A rainstorm and flood real-time monitoring system based on Internet of Things

Abstract

The application discloses a kind of rainstorm flood real-time monitoring system based on Internet of Things, it is related to rainstorm flood monitoring technical field;The application includes: topological construction module, rainstorm monitoring module and decision warning module;Topological construction module is based on city drainage pipe network data and constructs weighted drainage topological graph, and through multi-parameter multiplicative model and combines historical waterlogging event to carry out dynamic correction to drainage capacity;Rainstorm monitoring module fuses rainfall station and weather radar data, generates high spatio-temporal resolution rainfall intensity field, and constructs rainfall risk index reflecting instantaneous peak value and cumulative effect simultaneously;Decision warning module couples dynamic rainfall threat and drainage pipe network vulnerability and carries out coupling analysis, generates comprehensive risk index, and triggers flood simulation and graded early warning according to spatial continuity.This application can realize the early perception, accurate evaluation and reliable early warning of rainstorm flood risk, and is suitable for urban waterlogging prevention and control and emergency decision-making.

Description

Technical Field

[0001] This invention relates to the field of rainstorm and flood monitoring technology, and more specifically, to a real-time rainstorm and flood monitoring system based on the Internet of Things. Background Technology

[0002] Heavy rain and the resulting floods are characterized by their suddenness, severe destructiveness, and widespread impact. Short-duration heavy rainfall can cause urban flooding, road waterlogging, and backflow into drainage systems, leading to severe damage such as flooded houses and disruptions to critical infrastructure (power, communication, water supply, and transportation). River floods can cause dike breaches, farmland inundation, mudslides, and bridge collapses. Furthermore, floods often bring secondary disasters, increasing the difficulty of rescue efforts and the cost of recovery. Therefore, timely and accurate monitoring of heavy rain and floods is a prerequisite for early warning systems, emergency response organization, and loss reduction.

[0003] Current rainstorm and flood monitoring relies heavily on water level thresholds for triggering, which is a simple and passive response. It lacks prediction of short-term rainfall evolution, spatial distribution information, and multi-source data fusion, which often leads to delayed warnings, false alarms, and missed reports. It cannot provide graded and interpretable disposal suggestions and fails to meet the needs of rapid and refined rainstorm and flood emergency decision-making. Summary of the Invention

[0004] The main objective of this invention is to provide a real-time monitoring system for rainstorms and floods based on the Internet of Things (IoT) to overcome the problems mentioned in the background section.

[0005] To achieve the above objectives, an IoT-based real-time monitoring system for rainstorms and floods is provided. The system includes: a topology construction module, a rainstorm monitoring module, and a decision-making and early warning module.

[0006] Among them, the topology construction module is used to construct the weighted drainage network topology graph and calculate the prior edge weights of the drainage network;

[0007] The rainstorm monitoring module is used to acquire and fuse multi-source rainfall data to generate rainfall risk indicators;

[0008] The decision-making and early warning module is used to integrate rainfall risk indicators with the prior boundary weights of drainage pipe networks to generate a comprehensive risk index and trigger flood warnings.

[0009] Furthermore, a weighted drainage network topology graph is constructed and the prior edge weights of the drainage network are calculated, including:

[0010] S2-1: Obtain complete geographic information system data and engineering as-built drawings of the drainage network. Entities such as manholes, inspection wells, node confluences, storm drains, and pump station inlets / outlets in the urban drainage network are designated as nodes. Drainage pipes or channels connecting two adjacent nodes are abstracted as directed edges in the topology graph. The direction of the edge is defined as the dominant design direction of water flow. Each edge is assigned a unique ID and associated with its starting and ending nodes. Each node and edge stores its corresponding attribute information. Specific node attributes include node ID, geographic location, and type; edge attributes include edge ID, starting node, ending node, length, pipe diameter, pipe material, slope, and roughness coefficient. Traverse each edge and establish connections in the node list based on the starting and ending node IDs, forming a directed graph composed of a set of nodes and a set of edges, representing the direction of water flow.

[0011] S2-1, For each edge, under the condition of no abnormality, calculate the inherent drainage capacity of each edge under the design capacity during drainage, and form the static drainage edge weight under the initial ideal state;

[0012] S2-3: Obtain records of urban flooding events over the past N years, and analyze the frequency and severity of these events to obtain frequency and severity coefficients. Assign a weighting factor to each of the frequency and severity coefficients and perform linear weighted fusion to obtain the dynamic water accumulation coefficient.

[0013] S2-4 uses the dynamic water accumulation coefficient as a reduction factor to apply to the static drainage edge weight to obtain the dynamic prior edge weight. The drainage edge weight and prior edge weight of each edge are marked on the corresponding edge in the drainage topology graph and sent to the decision warning module.

[0014] Furthermore, the calculation of the drainage boundary weight includes:

[0015] Extract the pipe diameter, slope, roughness coefficient of the inner wall of the pipe, pipe length, and pipe service life, and normalize them to obtain the water flow fraction d. A Slope fraction d S Roughness fraction d b Length fraction d L and duration score d T The fractions are labeled using an index, denoted as di, where i represents the index of the fraction type, i∈{A,S,b,L,T}; a multiplicative power model is applied to each type of fraction. The drainage edge weight q is calculated; where η i Let f(x) be the weights of the scores of each type of edge, and let the sum of the weights be one.

[0016] Furthermore, the frequency coefficient calculation includes:

[0017] Historical flooding events are extracted, including: flooding time, water accumulation location, water depth, and water accumulation duration. Using a spatial geographic matching algorithm, the water accumulation location of each historical flooding event is matched to the nearest edge on the drainage topology map, thereby obtaining the historical flooding events for each edge.

[0018] Extract historical flooding events from the edges and divide them into the corresponding months of the year according to the time of flooding occurrence. This gives the monthly distribution of historical flooding events for each edge. Count the number of historical flooding events in each month and divide it by the total number of historical flooding events in the entire drainage topology graph to obtain the frequency coefficient of the edge in each month. Compare the monthly distribution of historical flooding events for each edge in the current month to obtain the frequency coefficient of each edge or node in the current month.

[0019] Furthermore, the severity coefficient calculation includes:

[0020] The average water depth and average water duration are calculated by averaging the historical waterlogging events along the edges. The maximum water depth and maximum waterlogging duration are selected from the entire drainage topology map, and the average water depth and average waterlogging duration are normalized by dividing them by the maximum water depth and maximum waterlogging duration, respectively. The normalized average water depth and average waterlogging duration are then treated as a two-dimensional point. The two-dimensional distance between this point and the origin (0,0) is calculated using the Euclidean distance formula. A nonlinear mapping formula is then used to quantify the effect of waterlogging time and depth, yielding a severity coefficient. The mapping formula is: , where H represents the two-dimensional distance, k represents the shape adjustment parameter, and the mapping formula is a monotonically increasing convex function shape.

[0021] Further, rainfall risk indicators include:

[0022] S6-1 acquires in real-time minute-level rainfall intensity sequences from ground rain gauges and the gridded reflectance field E(x, y, z, t) from a meteorological radar, where t is the time index, x and y represent latitude and longitude, y represents altitude, and E(x, y, z, t) represents the measured atmospheric reflectance intensity at each three-dimensional grid point indexed by (x, y, z, t) in space at time t. Using data from all available rain gauges at the same time, a functional relationship is established with the corresponding grid point reflectance of the radar for local dynamic calibration. Regression analysis is performed using a dynamic time window to obtain the calibrated radar-estimated rainfall rate field R(x, y, t), in millimeters per hour. For the window... For each rain gauge station with valid rainfall records within the area, and at each time point, the grid point in the radar data that is closest to the rain gauge station horizontally and has the lowest elevation angle vertically is found. Its reflectivity value E(t) is extracted. Using the matched data, regression analysis is performed on {E(t), r(t)} within the time window to generate a spatially continuous calibrated radar estimated rainfall rate field R(x, y), with units of millimeters per hour. Thus, the radar estimated rainfall field R(x, y, t) at each time t can be obtained. By using the optimal interpolation method to take the rainfall intensity sequence at the rain gauge station as the true value anchor point, the radar field is spatially optimally adjusted to obtain the instantaneous rainfall intensity field I(x, y, t), with units of millimeters per hour.

[0023] S6-2 divides the urban area into dynamic units Uj that match the topology of the drainage network, administrative grid, or natural watershed, where j represents the index of the unit; each unit contains all nodes and edges of the drainage network within its boundary; for each unit Uj, the instantaneous rainfall intensity field I(x, y, t) within the time window is obtained, and the instantaneous rainfall intensity corresponding to the grid point in the instantaneous rainfall intensity field I(x, y, t) that falls into unit Uj is taken as the instantaneous rainfall intensity data of the unit, and the maximum instantaneous rainfall intensity is selected from it;

[0024] S6-3: The arithmetic mean of the instantaneous rainfall intensity data within the unit is used to calculate the average rainfall intensity at each time t within the time window, and this is multiplied by the duration of the current time window to obtain the rainfall amount. The rainfall amounts of all time windows from the start of rainfall to the current time window are extracted. The difference between the midpoint of each time window and the midpoint of the current time window is calculated to obtain the interval time. A backtracking time threshold is preset, and time windows with interval times less than the backtracking time threshold are selected as the relevant windows of the current time window. The attenuation amount of all relevant windows is calculated using time decay, and then summed with the rainfall amount of the current time window to obtain the rainfall index. The rainfall risk indicators include the maximum instantaneous rainfall intensity and the rainfall index.

[0025] Furthermore, generating a comprehensive risk index and triggering flood warnings includes:

[0026] The rainfall index and the maximum instantaneous rainfall intensity are normalized, and then the normalized rainfall index and the maximum instantaneous rainfall intensity are used to obtain the rainfall threat index through a weighted geometric mean model with enhanced short-board effect. Thus, the rainfall threat index of each unit Uj can be obtained.

[0027] The inherent risk of poor drainage in each unit is quantified based on the weighted drainage network topology diagram, and the vulnerability index of the unit's pipe network is combined with the vulnerability of the pipe network to generate the final comprehensive risk index. From this, the comprehensive risk index Fj of all units Uj can be obtained, and a global judgment is made based on this, triggering a flood warning; the flood warning includes a red warning for waterlogging and a yellow warning for waterlogging.

[0028] Furthermore, the rainfall index and maximum instantaneous rainfall intensity are normalized, and the weighted geometric mean model that enhances the weakest link effect is as follows:

[0029] (1) The impact threshold and extreme threshold are preset to form a rainfall range. The impact threshold and extreme threshold are summed and divided by two to obtain the inflection point threshold. When the maximum instantaneous rainfall intensity is greater than the extreme threshold, the rainfall index is directly normalized to a value of one. If the maximum instantaneous rainfall intensity is within the rainfall range formed by the impact threshold and extreme threshold, then the inflection point threshold is obtained by formula. The maximum instantaneous rainfall intensity is normalized, where Imax is the extreme threshold, Imid is the inflection point threshold, and s is the steepness parameter; if the maximum instantaneous rainfall intensity is less than the influence threshold, the rainfall index is directly normalized to zero.

[0030] (2) The rainfall index is normalized using a power function with saturation characteristics: ,in Rnorm is a shape parameter that controls the rate at which the curve approaches 1; Rnorm is the normalized baseline value.

[0031] Weighted geometric mean model that amplifies the weakest link effect: Where θ1 and θ2 are dynamic weighting coefficients, which are not fixed values ​​and are calculated according to the formula. Y Imax Y represents the normalized maximum instantaneous rainfall intensity. R The normalized rainfall index.

[0032] Furthermore, the global assessment and triggering of flood warnings includes:

[0033] A flood simulation threshold is preset. The comprehensive risk index of each unit is compared with the flood simulation threshold. For each unit whose comprehensive risk index is greater than the flood simulation threshold, the region growth algorithm is started to find spatially adjacent or pipeline-connected units that meet the conditions. If the number of units in any grown region is ≥3, the grown region is marked as a waterlogging danger area and a red waterlogging warning is triggered. If the number of units in any grown region is <3, the grown region is marked as a local waterlogging point and a yellow waterlogging warning is triggered.

[0034] Furthermore, when a red or yellow alert for waterlogging is triggered, each edge within the corresponding area accumulates one waterlogging event and updates the topology building module.

[0035] The beneficial effects of this invention are:

[0036] (1) By abstracting the urban drainage network into a weighted directed topology graph and calculating the static drainage edge weights based on the multiplicative power model of pipe diameter, slope, roughness coefficient, length and service life, the engineering bottleneck effect of drainage capacity is effectively characterized, avoiding the problem of distortion of key constraints by traditional linear weighting; further, the edge weights are dynamically reduced by combining the frequency of historical waterlogging events and the severity of water accumulation, so that the drainage capacity assessment is transformed into operational credibility, realizing the structured and quantifiable expression of the vulnerability of the drainage network, and providing a priori basis with physical meaning and historical constraints for subsequent risk assessment;

[0037] (2) By integrating the high-precision point measurement data of rain gauges with the spatial continuity advantage of meteorological radar, a high spatiotemporal resolution instantaneous rainfall intensity field is constructed to enhance the ability to characterize the spatial heterogeneity of short-term heavy rainfall; at the same time, through time window and time decay mechanism, a rainfall index that takes into account both instantaneous peak impact and previous cumulative effect is constructed to effectively avoid the problem that simple threshold triggering cannot reflect the rainfall evolution process, so that rainfall risk assessment is in line with process perception and trend judgment, and provides a reliable basis for early warning and graded response;

[0038] (3) By integrating the normalized maximum instantaneous rainfall intensity and the rainfall index with a weighted geometric mean model enhanced by the short-board effect, it is ensured that a high rainfall threat is formed only when the peak intensity and cumulative intensity are simultaneously satisfied, thus effectively suppressing false alarms; furthermore, the dynamic rainfall threat is multiplicatively coupled with the vulnerability index of the unit drainage network to amplify the expression of the risk of rainfall impact and drainage capacity mismatch; combined with spatial continuity constraints and network connectivity, global scanning is used to trigger flood simulation, so that the early warning decision has spatial coverage, temporal stability and engineering interpretability, thereby improving the accuracy and practical value of urban waterlogging early warning. Attached Figure Description

[0039] The accompanying drawings, which form part of this invention, are used to provide a further understanding of the invention. The illustrative embodiments of the invention and their descriptions are used to explain the invention and do not constitute an undue limitation of the invention. In the drawings:

[0040] Figure 1 This is a schematic diagram of the system module connections of the present invention. Detailed Implementation

[0041] It should be noted that, unless otherwise specified, the embodiments and features described in the present invention can be combined with each other. The present invention will now be described in detail with reference to the accompanying drawings and embodiments.

[0042] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.

[0043] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate for the embodiments of the invention described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0044] To make the objectives and advantages of the present invention clearer, the present invention will be further described below with reference to embodiments; it should be understood that the specific embodiments described herein are merely for explaining the present invention and are not intended to limit the present invention.

[0045] Please see Figure 1 As shown, the present invention is a real-time monitoring system for rainstorms and floods based on the Internet of Things, comprising: a topology construction module, a rainstorm monitoring module, and an early warning decision module;

[0046] The topology construction module assigns static drainage edge weights based on the city's underground drainage network GIS data. Then, it reduces the weights of these static edge weights based on historical water accumulation data to obtain dynamic prior edge weights, and outputs a weighted drainage network topology map. It should be noted that the prior edge weights reflect a priori estimates of drainage smoothness; specifically, they include:

[0047] By acquiring complete geographic information system data and engineering as-built drawings of the drainage network through the Internet of Things (IoT), entities such as manholes, inspection wells, junction inlets, storm drains, and pump station inlets / outlets in the urban drainage network are designated as nodes. Drainage pipes or channels connecting two adjacent nodes are abstracted as directed edges in the topology graph. The direction of the edge is defined as the dominant design direction of water flow (from upstream to downstream). Each edge is assigned a unique ID and associated with its starting and ending nodes. Each node and edge stores its corresponding attribute information. Specific node attributes include node ID, geographical location (longitude, latitude, and elevation), and type (inspection well, junction, pump station, etc.). Edge attributes include edge ID, starting node, ending node, length, pipe diameter, pipe material, slope, and roughness coefficient. Each edge is traversed, and connections are established in the node list based on the starting and ending node IDs, forming a directed graph composed of a set of nodes and a set of edges, representing the direction of water flow.

[0048] For each edge, calculate the basic drainage generality index, reflecting the inherent drainage capacity of each edge under design capacity during drainage, assuming no anomalies, thus forming the initial ideal state of static drainage edge weight; the specific process is as follows:

[0049] The factors considered in this study include pipe diameter, slope, roughness coefficient of the pipe's inner wall, pipe length, and service life. Specifically, the pipe diameter determines the cross-sectional area of ​​the water passage, which is the primary factor determining flow capacity; a larger diameter indicates greater drainage potential. The slope determines the water flow velocity; a steeper slope results in higher flow velocity and smoother drainage. At the same slope and pipe diameter, increasing pipe length leads to increased head loss along the pipe, relatively decreasing drainage efficiency. Furthermore, the longer the pipe is used, the more easily deposits accumulate on the inner wall, weakening its drainage capacity.

[0050] The pipe diameter D is calculated using the cross-sectional area formula. The cross-sectional area A of the water passage is calculated. Because the effect of the cross-sectional area on the flow rate is power-law, and the diameter distribution may span orders of magnitude, a logarithmic transformation is performed on the cross-sectional area before normalization to obtain the water flow fraction. The specific calculation formula is as follows: Amax and Amin are the largest and smallest cross-sectional areas of water flow in the entire network, respectively.

[0051] The slope S is normalized using a formula to obtain the slope fraction. The specific formula is as follows: Where Smax and Smin are the maximum and minimum design slopes among all pipelines in the entire network, respectively;

[0052] The roughness coefficient b of the pipe inner wall is reverse normalized using a formula to obtain the roughness fraction. The specific formula is as follows: Where bmax and bmin are the largest and smallest internal wall roughness coefficients among all pipes in the entire network;

[0053] An increase in pipe length L leads to a rise in head loss along the pipe, resulting in a decrease in drainage efficiency. In this embodiment, the pipe length is logarithmically scaled and inversely normalized using a formula to reduce the long-tail effect, yielding a length fraction. The specific formula is as follows: Where Lmax and Lmin are the longest and shortest lengths of all pipes in the entire network, respectively;

[0054] The longer a pipe is used, the more easily it accumulates deposits and becomes damaged on its inner wall, leading to decreased flow. A time-based fraction is obtained by mapping this fraction using a linear formula. The specific formula is as follows: Tmax and Tmin represent the longest service life (i.e., the oldest pipeline) and the shortest service life (i.e., the newest pipeline) among all pipelines in the entire network, respectively.

[0055] The excess water content d A Slope fraction d S Roughness fraction d b Length fraction d L and duration score d T The fractions are labeled using an index, denoted as di, where i represents the index of the fraction type, i∈{A,S,b,L,T}; a multiplicative power model is applied to each fraction type. The drainage edge weight q is calculated; where η i The weights of each type of fraction on the edge are given, and the sum of the weights is one; the weights of each factor are set according to their physical dominance and sensitivity in drainage capacity, and the water flow fraction d. A Slope fraction d S Roughness fraction d b Length fraction d L and duration score d T The weights are set to 0.35, 0.30, 0.15, 0.10, and 0.10 respectively. The multiplicative power model is essentially a weighted geometric aggregation model, which has a clear advantage over linear weighted summation in drainage topology scenarios. Urban drainage capacity exhibits a significant bottleneck effect, meaning that any serious deficiency in any key factor will significantly reduce the actual drainage capacity of the entire pipe section. For example, even with a large pipe diameter, drainage will be severely limited if the slope is close to zero; a steep slope with severe old deposits will also restrict flow. This is the core advantage of the multiplicative model: when any indicator approaches 0, the overall weight decreases rapidly. If a linear summation model is used, even if one term is 0, other terms can still compensate. The multiplicative power model used in this embodiment has non-compensable characteristics, better reflects real engineering constraints, and is extremely important for quantifying the static drainage edge weights of each side in the drainage topology.

[0056] Obtain records of urban flooding events over the past N years (e.g., 5-10 years). Each flooding event includes: flooding occurrence time, location of water accumulation points (latitude and longitude), water depth (maximum depth during the flooding process), and duration of water accumulation. Using a spatial geographic matching algorithm, match the location of the water accumulation point for each historical flooding event to the nearest edge on the drainage topology map. This yields the historical flooding events for each edge. Note that an edge may correspond to zero, one, or more historical flooding events. For each edge, calculate its dynamic water accumulation coefficient based on the matched historical flooding events. Specifically, this includes extracting the historical flooding events for the edge and arranging them according to the flooding occurrence time. Historical flooding events are categorized into the corresponding months of the year (12 months). It should be noted that flooding events exhibit a clear seasonal attribute, therefore, categorizing them by month is a scientifically effective approach. This allows us to obtain the monthly distribution of historical flooding events for each edge. By counting the number of historical flooding events in each month and dividing it by the total number of historical flooding events in the entire drainage topology graph, we obtain the frequency coefficient of each edge in each month. The higher the frequency coefficient, the greater the probability that a node or edge will experience flooding in that month. By comparing the monthly distribution of historical flooding events for each edge in the current month, we can obtain the frequency coefficient of each edge or node in the current month.

[0057] The average water depth and average water duration are calculated by averaging the historical water depth and water duration of waterlogging events. The maximum water depth and maximum water duration are selected in the entire drainage topology map, and the average water depth and average water duration are normalized by dividing them by the maximum water depth and maximum water duration, respectively. The normalized average water depth and average water duration are then used as a two-dimensional point. The two-dimensional distance between this point and the origin (0,0) is calculated using the Euclidean distance formula. It should be noted that the two-dimensional distance comprehensively reflects the depth and duration of waterlogging; the larger the two-dimensional distance, the farther the state of the two-dimensional point is from the origin (normal state), and the more severe the waterlogging event. To compress and map the two-dimensional distance to the range [0,1], this embodiment uses a nonlinear mapping formula to quantify the effect of waterlogging time and depth, obtaining a severity coefficient. The mapping formula is: Where H represents the two-dimensional distance, k represents the shape adjustment parameter, and the mapping formula is a monotonically increasing convex function shape. In this example, k is taken as 1.5; when H=0, it means there is no water accumulation, and the severity coefficient is zero; when H= When the time indicates the most severe flooding in history, the severity coefficient is closer to 1.

[0058] A weighting factor is assigned to both the frequency coefficient and the severity coefficient. In this embodiment, the weighting factors for the frequency coefficient and the severity coefficient are set to 0.4 and 0.6, respectively. A dynamic water accumulation coefficient is obtained by linearly weighting and fusing the water accumulation frequency and severity coefficients with their corresponding weighting factors. This dynamic water accumulation coefficient is then used as a reduction factor on the static drainage edge weight q to obtain the dynamic prior edge weight W. The specific reduction model is as follows: Where C represents the dynamic water accumulation coefficient, and γ is the historical performance reduction intensity coefficient, 0<γ≤1, which is a coefficient that controls the correction of the theoretical value by the historical data. When γ=1, it means that the historical performance is fully accepted and full reduction is performed; when γ<1, it means that the reduction intensity is weakened and a balance is achieved between the historical performance and the theoretical design. In this embodiment, the value of γ is taken as 0.7. In the drainage topology diagram, the drainage edge weight and prior edge weight of each edge are marked on the corresponding edge and sent to the decision warning module.

[0059] By abstracting urban drainage networks into weighted directed topological graphs and calculating static drainage edge weights based on a multiplicative power model of pipe diameter, slope, roughness coefficient, length, and service life, the engineering bottleneck effect of drainage capacity is effectively characterized, avoiding the distortion of key constraints by traditional linear weighting. Furthermore, the edge weights are dynamically reduced by combining the frequency of historical flooding events and the severity of water accumulation, transforming drainage capacity assessment into operational credibility. This achieves a structured and quantifiable expression of the vulnerability of drainage networks, providing a priori basis with physical meaning and historical constraints for subsequent risk assessment.

[0060] The rainstorm monitoring module continuously monitors the current rainfall situation, generates rainfall risk indicators for triggering flood simulation, including rainfall index and maximum instantaneous rainfall intensity, and sends them to the decision-making and early warning module. The specific process is as follows:

[0061] Real-time acquisition of minute-level rainfall intensity sequences from ground rain gauges and a gridded reflectance field E(x, y, z, t) from a meteorological radar is performed, where t is the time index, x and y represent latitude and longitude, y represents altitude, and E represents the reflectance in each specific grid (i.e., a three-dimensional pixel) defined by (x, y, z, t), in decibels. Reflectance indicates the quantity and size of cloud and rain particles in the air at that location, indirectly representing rainfall potential. Therefore, E(x, y, z, t) represents the measured atmospheric reflectance intensity at each three-dimensional grid point indexed by (x, y, z, t) in space at time t. Using data from all available rain gauges at the same time, a functional relationship is established with the radar's corresponding grid point reflectance for local dynamic calibration. Regression analysis is performed using a time window (set to 1 hour in this example) to obtain the calibrated radar estimated rainfall rate field R(x, y, t), in millimeters per hour. It should be noted that the functional relationship used in this embodiment is in power-law form. Where Z is the linear reflectance, and α and β are empirical parameters whose values ​​are set according to rainfall type, season, and regional variations. At the current moment, data from a past time window (set to 1 hour in this embodiment) are selected. For each rain gauge station and time with valid rainfall records within the window, the grid point in the radar data that is closest to the rain gauge station horizontally and has the lowest elevation angle (usually best representing ground rainfall) is found, and its reflectance value E(t) is extracted. Using the matched data, regression analysis is performed on {E(t), r(t)} within the time window. This embodiment uses a piecewise linear regression model based on physical constraints, and its functional relationship is as follows: E th The reflectivity threshold is used to distinguish between stratiform and convective rainfall. Parameters λ1 and μ1 are conversion parameters for stratiform rainfall (usually stable, wide-ranging, and moderate in intensity), while λ2 and μ2 are conversion parameters for convective rainfall (usually severe, highly localized, and accompanied by strong convection). The parameters obtained by dynamic fitting are applied to the lowest elevation angle reflectivity plane field E(x, y) of the entire radar scanning area. Through the above functional relationship, a spatially continuous calibrated radar estimated rainfall rate field R(x, y) is generated, with units of millimeters per hour. This process reduces the impact of radar system errors and drop spectrum variations. This yields the radar-estimated rainfall field R(x, y, t) at each time t, providing a good spatial structure of rainfall. However, the absolute accuracy at specific points may deviate from the measured values ​​at ground rain gauges. Therefore, by using the optimal interpolation method to take the rainfall intensity sequence at the rain gauge as the true value anchor point, the radar field is spatially optimally adjusted to obtain a high spatiotemporal resolution instantaneous rainfall intensity field I(x, y, t), in millimeters per hour. The instantaneous rainfall intensity field I(x, y, t) simultaneously possesses the high point accuracy of rain gauge data and the spatial continuity details of radar data, providing data support for rainstorm and flood monitoring.

[0062] The urban area is divided into dynamic units Uj that match the topology of the drainage network, administrative grid, or natural watershed, where j represents the unit index. Each unit contains all nodes and edges of the drainage network within its boundary. For each unit Uj, the instantaneous rainfall intensity field I(x, y, t) within the time window is obtained. The instantaneous rainfall intensity corresponding to the grid point in the instantaneous rainfall intensity field I(x, y, t) falling into unit Uj is taken as the instantaneous rainfall intensity data of the unit, and the maximum instantaneous rainfall intensity is selected from it. Then, the instantaneous rainfall intensity data within the unit is arithmetically averaged to obtain the average rainfall intensity Iavg of the unit at each time t within the time window, and multiplied by the duration of the current time window to obtain the rainfall. The process involves extracting rainfall data from all time windows from the start of rainfall to the current time window. The difference between the midpoint of each time window and the midpoint of the current time window is used to calculate the interval ΔT. A preset backtracking time threshold is used; in this example, it is set to 48 hours. By default, rainfall before 48 hours has a negligible impact on rainfall within the current time window. Time windows with intervals less than the backtracking time threshold are selected as relevant windows for the current time window. The decay of all relevant windows is calculated using time decay, and this decay is summed with the rainfall in the current time window to calculate the rainfall index. The rainfall index reflects the cumulative effect of rainfall. The time decay formula is: , where P represents the rainfall in the relevant window;

[0063] By integrating high-precision point measurement data from rain gauges with the spatial continuity advantage of meteorological radar, a high spatiotemporal resolution instantaneous rainfall intensity field is constructed, enhancing the ability to characterize the spatial heterogeneity of short-duration heavy rainfall. Simultaneously, through time windows and time decay mechanisms, a rainfall index that takes into account both the instantaneous peak impact and the previous cumulative effect is constructed, effectively avoiding the problem that simple threshold triggering cannot reflect the rainfall evolution process. This makes rainfall risk assessment more aligned with process perception and trend judgment, providing a reliable basis for early warning and graded response.

[0064] The decision-making and early warning module integrates rainfall risk indicators and prior boundary weights of the drainage network to achieve a comprehensive analysis of dynamic rainfall threats and static network vulnerability, generating a final comprehensive risk index and performing a global scan to determine whether to trigger flood simulation; the specific process includes:

[0065] The rainfall index and maximum instantaneous rainfall intensity are normalized, and then the normalized rainfall index and maximum instantaneous rainfall intensity are used to obtain the rainfall threat index through a weighted geometric mean model with enhanced short-board effect. This yields the rainfall threat index for each unit Uj. The specific normalization and weighted geometric mean model with enhanced short-board effect are as follows:

[0066] (3) The impact threshold and extreme threshold are set to form a rainfall range. The impact threshold and extreme threshold are summed and divided by two to obtain the inflection point threshold. The specific impact threshold is usually taken as the maximum 5-minute rainfall intensity that can be smoothly discharged in the local drainage system design standard (e.g., 5-10 mm / h). If it is lower than this value, it is considered that there is no impact threat. The extreme threshold is taken as the local historical extreme short-term rainstorm record or the upper limit of the drainage system design (e.g., the 5-minute rainfall intensity corresponding to a 50-year return period, such as 80-100 mm / h). If it reaches this value, it is considered an extreme impact. When the maximum instantaneous rainfall intensity is greater than the extreme threshold, the rainfall index is directly normalized to a value of one. If the maximum instantaneous rainfall intensity is within the rainfall range formed by the impact threshold and extreme threshold, it is then calculated using the formula. The maximum instantaneous rainfall intensity is normalized, where Imax is the extreme threshold, Imid is the inflection point threshold, and s is the steepness parameter, controlling the rate of rise of the curve from 0 to 1. This normalization reflects the nonlinear response of the drainage system to rainfall impact: below a certain intensity, there is no effect; above a certain intensity, the effect increases rapidly and approaches the limit (i.e., 1); if the maximum instantaneous rainfall intensity is less than the impact threshold, the rainfall index is directly normalized to a value of zero.

[0067] (4) The rainfall index is normalized using a power function with saturation characteristics: ,in The shape parameter controls the rate at which the curve approaches 1; in this example, it is set to 1.5. Rnorm is the normalized baseline value, representing the typical cumulative rainfall required to bring the soil close to saturation or the drainage network to its design load. Specifically, it is converted into the equivalent 48-hour cumulative rainfall based on the local maximum soil water holding capacity or the design daily load of the drainage network; in this example, 80-120 mm is taken as the baseline. This normalization process reflects the law of diminishing marginal effects. When the cumulative rainfall increases from 0, its impact on soil saturation and drainage network load increases rapidly. When the cumulative rainfall reaches a very high level, it is close to saturation, and the marginal contribution of the newly added cumulative rainfall becomes smaller.

[0068] (5) Weighted geometric mean model with enhanced short-board effect: Where θ1 and θ2 are dynamic weighting coefficients, which are not fixed values ​​and are calculated according to the formula. Y Imax Y represents the normalized maximum instantaneous rainfall intensity. R The normalized rainfall index; compared with the arithmetic mean, the geometric mean requires both sub-indices to be high in order to obtain a high composite index. If either sub-indice approaches 0 (such as no instantaneous rainstorm or no previous accumulation), the rainfall threat index Q will rapidly approach 0, which is consistent with the physical reality that the risk of urban flooding requires the combined effect of peak and cumulative rainfall.

[0069] Based on the weighted drainage network topology diagram, the inherent risk of poor drainage in each unit is quantified, and the vulnerability index V of the unit network is calculated using the following formula: Where Mj represents the total number of drainage pipe network edges contained within the boundary of unit Uj, W(e) refers to the prior edge weight of edge e, and e represents the set of drainage pipe network edges contained within the boundary of unit Uj; the vulnerability index is the complement of the average of the prior edge weights (representing unobstructed flow) of all pipes within unit Uj. Therefore, the value range of the vulnerability index V is between [0,1]. The larger the value, the higher the overall vulnerability of the pipe network in unit Uj, the worse the drainage capacity, and the higher the risk of the same rainfall falling on the vulnerable area. By combining dynamic rainfall threats with static pipeline vulnerability, a final comprehensive risk index is generated, where σ is the vulnerability amplification coefficient (ranging from 0.5 to 1.0; in this embodiment, it is set to 0.7). This formula is a multiplicative model, indicating that high rainfall threats and pipeline vulnerability mutually amplify risks. From this, the comprehensive risk index Fj for all units Uj can be obtained, and a global judgment is made based on this. A flood simulation threshold is preset. The specific flood simulation threshold is set by collecting various data before and during typical urban flooding events over the past few years, using the system model to calculate the theoretical Fj value for each affected unit in the event, statistically analyzing the distribution of Fj values, and taking its 90th percentile as the flood simulation threshold. This ensures that when the future Fj value reaches this level, it means a risk situation similar to historical disasters has occurred; in this embodiment, it is set to 0.6. The comprehensive risk index Fj of each unit is compared with the flood simulation threshold. For each comprehensive risk index Fj greater than the flood simulation threshold... For units meeting the threshold, the region growth algorithm is activated to find spatially adjacent or pipe-connected units that meet the conditions. If the number of units in any grown region is ≥3, the grown region is marked as a flood-prone area and a red flood warning is triggered. At the same time, all edges included in the flood-prone area accumulate one flood event and are updated to the topology building module. If the number of units in any grown region is <3, the grown region is marked as a local flood point and a yellow flood warning is triggered. At the same time, all edges included in the local flood point accumulate one flood event and are updated to the topology building module. It should be noted that the number of units in the grown region is one or more. Spatially adjacent means that the physical boundaries of two units are in contact (shared edge or shared node). Pipe-connected means that there is at least one drainage pipe (edge) with its starting node in one unit and its ending node in another unit. This can be quickly determined by querying the edge attributes (connection relationship) in the weighted drainage network topology graph.

[0070] By fusing the normalized maximum instantaneous rainfall intensity with the rainfall index using a weighted geometric mean model enhanced by the weakest link effect, a high rainfall threat is ensured only when both peak intensity and cumulative intensity are met, effectively suppressing false alarms. Furthermore, the dynamic rainfall threat is multiplicatively coupled with the vulnerability index of the unit drainage network to amplify the expression of the risk of rainfall impact and drainage capacity mismatch. Combining spatial continuity constraints and network connectivity to conduct global scanning to trigger flood simulation, the early warning decision has spatial coverage, temporal stability, and engineering interpretability, improving the accuracy and practical value of urban flooding early warning.

[0071] The above are merely embodiments of the present invention and are not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principle of the present invention should be included within the scope of the claims of the present invention.

Claims

1. A real-time monitoring system for rainstorms and floods based on the Internet of Things, characterized in that, include: Topology building module, rainstorm monitoring module, and decision-making and early warning module; Among them, the topology construction module is used to construct the weighted drainage network topology graph and calculate the prior edge weights of the drainage network; The rainstorm monitoring module is used to acquire and fuse multi-source rainfall data to generate rainfall risk indicators; The decision-making and early warning module is used to integrate rainfall risk indicators with the prior boundary weights of drainage pipe networks to generate a comprehensive risk index and trigger flood warnings.

2. The real-time monitoring system for rainstorms and floods based on the Internet of Things according to claim 1, characterized in that, Constructing a weighted drainage network topology and calculating the prior edge weights of the drainage network includes: S2-1: Entities in the urban drainage network are treated as nodes; drainage pipes or channels connecting two adjacent nodes are abstracted as directed edges in the topology graph. The direction of the edge is defined as the dominant design direction of water flow. Each edge is assigned a unique ID and associated with its starting and ending nodes. Each node and edge stores its corresponding attribute information. Traverse each edge and establish a connection relationship in the node list based on the starting node ID and the ending node to form a directed graph consisting of a set of nodes and a set of edges, representing the direction of water flow. S2-1, For each edge, under the condition of no abnormality, calculate the inherent drainage capacity of each edge under the design capacity during drainage, and form the static drainage edge weight under the initial ideal state; S2-3: Obtain records of urban flooding events over the past N years, and analyze the frequency and severity of these events to obtain frequency and severity coefficients. Assign a weighting factor to each of the frequency and severity coefficients, and then perform linear weighted fusion to obtain the dynamic water accumulation coefficient. S2-4 uses the dynamic water accumulation coefficient as a reduction factor to apply to the static drainage edge weight to obtain the dynamic prior edge weight. The drainage edge weight and prior edge weight of each edge are marked on the corresponding edge in the drainage topology graph and sent to the decision warning module.

3. The real-time monitoring system for rainstorms and floods based on the Internet of Things according to claim 2, characterized in that, The calculation of drainage boundary weights includes: Extract the pipe diameter, slope, roughness coefficient of the inner wall of the pipe, pipe length, and pipe service life, and normalize them to obtain the water flow fraction d. A Slope fraction d S Roughness fraction d b Length fraction d L and duration score d T The fractions are labeled using an index, denoted as di, where i represents the index of the fraction type, i∈{A,S,b,L,T}; a multiplicative power model is applied to each type of fraction. The drainage edge weight q is calculated; where η i The weights of the scores of each type for the edge.

4. The real-time monitoring system for rainstorms and floods based on the Internet of Things according to claim 3, characterized in that, Frequency coefficient calculation includes: Historical flooding events are extracted, including: flooding time, water accumulation location, water depth, and water accumulation duration. Using a spatial geographic matching algorithm, the water accumulation location of each historical flooding event is matched to the nearest edge on the drainage topology map, thereby obtaining the historical flooding events for each edge. Historical flooding events are extracted from the edges and divided into the corresponding months of the year according to the time of flooding occurrence. This yields the monthly distribution of historical flooding events for each edge. The number of historical flooding events in each month is counted and divided by the total number of historical flooding events in the entire drainage topology graph to obtain the frequency coefficient of the edge in each month. The monthly distribution of historical flooding events for each edge in the current month is compared to match the frequency coefficient of each edge or node in the current month.

5. A real-time monitoring system for rainstorms and floods based on the Internet of Things according to claim 4, characterized in that, Severity factor calculation includes: The average water depth and average water duration of historical flooding events are calculated by averaging the water depth and water duration. The maximum water depth and maximum water duration are selected in the entire drainage topology map. The average water depth and average water duration are then normalized by dividing the average water depth and average water duration by the maximum water depth and maximum water duration, respectively. The normalized average water depth and average water duration are then used as a two-dimensional point. The two-dimensional distance between the two-dimensional point and the origin is calculated using the Euclidean distance formula. The effect of water time and depth is quantified using a nonlinear mapping formula to obtain the severity coefficient.

6. A real-time monitoring system for rainstorms and floods based on the Internet of Things according to claim 5, characterized in that, Rainfall risk indicators include: S6-1, real-time acquisition of minute-level rainfall intensity sequences from ground rain gauges and the gridded reflectance field E(x, y, z, t) from meteorological radar, where t is the time index, x and y represent latitude and longitude, z represents altitude, and E(x, y, z, t) represents the measured atmospheric reflectance intensity at each three-dimensional grid point indexed by (x, y, z, t) in space at time t. Using data from all available rain gauges at the same time, a functional relationship is established with the radar's corresponding grid point reflectance for local dynamic calibration. Regression analysis is performed using a dynamic time window to obtain the calibrated radar-estimated rainfall rate field R(x, y, t). For each rain gauge station and time with valid rainfall records within the window, find the grid point in the radar data that is closest to the rain gauge station in terms of horizontal position and has the lowest elevation angle in terms of height, and extract its reflectivity value E(t). Using the matched data, perform regression analysis on {E(t), r(t)} within the time window to generate a spatially continuous calibrated radar estimated rainfall rate field R(x, y). From this, the radar estimated rainfall field R(x, y, t) for each time t can be obtained. By using the optimal interpolation method to take the rainfall intensity sequence at the rain gauge station as the true value anchor point, the radar field is spatially optimally adjusted to obtain the instantaneous rainfall intensity field I(x, y, t). S6-2 divides the urban area into dynamic units Uj that match the topology of the drainage network, administrative grid, or natural watershed, where j represents the index of the unit; each unit contains all nodes and edges of the drainage network within its boundary; for each unit Uj, the instantaneous rainfall intensity field I(x, y, t) within the time window is obtained, and the instantaneous rainfall intensity corresponding to the grid point in the instantaneous rainfall intensity field I(x, y, t) that falls into the unit is taken as the instantaneous rainfall intensity data of the unit, and the maximum instantaneous rainfall intensity is selected from it; S6-3: The arithmetic mean of the instantaneous rainfall intensity data within the unit is used to calculate the average rainfall intensity at each time t within the time window, and this is multiplied by the duration of the current time window to obtain the rainfall amount. The rainfall amounts of all time windows from the start of rainfall to the current time window are extracted. The difference between the midpoint of each time window and the midpoint of the current time window is calculated to obtain the interval time. A backtracking time threshold is preset, and time windows with interval times less than the backtracking time threshold are selected as the relevant windows of the current time window. The attenuation amount of all relevant windows is calculated using time decay, and then summed with the rainfall amount of the current time window to obtain the rainfall index. The rainfall risk indicators include the maximum instantaneous rainfall intensity and the rainfall index.

7. A real-time monitoring system for rainstorms and floods based on the Internet of Things according to claim 6, characterized in that, Generating a comprehensive risk index and triggering a flood warning includes: The rainfall index and the maximum instantaneous rainfall intensity are normalized, and then the normalized rainfall index and the maximum instantaneous rainfall intensity are used to obtain the rainfall threat index through a weighted geometric mean model with enhanced short-board effect. Thus, the rainfall threat index of each unit can be obtained. The inherent risk of poor drainage in each unit is quantified based on the weighted drainage network topology diagram, and the vulnerability index of the unit's pipe network is calculated. This is then combined with the vulnerability of the pipe network to generate the final comprehensive risk index. From this, the comprehensive risk index of all units can be obtained, and a global judgment is made based on this index to trigger flood warnings. The flood warnings include red and yellow warnings for waterlogging.

8. A real-time monitoring system for rainstorms and floods based on the Internet of Things according to claim 7, characterized in that, The rainfall index and maximum instantaneous rainfall intensity are normalized, and the weighted geometric mean model that enhances the weakest link effect is as follows: A rainfall range is defined by a preset impact threshold and an extreme threshold. The impact threshold and the extreme threshold are summed and divided by two to obtain the inflection point threshold. When the maximum instantaneous rainfall intensity is greater than the extreme threshold, the rainfall index is directly normalized to a value of one. If the maximum instantaneous rainfall intensity is within the rainfall range defined by the impact threshold and the extreme threshold, the maximum instantaneous rainfall intensity is normalized. If the maximum instantaneous rainfall intensity is less than the impact threshold, the rainfall index is directly normalized to a value of zero. The rainfall index is normalized using a power function with saturation characteristics; The rainfall threat index is output through a weighted geometric mean model enhanced by the weakest link effect.

9. A real-time monitoring system for rainstorms and floods based on the Internet of Things according to claim 8, characterized in that, Global assessment and triggering of flood warnings include: A flood simulation threshold is preset. The comprehensive risk index of each unit is compared with the flood simulation threshold. For each unit whose comprehensive risk index is greater than the flood simulation threshold, the region growth algorithm is started to find spatially adjacent or pipeline-connected units that meet the conditions. If the number of units in any grown region is ≥3, the grown region is marked as a waterlogging danger area and a red waterlogging warning is triggered. If the number of units in any grown region is <3, the grown region is marked as a local waterlogging point and a yellow waterlogging warning is triggered.

10. A real-time monitoring system for rainstorms and floods based on the Internet of Things according to claim 9, characterized in that, When a red or yellow alert for waterlogging is triggered, each edge within the corresponding area accumulates one waterlogging event and updates the topology building module.

Images