Waterlogging prediction method based on behavior thermodynamic diagram and infiltration dynamic modeling

Abstract

The invention provides a waterlogging prediction method based on a behavior thermodynamic diagram and infiltration dynamic modeling, and the method comprises the steps: firstly dividing a space region based on topographic data, extracting a surface type, a soil attribute and vegetation information, and constructing an infiltration capacity model which can be dynamically adjusted along with rainfall and time; then combining the surface temperature and meteorological data to establish an evaporation model considering the urban heat island effect; further generating a behavior thermodynamic diagram by using the crowd trajectory data, and extracting a behavior feedback factor reflecting an abnormal movement mode; according to the method, rainfall, infiltration, evaporation, drainage capacity, behavior feedback factors and the like are fused, a machine learning model used for predicting the waterlogging probability of each region is constructed, a loss function is optimized by introducing a weight mechanism of the behavior feedback factors, and the response sensitivity of the model to a risk region is improved. The method gives consideration to the physical mechanism and social behavior characteristics, and improves the accuracy and timeliness of waterlogging prediction in a complex urban environment.

Description

Technical Field

[0001] The present invention belongs to the technical field of waterlogging prediction, and particularly relates to a waterlogging prediction method based on behavior heat maps and infiltration dynamic modeling. Background Art

[0002] With the continuous acceleration of the urbanization process, the problem of urban waterlogging has become increasingly prominent, becoming an important factor affecting the safe operation of cities and the quality of life of residents. Waterlogging not only causes traffic paralysis and property losses, but also seriously threatens public safety. Especially in the context of frequent extreme weather events, the prevention and control of waterlogging problems are particularly urgent.

[0003] Existing waterlogging prediction methods mainly rely on meteorological data and drainage system information, and judge waterlogging risks by analyzing rainfall processes and drainage capabilities. However, due to the complex and changeable urban terrain and significant differences in surface cover types, if these factors are not fully considered in the model, it often leads to a decrease in the accuracy of waterlogging process simulation. Differences in surface hardening degrees and soil infiltration capabilities in different regions directly affect the flow paths and accumulation speeds of rainwater. Ignoring these factors easily causes large deviations in the prediction of waterlogging risks by the model.

[0004] In addition, local microclimate changes (such as the urban heat island effect) caused by factors such as dense buildings and scarce green spaces in the urban environment will affect the evaporation and infiltration processes of surface water, and thus affect the formation and dissipation of waterlogging. If the prediction model fails to reflect such environmental dynamic factors, it is difficult to accurately predict the spatial and temporal distribution and severity of waterlogging.

[0005] In the prior art, Chinese Patent CN119378448A discloses a method for constructing an urban waterlogging prediction model, an urban waterlogging prediction method and a device, which relate to the technical field of data processing. The method for constructing an urban waterlogging prediction model includes: constructing an urban waterlogging process model coupled with an emergency drainage mode based on regional terrain data, hydrometeorological data and flood control response data; inputting rainfall data under different scenarios into the urban waterlogging process model to obtain waterlogging characteristic data; performing a correlation analysis on the rainfall data and the waterlogging characteristic data to determine rainfall parameters related to flood risks from the rainfall data; constructing sample data according to the rainfall parameters and the waterlogging characteristic data, and training a neural network model based on the sample data to obtain an urban waterlogging prediction model. Although this method considers factors such as terrain, hydrology and drainage facilities to a certain extent, its model lacks consideration of surface cover differences, local environmental changes and real-time dynamic responses, and is prone to insufficient adaptability of the prediction results to actual complex situations.

[0006] In addition, existing methods mainly rely on preset flood control response measures and drainage facility scheduling, lacking real-time reflection of the dynamic changes in the city (such as population distribution), which affects the real-time performance and accuracy of the prediction model.

[0007] Therefore, the existing technologies still have deficiencies in terms of comprehensive consideration of multiple factors and dynamic adaptation ability of the model, and it is difficult to meet the requirements of high precision and real-time for urban waterlogging risk management. Summary of the Invention

[0008] The purpose of the present invention is to overcome the deficiencies of the above-mentioned existing technologies and provide a waterlogging prediction method based on behavior heat map and infiltration dynamic modeling.

[0009] The purpose of the present invention can be achieved through the following technical solutions: The present invention provides a waterlogging prediction method based on behavior heat map and infiltration dynamic modeling, including the following steps: Collect multi-source data of the target area, including environmental data, meteorological data, terrain data, drainage pipe network data, population trajectory data, and surface temperature data; According to the terrain data, divide the space area, extract the surface cover type, soil property, and vegetation cover situation of each space area, and construct a rainwater infiltration capacity model that dynamically adjusts with time and rainfall conditions; Based on the surface temperature data and environmental data, establish an evaporation model reflecting the urban heat island effect, and calculate the evaporation capacity of each space area; Use the population trajectory data to generate a behavior heat map that changes with time, and use the degree of personnel aggregation and abnormal behavior changes in the behavior heat map as behavior feedback factors; Combine the rainwater infiltration capacity model, evaporation model, behavior feedback factors, meteorological data, and drainage pipe network data to construct a waterlogging prediction model; According to the waterlogging prediction model, output the waterlogging occurrence probability of each space area in the future time period.

[0010] Furthermore, the construction of the rainwater infiltration capacity model that dynamically adjusts with time and rainfall conditions specifically includes: According to the terrain data, divide the target area into several space areas; For each space area, based on the terrain data, extract the main surface types within the space area, including bare soil, grassland, concrete, asphalt, and masonry paving, and calculate the area proportion of each surface type within the space area ; For each surface type j , obtain its corresponding infiltration-related parameters, including: saturated hydraulic conductivity , capillary suction , saturated volumetric water content , volumetric water content of the previous time period ; among them, the saturated hydraulic conductivity refers to the surface type jThe amount of water that can infiltrate per unit time under fully saturated conditions, capillary suction Refers to the suction force exhibited when water is adsorbed in the capillary pores of the soil in the surface type j , saturated volumetric water content Refers to the surface type j The maximum amount of water that can be accommodated in a unit volume of soil in a fully saturated state, the volumetric water content of the previous time period Refers to the surface type j The actual water content of the soil at the previous moment; saturated hydraulic conductivity , capillary suction , saturated volumetric water content Obtained through geological data and soil databases, the volumetric water content of the previous time period Obtained through the measured data of soil moisture sensors; An improved algorithm based on the Green-Ampt model is adopted. According to the infiltration-related parameters of each spatial region, a dynamic evolution rainwater infiltration capacity model of the infiltration capacity of each spatial region changing with time is constructed to calculate the instantaneous infiltration rate at different times t as follows: where, i represents the instantaneous infiltration rate of the t th spatial region at time i , which represents the rate at which rainwater is absorbed by the surface of the t th spatial region per unit area of the surface at time , t represents the cumulative infiltration depth of the i th type of surface in the j th spatial region from the initial time to time j , obtained by integrating the instantaneous infiltration rate of the th type of surface with whether precipitation occurs over time, is the total number of surface types.

[0011] Furthermore, dividing the target area into several spatial regions according to the terrain data specifically includes: Calculating the slope and aspect of each grid cell in the digital elevation model according to the digital elevation model in the terrain data of the target area; Based on the slope and aspect, combined with the analysis of the water flow path, calculating the catchment area of each grid cell; By setting the basin area threshold and slope threshold, screening the spatial regions that meet the conditions: where,​ A set of spatial regions obtained by screening and partitioning the resultant terrain data represents the catchment area of a raster cell ; is the catchment area threshold ; represents the slope of a raster cell ; is the slope threshold

[0012] Furthermore, based on the surface temperature data and environmental data, an evaporation model reflecting the urban heat island effect is established, which specifically includes: Collect the surface temperature data and environmental data of each spatial region in the target area. The environmental data includes surface temperature, air temperature, wind speed, relative humidity, and solar radiation; According to the surface temperature data and environmental data, use the modified Penman-Monteith model to construct an evaporation model considering the urban heat island effect, and calculate the potential evaporation rate of each spatial region in the future time period , and the formula is as follows: where represents the potential evaporation rate of the i th spatial region at time t ; represents the slope of the saturation vapor pressure curve of the i th spatial region, calculated according to the air temperature; represents the net radiation of the i th spatial region, estimated according to the solar radiation and surface temperature; represents the soil heat flux of the i th spatial region; is the humidity constant; represents the average air temperature of the i th spatial region; represents the wind speed at a height of 2 meters above the ground of the i th spatial region, , respectively represent the saturation vapor pressure and actual vapor pressure of the i th spatial region

[0013] Furthermore, the generation of a behavior heat map that changes with time using the crowd trajectory data specifically includes: Collect the crowd trajectory data within the target area , and the trajectory data is obtained by mobile signal, WiFi detection, or video monitoring, and is used to represent the spatial position of an individual at time t ; Map the personnel trajectory data to multiple spatial regions divided by the target area, and count the number of people in each spatial region at any time. t number of people inside ; According to each spatial region at any time t number of people inside The area of each space region Calculate the i A region of space in time t Crowd density : Gaussian kernel function is used to calculate the crowd density Perform kernel density estimation and generate the moment t Behavior heat map: in, Indicates spatial location At the moment t The crowd behavior heat map value, Indicates i The coordinates of the centroid of the spatial region, represents the total number of spatial regions, is the kernel function, is the bandwidth parameter.

[0014] Furthermore, the behavior feedback factor calculation process is: Calculate the average density value of the historical behavior heat map , as a baseline reference; Based on the average density value of the current behavior heat map and the historical behavior heat map of the same period , calculate the behavioral anomaly factor: in, Indicates spatial location At the moment t behavioral abnormalities; Setting Thresholds , when the behavioral abnormality factor When , the crowd gathering intensity is extracted as the risk weighting factor: Risk-weighted factors By integrating and averaging according to the spatial area, we get i Behavioral feedback factor for each spatial region: in, Indicates iThe behavior feedback factor of a spatial region at a certain time t

[0015] Furthermore, the historical same period refers to the historical time period corresponding to the current time point in the time cycle.

[0016] Furthermore, by combining the rainwater infiltration capacity model, evaporation model, behavior feedback factor, meteorological data and drainage pipe network data, an urban waterlogging prediction model is constructed, which specifically includes: Obtain the rainfall intensity i of each spatial region at a certain time t , instantaneous infiltration rate , potential evaporation rate , and drainage capacity to estimate the surface water volume aggregation index per unit area of each spatial region i : Among them, represents the surface water volume aggregation index per unit area of the spatial region i , is the time interval, that is, the time length from the previous time step to the current time step, , respectively represent the rainfall intensity, instantaneous infiltration rate and potential evaporation rate of the spatial region i , is the drainage capacity of the spatial region i , obtained according to the drainage pipe network data of each spatial region i ; is the rainfall intensity of the spatial region i , obtained according to the meteorological data; represents the area of the spatial region i ; Taking the characteristic information of each spatial region at each time step as input, training samples are constructed. The characteristic information includes the estimated surface water volume aggregation index, behavior feedback factor and meteorological data, where the meteorological data includes the current rainfall intensity and the cumulative rainfall in the previous period of time; Whether waterlogging occurs in each spatial region is used as the supervised learning label, and the label is determined based on the historical waterlogging event data, and the historical waterlogging event data is obtained according to the remote sensing image and the water level sensor monitoring data; Construct an urban waterlogging prediction model based on a classification model to predict the probability of waterlogging occurrence. The classification model is a probability model, including a gradient boosting decision tree model; ​​The above-mentioned waterlogging prediction model takes the characteristic information of each spatial region as input, outputs the predicted value of the waterlogging occurrence probability at the corresponding moment, and trains the waterlogging prediction model through a loss function based on the output predicted value of the waterlogging probability to obtain the trained waterlogging prediction model.

[0017] Further, the drainage capacity is calculated through the Manning formula according to the drainage pipe network data of each spatial region. The formula is: where, is the number of drainage pipes in the spatial region i , represents the hydraulic radius of the j th drainage pipe, represents the hydraulic gradient of the j th drainage pipe, represents the cross-sectional area of flow of the j th drainage pipe, represents the Manning roughness coefficient of the j th drainage pipe.

[0018] Further, the loss function is: where, is the loss function, is the weight coefficient, is the adjustment parameter, is the behavior feedback factor of the spatial region i , represents the true label of waterlogging in the spatial region i at time t . 1 indicates waterlogging occurs, 0 indicates no waterlogging occurs, and the label is determined based on historical waterlogging event data; represents the predicted value of the waterlogging occurrence probability of the i th spatial region output by the waterlogging prediction model at time t ; represents the total number of spatial regions, represents the number of time steps in the training sample.

[0019] Compared with the prior art, the present invention has the following advantages: (1) By utilizing the digital elevation model in the terrain data, the present invention calculates the slope and aspect of each grid cell, and calculates the catchment area based on the slope and the analysis of the water flow path. The target area is screened and divided by using the basin area threshold and the slope threshold, which solves the problem in the prior art that the key areas prone to waterlogging in the target area are not effectively divided, resulting in a large amount of calculation for waterlogging prediction and difficulty in accurately identifying the waterlogging risk. By screening out the spatial areas with a large catchment area and a steep slope, the key areas prone to waterlogging in the target area are accurately located, avoiding redundant calculations for areas with little influence on water flow concentration or areas where waterlogging is not likely to occur, thereby significantly reducing the consumption of computing resources, improving the computing efficiency and response speed of the waterlogging prediction model. At the same time, this division method ensures that the waterlogging prediction model can focus on the areas where the hydrological process is active, enhances the pertinence and accuracy of the prediction, and improves the model's ability to identify waterlogging risks and its practical application value.

[0020] (2) The present invention calculates the instantaneous infiltration rate, aiming to solve the problems in traditional waterlogging simulation, such as rough modeling of the rainwater infiltration process, insufficient dynamic response ability, and inadequate consideration of surface heterogeneity. During the actual rainfall process, the infiltration capacity of the surface is constantly changing, which is comprehensively affected by the surface type, soil moisture state, and rainfall intensity. However, this point is not considered in the prior art. The present invention introduces a dynamic infiltration capacity modeling method based on the Green-Ampt model to calculate the instantaneous infiltration rate at each time point by dynamic update. Through the refined and dynamic calculation of the instantaneous infiltration rate, the present invention realizes the accurate simulation of the response of surface infiltration to rainfall, can more accurately judge the time point and intensity of surface runoff generation during rainfall, enhances the model's response ability to the dynamic behavior of rainwater, significantly improves the physical rationality of surface runoff simulation, reduces the error accumulation caused by inaccurate infiltration estimation, and thus improves the accuracy and reliability of the urban waterlogging prediction system as a whole, providing a more timely and spatially targeted decision-making basis for urban drainage planning, waterlogging warning, and emergency response.

[0021] (3) By constructing an evaporation model under the urban heat island effect based on surface temperature data and environmental data, the present invention aims to solve the problem of neglecting or roughly dealing with the evaporation process in traditional waterlogging simulation, especially in areas with significant urban heat island effect where the evaporation capacity changes drastically and has an important impact on the surface water balance. However, the existing technologies do not consider the situation of rainwater evaporation when considering waterlogging, resulting in insufficient simulation accuracy of the key processes of the surface water cycle. The present invention collects multi-source environmental data such as surface temperature, air temperature, wind speed, relative humidity, and solar radiation, and uses the modified Penman-Monteith model to establish an evaporation model reflecting the urban heat island effect, dynamically calculating the potential evaporation rate of each spatial region in the future time period. By introducing high-precision dynamic evaporation modeling, the present invention effectively enhances the simulation ability of the surface water evaporation process, providing key support for accurately estimating the net surface water accumulation, judging the generation of local runoff, and the evolution of waterlogging. This method improves the physical completeness and parameter adaptability of the entire waterlogging simulation system, significantly enhances the prediction accuracy and spatio-temporal resolution ability of the model in complex urban environments, and provides more reliable technical support for urban water environment management, extreme rainfall warning, and drainage scheduling decision-making.

[0022] (4) The behavior feedback factor calculated by the present invention is an index used to quantify the abnormal degree of crowd behavior, reflecting the deviation degree of the crowd from the historical behavior pattern within a specific time and space range. Based on the behavior heat map generated from crowd trajectories, by comparing the crowd density distribution at the current moment with the benchmark density distribution in the same historical period, abnormal behaviors such as crowd gathering or avoidance are identified, thereby inferring that the area may be experiencing abnormal environmental or social events. In extreme weather or sudden waterlogging situations, crowd behavior often reacts prior to physical changes, such as phenomena like detouring, gathering, staying, and centralized evacuation. Therefore, the behavior feedback factor has strong sensitivity and foresight in both spatial and temporal dimensions, capable of promptly capturing areas in the city where risk events may be occurring or about to occur. As part of the waterlogging prediction model, it can effectively reflect the location and evolution trend of potential waterlogging risk areas, make up for the deficiencies of traditional hydrological models in real-time and sudden event identification, improve the model's response ability to the dynamic operation of the city, and contribute to more timely and accurate risk warning and dispatching response.

[0023] (5) By constructing an urban waterlogging prediction model that integrates a rainwater infiltration capacity model, an evaporation model, a behavior feedback factor, meteorological data, and drainage pipe network data, the present invention solves the problems of incomplete physical mechanisms, insufficient response to dynamic changes, and lack of crowd behavior perception ability in existing urban waterlogging predictions. Traditional models often only consider rainfall intensity and drainage capacity, while ignoring the dynamic evolution of key water cycle processes such as surface infiltration and evaporation, resulting in inaccurate estimation of water volume surplus and prone to prediction errors. The present invention introduces high-resolution spatio-temporal dynamic instantaneous infiltration rate and potential evaporation rate, effectively supplementing the physical elements of the surface water volume balance process modeling and enhancing the authenticity and sensitivity of the surface water volume aggregation index per unit area. At the same time, the behavior feedback factor reflects the reaction characteristics of the crowd in sudden waterlogging events, realizing enhanced identification of abnormal areas and helping to capture waterlogging signs that are difficult to directly obtain from ground perception signals.

[0024] (6) The loss function designed by the present invention solves the problems of sample imbalance and insufficient identification of risk areas in traditional urban waterlogging predictions by introducing a weight coefficient based on the behavior feedback factor. Urban waterlogging events usually have a low occurrence probability, resulting in extremely uneven distribution of positive and negative samples, which easily makes the model tend to predict the "no waterlogging" category, thus ignoring potential high-risk areas. By integrating the behavior feedback factor into the weight, the sample loss weight corresponding to areas with a higher probability of waterlogging or abnormal crowd movement is amplified, and the model pays more attention to the prediction accuracy of these key spatio-temporal points during the training process. The adjustment parameter α in the weight coefficient can flexibly control the influence intensity of the behavior feedback factor on the loss function, realizing the sensitive response of the model to the risk areas guided by abnormal crowd behavior. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] Figure 1 is a diagram of the urban waterlogging prediction method according to an embodiment of the present invention; Figure 2 is a model diagram of the urban waterlogging prediction system according to an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0026] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are some, but not all, of the embodiments of the present invention. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0027] This embodiment provides an urban waterlogging prediction method based on behavior heatmap and infiltration dynamic modeling, as Figure 1 shown, including the following steps: Step S1: Collect multi-source data of the target area, including environmental data, meteorological data, terrain data, drainage network data, population trajectory data, and surface temperature data. Among them, environmental data includes meteorological elements such as air temperature, wind speed, relative humidity, and solar radiation, as well as relevant environmental characteristics reflecting the urban heat island effect. The environmental data can be obtained through multiple channels such as meteorological monitoring stations, environmental monitoring equipment, remote sensing satellites, and ground sensors. Meteorological data mainly includes the current rainfall intensity and the cumulative rainfall in the past period of time, which is usually collected in real time through meteorological radars, rain gauges, meteorological satellites, and meteorological forecasting systems. Terrain data includes information such as digital elevation model (DEM), slope, aspect, and catchment area. Using these data, the target area can be divided into multiple spatial regions, and further, the surface cover types (such as bare soil, grassland, concrete, asphalt, masonry paving), soil properties, and vegetation cover conditions within each spatial region can be extracted. These terrain data are generally obtained through lidar (LiDAR), aerial photogrammetry, remote sensing satellite images, and geographic information system (GIS) databases. Drainage network data covers information such as the number, location, pipe diameter, cross-sectional area, hydraulic radius, hydraulic slope, and Manning roughness coefficient of drainage pipes. The drainage network data is obtained from urban drainage network design drawings, network management systems, on-site measured data, and urban infrastructure databases. Population trajectory data represents the spatial positions of individuals in the target area at different time points, which is used to reflect the movement and aggregation of the population. This data can be collected through mobile communication signals, WiFi detection devices, and video surveillance systems. Surface temperature data is the surface heat condition information collected through remote sensing satellite infrared thermal imaging equipment, ground thermal cameras, and environmental monitoring stations. By collecting the above multi-source data, the natural environment, meteorological conditions, terrain features, drainage capacity, and population behavior characteristics of the target area can be detailedly reflected, providing a comprehensive and accurate data basis for the subsequent construction of dynamic rainwater infiltration capacity models, evaporation models, behavior feedback factors, and waterlogging prediction models.

[0028] Step S2: Divide the spatial regions according to the terrain data, extract the surface cover types, soil properties, and vegetation cover conditions of each spatial region, and construct a rainwater infiltration capacity model that dynamically adjusts with time and rainfall conditions. Step S2 specifically includes: According to the terrain data, divide the target area into several spatial regions, specifically including: According to the digital elevation model in the terrain data of the target area, calculate the slope and aspect of each grid cell in the digital elevation model. Based on the slope and aspect, combined with the water flow path analysis, calculate the catchment area of each grid cell. By setting the basin area threshold and slope threshold, screen the spatial regions that meet the conditions: Among them, is the set of spatial regions obtained by screening and dividing the result terrain data, represents the grid cell 's catchment area, is the threshold of the basin area, represents the grid cell 's slope, represents the slope threshold.

[0029] For each spatial region, based on the terrain data, extract the main surface types within the spatial region, including bare soil, grassland, concrete, asphalt, and masonry paving, and calculate the area proportion of each surface type within the spatial region ; For each surface type j , obtain its corresponding infiltration-related parameters, including: saturated hydraulic conductivity , capillary suction , saturated volumetric water content , volumetric water content in the previous time period ; among them, the saturated hydraulic conductivity refers to the amount of water that the surface type j can infiltrate per unit time under fully saturated conditions, the capillary suction refers to the suction force exhibited when water is adsorbed in the soil capillary pores of the surface type j , the saturated volumetric water content refers to the maximum amount of water that the surface type j can hold per unit volume of soil in a fully saturated state, and the volumetric water content in the previous time period refers to the actual water content of the soil of the surface type j at the previous moment; the saturated hydraulic conductivity , capillary suction , saturated volumetric water content are obtained through geological data and soil databases, and the volumetric water content in the previous time period is obtained through measured soil moisture sensor data; Adopt an improved algorithm based on the Green-Ampt model. According to the infiltration-related parameters of each spatial region, construct a dynamic evolution rainwater infiltration capacity model of the infiltration capacity of each spatial region over time, and calculate the instantaneous infiltration rate t at different times , and the calculation method is as follows: Among them, represents the instantaneous infiltration rate of the i th spatial region at time t , representing thei At a certain moment in time, the rate at which rainwater is absorbed by the surface of a unit area of the ground in a spatial region t represents the cumulative infiltration depth of the -th type of surface in the t -th spatial region from the initial moment to time i , which is obtained by integrating the instantaneous infiltration rate of the j -th type of surface and whether it is raining over time. j Here, is the total number of surface types.

[0030] Since the target area is usually vast and has complex and diverse terrains, directly predicting waterlogging for the entire area requires detailed simulation of rainwater infiltration and hydrological processes for each spatial unit. This full-area, high-resolution modeling method has an extremely large computational workload, consumes a very high amount of computing resources during model training and prediction, is difficult to achieve real-time or rapid prediction, and limits the practicality and response speed of the waterlogging warning system. To address this technical challenge, in step S2 of this embodiment, high-precision terrain data in the digital elevation model (DEM) is first used to reasonably divide the target area. By calculating the slope and aspect of each grid cell and combining the analysis of the water flow path and catchment area, spatial regions with significant hydrological impacts are selected, and regions with a relatively low likelihood of waterlogging occurrence are excluded. This division strategy can effectively reduce the spatial dimension of the model, reduce redundant calculations, and thus significantly compress the data processing scale and computational burden. Within each divided spatial region, considering the surface cover types (such as bare soil, grassland, concrete, etc.), soil physical properties, and vegetation cover in the region, an improved Green-Ampt infiltration model is used to dynamically calculate the rainwater infiltration capacity. This model comprehensively considers parameters such as the saturated hydraulic conductivity, capillary suction, and volumetric water content of each surface type, and introduces the soil water content state in the previous time period to achieve a dynamic response of the infiltration process to time and rainfall conditions. By calculating the area weighting of different surface types, the heterogeneity of rainwater infiltration characteristics within the spatial region can be accurately reflected. Through spatial division based on terrain data, the number of spatial units that the model needs to process is significantly reduced, and the overall computational complexity is lowered; secondly, the dynamic rainwater infiltration capacity model can capture changes in rainfall intensity and soil moisture state in real time, improving the timeliness and accuracy of the infiltration process; finally, considering the heterogeneity of terrain, hydrology, and surface cover comprehensively effectively improves the spatial resolution and prediction accuracy of the waterlogging prediction model. Overall, the technical bottleneck of excessive computational workload for waterlogging modeling in a large-scale target area is solved, and the prediction efficiency and application feasibility are improved.

[0031] Step S3: Based on the surface temperature data and environmental data, establish an evaporation model reflecting the urban heat island effect and calculate the evaporation capacity of each spatial region; Step S3 specifically includes: Collect the surface temperature data and environmental data of each spatial area in the target area. The environmental data includes surface temperature, air temperature, wind speed, relative humidity, and solar radiation; According to the surface temperature data and environmental data, use the modified Penman-Monteith model to construct an evaporation model considering the urban heat island effect, and calculate the potential evaporation rate of each spatial area in the future time period , and the formula is as follows: Among them, represents the i potential evaporation rate of the t spatial area at time represents the slope of the saturated vapor pressure curve of the i spatial area, calculated according to the air temperature; represents the i net radiation of the spatial area, estimated according to the solar radiation and surface temperature; i represents the soil heat flux of the spatial area; is the humidity constant; i represents the average air temperature of the i spatial area; , respectively represent the i saturated vapor pressure and actual vapor pressure of the

[0032] Step S3 is to accurately reflect the impact of the urban heat island effect on the evaporation process, thereby improving the prediction accuracy of the hydrological model for evaporation. The urban heat island effect causes the surface temperature in urban areas to be generally higher than that in the surrounding suburbs. This temperature difference significantly affects the evaporation rate and is directly related to the simulation of the rainwater moisture cycle and soil moisture dynamics. The spatial scope of the target area is extensive and the surface types are complex, with significant spatial differences in surface temperature and environmental conditions. Simply using a unified evaporation model cannot accurately reflect the evaporation characteristics of different spatial regions. Therefore, in this embodiment, high-resolution surface temperature data and environmental data (including air temperature, wind speed, relative humidity, and solar radiation) are collected to achieve a detailed characterization of the environmental conditions in each spatial region. Based on these data, a modified Penman-Monteith model is used, combined with the correction of the urban heat island effect on surface temperature and radiation, to establish a dynamic evaporation model. This model not only considers traditional meteorological factors but also incorporates the impact of the urban heat island effect on net radiation and soil heat flux, making the calculation of potential evaporation rate more in line with the actual situation of the urban environment. By accurately calculating the potential evaporation rate in each spatial region, the model can reflect the evaporation differences under different terrains and coverage types, improving the response ability of hydrological simulation to the water evaporation process. Through detailed spatial zoning combined with multi-source environmental data, the spatial resolution and time dynamic response ability of the evaporation model are improved; and the introduction of corrections related to the urban heat island effect solves the error problem caused by ignoring the high-temperature effect in the evaporation calculation of traditional models in urban areas; overall, the accuracy of rainwater cycle simulation is enhanced, providing more scientific support for urban water resource management and urban flooding risk prediction.

[0033] Step S4: Generate a behavior heat map that changes over time using crowd trajectory data, and use the degree of personnel aggregation and abnormal behavior changes in the behavior heat map as behavior feedback factors; Step S4 specifically includes: Collect crowd trajectory data within the target area , and the trajectory data is obtained by mobile signals, WiFi detection, or video monitoring, and is used to represent the spatial position of an individual at a certain time t ; Map the personnel trajectory data to multiple spatial regions divided in the target area, and count the number of people in each spatial region at any given time t ; ; According to the number of people in each spatial region at any given time t and the area of each spatial region , calculate the crowd density of the th i spatial region at time t : : Use a Gaussian kernel function to estimate the crowd density and generate a behavior heat map at a certain moment t : Among them, represents the value of the behavior heat map of the crowd at the spatial position at the moment t ; represents the centroid coordinates of the i th spatial region, represents the total number of spatial regions, is the kernel function, is the bandwidth parameter.

[0034] The calculation process of the behavior feedback factor is as follows: Calculate the average density value of the behavior heat map in the same historical period , which is used as the baseline reference; the same historical period refers to the historical time period corresponding to the current time point in the time cycle.

[0035] Based on the average density value of the behavior heat map at the current moment and the behavior heat map in the same historical period , calculate the behavior anomaly factor: Among them, represents the behavior anomaly factor at the spatial position at the moment t ; Set a threshold . When the behavior anomaly factor , extract the crowd aggregation intensity as the risk weighting factor: Integrate and average the risk weighting factor by spatial region to obtain the behavior feedback factor of the i th spatial region: Among them, represents the behavior feedback factor of the i th spatial region at time t .

[0036] The core purpose of calculating the behavior feedback factor in step S4 of this embodiment is to indirectly reflect the potential risk of waterlogging and its scope of influence by dynamically monitoring the degree of crowd gathering and abnormal behavior in the target area. When waterlogging occurs, ground waterlogging, traffic blockage and damage to public facilities usually cause crowds to gather in limited safe areas or abnormal evacuation and retention behaviors. Traditional waterlogging risk assessment mostly relies on static meteorological, hydrological and topographic data, which is difficult to timely reflect the complex crowd response dynamics caused by waterlogging, thereby affecting the effective formulation and implementation of prevention and control measures. By collecting and analyzing real-time crowd trajectory data and calculating the behavior feedback factor, it is possible to keenly capture the abnormal gathering or dispersion behavior of the crowd caused by waterlogging. For example, in areas with severe waterlogging, the behavior feedback factor will show an abnormal peak value that is significantly deviated from the historical normal level. This abnormal change is not only a reflection of the crowd response, but also an important indicator signal of the occurrence and development of waterlogging. The introduction of the behavior feedback factor makes up for the deficiency of waterlogging impact assessment based solely on environmental and hydrological models, and realizes the dynamic coupling monitoring of the human-water-environment system. The calculation method of the behavioral feedback factor can accurately locate the crowd gathering hotspots and abnormal areas in high-risk areas of waterlogging through kernel density estimation and smoothing processing; through the comparison of historical data for the same period, the risk threshold is dynamically adjusted to reduce the false alarm rate and improve the timeliness and accuracy of detection. These feedback factors are integrated into the waterlogging risk assessment model as important behavioral input parameters to assist in judging the spatial distribution and severity of waterlogging events and improve the scientificity and practicality of the overall risk warning. The behavioral feedback factor of this embodiment not only realizes the accurate quantification of the dynamic behavior of the crowd, but more importantly, by reflecting the behavioral anomalies caused by waterlogging, it establishes the intrinsic connection between crowd behavior and waterlogging risk, effectively solves the problem of insufficient dynamic response of traditional models to the impact of waterlogging, significantly enhances the ability to identify waterlogging risks, warn and respond to emergencies, and improves the overall level and management efficiency of urban flood control and disaster reduction.

[0037] Step S5: constructing a waterlogging prediction model by combining the rainwater infiltration capacity model, the evaporation model, the behavioral feedback factor, the meteorological data and the drainage network data; Step S5 specifically includes: Get each space area i In time t The rainfall intensity , instantaneous infiltration rate Potential evaporation rate , drainage capacity , estimate each spatial region i The concentration index of surface water per unit area : in, Represents a spatial region iSurface water volume aggregation index per unit area is the time interval, i.e., the time length from the previous time step to the current time step , respectively represent the rainfall intensity, instantaneous infiltration rate and potential evaporation rate of the spatial region i ; is the drainage capacity of the spatial region i and is obtained according to the drainage pipe network data of each spatial region i ; is the rainfall intensity of the spatial region i and is obtained according to meteorological data represents the area of the spatial region i ; Drainage capacity is calculated by the Manning formula according to the drainage pipe network data of each spatial region. The formula is: where is the number of drainage pipes in the spatial region i ; represents the hydraulic radius of the j th drainage pipe ; j represents the hydraulic gradient of the th drainage pipe j ; represents the cross-sectional area of the j th drainage pipe

[0038] Using the characteristic information of each spatial region at each time step as input, training samples are constructed. The characteristic information includes the estimated surface water volume aggregation index, behavior feedback factor and meteorological data, where the meteorological data includes the current rainfall intensity and the cumulative rainfall in the previous period Whether waterlogging occurs in each spatial region is used as the supervised learning label, and the label is determined according to the historical waterlogging event data, which is obtained according to remote sensing images and water level sensor monitoring data Construct an waterlogging prediction model based on a classification model to predict the probability of waterlogging occurrence. The classification model is a probability model, including a gradient boosting decision tree model The waterlogging prediction model takes the characteristic information of each spatial region as input and outputs the predicted value of the waterlogging occurrence probability at the corresponding moment. The waterlogging prediction model is trained through a loss function according to the output waterlogging probability prediction value to obtain the trained waterlogging prediction model

[0039] The loss function is: where is the loss function, is the weight coefficient, is the adjustment parameter, is the spatial region i 's behavior feedback factor, represents the spatial region i at time t 's real label of waterlogging. 1 indicates waterlogging occurs, 0 indicates no waterlogging occurs. The label is determined based on historical water accumulation event data; represents the i -th spatial region output by the waterlogging prediction model at time t 's predicted value of the probability of waterlogging occurrence; represents the total number of spatial regions, represents the number of time steps in the training samples.

[0040] The present invention calculates the surface water volume aggregation index, mainly to accurately reflect the dynamic change of surface water in each spatial region at a specific time, so as to provide a scientific basis for waterlogging risk prediction. The occurrence of urban waterlogging is essentially the result that surface water exceeds the carrying capacity of the drainage system and cannot be drained in time, resulting in waterlogging flooding the ground. Therefore, the accumulation process of surface water volume is directly related to the formation mechanism of waterlogging and is the core physical quantity for judging waterlogging risk. Specifically, the surface water volume aggregation index quantitatively expresses the net increase in water volume per unit area of the surface per unit time by comprehensively considering four key factors: rainfall input (rainfall intensity), rainwater infiltration consumption (instantaneous infiltration rate), evaporation loss (potential evaporation rate), and drainage system drainage capacity (drainage capacity). This index reflects the inflow and outflow process of rainfall water into the urban surface and can dynamically capture the surplus or deficit state of water volume. During the urban rainstorm process, if the surface water volume aggregation index of a certain spatial region is positive and the value is large for a long time, it indicates that the rainfall exceeds the sum of the infiltration capacity, evaporation loss, and drainage capacity of this region, and water begins to accumulate on the surface, which is exactly the precursor of waterlogging occurrence. By real-time monitoring and calculating this index, it is possible to detect the regions and time points where water volume abnormally accumulates earlier, and provide a quantitative early warning signal for waterlogging risk from the physical mechanism.

[0041] However, the accumulation process of surface water volume is indeed the direct physical mechanism for the formation of waterlogging. However, there are obvious limitations in directly judging whether waterlogging occurs based solely on the surface water volume aggregation index. Because the occurrence of urban waterlogging depends not only on the amount of surface water volume, but also on various complex factors such as the state of the drainage system, topography, sewer blockage, soil bearing capacity, and human activities. A single water volume index cannot comprehensively reflect the interaction of these complex factors and is difficult to accurately distinguish the waterlogging risks under different conditions. The surface water volume aggregation index itself reflects the net accumulation of water volume per unit time, but the specific manifestation and severity of waterlogging also depend on dynamic processes such as water flow convergence, drainage paths, surface structures, and temporary water flow blockages. Simple threshold judgments often ignore these non-linear and spatio-temporal complexities and are prone to misjudgments or missed judgments.

[0042] Adopting a probabilistic prediction model can quantify the waterlogging risk, provide prediction results of different risk levels, and facilitate urban managers to formulate differentiated emergency plans and preventive measures according to actual needs, rather than a simple binary judgment of "occurred" or "not occurred", which improves the scientificity and practical value of waterlogging prevention and control.

[0043] The behavior feedback factor reflects the significant difference between the spatial distribution and movement patterns of people in a certain area and the historical normal state. When waterlogging occurs, the surface water accumulation hinders the normal traffic and walking paths, resulting in people being forced to change their established movement routes, and even congestion, detours, or concentrated shelters occur. This abnormal movement pattern is not only a direct reflection of the occurrence of waterlogging but also a specific manifestation of the impact of waterlogging on the use of urban space. Therefore, by analyzing the abnormal movement changes in the crowd trajectory data, the spatial environmental changes and traffic obstacles caused by waterlogging can be indirectly identified. In other words, the behavior feedback factor reveals the abnormal flow of people caused by waterlogging. This abnormal movement pattern, as the "social perception signal" of the occurrence of waterlogging, supplements the waterlogging prediction based solely on hydrometeorological and drainage system parameters, enabling the model to capture the waterlogging risk more comprehensively and accurately, especially in rapidly changing and complex urban environments. In the present invention, by adding the behavior feedback factor to the weight coefficient of the loss function, since the probability of waterlogging events is usually low, resulting in a highly unbalanced distribution of positive and negative samples, it is easy for the model to be biased towards predicting the "no waterlogging" category, thus ignoring potential high-risk areas. By integrating the behavior feedback factor into the weight, the sample loss weights corresponding to areas with a higher probability of waterlogging occurrence or abnormal movement of the crowd are amplified, and the model pays more attention to the prediction accuracy of these key spatio-temporal points during the training process. The adjustment parameter α in the weight coefficient can flexibly control the influence intensity of the behavior feedback factor on the loss function, realizing the sensitive response of the model to the risk areas with abnormal guidance of crowd behavior.

[0044] Step S6: Output the waterlogging occurrence probability of each spatial area in the future time period according to the waterlogging prediction model; Step S6 specifically includes: Input the current and historical characteristic information of each spatial region, including rainfall intensity, rainwater infiltration capacity, potential evaporation rate, drainage capacity, behavior feedback factor, and relevant meteorological data. The model outputs the corresponding probability of waterlogging occurrence for each spatial region within a preset time period in the future. This probability reflects the likelihood of waterlogging occurring in this region due to the superposition of multiple factors such as rainfall, hydrological conditions, drainage status, and population behavior within a specific time window.

[0045] By analyzing the probability values of consecutive time steps, dynamic monitoring and early warning of waterlogging risks can be achieved, supporting decision-makers to take targeted measures in urban management and disaster prevention and mitigation in a timely manner, and effectively reducing the losses caused by waterlogging.

[0046] This embodiment also provides a waterlogging prediction system based on behavior heat maps and infiltration dynamic modeling, as Figure 2 shown, including the following main modules: Data acquisition module, used to obtain multi-source data of the target area in real time, including terrain data, soil and vegetation coverage information, surface temperature and environmental meteorological data, population trajectory data, rainfall data, and drainage pipe network information; Spatial division and parameter extraction module, divides several spatial regions according to terrain data, extracts the surface coverage type, soil properties, and vegetation conditions of each region, and dynamically adjusts the rainwater infiltration capacity model in combination with sensor data to construct an infiltration dynamic model reflecting the change of rainfall and time; Evaporation calculation module, based on the collected surface temperature and environmental data, uses the modified Penman-Monteith model to establish an evaporation model considering the urban heat island effect, and calculates the potential evaporation rate of each spatial region; Behavior heat map generation module, uses population trajectory data to generate a behavior heat map that changes with time through kernel density estimation technology, and then extracts the behavior feedback factor to reflect the abnormal movement of the population; Waterlogging prediction model construction module, fuses the infiltration model, evaporation model, behavior feedback factor, meteorological data, and drainage pipe network information, and trains the waterlogging prediction model based on machine learning methods to realize the dynamic prediction of the probability of waterlogging occurrence in each spatial region; Prediction result output module, according to the output of the waterlogging prediction model, provides the waterlogging risk probability of each spatial region in the future time period, supporting urban managers to carry out early warning and emergency response.

[0047] Through multi-source data fusion and dynamic modeling, this system effectively improves the accuracy and timeliness of waterlogging prediction. Especially in a complex urban environment, it can comprehensively reflect the hydrological process and population behavior changes, and improve the waterlogging risk identification and early warning ability.

[0048] If the above functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or a part of this technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to enable a computer device (which may be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in various embodiments of the present invention. The foregoing storage medium includes: various media that can store program codes, such as USB flash drives, mobile hard disks, read-only memories (ROM, Read-Only Memory), random access memories (RAM, Random Access Memory), magnetic disks, or optical discs.

[0049] The above is only the specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art within the technical scope disclosed by the present invention can easily think of various equivalent modifications or substitutions, and these modifications or substitutions should all be covered within the protection scope of the present invention. Therefore, the protection scope of the present invention shall be subject to the protection scope of the claims.

Claims

1. An urban waterlogging prediction method based on behavior heat maps and infiltration dynamic modeling, characterized in that, It includes the following steps: Collect multi-source data of the target area, including environmental data, meteorological data, terrain data, drainage network data, population trajectory data, and surface temperature data; Divide the spatial area according to the terrain data, extract the surface cover type, soil properties, and vegetation cover of each spatial area, and construct a rainwater infiltration capacity model that dynamically adjusts with time and rainfall conditions; Based on the surface temperature data and environmental data, establish an evaporation model reflecting the urban heat island effect, and calculate the evaporation capacity of each spatial area; Use the population trajectory data to generate a behavior heat map that changes with time, and regard the degree of personnel aggregation and abnormal behavior changes in the behavior heat map as behavior feedback factors; Combine the rainwater infiltration capacity model, evaporation model, behavior feedback factors, meteorological data, and drainage network data to construct an urban waterlogging prediction model; According to the urban waterlogging prediction model, output the urban waterlogging occurrence probability of each spatial area in the future time period.

2. The method for predicting waterlogging based on behavior heat map and infiltration dynamic modeling according to claim 1, characterized in that, The construction of the rainwater infiltration capacity model that dynamically adjusts with time and rainfall conditions specifically includes: According to the terrain data, divide the target area into several spatial areas; For each spatial region, based on the terrain data, extract the main surface types within the spatial region, including bare soil, grassland, concrete, asphalt, and masonry paving, and calculate the area proportion of each surface type within the spatial region ; For each type of surface j , obtain the infiltration-related parameters corresponding to it, including: saturated hydraulic conductivity , capillary suction , saturated volumetric water content , volumetric water content in the previous time period ; among them, the saturated hydraulic conductivity refers to the amount of water that can infiltrate per unit time under fully saturated conditions for the surface type j , the capillary suction refers to the magnitude of the suction force when water is adsorbed in the soil capillary pores of the surface type j , the saturated volumetric water content refers to the maximum amount of water that can be contained in a unit volume of soil under fully saturated conditions for the surface type j , and the volumetric water content in the previous time period refers to the actual water content of the soil for the surface type j at the previous moment; the saturated hydraulic conductivity , capillary suction , saturated volumetric water content are obtained through geological data and soil databases, and the volumetric water content in the previous time period is obtained through the measured data of soil moisture sensors; An improved algorithm based on the Green-Ampt model is adopted. According to the infiltration-related parameters of each spatial region, a rainwater infiltration capacity model with dynamic evolution of the infiltration capacity varying with time in each spatial region is constructed, and the instantaneous infiltration rate at different times is calculated. t The instantaneous infiltration rate is calculated as follows: Among them, represents the i instantaneous infiltration rate of the t th spatial region at time i which means the rate at which rainwater is absorbed by the surface of this region per unit area of the surface at time t . represents the cumulative infiltration depth of the t th type of surface in the i spatial region from the initial time to time j , which is obtained by integrating the instantaneous infiltration rate of the j th type of surface and whether there is precipitation over time, being the total number of surface types.

3. The method for predicting waterlogging based on a behavior heat map and infiltration dynamic modeling according to claim 2, wherein The step of dividing the target area into several spatial areas according to the terrain data specifically includes: According to the digital elevation model in the terrain data of the target area, calculate the slope and aspect of each grid cell in the digital elevation model; Based on the slope and aspect, combined with the water flow path analysis, calculate the catchment area of each grid cell; By setting the basin area threshold and slope threshold, screen the spatial areas that meet the conditions: Among them, is a set of spatial regions obtained by screening and dividing the result terrain data, represents the catchment area of the raster cell , is the threshold of the catchment area, represents the slope of the raster cell , represents the slope threshold.

4. The method for predicting waterlogging based on behavior heat map and infiltration dynamic modeling according to claim 1, characterized in that, The establishment of an evaporation model reflecting the urban heat island effect based on the surface temperature data and environmental data specifically includes: Collect the surface temperature data and environmental data of each spatial area of the target area. The environmental data includes surface temperature, air temperature, wind speed, relative humidity, and solar radiation; According to the surface temperature data and environmental data, a modified Penman-Monteith model is used to construct an evaporation model considering the urban heat island effect, and the potential evaporation rate of each spatial region in the future time period is calculated. , and the formula is as follows: Among them, represents the potential evaporation rate of the i th spatial region at time t ; represents the slope of the saturation vapor pressure curve of the i th spatial region, calculated based on air temperature; represents the net radiation of the i th spatial region, estimated based on solar radiation and surface temperature; represents the soil heat flux of the i th spatial region; is the humidity constant; represents the average air temperature of the i th spatial region; represents the wind speed at a height of 2 meters above the ground in the i th spatial region, , respectively represent the saturation vapor pressure and the actual vapor pressure of the i th spatial region.

5. A method for predicting waterlogging based on a behavior heat map and infiltration dynamic modeling according to claim 1, characterized in that, The use of population trajectory data to generate a behavior heat map that changes with time specifically includes: Collect the trajectory data of the people in the target area , where the trajectory data is obtained by mobile signals, WiFi detection or video monitoring, and is used to represent the spatial position of an individual at a time t ; Map the personnel trajectory data to multiple spatial regions divided in the target area, and count the number of people in each spatial region at any given time t within ; According to the number of people in each spatial region at any given time t and the area of each spatial region calculate the population density of the nth i spatial region at time t : ​ Use the Gaussian kernel function for the crowd density to perform kernel density estimation and generate the t behavior heat map at a certain moment: Among them, represents the spatial position at the moment t of the crowd behavior heat map value, represents the centroid coordinates of the i th spatial region, represents the total number of spatial regions, is the kernel function, is the bandwidth parameter.

6. The method for predicting waterlogging based on behavior heat map and infiltration dynamic modeling according to claim 1 or 5, characterized in that The calculation process of the behavior feedback factor is: Calculate the average density value of the historical behavior heat map for the same period as the baseline reference; , as the baseline reference; Based on the average density values of the current moment behavior heat map and the historical same period behavior heat map , calculate the behavior anomaly factor: Among them, represents the spatial position at the moment t of the behavior anomaly factor; Set a threshold When the behavior anomaly factor is present, extract the crowd gathering intensity as the risk weighting factor: Integrate and average the risk-weighted factor by spatial region to obtain the behavior feedback factor for the i th spatial region: Among them, represents the i th spatial region's behavior feedback factor at time t .

7. A method for predicting waterlogging based on a behavior heat map and infiltration dynamic modeling according to claim 6, characterized in that The historical same period refers to the historical time period corresponding to the current time point in the time cycle.

8. The method for predicting waterlogging based on behavior heat map and infiltration dynamic modeling according to claim 1, wherein The combination of the rainwater infiltration capacity model, evaporation model, behavior feedback factors, meteorological data, and drainage network data to construct an urban waterlogging prediction model specifically includes: Obtain each spatial region i At time t Rainfall intensity 、Instantaneous infiltration rate 、Potential evaporation rate 、Drainage capacity Estimate each spatial region i Surface water volume aggregation index per unit area : Among them, represents the surface water volume aggregation index per unit area of the spatial region i , is the time interval, that is, the time length from the previous time step to the current time step, 、 respectively represent the rainfall intensity, instantaneous infiltration rate and potential evaporation rate of the spatial region i , is the drainage capacity of the spatial region i , obtained from the drainage network data of each spatial region i ; is the rainfall intensity of the spatial region i , obtained from meteorological data; represents the area of the spatial region i . Use the characteristic information of each spatial area at each time step as input to construct a training sample. The characteristic information includes the estimated surface water volume aggregation index, behavior feedback factors, and meteorological data, where the meteorological data includes the current rainfall intensity and the cumulative rainfall in the previous period; Regard whether waterlogging occurs in each spatial area as a supervised learning label. The label is determined based on historical waterlogging event data, and the historical waterlogging event data is obtained according to remote sensing images and water level sensor monitoring data; Construct an urban waterlogging prediction model based on a classification model to predict the urban waterlogging occurrence probability. The classification model is a probability model, including a gradient boosting decision tree model; The urban waterlogging prediction model takes the characteristic information of each spatial area as input, outputs the urban waterlogging occurrence probability prediction value corresponding to the moment, and trains the urban waterlogging prediction model through a loss function according to the output urban waterlogging probability prediction value to obtain the trained urban waterlogging prediction model.

9. A method for predicting waterlogging based on behavioral heatmaps and infiltration dynamic modeling according to claim 8, characterized in that The drainage capacity Calculated according to the drainage pipe network data of each spatial area through the Manning formula, the formula is: Among them, is the number of drainage pipes in the spatial region i . represents the hydraulic radius of the j th drainage pipe, represents the hydraulic gradient of the j th drainage pipe, represents the cross-sectional area of flow of the j th drainage pipe, represents the Manning roughness coefficient of the j th drainage pipe.

10. A method for predicting waterlogging based on behavior heatmaps and infiltration dynamic modeling according to claim 8, characterized in that, The loss function is: Among them, is the loss function, is the weight coefficient, is the adjustment parameter, is the behavior feedback factor of the spatial region i ; represents the actual waterlogging label of the spatial region i at time t . 1 indicates the occurrence of waterlogging, and 0 indicates the non-occurrence of waterlogging. The label is determined based on historical waterlogging event data; represents the predicted value of the probability of waterlogging occurrence in the i -th spatial region output by the waterlogging prediction model at time t ; represents the total number of spatial regions, represents the number of time steps in the training sample.

Images