote sensing coordinates as first coordinates, and mark the site coordinates that are different from the radar coordinates as second coordinates; calculate the rainfall amount of each first coordinate based on the real-time remote sensing data using the spatial difference method; calculate the rainfall amount of each second coordinate based on the projected radar data using the spatial difference method; add each first coordinate and the corresponding rainfall amount to the real-time remote sensing data to generate remote sensing distribution data; add each second coordinate and the corresponding rainfall amount to the projected radar data to generate radar distribution data;

[0022] Generate rainfall processing data based on remote sensing distribution data, radar distribution data and monitoring distribution data.

[0023] Furthermore, the method for forming fused rainfall data includes:

[0024] The rainfall corresponding to the same geographical coordinates in the remote sensing distribution data, radar distribution data and monitoring distribution data is taken as a group of data sets, and the data sets correspond to the geographical coordinates one by one; a weight set is preset, and the weight set includes weight coefficients corresponding to the remote sensing distribution data, radar distribution data and monitoring distribution data; each data in each data set is multiplied by the corresponding weight coefficient in the weight set, and the results are added in sequence to obtain the fused rainfall corresponding to each data set; based on the fused rainfall corresponding to each data set, fused rainfall data is formed.

[0025] Furthermore, the hydrological environment data includes meteorological data, topographic data, soil and vegetation data, watershed hydrological data and groundwater data;

[0026] Methods for constructing hydrodynamic models include:

[0027] Construct a surface runoff model, the input of which is the fused rainfall data, and the output is the net rainfall; construct a shallow water dynamics model, the input of which is the net rainfall, and the output is the water flow characteristic data, which includes the water depth and water flow velocity; couple the surface runoff model with the shallow water dynamics model to construct a hydrodynamic model;

[0028] The steps to construct a shallow water dynamics model include:

[0029] Step S401: using the Saint-Venant equations as basic equations, where the Saint-Venant equations include a continuity equation and a momentum equation;

[0030] Step S402: converting hydrological environment data into model parameters;

[0031] Step S403: Divide the city into grid cells, each of which corresponds to a monitoring site and a set of state variables, including water depth and water flow velocity;

[0032] Step S404: setting boundary conditions, which include upstream boundary conditions, downstream boundary conditions, and lateral boundary conditions; wherein the upstream boundary condition is a depth time series, the downstream boundary condition is a depth-velocity relationship curve, and the lateral boundary condition is net rainfall;

[0033] Step S405: setting initial conditions for each grid cell in turn, the initial conditions including initial water depth and initial water flow velocity;

[0034] Step S406: Based on the model parameters, boundary conditions and initial conditions, a numerical calculation method is used to solve the Saint-Venant equations to construct a shallow water dynamics model;

[0035] The fused rainfall data is input into the hydrological dynamics model to calculate the water flow characteristic data corresponding to each monitoring station.

[0036] Furthermore, the steps of intelligently assessing flood risk levels in different urban areas include:

[0037] Step S601: Divide the city into regions based on water flow characteristic data and obtain urban areas;

[0038] Step S602: Obtain the rainfall corresponding to each urban area based on the fused rainfall data; count the number of rainfall corresponding to each urban area and mark it as rainfall points; add the rainfall of each urban area in sequence and divide it by the corresponding rainfall point number to obtain the average rainfall of each urban area;

[0039] Step S603: construct multiple fuzzy sets for each of the water depth, water flow velocity, and average rainfall. Step S604: convert the fused feature data corresponding to each urban area into the membership degree of each corresponding fuzzy set using fuzzification technology. The fused feature data includes the water depth, water flow velocity, and average rainfall.

[0040] Step S605: defining fuzzy rules;

[0041] Step S606: Match each set of fuzzified fusion feature data with fuzzy rules, and perform fuzzy reasoning using a fuzzy reasoning method to obtain fuzzy reasoning results corresponding to each urban area. The fuzzy reasoning results include the membership degree corresponding to each flood risk level, which includes extreme risk, high risk, medium risk, and low risk.

[0042] Step S607: Compare each degree of membership in each fuzzy inference result, and take the flood risk level corresponding to the degree of membership with the largest value as the flood risk level of the corresponding urban area.

[0043] Furthermore, in step S601, obtaining The steps for each city region include:

[0044] Step S610: Take each set of water flow characteristic data as a node and randomly select nodes as the center point, and the nodes that are not the center point as the sample point. is an integer greater than 1;

[0045] Step S611: According to The center points construct the corresponding For each group, calculate the Euclidean distance from each sample point to each center point in turn and mark it as point distance;

[0046] Step S612: Preset a distance threshold, compare each point distance corresponding to each sample point with the distance threshold, mark the center point corresponding to the point distance whose value is less than the distance threshold as a classification point, and do not mark the point distance whose value is greater than or equal to the distance threshold. Then, classify each sample point into the group corresponding to the corresponding classification point.

[0047] Step S613: Calculate the new center point corresponding to each group, and replace the center point of each group with the corresponding new center point;

[0048] Step S614: loop through steps S611 to S613 until the new center point corresponding to each group calculated in step S613 is consistent with the corresponding new center point calculated in the previous loop, and the loop ends. groups;

[0049] Step S615: Perform connectivity analysis on the nodes in each group to obtain A collection of regions; according to A collection of regions, dividing the city into urban areas;

[0050] In step S613, the method for calculating the new center point includes:

[0051] Count the number of nodes in each group and mark it as the number of nodes; add up the nodes in each group in sequence and divide it by the corresponding number of nodes to obtain the average point corresponding to each group; calculate the point distance from each node in each group to the corresponding average point and mark it as the average distance; compare the same average distances of corresponding groups, and take the node corresponding to the smallest average distance as the new center point of the corresponding group.

[0052] Furthermore, in step S615, obtaining The steps for a region collection include:

[0053] Step S621: randomly select a node that is not marked as a selected node and mark it as the current node;

[0054] Step S622: construct a region set and add the current node to the region set;

[0055] Step S623: Obtain the neighboring nodes of the current node, mark the neighboring nodes that belong to the same group as the current node as similar nodes, and add the similar nodes to the region set;

[0056] Step S624: Obtain the neighboring nodes of the same type of node and mark them as the same neighboring nodes; mark the same neighboring nodes that belong to the same group as the current node as the same type of nodes, and add the same type of nodes to the region set;

[0057] Step S625: Loop step S624 until the groups corresponding to the adjacent nodes of all nodes in the region set are different from the group corresponding to the current node, then the loop ends, obtains a region set, and marks all nodes in the region set as selected nodes;

[0058] Step S626: Loop through steps S621 to S625 until all nodes are marked as selected nodes. The loop ends and the result is obtained. A collection of regions.

[0059] Furthermore, the method for dynamically formulating differentiated emergency response plans includes:

[0060] Obtain the number of monitoring stations in each urban area and mark it as the monitoring number; take the monitoring number and flood risk level of each urban area as a set of regional characteristic data; obtain the corresponding regional characteristic data of each set from the pre-built historical disaster database. Group rescue feature data, is an integer greater than 1; rescue feature data includes the number of rescued people and the quantity of various rescue materials; according to the corresponding regional feature data of each group The rescue feature data of each group are obtained, and the regional range corresponding to each group of regional feature data is obtained; a value is randomly selected from each range within each regional range, and a set of candidate sets is constructed. Group alternative sets; add up the number of rescued people in each alternative set in sequence to obtain the total number of rescued people; add up the number of materials with the same corresponding rescue materials in each alternative set in sequence to obtain the total number of materials corresponding to each rescue material; obtain allocation data, which includes the number of allocated people and the allocated quantity of each rescue material; compare the total number of rescued people with the allocated number in each alternative set, and compare each total number of materials with the corresponding allocated quantity; mark the alternative sets whose total number of rescued people is less than or equal to the allocated number and whose total number of materials is less than or equal to the corresponding allocated quantity as rescue sets; alternative sets whose total number of rescued people is greater than the allocated number or whose total number of materials is greater than the corresponding allocated quantity will not be marked;

[0061] Different digital labels are set for different flood risk levels and marked as level labels; the flood risk levels in all regional feature data are replaced with corresponding level labels, and added to each rescue set in turn to obtain a rescue feature set; each rescue feature set is input into the trained effect prediction model to predict the corresponding rescue data; the rescue data includes rescue efficiency, casualty rate and response cost; the effect prediction model is a deep neural network model; a preset proportion set includes proportional coefficients corresponding to rescue efficiency, casualty rate and response cost; the rescue efficiency, casualty rate and response cost corresponding to each rescue set are multiplied by the corresponding proportional coefficient in the proportion set, and the efficiency coefficient, casualty coefficient and cost coefficient are obtained in turn; the efficiency coefficient is subtracted from the casualty coefficient, and then the cost coefficient is subtracted to obtain the rescue effect; the rescue feature set with the largest rescue effect is marked as the optimal feature set, and the rescue set in the optimal feature set is used as the emergency response plan.

[0062] The technical effects and advantages of the intelligent flood perception and decision-making method based on multi-source rainfall data fusion of the present invention are as follows:

[0063] By building a spatially integrated rainfall monitoring network, it is possible to integrate rainfall monitoring data from multiple sources, including rain gauges, weather radars, and remote sensing satellites, to achieve comprehensive, real-time monitoring of rainfall within the city and improve the spatiotemporal resolution and continuity of rainfall information. Time alignment and spatial registration techniques are used to effectively address time differences and coordinate inconsistencies between data sources from different monitoring methods, ensuring the spatiotemporal consistency of the fused rainfall data. A hydrodynamic model driven by fused rainfall data accurately simulates urban surface hydrological processes and extracts flow characteristics, laying a reliable foundation for flood risk assessment. A flood risk assessment system based on fuzzy comprehensive evaluation is constructed to intelligently assess flood risk levels in different urban areas, providing a scientific basis for emergency response. In combination with a historical disaster database, a multi-objective decision-making optimization algorithm is used to dynamically formulate differentiated emergency response plans. This approach minimizes casualties while ensuring rescue efficiency and reducing response costs, significantly improving the accuracy and effectiveness of emergency rescue. Intelligent perception and response to urban flood disasters can effectively support emergency response needs in a dynamic environment, enhance urban flood prevention and control capabilities and disaster resilience, and significantly improve the scientific nature and efficiency of flood disaster response. BRIEF DESCRIPTION OF THE DRAWINGS

[0064] Figure 1 This is a flow chart of an intelligent flood perception and decision-making method based on multi-source rainfall data fusion according to Example 1 of the present invention. DETAILED DESCRIPTION

[0065] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only 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 making creative efforts are within the scope of protection of the present invention.

[0066] Example 1

[0067] See also Figure 1 As shown, the intelligent flood perception and decision-making method based on multi-source rainfall data fusion described in this embodiment includes:

[0068] S1: Use multi-source data acquisition technology to build a spatially integrated rainfall monitoring network and collect multi-dimensional rainfall datasets in real time.

[0069] Methods for constructing a spatially integrated rainfall monitoring network include:

[0070] Those skilled in the art conduct a comprehensive analysis of factors such as the city's geographical environment (i.e., the city's natural topographical features, including ups and downs, low-lying areas, and river distribution), climate characteristics (i.e., the city's climate type and meteorological characteristics such as rainfall and rainfall cycle), drainage system (i.e., various facilities built in the city to guide rainwater outflow, such as drainage pipe networks and drainage pumping stations), flood history (i.e., the frequency and duration of flood disasters in the city in the past), and infrastructure layout (i.e., the distribution of important facilities in the city, such as transportation hubs, bridges, tunnels, etc.), and set up a comprehensive plan based on actual conditions. monitoring stations, is an integer greater than 1, and each monitoring site can be equipped with a rain gauge to monitor rainfall; Randomly select monitoring sites monitoring sites, and as a set of sites, a total of Group site collections, , ;

[0071] Obtain the city range and the monitoring range of each monitoring station. The city range is the administrative boundary of the city and is obtained using open source map data (such as Amap and Baidu Map). The monitoring range is determined by technicians in this field based on the measurement range of the rain gauge, that is, the monitoring range is equal to the measurement range. GIS tools (such as ArcGIS and QGIS) are used to merge the monitoring range of each monitoring station in each station set to obtain the total monitoring range of each station set. GIS tools are used to perform an overlay analysis of the total monitoring range of each station set with the city range (i.e., the total monitoring range and the city range are overlapped and the spatial relationship between them is compared). Station sets with a total monitoring range greater than or equal to the city range are marked as candidate sets (i.e., the total monitoring range can completely cover the entire city range, and each spatial location within the city range is covered by the monitoring range of at least one monitoring station). Station sets with a total monitoring range smaller than the city range are not marked. The number of monitoring stations in each candidate set is counted and marked as the number of stations. The candidate set with the smallest number of stations is marked as the screening set. A rain gauge is set for each monitoring station in the screening set, and a spatially integrated rainfall monitoring network is constructed in combination with meteorological radar and remote sensing satellites.

[0072] Multidimensional rainfall datasets include The monitoring data is the rainfall collected by the monitoring station, the radar data is the rainfall collected by the meteorological radar, and the remote sensing data is the rainfall collected by the remote sensing satellite. It should be noted that the monitoring data can provide high-precision rainfall information, but the spatial coverage is limited and can only represent the rainfall conditions within the corresponding monitoring range of the monitoring station. Radar data can provide mesoscale, quasi-real-time rainfall distribution information with high spatial resolution, but will be affected by terrain obstruction, reducing the estimation accuracy of long-distance rainfall. Remote sensing data provides large-scale, full-coverage rainfall distribution information without being restricted by terrain, but has relatively low temporal resolution and is difficult to effectively capture short-term heavy rainfall events. Therefore, it is necessary to effectively integrate the monitoring data, radar data and remote sensing data to overcome the limitations of a single monitoring method and realize comprehensive monitoring of rainfall, thereby improving the spatiotemporal continuity and integrity of rainfall.

[0073] S2: Temporal alignment and spatial registration of multidimensional rainfall datasets to generate rainfall processing data.

[0074] Methods for temporal alignment of multidimensional rainfall datasets include:

[0075] Obtain the acquisition time of the monitoring data in the multidimensional rainfall dataset and mark it as the current time. The current time is obtained through the real-time clock module built into the rain gauge; obtain the acquisition time of the radar data in the multidimensional rainfall dataset and mark it as the radar time. The radar time is obtained through the real-time clock module built into the weather radar; obtain the acquisition time of the remote sensing data in the multidimensional rainfall dataset and mark it as the remote sensing time. The remote sensing time is obtained through the real-time clock module built into the remote sensing satellite; subtract the radar time from the current time to obtain the radar difference time; subtract the remote sensing time from the current time to obtain the remote sensing difference time; obtain historical data from the rainfall database of the city meteorological bureau. The historical data includes Set of historical radar data and The historical remote sensing data is the historical radar data, the historical remote sensing data is the remote sensing data obtained at the historical moment. is an integer greater than 1; it should be noted that, The acquisition time corresponding to the set of historical radar data is continuous, and The latest acquisition time in the historical radar data set is adjacent to the radar time; that is, The historical radar data is collected continuously before the weather radar collects radar data. Group radar data; The collection time corresponding to the set of historical remote sensing data is continuous, and The latest acquisition time in the historical remote sensing data of the group is adjacent to the remote sensing time; that is, A set of historical remote sensing data is the data collected continuously before the remote sensing satellite collects remote sensing data. Group remote sensing data;

[0076] Set the preset radar sampling period, radar difference time and The historical radar data of the group are input into the trained radar prediction model in turn to predict the corresponding real-time radar data. The real-time radar data is the radar data corresponding to the current time. The preset remote sensing sampling period, remote sensing difference time and The historical remote sensing data of the group are input into the trained remote sensing prediction model in sequence to predict the corresponding real-time remote sensing data. The real-time remote sensing data is the remote sensing data corresponding to the current time. The radar sampling period is the time interval between each collection of radar data, and the remote sensing sampling period is the time interval between each collection of remote sensing data. The radar sampling period and the remote sensing sampling period are both pre-set by technical personnel in this field according to actual conditions. The radar prediction model and the remote sensing prediction model are both deep neural network models. The deep neural network model is an existing technology, and the specific training method will not be elaborated here.

[0077] It should be understood that the purpose of time alignment is to make up for the time differences between different data sources (such as monitoring data, radar data and remote sensing data); since the sampling period of radar data and remote sensing data is long, the collection time of radar data and remote sensing data is inconsistent with the collection time of monitoring data, resulting in time mismatch between different data sources; through time alignment, it can ensure that the rainfall from different sources is comparable within the same time frame, thereby improving the temporal continuity of rainfall.

[0078] Methods for spatial registration of multidimensional rainfall datasets include:

[0079] Obtain the geographic coordinates (including longitude and latitude) of the monitoring site corresponding to each monitoring data in the multidimensional rainfall dataset and mark them as site coordinates. The site coordinates are obtained by a person skilled in the art through field measurements; combine the monitoring data and site coordinates of each monitoring site to generate monitoring distribution data; project the real-time radar data into the WGS84 coordinate system to obtain projected radar data; projection methods such as equidistant conic projection, equidistant cylindrical projection, Mercator projection, etc. are used to ensure that the spatial positions of the monitoring distribution data, real-time radar data, and real-time remote sensing data are consistent; it should be understood that since the coordinate systems of the monitoring distribution data and the real-time remote sensing data are the WGS84 coordinate system, there is no need to project the monitoring distribution data and the real-time remote sensing data; only the real-time radar data needs to be projected into the WGS84 coordinate system;

[0080] Obtain the geographic coordinates corresponding to each rainfall amount in the real-time remote sensing data and mark them as remote sensing coordinates; obtain the geographic coordinates corresponding to each rainfall amount in the projected radar data and mark them as radar coordinates; mark the site coordinates that are different from the remote sensing coordinates as the first coordinates, and mark the site coordinates that are different from the radar coordinates as the second coordinates; calculate the rainfall amount for each first coordinate based on the real-time remote sensing data using a spatial difference method (such as inverse distance weighted IDW, Kriging interpolation, Spline interpolation, etc.); calculate the rainfall amount for each second coordinate based on the projected radar data using a spatial difference method; add each first coordinate and the corresponding rainfall amount to the real-time remote sensing data to generate remote sensing distribution data; add each second coordinate and the corresponding rainfall amount to the projected radar data to generate radar distribution data.

[0081] Generate rainfall processing data based on remote sensing distribution data, radar distribution data and monitoring distribution data.

[0082] S3: Use data fusion algorithm to perform multi-dimensional data fusion on rainfall processing data to form fused rainfall data.

[0083] Methods for forming fused rainfall data include:

[0084] The rainfall corresponding to the same geographical coordinates in the remote sensing distribution data, radar distribution data and monitoring distribution data is taken as a group of data sets, and the data sets correspond to the geographical coordinates one by one; a weight set is preset, and the weight set includes weight coefficients corresponding to the remote sensing distribution data, radar distribution data and monitoring distribution data. The weight set is pre-set by technical personnel in this field based on the actual rainfall monitoring accuracy of rain gauges, meteorological radars and remote sensing satellites; each data in each group of data sets is multiplied by the corresponding weight coefficient in the weight set, and the results are added in sequence to obtain the fused rainfall corresponding to each group of data sets; based on the fused rainfall corresponding to each group of data sets, fused rainfall data is formed.

[0085] S4: Obtain hydrological environment data and build a hydrological dynamics model based on the hydrological environment data.

[0086] Hydrological environment data refers to all key environmental factors that affect hydrological processes, which directly affect hydrological processes such as water flow, runoff, groundwater, and evaporation; hydrological environment data include meteorological data, topographic data, soil and vegetation data, watershed hydrological data, and groundwater data; meteorological data such as rainfall, temperature, wind speed, humidity, etc.; topographic data such as slope, altitude, watershed shape, surface roughness, etc.; soil and vegetation data such as soil type, soil moisture, soil hydraulic conductivity, vegetation cover type, etc.; watershed hydrological data such as river flow, reservoir water level, river section, etc.; groundwater data such as groundwater level, groundwater flow, permeability, etc.; hydrological environment data are obtained by technical personnel in this field through a combination of field monitoring, hydrological data collection, remote sensing technology, geographic information systems, etc.

[0087] Methods for constructing hydrodynamic models include:

[0088] Construct a surface runoff model (such as the SCS-CN model, Green-Ampt infiltration model, etc.). The input of the surface runoff model is the fused rainfall data, and the output is the net rainfall (i.e., the amount of water that actually forms surface runoff). Construct a shallow water dynamics model. The input of the shallow water dynamics model is the net rainfall, and the output is water flow characteristic data, which includes water accumulation depth and water flow velocity. Couple the surface runoff model with the shallow water dynamics model to construct a hydrodynamic model.

[0089] The steps to construct a shallow water dynamics model include:

[0090] Step S401: Using the Saint-Venant equations (shallow water equations) as basic equations to describe surface water flow; the Saint-Venant equations include the continuity equation (i.e., the mass conservation equation) and the momentum equation (i.e., the momentum conservation equation); the Saint-Venant equations are known in the art, and the specific equations corresponding to the continuity equation and the momentum equation are not described in detail here;

[0091] Step S402: Those skilled in the art convert hydrological environment data into model parameters by systematically processing, professionally analyzing, and scientifically converting the hydrological environment; the model parameters include slope coefficient, Manning roughness coefficient, digital elevation model, permeability, rainfall, etc.;

[0092] Step S403: Divide the city into grid cells, each of which corresponds to a monitoring site and a set of state variables, including water depth and water flow velocity;

[0093] Step S404: Boundary conditions are set. Boundary conditions include upstream boundary conditions, downstream boundary conditions, and lateral boundary conditions. The upstream boundary condition is a depth time series (i.e., the time-varying water depth), the downstream boundary condition is a depth-flow velocity relationship curve (i.e., the functional relationship between water depth and water flow velocity), and the lateral boundary condition is the net rainfall. Boundary conditions are set by those skilled in the art using a combination of methods such as field observations, theoretical calculations, and numerical simulations.

[0094] Step S405: Initial conditions are set for each grid cell in sequence. The initial conditions include initial water depth and initial water flow velocity. The initial conditions are obtained by those skilled in the art through field measurements or empirical calculations.

[0095] Step S406: Based on the model parameters, boundary conditions, and initial conditions, a numerical calculation method (such as finite difference and finite volume) is used to solve the Saint-Venant equations and construct a shallow water dynamics model. The shallow water dynamics model can simulate and output the water depth and water flow velocity of each grid cell that changes with time.

[0096] S5: Drive the hydrological dynamics model based on the fused rainfall data to simulate the water flow evolution process and obtain water flow characteristic data.

[0097] The fused rainfall data is input into the hydrological dynamics model to calculate the water flow characteristic data corresponding to each monitoring station.

[0098] S6: Combine water flow characteristic data and integrated rainfall data to build a flood risk assessment system based on fuzzy comprehensive evaluation to intelligently assess the flood risk levels of different urban areas.

[0099] The steps for intelligently assessing flood risk levels in different urban areas include:

[0100] Step S601: Divide the city into regions based on water flow characteristic data and obtain urban areas;

[0101] Step S602: Obtain the rainfall corresponding to each urban area based on the fused rainfall data; count the number of rainfall corresponding to each urban area and mark it as rainfall points; add the rainfall of each urban area in sequence and divide it by the corresponding rainfall point number to obtain the average rainfall of each urban area;

[0102] Step S603: construct multiple fuzzy sets for the water depth, water flow velocity, and average rainfall respectively; for example, the fuzzy sets corresponding to the water depth include deep water, medium water, shallow water, etc., and the fuzzy sets corresponding to the water flow velocity include fast speed, medium speed, slow speed, etc.;

[0103] Step S604: The fused feature data corresponding to each urban area is converted into the membership of each corresponding fuzzy set using fuzzification technology. The fused feature data includes water depth, water flow velocity, and average rainfall. Fuzzification technology is the process of converting precise numerical values into the membership corresponding to the fuzzy set. Fuzzification technologies include triangular membership function and trapezoidal membership function. For example, if the water depth value is low, it is inferred that the membership of shallow water is 0.9, the membership of medium water is 0.1, and the membership of deep water is 0.

[0104] Step S605: Define fuzzy rules. The fuzzy rules are defined based on expert knowledge or relevant literature. For example, if the water accumulation is deep, the velocity is fast, and the rainfall is heavy, then the probability of inferring that the corresponding urban area is at extreme risk is high; if the water accumulation is shallow, the velocity is slow, and the rainfall is light, then the probability of inferring that the corresponding urban area is at low risk is high.

[0105] Step S606: Each set of fuzzified fusion feature data is matched with fuzzy rules, and fuzzy reasoning is performed using a fuzzy reasoning method (such as the Mamdani fuzzy reasoning model or the Sugeno fuzzy reasoning model) to obtain a fuzzy reasoning result corresponding to each urban area. The fuzzy reasoning result includes a membership degree corresponding to each flood risk level. The flood risk levels include extreme risk, high risk, medium risk, and low risk. For example, the fuzzy reasoning result has a membership degree of 0.3 for extreme risk, 0.0 for high risk, 0.1 for medium risk, and 0 for low risk.

[0106] Step S607: Compare each degree of membership in each fuzzy inference result, and take the flood risk level corresponding to the degree of membership with the largest value as the flood risk level of the corresponding urban area.

[0107] In the above step S601, obtain The steps for each city region include:

[0108] Step S610: Take each set of water flow characteristic data as a node and randomly select nodes as the center point, and the nodes that are not the center point as the sample point. is an integer greater than 1, The specific value of is preset by those skilled in the art according to actual conditions;

[0109] Step S611: According to The center point constructs the corresponding For each group, the Euclidean distance from each sample point to each center point is calculated in turn and marked as point distance. The calculation method of Euclidean distance is an existing technology, and the specific calculation process is not described in detail here.

[0110] Step S612: Preset a distance threshold, compare each point distance corresponding to each sample point with the distance threshold, mark the center point corresponding to the point distance whose value is less than the distance threshold as a classification point, and do not mark the point distance whose value is greater than or equal to the distance threshold. Then, classify each sample point into the group corresponding to the corresponding classification point. The distance threshold is preset by those skilled in the art based on actual conditions.

[0111] Step S613: Calculate the new center point corresponding to each group, and replace the center point of each group with the corresponding new center point;

[0112] Step S614: loop through steps S611 to S613 until the new center point corresponding to each group calculated in step S613 is consistent with the corresponding new center point calculated in the previous loop, and the loop ends. groups;

[0113] Step S615: Perform connectivity analysis on the nodes in each group to obtain A collection of regions; according to A collection of regions, dividing the city into There are city areas, and city areas correspond one to one with region sets.

[0114] In the above step S613, the method for calculating the new center point includes:

[0115] Count the number of nodes in each group and mark it as the number of nodes; add up the nodes in each group in sequence and divide it by the corresponding number of nodes to obtain the average point corresponding to each group; calculate the point distance from each node in each group to the corresponding average point and mark it as the average distance; compare the same average distances of corresponding groups, and take the node corresponding to the smallest average distance as the new center point of the corresponding group.

[0116] In the above step S615, obtain The steps for a region collection include:

[0117] Step S621: randomly select a node that is not marked as a selected node and mark it as the current node;

[0118] Step S622: construct a region set and add the current node to the region set;

[0119] Step S623: Obtain the neighboring nodes of the current node, mark the neighboring nodes that belong to the same group as the current node as similar nodes, and add the similar nodes to the region set;

[0120] Step S624: Obtain the neighboring nodes of the same type of node and mark them as the same neighboring nodes; mark the same neighboring nodes that belong to the same group as the current node as the same type of nodes, and add the same type of nodes to the region set;

[0121] Step S625: Loop step S624 until the groups corresponding to the adjacent nodes of all nodes in the region set are different from the group corresponding to the current node, then the loop ends, obtains a region set, and marks all nodes in the region set as selected nodes;

[0122] Step S626: Loop through steps S621 to S625 until all nodes are marked as selected nodes. The loop ends and the result is obtained. A collection of regions.

[0123] S7: Based on the flood risk levels of different urban areas and combined with a pre-built historical disaster database, a multi-objective decision-making optimization algorithm is used to dynamically formulate differentiated emergency response plans.

[0124] Methods for dynamically developing differentiated emergency response plans include:

[0125] Obtain the number of monitoring stations in each urban area and mark it as the monitoring number; take the monitoring number and flood risk level of each urban area as a set of regional characteristic data, and the regional characteristic data corresponds to the urban area one by one; obtain the corresponding regional characteristic data of each set from the pre-built historical disaster database. Group rescue feature data, is an integer greater than 1; rescue feature data includes the number of rescued people and the quantity of various rescue materials, such as lifeboats, life jackets, sandbags, drainage pumps, etc.; the historical disaster database is constructed by those skilled in the art by collecting the rescue feature data corresponding to the urban areas with different regional feature data when historical flood disasters occurred; according to the corresponding data of each group of regional feature data The rescue feature data of each group is used to obtain the area range corresponding to each group of area feature data. The area range includes the number of people and the material range. The maximum value of the number of people is The maximum number of rescuers in the group rescue feature data, the minimum value of the number range is The minimum number of rescuers in the group rescue feature data; the material range includes the quantity range corresponding to each rescue material, and the maximum value of the quantity range is The maximum quantity of rescue materials corresponding to the rescue characteristics data of the group, and the minimum value of the quantity range is The minimum quantity of rescue supplies corresponding to the group rescue feature data;

[0126] From each range in each region, a value is randomly selected and a set of alternative sets is constructed. Group alternative sets, The set of group alternatives is different. is an integer greater than 1; the number of rescued people in each alternative set is added up in sequence to obtain the total number of rescued people; the number of materials with the same corresponding rescue materials in each alternative set is added up in sequence to obtain the total number of materials corresponding to each rescue material; according to the city's emergency resource management system, the deployment data is obtained, and the deployment data includes the number of deployed people and the deployment quantity of each rescue material; among which the number of deployed people is the total number of rescue personnel that the city can mobilize, and the deployment quantity is the total number of various rescue materials that the city can mobilize; in each alternative set, the total number of rescued people is compared with the number of deployed people, and each total quantity of materials is compared with the corresponding deployment quantity; the alternative sets with a total number of rescued people less than or equal to the number of deployed people and each total quantity of materials less than or equal to the corresponding deployment quantity are all marked as rescue sets; the alternative sets with a total number of rescued people greater than the number of deployed people, or with a total quantity of materials greater than the corresponding deployment quantity are not marked;

[0127] Different digital labels are set for different flood risk levels and marked as level labels; the flood risk levels in all regional feature data are replaced with corresponding level labels, and added to each rescue set in turn to obtain a rescue feature set; each rescue feature set is input into the trained effect prediction model to predict the corresponding rescue data; the rescue data includes rescue efficiency, casualty rate and response cost; the effect prediction model is a deep neural network model; a preset proportion set includes proportional coefficients corresponding to rescue efficiency, casualty rate and response cost, and the proportion set is pre-set by technical personnel in this field according to actual conditions; the rescue efficiency, casualty rate and response cost corresponding to each rescue set are multiplied by the corresponding proportional coefficient in the proportion set, and the efficiency coefficient, casualty coefficient and cost coefficient are obtained in turn; the efficiency coefficient is subtracted from the casualty coefficient, and then the cost coefficient is subtracted to obtain the rescue effect; the rescue effects corresponding to each rescue feature set are compared, and the rescue feature set with the largest rescue effect is marked as the best feature set; the rescue set in the best feature set is used as the emergency response plan.

[0128] This embodiment builds a spatially integrated rainfall monitoring network that integrates rainfall monitoring data from multiple sources, including rain gauges, weather radars, and remote sensing satellites, to achieve comprehensive, real-time monitoring of rainfall within the city and improve the spatiotemporal resolution and continuity of rainfall information. It employs time alignment and spatial registration techniques to effectively address time differences and coordinate inconsistencies between data sources from different monitoring methods, ensuring the spatiotemporal consistency of the fused rainfall data. A hydrodynamic model driven by the fused rainfall data accurately simulates urban surface hydrological processes and extracts flow characteristic data, laying a reliable foundation for flood risk assessment. A flood risk assessment system based on fuzzy comprehensive evaluation is constructed to intelligently assess flood risk levels in different urban areas, providing a scientific basis for emergency response. In conjunction with a historical disaster database, a multi-objective decision-making optimization algorithm is used to dynamically formulate differentiated emergency response plans. This approach minimizes casualties while ensuring rescue efficiency and reducing response costs, significantly improving the accuracy and effectiveness of emergency rescue. Intelligent perception and response to urban flood disasters effectively supports emergency response needs in dynamic environments, enhances urban flood prevention and control capabilities, and improves disaster resilience, thereby significantly enhancing the scientific nature and efficiency of flood disaster response.

[0129] Example 2

[0130] This application also provides an electronic device. The electronic device may include one or more processors and one or more memories. The memories may store computer-readable code that, when executed by the one or more processors, may implement the intelligent flood perception and decision-making method using multi-source rainfall data fusion as described above.

[0131] The method or system according to the embodiment of the present application can also be implemented with the aid of the architecture of the electronic device shown in the present application. The electronic device may include a bus, one or more CPUs, ROM, RAM, a communication port connected to a network, input / output, a hard disk, etc. The storage device in the electronic device, such as a ROM or a hard disk, can store the intelligent flood perception decision-making method based on the fusion of multi-source rainfall data provided in the present application. Furthermore, the electronic device may also include a user interface. Of course, the architecture shown in the present application is only exemplary. When implementing different devices, one or more components in the electronic device shown in the present application may be omitted according to actual needs.

[0132] Example 3

[0133] As shown, one embodiment of the present application discloses a computer-readable storage medium. The computer-readable storage medium stores computer-readable instructions. When executed by a processor, the computer-readable instructions can execute the intelligent flood awareness and decision-making method based on multi-source rainfall data fusion according to the embodiment of the present application described with reference to the above figures. The storage medium includes, but is not limited to, volatile memory and / or non-volatile memory. Volatile memory may include, for example, random access memory (RAM) and cache memory. Non-volatile memory may include, for example, read-only memory (ROM), a hard disk, and flash memory.

[0134] Furthermore, according to embodiments of the present application, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, the present application provides a non-transitory machine-readable storage medium storing machine-readable instructions capable of being executed by a processor to perform the steps corresponding to the method provided herein, such as a method for intelligent flood awareness and decision-making based on multi-source rainfall data fusion. When executed by a central processing unit (CPU), this computer program performs the functions defined in the method provided herein.

[0135] The foregoing description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art will be able to modify the technical solutions described in the foregoing embodiments or to substitute equivalents for some of the technical features. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention shall be included within the scope of protection of the present invention.

[0136] It should be noted that, in this document, the terms "comprises," "includes," or any other variations thereof are intended to encompass non-exclusive inclusion, such that a process, method, article, or apparatus comprising a series of elements includes not only those elements but also other elements not explicitly listed, or elements inherent to such process, method, article, or apparatus. In the absence of further limitations, an element defined by the phrase "comprising a ..." does not exclude the presence of other identical elements in the process, method, article, or apparatus comprising the element.

[0137] In the description of the present invention, it should be understood that the terms "first", "second", etc. are only used to distinguish the descriptions and cannot be understood as indicating or implying relative importance.

[0138] In the description of the present invention, unless otherwise specified, "plurality" means two or more.

[0139] In the description of the present invention, “several” means one or more, and “a large number” means two or more.

[0140] Throughout this specification, reference to terms such as "one embodiment," "some embodiments," "examples," "specific examples," or "some examples" means that a specific feature, structure, material, or characteristic described in conjunction with that embodiment or example is included in at least one embodiment or example of the present invention. In this specification, schematic representations of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in any one or more embodiments or examples.

[0141] The formulas in this manual are all dimensionless and calculated using numerical values. The formulas are obtained by collecting a large amount of data and performing software simulation to obtain the most recent real situation. The preset parameters and thresholds in the formulas are set by technicians in this field based on actual conditions.

[0142] While embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions, and variations may be made to the embodiments without departing from the principles and spirit of the invention, and that the scope of the invention is defined by the claims and their equivalents.

Claims

1. An intelligent flood perception and decision-making method based on multi-source rainfall data fusion, characterized by: include: S1: Using multi-source data acquisition technology, we build a spatially integrated rainfall monitoring network to collect multi-dimensional rainfall datasets in real time; S2: Temporal alignment and spatial registration of multidimensional rainfall datasets to generate rainfall processing data; S3: Using data fusion algorithm, multi-dimensional data fusion is performed on rainfall processing data to form fused rainfall data; S4: Acquire hydrological environment data and build a hydrological dynamics model based on the hydrological environment data; S5: Drive the hydrological dynamics model based on the integrated rainfall data to simulate the water flow evolution process and obtain water flow characteristic data; S6: Combine water flow characteristic data and integrated rainfall data to build a flood risk assessment system based on fuzzy comprehensive evaluation to intelligently assess the flood risk level of different urban areas; The steps of intelligently assessing flood risk levels in different urban areas include: Step S601: Divide the city into regions based on water flow characteristic data and obtain urban areas; Step S602: Obtain the rainfall corresponding to each urban area based on the fused rainfall data; count the number of rainfall corresponding to each urban area and mark it as rainfall points; add the rainfall of each urban area in sequence and divide it by the corresponding rainfall point number to obtain the average rainfall of each urban area; Step S603: construct multiple fuzzy sets for each of the water depth, water flow velocity, and average rainfall. Step S604: convert the fused feature data corresponding to each urban area into the membership degree of each corresponding fuzzy set using fuzzification technology. The fused feature data includes the water depth, water flow velocity, and average rainfall. Step S605: defining fuzzy rules; Step S606: Match each set of fuzzified fusion feature data with fuzzy rules, and perform fuzzy reasoning using a fuzzy reasoning method to obtain fuzzy reasoning results corresponding to each urban area. The fuzzy reasoning results include the membership degree corresponding to each flood risk level, which includes extreme risk, high risk, medium risk, and low risk. Step S607: Compare each degree of membership in each fuzzy inference result, and use the flood risk level corresponding to the degree of membership with the largest value as the flood risk level of the corresponding urban area; S7: Based on the flood risk levels of different urban areas and combined with a pre-built historical disaster database, a multi-objective decision-making optimization algorithm is used to dynamically formulate differentiated emergency response plans.

2. The intelligent flood perception and decision-making method based on multi-source rainfall data fusion according to claim 1 is characterized in that: The method for constructing a spatially integrated rainfall monitoring network includes: set up monitoring stations, is an integer greater than 1; Randomly select monitoring sites monitoring sites, and as a set of sites, a total of Group site collections, , ; Obtain the city range and the monitoring range of each monitoring station, merge the monitoring range of each monitoring station in each group of station sets, and obtain the total monitoring range of each station set; superimpose the total monitoring range of each station set with the city range, mark the station sets whose total monitoring range is greater than or equal to the city range as candidate sets, and do not mark the station sets whose total monitoring range is less than the city range; count the number of monitoring stations in each candidate set and mark it as the number of stations, and mark the candidate set with the smallest number of stations as the screening set; set a rain gauge for each monitoring station in the screening set, and build a spatial integrated rainfall monitoring network by combining meteorological radar and remote sensing satellite; Multidimensional rainfall datasets include The data consists of monitoring data, radar data and remote sensing data; the monitoring data is the rainfall collected by monitoring stations, the radar data is the rainfall collected by meteorological radars, and the remote sensing data is the rainfall collected by remote sensing satellites.

3. The intelligent flood perception and decision-making method based on multi-source rainfall data fusion according to claim 2 is characterized in that: Methods for temporal alignment of multidimensional rainfall datasets include: Get the acquisition time of monitoring data in the multidimensional rainfall dataset and mark it as the current time; get the acquisition time of radar data in the multidimensional rainfall dataset and mark it as radar time; get the acquisition time of remote sensing data in the multidimensional rainfall dataset and mark it as remote sensing time; subtract radar time from current time to get radar difference time; subtract remote sensing time from current time to get remote sensing difference time; get historical data, which includes Set of historical radar data and The historical remote sensing data is the historical radar data, the historical remote sensing data is the remote sensing data obtained at the historical moment. is an integer greater than 1; Set the preset radar sampling period, radar difference time and The historical radar data of the group are input into the trained radar prediction model in turn to predict the corresponding real-time radar data. The real-time radar data is the radar data corresponding to the current time. The preset remote sensing sampling period, remote sensing difference time and The historical remote sensing data of the group are input into the trained remote sensing prediction model in turn to predict the corresponding real-time remote sensing data. The real-time remote sensing data is the remote sensing data corresponding to the current time. Both the radar prediction model and the remote sensing prediction model are deep neural network models.

4. The intelligent flood perception and decision-making method based on multi-source rainfall data fusion according to claim 3 is characterized in that: Methods for spatial registration of multidimensional rainfall datasets include: Obtain the geographic coordinates of the monitoring site corresponding to each monitoring data in the multidimensional rainfall dataset and mark them as site coordinates; combine the monitoring data and site coordinates of each monitoring site to generate monitoring distribution data; project the real-time radar data to the WGS84 coordinate system to obtain projected radar data; obtain the geographic coordinates corresponding to each rainfall amount in the real-time remote sensing data and mark them as remote sensing coordinates; obtain the geographic coordinates corresponding to each rainfall amount in the projected radar data and mark them as radar coordinates; mark the site coordinates that are different from the remote sensing coordinates as first coordinates, and mark the site coordinates that are different from the radar coordinates as second coordinates; calculate the rainfall amount of each first coordinate based on the real-time remote sensing data using the spatial difference method; calculate the rainfall amount of each second coordinate based on the projected radar data using the spatial difference method; add each first coordinate and the corresponding rainfall amount to the real-time remote sensing data to generate remote sensing distribution data; add each second coordinate and the corresponding rainfall amount to the projected radar data to generate radar distribution data; Generate rainfall processing data based on remote sensing distribution data, radar distribution data and monitoring distribution data.

5. The intelligent flood perception and decision-making method based on multi-source rainfall data fusion according to claim 4 is characterized in that: The method for forming fused rainfall data comprises: The rainfall corresponding to the same geographical coordinates in the remote sensing distribution data, radar distribution data and monitoring distribution data is taken as a group of data sets, and the data sets correspond to the geographical coordinates one by one; a weight set is preset, and the weight set includes weight coefficients corresponding to the remote sensing distribution data, radar distribution data and monitoring distribution data; each data in each data set is multiplied by the corresponding weight coefficient in the weight set, and the results are added in sequence to obtain the fused rainfall corresponding to each data set; based on the fused rainfall corresponding to each data set, fused rainfall data is formed.

6. The intelligent flood perception and decision-making method based on multi-source rainfall data fusion according to claim 5 is characterized in that: The hydrological environment data includes meteorological data, topographic data, soil and vegetation data, watershed hydrological data and groundwater data; Methods for constructing hydrodynamic models include: Construct a surface runoff model, the input of which is the fused rainfall data, and the output is the net rainfall; construct a shallow water dynamics model, the input of which is the net rainfall, and the output is the water flow characteristic data, which includes the water depth and water flow velocity; couple the surface runoff model with the shallow water dynamics model to construct a hydrodynamic model; The steps to construct a shallow water dynamics model include: Step S401: using the Saint-Venant equations as basic equations, where the Saint-Venant equations include a continuity equation and a momentum equation; Step S402: converting hydrological environment data into model parameters; Step S403: Divide the city into grid cells, each of which corresponds to a monitoring site and a set of state variables, including water depth and water flow velocity; Step S404: setting boundary conditions, which include upstream boundary conditions, downstream boundary conditions, and lateral boundary conditions; wherein the upstream boundary condition is a depth time series, the downstream boundary condition is a depth-velocity relationship curve, and the lateral boundary condition is net rainfall; Step S405: setting initial conditions for each grid cell in turn, the initial conditions including initial water depth and initial water flow velocity; Step S406: Based on the model parameters, boundary conditions and initial conditions, a numerical calculation method is used to solve the Saint-Venant equations to construct a shallow water dynamics model; The fused rainfall data is input into the hydrological dynamics model to calculate the water flow characteristic data corresponding to each monitoring station.

7. The intelligent flood perception and decision-making method based on multi-source rainfall data fusion according to claim 6 is characterized in that: In step S601, obtain The steps for each city region include: Step S610: Take each set of water flow characteristic data as a node and randomly select nodes as the center point, and the nodes that are not the center point as the sample point. is an integer greater than 1; Step S611: According to The center points construct the corresponding For each group, calculate the Euclidean distance from each sample point to each center point in turn and mark it as point distance; Step S612: Preset a distance threshold, compare each point distance corresponding to each sample point with the distance threshold, mark the center point corresponding to the point distance whose value is less than the distance threshold as a classification point, and do not mark the point distance whose value is greater than or equal to the distance threshold. Then, classify each sample point into the group corresponding to the corresponding classification point. Step S613: Calculate the new center point corresponding to each group, and replace the center point of each group with the corresponding new center point; Step S614: loop through steps S611 to S613 until the new center point corresponding to each group calculated in step S613 is consistent with the corresponding new center point calculated in the previous loop, and the loop ends. groups; Step S615: Perform connectivity analysis on the nodes in each group to obtain A collection of regions; according to A collection of regions, dividing the city into urban areas; In step S613, the method for calculating the new center point includes: Count the number of nodes in each group and mark it as the number of nodes; add up the nodes in each group in sequence and divide it by the corresponding number of nodes to obtain the average point corresponding to each group; calculate the point distance from each node in each group to the corresponding average point and mark it as the average distance; compare the same average distances of corresponding groups, and take the node corresponding to the smallest average distance as the new center point of the corresponding group.

8. The intelligent flood perception and decision-making method based on multi-source rainfall data fusion according to claim 7 is characterized in that: In step S615, obtain The steps for a region collection include: Step S621: randomly select a node that is not marked as a selected node and mark it as the current node; Step S622: construct a region set and add the current node to the region set; Step S623: Obtain the neighboring nodes of the current node, mark the neighboring nodes that belong to the same group as the current node as similar nodes, and add the similar nodes to the region set; Step S624: Obtain the neighboring nodes of the same type of node and mark them as the same neighboring nodes; mark the same neighboring nodes that belong to the same group as the current node as the same type of nodes, and add the same type of nodes to the region set; Step S625: Loop step S624 until the groups corresponding to the adjacent nodes of all nodes in the region set are different from the group corresponding to the current node, then the loop ends, obtains a region set, and marks all nodes in the region set as selected nodes; Step S626: Loop through steps S621 to S625 until all nodes are marked as selected nodes. The loop ends and the result is obtained. A collection of regions.

9. The intelligent flood perception and decision-making method based on multi-source rainfall data fusion according to claim 8 is characterized in that: The method for dynamically formulating differentiated emergency response plans includes: Obtain the number of monitoring stations in each urban area and mark it as the monitoring number; take the monitoring number and flood risk level of each urban area as a set of regional characteristic data; obtain the corresponding regional characteristic data of each set from the pre-built historical disaster database. Group rescue feature data, is an integer greater than 1; rescue feature data includes the number of rescued people and the quantity of various rescue materials; according to the corresponding regional feature data of each group The rescue feature data of each group are obtained, and the regional range corresponding to each group of regional feature data is obtained; a value is randomly selected from each range within each regional range, and a set of candidate sets is constructed. Group alternative sets; add up the number of rescued people in each alternative set in sequence to obtain the total number of rescued people; add up the number of materials with the same corresponding rescue materials in each alternative set in sequence to obtain the total number of materials corresponding to each rescue material; obtain allocation data, which includes the number of allocated people and the allocated quantity of each rescue material; compare the total number of rescued people with the allocated number in each alternative set, and compare each total number of materials with the corresponding allocated quantity; mark the alternative sets whose total number of rescued people is less than or equal to the allocated number and whose total number of materials is less than or equal to the corresponding allocated quantity as rescue sets; alternative sets whose total number of rescued people is greater than the allocated number or whose total number of materials is greater than the corresponding allocated quantity will not be marked; Different digital labels are set for different flood risk levels and marked as level labels; the flood risk levels in all regional feature data are replaced with corresponding level labels, and added to each rescue set in turn to obtain a rescue feature set; each rescue feature set is input into the trained effect prediction model to predict the corresponding rescue data; the rescue data includes rescue efficiency, casualty rate and response cost; the effect prediction model is a deep neural network model; a preset proportion set includes proportional coefficients corresponding to rescue efficiency, casualty rate and response cost; the rescue efficiency, casualty rate and response cost corresponding to each rescue set are multiplied by the corresponding proportional coefficient in the proportion set, and the efficiency coefficient, casualty coefficient and cost coefficient are obtained in turn; the efficiency coefficient is subtracted from the casualty coefficient, and then the cost coefficient is subtracted to obtain the rescue effect; the rescue feature set with the largest rescue effect is marked as the optimal feature set, and the rescue set in the optimal feature set is used as the emergency response plan.

Images