Debris flow dynamic early warning method and system based on sponge city facilities

Abstract

The invention relates to the related technical field of geological disaster early warning, in particular to a sponge city facility-based debris flow dynamic early warning method and system, and the method comprises the steps: setting monitoring nodes, obtaining monitoring data, generating a monitoring network topological graph, configuring a grading early warning mechanism, and superposing water catchment paths among facilities through a GIS according to the sponge city and sponge city facility distribution; and marking the area of the threatened region, and displaying and outputting. The technical problems that a conventional debris flow dynamic early warning method depends on single environment parameter monitoring, debris flow risk changes caused by facility operation state changes cannot be accurately captured, and debris flow risk early warning accuracy is limited are solved, and spatial association between various facilities and risks is determined. The technical effects of configuring a hierarchical early warning mechanism through time domain and spatial feature analysis, dynamically tracking the time evolution and spatial diffusion trend of the risk, displaying the threatened area in the topological graph, improving the accuracy of debris flow risk early warning, and providing support for improving the pertinence of emergency response are achieved.

Description

Technical Field

[0001] This invention relates to the field of geological disaster early warning technology, specifically to a dynamic early warning method and system for debris flows based on sponge city facilities. Background Technology

[0002] Sponge city facilities, as an important means of dealing with urban flooding and regulating water circulation, play a key role in retaining rainwater, reducing flood peaks, and conserving groundwater. In mountainous cities or hilly areas, sponge city facilities are integrated with the natural terrain, and their water storage and infiltration capacity directly affects slope stability and is closely related to the dynamic risk of debris flows.

[0003] Current debris flow dynamic early warning systems based on sponge city facilities rely heavily on rainfall, which fails to reflect the impact of facility operation status on debris flow risk. Furthermore, they do not consider the dynamic performance differences of different types of facilities, such as those for retention and discharge, making it impossible to accurately locate threatened areas.

[0004] In summary, existing technologies suffer from the technical problem that conventional debris flow dynamic early warning methods rely on monitoring a single environmental parameter, which cannot accurately capture changes in debris flow risk caused by changes in facility operation status, thus limiting the accuracy of debris flow risk early warning. Summary of the Invention

[0005] This application provides a method and system for dynamic early warning of debris flows based on sponge city facilities. It aims to solve the technical problem that conventional dynamic early warning methods for debris flows in the prior art rely on monitoring a single environmental parameter, which cannot accurately capture changes in debris flow risk caused by changes in the operating status of facilities, thus limiting the accuracy of debris flow risk early warning.

[0006] In view of the above problems, the technical solution to achieve the present application is as follows:

[0007] In a first aspect, this application provides a dynamic early warning method for debris flows based on sponge city facilities. The method includes: setting up monitoring nodes according to the distribution of sponge cities and their facilities, and acquiring monitoring data including rainfall, facility water storage depth, soil moisture content, and slope displacement; generating a monitoring network topology map based on the distribution of sponge city facilities and the monitoring nodes, the monitoring network topology map having markers for infiltration facilities, retention facilities, purification facilities, and emission facilities; performing temporal and spatial feature analysis on the monitoring nodes and monitoring data, and configuring a tiered early warning mechanism; based on the tiered early warning mechanism, marking the threatened area area by overlaying water runoff paths between facilities using GIS, and displaying the impact range corresponding to the threatened area area in the monitoring network topology map.

[0008] Preferably, the monitoring data is subjected to feature processing to determine the first debris flow risk association feature and the second debris flow risk association feature; based on the monitoring network topology, and combined with the first debris flow risk association feature and the second debris flow risk association feature, a risk feature matrix is ​​constructed.

[0009] Preferably, the monitoring data is processed to obtain basic feature information; based on the soil moisture conduction coefficient in the basic feature information, a first debris flow risk association feature is determined that is associated with regional rainfall information. The first debris flow risk association feature includes infiltration attenuation rate, runoff retention rate, and sponge load rate.

[0010] Preferably, based on the facility carrying capacity in the basic feature information, a second debris flow risk association feature related to the operating status of sponge city facilities is determined. The second debris flow risk association feature includes water accumulation growth rate, soil saturation duration, and slope displacement acceleration. The first debris flow risk association feature and the second debris flow risk association feature are spatially mapped according to the confluence risk zone to construct a risk feature matrix.

[0011] Preferably, based on the monitoring network topology and combined with the facility water saturation, runoff risk zones are divided; the row vectors of the risk feature matrix correspond to each runoff risk zone, and the column vectors include the coupling value of infiltration attenuation rate and water accumulation growth rate, the ratio of runoff retention rate to soil saturation duration, and the product of sponge load rate and slope displacement acceleration.

[0012] Preferably, the critical conditions for obtaining the historical debris flow event dataset of the sponge city are determined; based on the critical conditions, and combined with the infiltration facility markers, retention facility markers, purification facility markers, and emission facility markers in the monitoring network topology diagram, initial triggering conditions are set.

[0013] Preferably, based on the initial triggering conditions and the operating status of sponge city facilities, key performance indicators of the facilities are extracted, fuzzy logic is used to infer the health status of the facilities, and the threshold value of the initial triggering conditions is dynamically updated; at the same time, time-domain feature analysis and spatial feature analysis are performed on monitoring nodes and monitoring data to determine risk diffusion pointers and facility association index items.

[0014] Preferably, in the confluence risk zone of the monitoring network topology map, the spatial distribution contour weights of the risk probability are dynamically adjusted based on the time-series risk prediction model; when the slope displacement rate is detected to exceed the preset safety threshold, the system is recalibrated to optimize the matching degree between the spatial distribution contours of the risk probability and the debris flow risk status; based on the threshold value of the dynamically updated initial triggering condition, a graded early warning mechanism is configured by referring to the risk diffusion pointer and facility association index.

[0015] Preferably, a spatiotemporal graph convolutional network is used as the model architecture, and the historical debris flow event dataset of the sponge city is used for training. The loss function adopts weighted cross-entropy loss, and a time-series risk prediction model is set. The nodes of the spatiotemporal graph convolutional network represent sponge city facilities, and the edges of the spatiotemporal graph convolutional network represent water runoff paths between facilities. The risk feature matrix is ​​converted into node feature vectors and edge feature vectors.

[0016] In a second aspect, this application provides a dynamic early warning system for debris flows based on sponge city facilities. The system includes: a monitoring node setting module: setting monitoring nodes according to the distribution of sponge cities and sponge city facilities, and acquiring monitoring data including rainfall, facility water storage depth, soil moisture content, and slope displacement; a monitoring network topology map generation module: generating a monitoring network topology map based on the distribution of sponge city facilities and the monitoring nodes, the monitoring network topology map having markers for infiltration facilities, retention facilities, purification facilities, and emission facilities; a feature analysis module: performing temporal and spatial feature analysis on the monitoring nodes and monitoring data, and configuring a tiered early warning mechanism; and a display output module: based on the tiered early warning mechanism, marking the threatened area area by overlaying water runoff paths between facilities using GIS, and displaying the impact range corresponding to the threatened area area in the monitoring network topology map.

[0017] In summary, one or more technical solutions provided in this application achieve the following technical effects: determining the spatial correlation between various facilities and risks; configuring a graded early warning mechanism through temporal and spatial feature analysis; dynamically tracking the temporal evolution and spatial diffusion trend of risks; displaying threatened areas on a topology map; improving the accuracy of debris flow risk early warning; and providing support for enhancing the pertinence of emergency response. Attached Figure Description

[0018] Figure 1 This application provides a flowchart illustrating a dynamic early warning method for debris flows based on sponge city facilities.

[0019] Figure 2 This application provides a structural schematic diagram of a debris flow dynamic early warning system based on sponge city facilities.

[0020] Explanation of reference numerals in the attached diagram: Monitoring node setting module M100, monitoring network topology map generation module M200, feature analysis module M300, display output module M400. Detailed Implementation

[0021] Example 1: The present application will be described in detail below with reference to the accompanying drawings, as follows... Figure 1 As shown, this application provides a dynamic early warning method for debris flows based on sponge city facilities, wherein the method includes:

[0022] S1: Based on the distribution of sponge city and sponge city facilities, set up monitoring nodes to obtain monitoring data including rainfall, facility water storage depth, soil moisture content and slope displacement; S2: Based on the distribution of sponge city facilities and the monitoring nodes, generate a monitoring network topology map, which includes markers for infiltration facilities, retention facilities, purification facilities and emission facilities.

[0023] Specifically, sponge city facilities refer to urban facilities that achieve rainwater retention, storage, purification, and utilization through natural accumulation, infiltration, and purification. Common infiltration facilities include permeable paving and rain gardens; retention facilities include sunken green spaces and rainwater storage ponds; purification facilities include vegetated buffer zones and artificial wetlands; and discharge facilities include rainwater pipe networks and pumping stations. Monitoring nodes refer to specific locations within the sponge city facility distribution area used to collect various monitoring data. They are usually equipped with sensors to obtain data such as rainfall, facility water storage depth, soil moisture content, and slope displacement in real time. The monitoring network topology map is used to show the connection relationship between monitoring nodes and the distribution of sponge city facilities. Different labels are used to distinguish infiltration, retention, purification, and discharge facilities, making it easy to intuitively understand the type and location of the facilities.

[0024] Implementation steps: Based on the distribution of sponge city facilities, monitoring nodes are set up at key locations. These monitoring nodes collect data in real time, such as rainfall, water storage depth of the facilities, soil moisture content, and slope displacement, through sensors. For example, monitoring nodes are set up in permeable pavement areas to obtain rainfall and soil moisture content data, and nodes are set up near rainwater storage ponds to monitor the water storage depth of the facilities. Based on the distribution of sponge city facilities and the location of monitoring nodes, a monitoring network topology map is generated. The monitoring network topology map uses different colors or shapes to distinguish between infiltration facilities, retention facilities, purification facilities, and emission facilities. In the above steps, by setting up monitoring nodes, key data affecting debris flow risk can be comprehensively obtained, ensuring that the early warning system can perceive environmental changes in real time.

[0025] S3: Perform temporal and spatial feature analysis on monitoring nodes and monitoring data, and configure a hierarchical early warning mechanism; S4: Based on the hierarchical early warning mechanism, mark the area of ​​the threatened area by overlaying water flow paths between facilities in GIS, and display and output the impact range corresponding to the area of ​​the threatened area in the monitoring network topology map.

[0026] Specifically, temporal characteristic analysis refers to analyzing the changing patterns of monitoring data over time, including characteristics such as data trends, periodicity, and abrupt changes. For example, analyzing the changing trend of rainfall over time, or the changes in the water storage depth of facilities over different time periods. Spatial characteristic analysis refers to analyzing the distribution patterns of monitoring data in geographic space, including characteristics such as spatial correlation and spatial clustering. For example, analyzing the distribution of soil moisture content in different geographical locations, or the spatial changing trend of slope displacement. The graded early warning mechanism refers to classifying early warning information into different levels according to the severity of debris flow risk, and taking corresponding early warning measures for different levels to improve the flexibility and pertinence of early warning. GIS (Geographic Information System) is a computer system used to collect, store, manage, analyze, and display geospatial data. GIS is used to overlay water runoff paths between facilities, mark the area of ​​threatened areas, and display the scope of impact in the monitoring network topology map.

[0027] Execution steps: Perform time series analysis on the data acquired by the monitoring nodes. Specifically, by analyzing the time series of rainfall, the intensity and duration of rainfall can be identified; by analyzing the time series of water storage depth in facilities, it can be determined whether the facilities are close to saturation, thereby dynamically monitoring the changing trend of debris flow risk; perform spatial distribution analysis on the monitoring data. Specifically, by analyzing the spatial distribution of soil moisture content, it can be determined which areas have high soil moisture content, which may increase the risk of debris flow; by analyzing the spatial distribution of slope displacement, unstable areas of the slope can be identified, thereby determining the spatial distribution range of debris flow risk.

[0028] Based on the results of temporal and spatial characteristic analysis, and combined with the critical conditions of historical debris flow event datasets, a tiered early warning mechanism is established. Specifically, a high-risk warning is triggered when rainfall exceeds a certain threshold and the water storage depth of facilities approaches saturation; when slope displacement exceeds a safety threshold, the spatial distribution contour lines of risk probability are recalibrated, and the warning level is dynamically updated. Using GIS technology, the water runoff paths between facilities are overlaid onto the monitoring network topology map. By analyzing the water runoff paths, the area of ​​the threatened region can be determined, and these regions can be marked and displayed on the topology map. Specifically, risk areas are marked with yellow areas on the topology map, thus providing intuitive visual information for emergency response.

[0029] In the above steps, the temporal evolution and spatial diffusion patterns of debris flow risks can be dynamically captured through temporal and spatial feature analysis. The configuration of the hierarchical early warning mechanism enables the early warning system to adjust the early warning level in a timely manner according to changes in risk, thereby improving the accuracy and timeliness of the early warning. The application of GIS technology enables the visualization of early warning information. Specifically, during rainfall, temporal and spatial feature analysis can monitor in real time whether the water storage depth of facilities under the sponge city infrastructure is close to saturation and whether the slope displacement is abnormal. At this time, the system automatically triggers a high-risk early warning and marks the area as a threatened area on the topology map, thereby enabling timely disaster prevention and mitigation measures to be taken.

[0030] Furthermore, based on the distribution of the sponge city facilities and the monitoring nodes, a monitoring network topology map is generated. The method of this application also includes:

[0031] The monitoring data is processed to determine the first debris flow risk association feature and the second debris flow risk association feature; based on the monitoring network topology, a risk feature matrix is ​​constructed by combining the first debris flow risk association feature and the second debris flow risk association feature.

[0032] Specifically, feature processing refers to the preprocessing and transformation of raw monitoring data to extract key features that reflect debris flow risk. This includes data standardization, feature extraction, and feature selection, with the aim of transforming complex monitoring data into a more easily analyzed and processed format. The first debris flow risk correlation feature is related to regional rainfall information, primarily reflecting the impact of rainfall on debris flow risk. Examples include infiltration attenuation rate, runoff retention rate, and sponge city load rate. These features can quantify the impact of rainfall on soil infiltration, runoff formation, and facility load. The second debris flow risk correlation feature is related to the operation of sponge city facilities. The row-state correlation features mainly reflect the impact of facility operation status on debris flow risk, such as water accumulation growth rate, soil saturation duration, and slope displacement acceleration. These features can quantify the facility's water storage capacity, soil saturation degree, and slope stability. The risk feature matrix is ​​used to store and represent the first and second debris flow risk correlation features. The row vectors of the matrix correspond to each runoff risk zone, and the column vectors contain values ​​of different features, such as the coupling value of infiltration attenuation rate and water accumulation growth rate, the ratio of runoff retention rate to soil saturation duration, and the product of sponge load rate and slope displacement acceleration.

[0033] Execution steps: Preprocess the monitoring data, including data cleaning, missing value handling, and data standardization. For example, standardize the rainfall data to the [0, 1] interval for comparison with other features. Extract the first debris flow risk-related features, such as infiltration attenuation rate, runoff retention rate, and sponge load rate. For example, the infiltration attenuation rate can be determined by the ratio of soil moisture conduction coefficient to rainfall. Extract the second debris flow risk-related features, such as water accumulation growth rate, soil saturation duration, and slope displacement acceleration. For example, the water accumulation growth rate can be determined by the rate of change of facility water storage depth.

[0034] Based on the monitoring network topology and combined with the water saturation of the facilities, runoff risk zones are divided. For example, based on the water depth of the facilities and the soil moisture content, urban areas are divided into different runoff risk zones. A risk feature matrix is ​​constructed, with the row vectors of the risk feature matrix corresponding to each runoff risk zone, and the column vectors of the risk feature matrix containing the values ​​of the first debris flow risk correlation feature and the second debris flow risk correlation feature. Specifically, each row of the matrix represents a runoff risk zone, and each column represents a risk correlation feature, such as the coupling value of infiltration attenuation rate and water accumulation growth rate, the ratio of runoff retention rate to soil saturation duration, and the product of sponge load rate and slope displacement acceleration.

[0035] In the above steps, by performing feature processing on the monitoring data and constructing a risk feature matrix, the complex monitoring data is transformed into key features that can directly reflect the debris flow risk. This not only quantifies the impact of rainfall and facility operation status on debris flow risk, but also, through the form of a risk feature matrix, specifically, during rainfall, through feature processing, determines the first and second debris flow risk correlation feature values ​​of each confluence risk zone. The risk feature matrix intuitively displays the changes in these feature values, thereby providing a data foundation for the configuration of a dynamic early warning mechanism.

[0036] Furthermore, the method used in this application to determine the risk-related characteristics of the first debris flow and the risk-related characteristics of the second debris flow includes:

[0037] The monitoring data is processed to obtain basic feature information; based on the soil moisture conduction coefficient in the basic feature information, a first debris flow risk association feature is determined that is associated with regional rainfall information. The first debris flow risk association feature includes infiltration attenuation rate, runoff retention rate, and sponge load rate.

[0038] Specifically, basic feature information refers to the preliminary feature set extracted from the original monitoring data through feature processing. These features can reflect the basic attributes and status of the monitoring data, such as rainfall, facility water storage depth, soil moisture content, and slope displacement after preliminary processing. Soil moisture conduction coefficient refers to the ability of soil moisture to be conducted between different locations, reflecting the soil's permeability and is an important parameter for assessing soil permeability and debris flow risk.

[0039] The first feature of debris flow risk correlation with regional rainfall information reflects the impact of rainfall on debris flow risk. Specifically, it includes: infiltration attenuation rate, which refers to the degree of attenuation of soil's ability to infiltrate through rainfall, reflecting the change in soil's infiltration capacity during continuous rainfall; runoff retention rate, which refers to the degree of retention of rainfall when it forms runoff on the soil surface, reflecting the soil's buffering capacity for runoff; and sponge city load rate, which refers to the load that sponge city facilities bear during rainfall, reflecting the facilities' water storage capacity and operational status.

[0040] Execution steps: The monitoring data undergoes feature processing, including data cleaning, standardization, and feature extraction. For example, rainfall data is standardized to the [0, 1] interval, and soil moisture content data is converted to relative humidity values. The processed data is defined as basic feature information, including rainfall, facility water storage depth, soil moisture content, and slope displacement. The soil moisture conduction coefficient, which reflects the soil's permeability, is extracted from this basic feature information. Based on the soil moisture conduction coefficient and regional rainfall, the first debris flow risk correlation characteristics are determined. Further, the soil moisture conduction coefficient is denoted as K, the regional rainfall as R, and the infiltration attenuation rate... This reflects the degree of soil infiltration capacity decay during rainfall. When the rainfall R increases, the infiltration decay rate α decreases, indicating a decline in soil infiltration capacity. Using a similar method, the runoff retention rate and sponge city load rate are determined. The runoff retention rate reflects the soil's ability to retain runoff, while the sponge city load rate reflects the load level of sponge city facilities.

[0041] In the above steps, feature processing transforms complex monitoring data into key features that directly reflect debris flow risk. Specifically, the infiltration attenuation rate quantifies the change in soil permeability during rainfall, helping to assess whether the soil is close to saturation; the runoff retention rate quantifies the soil's buffering capacity against runoff, helping to assess the risk of runoff formation; and the sponge city load rate quantifies the operational status of sponge city facilities, helping to assess whether the facilities are close to saturation, thus affecting debris flow risk. The dynamic changes of these feature values ​​provide a scientific basis for the early warning system, enabling timely adjustment of early warning levels and improving the accuracy and timeliness of early warnings.

[0042] Furthermore, the method of this application includes:

[0043] Based on the facility carrying capacity in the basic feature information, a second debris flow risk association feature is determined that is associated with the operating status of sponge city facilities. The second debris flow risk association feature includes water accumulation growth rate, soil saturation duration, and slope displacement acceleration. The first debris flow risk association feature and the second debris flow risk association feature are spatially mapped according to the confluence risk zone to construct a risk feature matrix.

[0044] Specifically, the facility carrying capacity refers to the upper limit of the capacity of sponge city facilities to receive and store rainwater. Further, based on the facility carrying capacity, the facility carrying margin is determined. The facility carrying margin refers to the remaining capacity of sponge city facilities to continue to receive and store rainwater under the current operating state, reflecting the operating state of the facilities and is a key parameter for assessing whether the facilities can effectively mitigate debris flow risks. The second debris flow risk correlation characteristics are related to the operating state of sponge city facilities, mainly reflecting the impact of facility operating state on debris flow risk. Specifically, the water accumulation growth rate refers to the rate of change of water depth in the facility over time, reflecting the water storage speed of the facility during rainfall. Soil saturation duration refers to the time required for soil to reach saturation from the start of absorption, reflecting the soil's permeability and degree of saturation; slope displacement acceleration refers to the rate of change of slope displacement velocity, reflecting changes in slope stability; spatial mapping refers to associating feature values ​​with geographic spatial locations. In this process, each feature value is assigned to its corresponding spatial location to analyze and display changes in feature values ​​spatially; the risk feature matrix is ​​used to store and represent the values ​​of the first and second debris flow risk association features in different confluence risk zones. The row vectors of the matrix correspond to each confluence risk zone, and the column vectors contain the values ​​of different features.

[0045] Execution steps: First, extract and determine the facility carrying capacity from basic feature information. The difference between the facility carrying capacity and the current equipment water storage capacity is used as the facility carrying margin, reflecting the facility's remaining water storage capacity. Based on the facility carrying margin and relevant monitoring data, determine the characteristics associated with the second debris flow risk, specifically including: the rate of water accumulation growth. Wherein, ΔH is the change in water depth within the facility, and Δt is the time interval, reflecting the water storage rate of the facility during rainfall. Similarly, the soil saturation duration and slope displacement acceleration are determined. The soil saturation duration reflects the time required for the soil to reach saturation from its current state, and the slope displacement acceleration reflects the rate of change of slope displacement velocity, which is an important indicator for assessing slope stability.

[0046] Based on the monitoring network topology and the water saturation of the facilities, runoff risk zones are divided. Specifically, urban areas are divided into different runoff risk zones according to the water depth of the facilities and the soil moisture content. The first debris flow risk correlation feature, including infiltration attenuation rate, runoff retention rate, and sponge city load rate, and the second debris flow risk correlation feature, including water accumulation growth rate, soil saturation duration, and slope displacement acceleration, are spatially mapped according to the runoff risk zones. Specifically, each feature value is assigned to its corresponding runoff risk zone. A risk feature matrix is ​​constructed, where the row vectors of the matrix correspond to each runoff risk zone, and the column vectors contain the values ​​of different features. Specifically, each row of the matrix represents a runoff risk zone, and each column represents a risk correlation feature, such as the coupling value of infiltration attenuation rate and water accumulation growth rate, the ratio of runoff retention rate to soil saturation duration, and the product of sponge city load rate and slope displacement acceleration.

[0047] In the above steps, by determining the second debris flow risk correlation characteristics and constructing a risk characteristic matrix, the facility operation status is directly linked to debris flow risk and mapped spatially. Specifically, the water accumulation growth rate can quantify the water storage speed of the facility during rainfall, helping to assess whether the facility can effectively regulate rainwater; soil near saturation increases the risk of runoff formation, and soil saturation duration can quantify the time required for the soil to reach saturation, helping to assess the soil's permeability and saturation level; decreased slope stability increases debris flow risk, and slope displacement acceleration can quantify the rate of change of slope displacement velocity, helping to assess slope stability. The dynamic changes of these characteristic values ​​provide a scientific basis for debris flow early warning, enabling timely adjustment of warning levels and improving the accuracy and timeliness of early warnings.

[0048] Furthermore, spatial mapping is performed according to confluence risk areas to construct a risk characteristic matrix. The method in this application includes:

[0049] Based on the monitoring network topology and combined with the facility's water saturation, runoff risk zones are divided. The row vectors of the risk feature matrix correspond to each runoff risk zone, and the column vectors contain the coupling value of infiltration attenuation rate and water accumulation growth rate, the ratio of runoff retention rate to soil saturation duration, and the product of sponge load rate and slope displacement acceleration.

[0050] Specifically, the monitoring network topology map is used to display the connection relationships between monitoring nodes and the distribution of sponge city facilities. Different labels are used to distinguish between infiltration, retention, purification, and emission facilities, making it easy to intuitively understand the type and location of the facilities. The facility water saturation refers to the ratio of the current water storage capacity of the sponge city facility to its maximum water storage capacity, reflecting the operational status of the facility. It is used to assess whether the facility is close to saturation, thus affecting its ability to regulate rainwater. The runoff risk zone refers to the area divided according to the facility water saturation and monitoring data. It is used to assess the debris flow risk in different areas. It is usually divided according to the distribution and operational status of the facilities for more accurate risk assessment. The risk feature matrix is ​​used to store and represent the risk characteristics of different runoff risk zones. The row vectors of the matrix correspond to each runoff risk zone, and the column vectors contain the values ​​of different features. These feature values ​​are used to assess debris flow risk.

[0051] Execution steps: Based on the monitoring network topology and the facility water saturation, the urban area is divided into different runoff risk zones. For example, according to the facility water saturation, the area is divided into low-risk zones (facility water saturation <30%), medium-risk zones (facility water saturation between 30% and 70%), and high-risk zones (facility water saturation >70%). This zone division considers the distribution and operational status of the facilities, ensuring more accurate risk assessment for each zone. For example, if zone A has high facility water saturation and is located above a slope, then zone A is classified as a high-risk zone. The row vectors of the risk feature matrix correspond to each runoff risk zone, and the column vectors contain the following three coupled feature values: specifically, the infiltration attenuation rate α and the water accumulation growth rate r. water The coupling value can be characterized as α×r water This coupling value reflects the combined effect of soil permeability and facility water storage rate. If the coupling characteristic value of the permeability decay rate and water accumulation growth rate in the high-risk area increases significantly, it indicates that the soil permeability and facility water storage rate in the area cannot effectively cope with the current rainfall situation, and the risk of debris flow increases.

[0052] Similar to the coupling characteristic values ​​of the infiltration attenuation rate and water accumulation growth rate corresponding to high-risk areas, the ratio of runoff retention rate to soil saturation duration and the product of sponge load rate and slope displacement acceleration were determined. The ratio of runoff retention rate to soil saturation duration reflects the combined influence of soil buffering capacity on runoff and soil saturation level. If the ratio of runoff retention rate to soil saturation duration decreases significantly, it indicates that the soil is close to saturation and the runoff buffering capacity is reduced, increasing the risk of debris flow. The product of sponge load rate and slope displacement acceleration reflects the combined influence of facility load and slope stability changes. If the product of sponge load rate and slope displacement acceleration increases significantly, it indicates that the facility load is too heavy and the slope stability is reduced, increasing the risk of debris flow.

[0053] In the above steps, during rainfall, by dividing runoff risk zones and constructing a risk feature matrix, monitoring data and facility operation status are spatially and characteristically integrated, providing a basis for dynamic assessment of debris flow risk. Specifically: dividing runoff risk zones based on facility water saturation and monitoring data allows for more accurate location of high-risk areas, facilitating targeted disaster prevention and mitigation measures. For example, high-risk areas may require more frequent monitoring and stricter early warning measures. Constructing a risk feature matrix, by coupling different feature values, allows for a more comprehensive assessment of debris flow risk in each runoff risk zone. These coupled feature values ​​comprehensively consider multiple factors such as soil permeability, facility water storage rate, soil saturation, facility load, and slope stability, making risk assessment more scientific and accurate.

[0054] Furthermore, in addition to configuring a tiered early warning mechanism, the method in this application also includes:

[0055] Obtain the critical conditions of the historical debris flow event dataset of the sponge city; based on the critical conditions, and in conjunction with the infiltration facility markers, retention facility markers, purification facility markers, and emission facility markers in the monitoring network topology diagram, set the initial trigger conditions.

[0056] Specifically, the critical conditions in the historical debris flow event dataset refer to the key thresholds for triggering debris flows determined by analyzing historical debris flow event data. Critical conditions typically include factors such as rainfall intensity, rainfall duration, terrain slope, and soil moisture content. Facility markings in the monitoring network topology map refer to the marking of different types of sponge city facilities in the monitoring network topology map, including infiltration facilities, retention facilities, purification facilities, and emission facilities, to distinguish the functions and locations of different facilities. Initial triggering conditions refer to the initial activation conditions of the early warning system set based on the critical conditions and facility markings, used to determine when to activate the early warning mechanism to deal with potential debris flow risks.

[0057] Execution steps: First, obtain critical conditions. Specifically, analyze historical debris flow event datasets from sponge city projects to determine key thresholds for triggering debris flows. Further, use statistical methods to determine critical values ​​such as rainfall intensity, duration, and terrain slope. For example, based on historical data, determine a critical value of 20 mm for 1-hour rainfall, 3 hours for rainfall duration, and 5 mm / hour for slope displacement. Second, set initial triggering conditions. Specifically, combine facility markers in the monitoring network topology map with the critical conditions to set initial triggering conditions. For example, if a facility in a certain area is marked as a storage facility and the rainfall intensity exceeds 20 mm / hour, a low-risk warning is triggered; if the facility is marked as a discharge facility and the slope displacement acceleration exceeds 5 mm / hour, a high-risk warning is triggered. In the above steps, obtaining critical conditions from historical debris flow event datasets and setting initial triggering conditions in conjunction with facility markers in the monitoring network topology map provides a scientific basis and activation mechanism for the debris flow early warning system, improving the accuracy and timeliness of debris flow warnings.

[0058] Furthermore, in setting initial triggering conditions, the method of this application also includes:

[0059] Based on the initial triggering conditions and the operational status of sponge city facilities, key performance indicators of the facilities are extracted, and fuzzy logic is used to infer the health status of the facilities to dynamically update the threshold values ​​of the initial triggering conditions. At the same time, time-domain feature analysis and spatial feature analysis are performed on monitoring nodes and monitoring data to determine risk diffusion indicators and facility-related index items.

[0060] Specifically, key performance indicators (KPIs) reflect the critical parameters of sponge city facilities' operational status, such as water storage depth, soil moisture content, and slope displacement, used to assess the facilities' health and operational efficiency. Fuzzy logic reasoning is used to assess the facilities' health based on KPIs and dynamically adjust the threshold values ​​of initial triggering conditions. Risk diffusion indicators assess the trend of debris flow risk diffusion, such as the spatial distribution contour weights of risk probability and risk diffusion speed, used to dynamically monitor risk trends. Furthermore, temporal characteristics are specifically represented by hourly cumulative rainfall, hourly rainfall intensity, and the coefficient of variation of hourly rainfall intensity (coefficient of variation = standard deviation / mean), reflecting rainfall uniformity. Spatial characteristics are specifically represented by the catchment area, slope, and runoff coefficient corresponding to drainage flow around the facilities. Facility association indexes represent the relationships between different facilities, such as catchment paths and distances between facilities, used to assess the mutual influence between facilities.

[0061] Execution steps: Based on the initial triggering conditions and the operational status of sponge city facilities, extract key performance indicators of the facilities. Specifically, for water retention facilities, extract the facility's water storage depth and soil moisture content; for discharge facilities, extract the slope displacement and facility operational status. For example, for water retention facilities, extract their current water storage depth as H. current The soil moisture content is S current Based on the extracted key performance indicators, fuzzy logic is used to infer the facility's health status, and the membership degree of the facility to each fuzzy set is calculated based on the values ​​of the key performance indicators. For example, if the facility's water storage depth H... current Approaching the maximum water storage depth H max If the soil moisture content S current Approaching saturation value S saturation If the facility's health status is "unhealthy", the membership degree will also tilt towards "unhealthy". Based on the result of fuzzy logic reasoning, the threshold value of the initial triggering condition is dynamically updated. For example, if the facility's health status is "unhealthy", the threshold value of the initial triggering condition is reduced to improve the sensitivity of the warning.

[0062] Temporal feature analysis was performed on the monitoring nodes and monitoring data to extract the temporal variation characteristics of the data, such as the changing trend of rainfall and the rate of change of water storage depth in facilities. Spatial feature analysis was performed on the monitoring data to extract the spatial distribution characteristics of the data, such as the spatial distribution of soil moisture content and the spatial distribution of slope displacement. For example, through temporal feature analysis, it was found that the rainfall in a certain area increased sharply in a short period of time, and the rate of change of water storage depth in facilities also increased significantly, indicating that the risk of debris flow in that area was increasing. Through spatial feature analysis, it was found that the soil moisture content in a certain area was close to saturation, and the slope displacement showed a spatial clustering trend, indicating that the risk of debris flow in that area was high.

[0063] Based on the results of temporal and spatial characteristic analysis, risk diffusion indicators are determined, such as the spatial distribution contour weights of risk probability and the risk diffusion rate. For example, by analyzing the spatial distribution contours of risk probability, it is found that the risk probability increases significantly in a certain direction, indicating that the debris flow risk is spreading in that direction. Facility association index items are determined, such as the water catchment paths between facilities and the distance between facilities. For example, by analyzing the water catchment paths between facilities, it is found that the water catchment path of a certain facility passes through multiple other facilities, indicating that the facility has a significant impact on other facilities.

[0064] In the above steps, key performance indicators of the facilities are extracted, fuzzy logic is used to infer the health status of the facilities, and the threshold values ​​of the initial triggering conditions are dynamically updated. This allows the early warning system to flexibly adjust the early warning threshold according to the actual operating status of the facilities, improving the accuracy and timeliness of the early warning. At the same time, through temporal and spatial feature analysis, it was found that the rainfall in the area where the sponge city facilities are located has increased sharply, and the risk probability has spread significantly in a certain direction. The risk diffusion pointer and facility association index are determined, which can dynamically monitor the changing trend of risk and the mutual influence between facilities, further improving the accuracy and timeliness of the early warning and providing a more comprehensive scientific basis for the dynamic assessment and early warning of debris flow risk.

[0065] Furthermore, the method of this application includes:

[0066] In the confluence risk zone of the monitoring network topology, the spatial distribution contour weights of the risk probability are dynamically adjusted based on the time-series risk prediction model. When the slope displacement rate is detected to exceed the preset safety threshold, the system is recalibrated to optimize the matching degree between the spatial distribution contours of the risk probability and the debris flow risk status. Based on the threshold value of the dynamically updated initial triggering condition, a graded early warning mechanism is configured by referring to the risk diffusion pointer and facility association index.

[0067] Specifically, a time-series risk prediction model refers to a risk prediction model based on time-series data, used to predict changes in debris flow risk over time. It typically combines historical data and real-time monitoring data, using machine learning or statistical methods to dynamically adjust the risk prediction results. Spatial distribution contour weights of risk probability refer to the weight coefficients of risk probability contour lines on a spatial distribution map, used to adjust the distribution of contour lines to more accurately reflect the actual risk situation. Slope displacement rate refers to the rate of change of slope displacement velocity, an important indicator for assessing slope stability; when the slope displacement rate exceeds a preset safety threshold, it indicates that slope stability may be threatened. Risk diffusion indicators are indicators used to assess the diffusion trend of debris flow risk, such as the rate of change of risk probability and the rate of risk diffusion. Facility association index items are index items used to represent the relationships between different facilities, such as water catchment paths between facilities and distances between facilities. A graded early warning mechanism refers to classifying early warning information into different levels based on the severity of debris flow risk and taking corresponding early warning measures for each level.

[0068] Execution steps: In the confluence risk zone of the monitoring network topology map, the spatial distribution contour weights of the risk probability are dynamically adjusted based on the time-series risk prediction model. Specifically, a spatiotemporal graph convolutional network model is trained using a historical debris flow event dataset, and the weighted cross-entropy loss function is used to optimize the model parameters. For example, the model predicts the debris flow risk probability in the confluence risk zone, and by adjusting the contour weights W, the spatial distribution of the risk probability becomes more accurate. Furthermore, when the slope displacement rate is detected to exceed a preset safety threshold, the spatial distribution contours of the risk probability are recalibrated to achieve optimal matching with the actual debris flow risk situation. Specifically, if the slope displacement rate a... displacement Exceeding the preset safety threshold a threshold Then, the contour line weights W are recalibrated to make the spatial distribution of risk probability more accurate. The specific formula is as follows: Among them, W old The contour line weights, W, are used to characterize the original risk probability. new Contour weights are used to characterize the calibrated risk probability; this dynamic calibration ensures that the spatial distribution contour lines of the risk probability accurately reflect the current debris flow risk situation.

[0069] Based on the threshold values ​​of the dynamically updated initial triggering conditions, and in accordance with the risk diffusion pointer and facility association index, a tiered early warning mechanism is configured. Specifically, based on the dynamically adjusted risk probability spatial distribution contour lines, and in conjunction with the risk diffusion pointer and facility association index, the early warning information is divided into multiple levels. Furthermore, the facility association index represents the water runoff path between facilities. By combining the facility association index, the mutual influence between different facilities is assessed, and the early warning mechanism is optimized.

[0070] In the above steps, by dynamically adjusting the spatial distribution contour line weights of risk probability, recalibrating the spatial distribution contour lines of risk probability, and configuring a graded early warning mechanism, the early warning system can dynamically adjust the early warning level based on real-time monitoring data and facility operation status, thereby improving the accuracy and timeliness of early warnings. Specifically, dynamically adjusting the spatial distribution contour line weights of risk probability, through a time-series risk prediction model combined with historical and real-time monitoring data, dynamically adjusts the spatial distribution of risk probability to more accurately reflect the actual risk situation; recalibrating the spatial distribution contour lines of risk probability, when the slope displacement rate exceeds the preset safety threshold, promptly adjusts the spatial distribution of risk probability to ensure the accuracy and reliability of early warning information; and evaluating the mutual influence between different facilities based on the dynamically adjusted risk probability and facility association index items, further optimizing the early warning mechanism, configuring a graded early warning mechanism, and improving the accuracy and timeliness of debris flow early warnings, enabling the early warning system to take corresponding early warning measures according to the severity of the risk, providing a scientific basis for disaster prevention and mitigation.

[0071] Furthermore, the method of this application includes:

[0072] A spatiotemporal graph convolutional network is adopted as the model architecture and trained using the historical debris flow event dataset of the sponge city. The loss function adopts weighted cross-entropy loss, and a time-series risk prediction model is set. The nodes of the spatiotemporal graph convolutional network represent sponge city facilities, and the edges of the spatiotemporal graph convolutional network represent water runoff paths between facilities. The risk feature matrix is ​​converted into node feature vectors and edge feature vectors.

[0073] Specifically, the spatiotemporal graph convolutional network is a graph convolutional network that combines temporal and spatial information to model the spatiotemporal relationships between sponge city facilities in order to predict debris flow risk. The node feature vector represents the feature vector of the sponge city facility, containing information such as the facility's operating status, soil moisture content, and water storage depth, used to describe the attributes of each facility. The edge feature vector represents the feature vector of the water runoff path between facilities, containing information such as the length, slope, and flow rate of the water runoff path, used to describe the relationship between facilities. The weighted cross-entropy loss is a loss function used to handle class imbalance problems and improve the model's ability to predict different risk levels.

[0074] Execution steps: Construct a spatiotemporal graph convolutional network using a historical debris flow event dataset for sponge cities. In the spatiotemporal graph convolutional network, nodes represent sponge city facilities, and edges represent water runoff paths between facilities. For example, nodes can be facilities such as rain gardens, sunken green spaces, and permeable paving, while edges can be drainage pipes or surface runoff paths connecting these facilities. Train the spatiotemporal graph convolutional network using the historical debris flow event dataset, and optimize the model parameters using a weighted cross-entropy loss function. The weighted cross-entropy loss function can handle the class imbalance problem between different risk levels and improve the prediction accuracy of the model. For example, if high-risk events are less common in the dataset, the weighted cross-entropy loss function can give high-risk events higher weights, making the model focus more on predicting high-risk events.

[0075] The risk feature matrix is ​​transformed into node feature vectors and edge feature vectors. The node feature vectors contain information such as the facility's operational status, soil moisture content, and water storage depth; the edge feature vectors contain information such as the length, slope, and flow rate of the water catchment path. For example, for a node associated with a sponge city facility, its feature vector can be represented as v. node =[H current ,S current ,γ], where H current It refers to the water storage depth of sponge city facilities, S current γ represents the soil moisture content of the sponge city infrastructure, and γ represents the sponge load rate; for an edge associated with a water catchment path, its eigenvector can be represented as v edge= [L,G,Q], where L is the length of the water flow path, G is the slope associated with the water flow path, and Q is the flow rate associated with the water flow path. The edge weights of the spatiotemporal graph convolutional network are related to the proportion of the water flow area corresponding to the water flow path.

[0076] In the above steps, a spatiotemporal graph convolutional network is constructed and trained, and then applied to a time-series risk prediction model to support the dynamic prediction of debris flow risk. Specifically, by representing sponge city facilities as nodes and the water runoff paths between facilities as edges, the spatiotemporal relationships between facilities can be comprehensively modeled, and a spatiotemporal graph convolutional network is constructed to support risk prediction. The model parameters of the time-series risk prediction model are optimized using a weighted cross-entropy loss function, which can effectively handle the class imbalance problem, improve the model's ability to predict different risk levels, and ensure the accuracy and reliability of the early warning system. The risk feature matrix is ​​converted into node feature vectors and edge feature vectors, enabling the model to better utilize the feature information of facilities and water runoff paths. For example, if the soil moisture content of a facility is close to saturation and the flow of its water runoff path increases, the time-series risk prediction model will update the risk prediction in a timely manner and issue a corresponding early warning signal through the early warning system, thereby further improving the accuracy of risk prediction and enhancing the precision and timeliness of debris flow early warning.

[0077] In summary, the beneficial effects of the embodiments of this application are:

[0078] This application provides a dynamic debris flow early warning method and system based on sponge city facilities. By setting monitoring nodes according to the distribution of sponge city facilities and infrastructure, monitoring data including rainfall, facility water storage depth, soil moisture content, and slope displacement are acquired. A monitoring network topology map is generated based on the distribution of sponge city facilities and monitoring nodes, with markers for infiltration, retention, purification, and emission facilities. Temporal and spatial feature analyses are performed on the monitoring nodes and data, and a tiered early warning mechanism is configured. Based on this tiered early warning mechanism, water runoff paths between facilities are overlaid using GIS to mark the area of ​​threatened regions, and the impact range corresponding to the threatened area is displayed on the monitoring network topology map. This application achieves the technical effect of determining the spatial correlation between various facilities and risks by providing a dynamic debris flow risk early warning method and system based on sponge city facilities. Through temporal and spatial feature analysis, a tiered early warning mechanism is configured to dynamically track the temporal evolution and spatial diffusion trend of risks, and threatened areas are displayed on the topology map, thus improving the accuracy of debris flow risk early warning and providing support for enhancing the targeting of emergency response.

[0079] Example 2 is based on the same inventive concept as the debris flow dynamic early warning method based on sponge city facilities in the previous examples, such as... Figure 2 As shown in the embodiment of this application, a debris flow dynamic early warning system based on sponge city facilities is provided, wherein the system includes:

[0080] Monitoring Node Setting Module M100: Based on the distribution of sponge city and sponge city facilities, monitoring nodes are set up to obtain monitoring data including rainfall, facility water storage depth, soil moisture content and slope displacement.

[0081] Monitoring network topology generation module M200: Based on the distribution of the sponge city facilities and the monitoring nodes, it generates a monitoring network topology map, which includes markers for infiltration facilities, storage facilities, purification facilities, and emission facilities.

[0082] Feature Analysis Module M300: Performs temporal and spatial feature analysis on monitoring nodes and monitoring data, and configures a hierarchical early warning mechanism.

[0083] Display output module M400: Based on the hierarchical early warning mechanism, it marks the area of ​​the threatened area by overlaying water flow paths between facilities in GIS, and displays the impact range corresponding to the threatened area in the monitoring network topology map.

[0084] Furthermore, the monitoring network topology generation module M200 is also used to perform the following method:

[0085] The monitoring data is processed to determine the first debris flow risk association feature and the second debris flow risk association feature; based on the monitoring network topology, a risk feature matrix is ​​constructed by combining the first debris flow risk association feature and the second debris flow risk association feature.

[0086] Furthermore, the monitoring network topology generation module M200 is also used to perform the following method:

[0087] The monitoring data is processed to obtain basic feature information; based on the soil moisture conduction coefficient in the basic feature information, a first debris flow risk association feature is determined that is associated with regional rainfall information. The first debris flow risk association feature includes infiltration attenuation rate, runoff retention rate, and sponge load rate.

[0088] Furthermore, the monitoring network topology generation module M200 is also used to perform the following method:

[0089] Based on the facility carrying capacity in the basic feature information, a second debris flow risk association feature is determined that is associated with the operating status of sponge city facilities. The second debris flow risk association feature includes water accumulation growth rate, soil saturation duration, and slope displacement acceleration. The first debris flow risk association feature and the second debris flow risk association feature are spatially mapped according to the confluence risk zone to construct a risk feature matrix.

[0090] Furthermore, the monitoring network topology generation module M200 is also used to perform the following method:

[0091] Based on the monitoring network topology and combined with the facility's water saturation, runoff risk zones are divided. The row vectors of the risk feature matrix correspond to each runoff risk zone, and the column vectors contain the coupling value of infiltration attenuation rate and water accumulation growth rate, the ratio of runoff retention rate to soil saturation duration, and the product of sponge load rate and slope displacement acceleration.

[0092] Furthermore, the feature analysis module M300 is also used to perform the following methods:

[0093] Obtain the critical conditions of the historical debris flow event dataset of the sponge city; based on the critical conditions, and in conjunction with the infiltration facility markers, retention facility markers, purification facility markers, and emission facility markers in the monitoring network topology diagram, set the initial trigger conditions.

[0094] Furthermore, the feature analysis module M300 is also used to perform the following methods:

[0095] Based on the initial triggering conditions and the operational status of sponge city facilities, key performance indicators of the facilities are extracted, and fuzzy logic is used to infer the health status of the facilities to dynamically update the threshold values ​​of the initial triggering conditions. At the same time, time-domain feature analysis and spatial feature analysis are performed on monitoring nodes and monitoring data to determine risk diffusion indicators and facility-related index items.

[0096] Furthermore, the feature analysis module M300 is also used to perform the following methods:

[0097] In the confluence risk zone of the monitoring network topology, the spatial distribution contour weights of the risk probability are dynamically adjusted based on the time-series risk prediction model. When the slope displacement rate is detected to exceed the preset safety threshold, the system is recalibrated to optimize the matching degree between the spatial distribution contours of the risk probability and the debris flow risk status. Based on the threshold value of the dynamically updated initial triggering condition, a graded early warning mechanism is configured by referring to the risk diffusion pointer and facility association index.

[0098] Furthermore, the feature analysis module M300 is also used to perform the following methods:

[0099] A spatiotemporal graph convolutional network is adopted as the model architecture and trained using the historical debris flow event dataset of the sponge city. The loss function adopts weighted cross-entropy loss, and a time-series risk prediction model is set. The nodes of the spatiotemporal graph convolutional network represent sponge city facilities, and the edges of the spatiotemporal graph convolutional network represent water runoff paths between facilities. The risk feature matrix is ​​converted into node feature vectors and edge feature vectors.

[0100] In summary, any step can be stored as a computer instruction or program in an unrestricted computer memory and can be called and recognized by an unrestricted computer processor; no further restrictions are imposed here.

[0101] Furthermore, the above technical solutions only embody the preferred technical solutions of the embodiments of this application. Any changes that those skilled in the art may make to certain parts of these solutions embody the novel principles of the embodiments of this application. Obviously, those skilled in the art can make various modifications and variations to this application without departing from the scope of this application.

Claims

1. A dynamic early warning method for debris flows based on sponge city facilities, characterized in that, The method includes: Based on the distribution of sponge cities and sponge city facilities, monitoring nodes are set up to obtain monitoring data including rainfall, facility water storage depth, soil moisture content, and slope displacement. Based on the distribution of the sponge city facilities and the monitoring nodes, a monitoring network topology map is generated. The monitoring network topology map includes markers for infiltration facilities, storage facilities, purification facilities, and emission facilities. Perform temporal and spatial feature analysis on monitoring nodes and monitoring data, and configure a hierarchical early warning mechanism; Based on the aforementioned hierarchical early warning mechanism, the area of ​​the threatened region is marked by overlaying water runoff paths between facilities in GIS, and the impact range corresponding to the threatened region area is displayed and output in the monitoring network topology map.

2. The debris flow dynamic early warning method based on sponge city facilities as described in claim 1, characterized in that, Based on the distribution of the sponge city facilities and the monitoring nodes, a monitoring network topology map is generated. The method further includes: The monitoring data is subjected to feature processing to determine the first debris flow risk correlation feature and the second debris flow risk correlation feature; Based on the monitoring network topology, and combined with the first debris flow risk correlation feature and the second debris flow risk correlation feature, a risk feature matrix is ​​constructed.

3. The debris flow dynamic early warning method based on sponge city facilities as described in claim 2, characterized in that, The method for determining the first debris flow risk association characteristics and the second debris flow risk association characteristics includes: The basic feature information is defined by performing feature processing on the monitoring data. Based on the soil moisture conduction coefficient in the basic feature information, a first debris flow risk association feature is determined that is associated with regional rainfall information. The first debris flow risk association feature includes infiltration attenuation rate, runoff retention rate, and sponge load rate.

4. The debris flow dynamic early warning method based on sponge city facilities as described in claim 3, characterized in that, The method includes: Based on the facility carrying capacity in the basic feature information, a second debris flow risk association feature is determined that is associated with the operating status of sponge city facilities. The second debris flow risk association feature includes water accumulation growth rate, soil saturation duration, and slope displacement acceleration. The first debris flow risk correlation feature and the second debris flow risk correlation feature are spatially mapped according to the confluence risk zone to construct a risk feature matrix.

5. The debris flow dynamic early warning method based on sponge city facilities as described in claim 4, characterized in that, The method involves spatial mapping based on confluence risk zones to construct a risk characteristic matrix, including: Based on the monitoring network topology and the facility's water saturation, confluence risk zones are delineated. The row vectors of the risk feature matrix correspond to each runoff risk zone, and the column vectors contain the coupling value of infiltration attenuation rate and water accumulation growth rate, the ratio of runoff retention rate to soil saturation duration, and the product of sponge load rate and slope displacement acceleration.

6. The debris flow dynamic early warning method based on sponge city facilities as described in claim 2, characterized in that, The method further includes configuring a tiered early warning mechanism: Critical conditions for obtaining the historical debris flow event dataset of the sponge city; Based on the critical conditions, and in conjunction with the markers for infiltration facilities, storage facilities, purification facilities, and emission facilities in the monitoring network topology diagram, initial triggering conditions are set.

7. The debris flow dynamic early warning method based on sponge city facilities as described in claim 6, characterized in that, The method for setting initial trigger conditions further includes: Based on the initial triggering conditions and the operating status of sponge city facilities, key performance indicators of the facilities are extracted, fuzzy logic is used to infer the health status of the facilities, and the threshold values ​​of the initial triggering conditions are dynamically updated. Simultaneously, temporal and spatial characteristic analyses are performed on monitoring nodes and monitoring data to determine risk diffusion indicators and facility-related index items.

8. The debris flow dynamic early warning method based on sponge city facilities as described in claim 7, characterized in that, The method includes: In the confluence risk zone of the monitoring network topology, the spatial distribution contour weights of the risk probability are dynamically adjusted based on the time-series risk prediction model. When the slope displacement rate is detected to exceed the preset safety threshold, recalibration is performed to optimize the matching degree between the spatial distribution contour lines of risk probability and the debris flow risk status. Based on the dynamically updated threshold values ​​of the initial triggering conditions, and in accordance with the risk diffusion pointer and facility association index, a tiered early warning mechanism is configured.

9. The debris flow dynamic early warning method based on sponge city facilities as described in claim 8, characterized in that, The method includes: A spatiotemporal graph convolutional network was adopted as the model architecture, and the model was trained using the historical debris flow event dataset of the sponge city. The loss function adopted was weighted cross-entropy loss, and a time series risk prediction model was set up. The nodes of the spatiotemporal graph convolutional network represent sponge city facilities, and the edges of the spatiotemporal graph convolutional network represent water runoff paths between facilities. The risk feature matrix is ​​converted into node feature vectors and edge feature vectors.

10. A debris flow dynamic early warning system based on sponge city facilities, characterized in that, The system is used for implementing the debris flow dynamic early warning method based on sponge city facilities according to any one of claims 1-9, wherein the system comprises: Monitoring node setting module: Based on the distribution of sponge city and sponge city facilities, monitoring nodes are set to obtain monitoring data including rainfall, facility water storage depth, soil moisture content and slope displacement. Monitoring network topology generation module: Based on the distribution of the sponge city facilities and the monitoring nodes, a monitoring network topology map is generated. The monitoring network topology map includes markers for infiltration facilities, storage facilities, purification facilities, and emission facilities. Feature analysis module: Performs temporal and spatial feature analysis on monitoring nodes and monitoring data, and configures a hierarchical early warning mechanism; Display output module: Based on the hierarchical early warning mechanism, the threatened area is marked by overlaying water runoff paths between facilities in GIS, and the impact range corresponding to the threatened area is displayed and output in the monitoring network topology map.

Images