Mountain house rainwater infiltration risk dynamic assessment method based on multi-source data fusion

Abstract

The present application relates to a kind of mountainous area house rainwater infiltration risk dynamic evaluation method based on multi-source data fusion, belong to infiltration risk evaluation technical field.It includes the following steps: collection meteorological data, terrain data, soil data and house data;Meteorological data and terrain data are combined, and the slope enhancement factor is used to correct the rainfall of house, and the corrected meteorological data is obtained;Compensate soil saturation using model to obtain compensated soil data;All data are integrated to obtain the corrected data set;Based on the corrected data set, calculate rainfall threat index, topographic water storage index and building vulnerability index, using adaptive weighting mechanism dynamically adjusts the weight of each index, and the comprehensive risk value is obtained by weighting fusion to each index;Comprehensive risk value is compared with preset condition, triggers early warning mechanism, and the weight of each index is optimized by integrating resident feedback.The present application can improve the accuracy of mountainous area house rainwater infiltration risk dynamic evaluation.

Description

Technical Field

[0001] This invention belongs to the field of infiltration risk assessment technology, specifically relating to a dynamic assessment method for rainwater infiltration risk of houses in mountainous areas based on multi-source data fusion. Background Technology

[0002] Mountainous areas serve as crucial ecological barriers and densely populated residential regions in my country, and the safety of their buildings directly impacts the safety of residents' lives and property, as well as regional stability and development. However, the complex terrain and unique climate of mountainous areas make rainwater infiltration a major natural hazard threatening building structures. Rainwater infiltration not only causes damp walls and leaking roofs, but in severe cases, it can lead to foundation softening, wall cracking, and even house collapse, posing a significant threat to the safety of residents in mountainous areas. Therefore, accurately assessing and providing early warnings of rainwater infiltration risks in mountainous buildings is of vital practical importance for developing targeted protective measures and reducing disaster losses.

[0003] To assess the risk of rainwater infiltration in mountainous buildings, existing technologies mostly employ static models. The core idea is to calculate the risk level based on historical rainfall data, fixed topographic parameters, soil saturation thresholds, and average building structural parameters using empirical formulas or statistical models. Static models struggle to adapt to the dynamic evolution of rainwater infiltration risk in mountainous areas: firstly, they ignore real-time spatial differences in rainfall intensity and runoff velocity caused by topographic microclimates, failing to capture the instantaneous high risk caused by sudden, localized heavy rainfall; secondly, static models use fixed soil saturation thresholds and building structural parameters, failing to consider the dynamic changes in infiltration time due to soil saturation lag, and the individual specificity of risk due to differences in building structures. This simplified treatment of the "multi-factor coupling dynamics" makes static models unable to track the dynamic evolution of risk in real time, ultimately leading to poor timeliness and high false alarm rates in risk warnings. Summary of the Invention

[0004] To address the aforementioned problems, this invention provides a dynamic assessment method for rainwater infiltration risk in mountainous buildings based on multi-source data fusion.

[0005] To achieve the above objectives, the present invention employs the following technical solution:

[0006] This invention provides a method for dynamic assessment of rainwater infiltration risk in mountainous buildings based on multi-source data fusion, comprising the following steps:

[0007] S1. Collect meteorological data, topographic data, soil data, and building data;

[0008] S2. Combine meteorological data with topographic data, calculate the slope enhancement factor to correct the rainfall received by the houses, and obtain the corrected meteorological data; use a model to compensate for soil saturation based on soil type, and obtain the compensated soil data; integrate all data into a unified house-level unit to obtain the corrected dataset.

[0009] Furthermore, the correction of the house's rainfall amount includes the following steps:

[0010] The weather station's grid rainfall data is located at the house coordinates, the house's center slope percentage is extracted, and a correction value is added proportionally to the slope percentage to obtain the terrain enhancement coefficient. The actual rainfall measured by the weather station is multiplied by the terrain enhancement coefficient to obtain the actual rainfall that the house can withstand. The rainfall increase predicted by the meteorological satellite is then superimposed to generate a time series updated every five minutes to obtain the corrected meteorological data.

[0011] Furthermore, the compensation for soil saturation includes the following steps:

[0012] Based on the soil type, a lag time constant is set, and combined with the rate of change of current soil moisture, dynamic attenuation compensation is performed on the real-time monitored saturation. If an upward trend in moisture content is detected, the compensation amount is increased based on the real-time value. If the moisture content decreases, the current value is maintained until the compensation period ends.

[0013] Furthermore, the obtained corrected dataset includes the following steps:

[0014] The 3D model of the drone was analyzed, the roof material was identified through texture features, the address type and risk level were associated, and the houses were divided into age groups according to the year to obtain structured house data;

[0015] Based on topographic data, corrected meteorological data, compensated soil data, and structured house data, all data timestamps are unified to a predetermined time interval. All data are bound by house IDs, and soil data is matched to the nearest house. Topographic data is converted into a polar coordinate grid centered on the house.

[0016] The corrected meteorological data was converted into a time series matrix, the topographic data into a raster file, and the soil data and building data were encapsulated into JSON and relational tables, respectively, to obtain a standardized dataset with spatiotemporal alignment.

[0017] S3. Based on the corrected dataset, calculate the rainfall threat index, topographic water storage index, and building vulnerability index. Use an adaptive weighting mechanism to dynamically adjust the weights of each index and perform weighted fusion of each index to obtain a comprehensive risk value.

[0018] Furthermore, the rainfall threat index is calculated by dividing the real-time rainfall intensity by the critical rainfall intensity threshold and multiplying it by a factor that takes into account future rainfall increases.

[0019] Furthermore, the topographic water storage index is calculated from the return path density, soil saturation, and drainage smoothness; the drainage smoothness is assigned a value according to the type of drainage facility.

[0020] Furthermore, the building vulnerability index is the product of the historical probability of roof permeability and the foundation impermeability grade.

[0021] Furthermore, the adaptive weighting mechanism dynamically adjusts the weights of each index by including the following steps:

[0022] When soil saturation exceeds a predetermined threshold, the weight of the topographic water storage index is increased, while the weights of the rainfall threat index and the building vulnerability index are reduced proportionally.

[0023] When rainfall intensity exceeds a predetermined threshold, the weight of the rainfall threat index is increased, while the weights of the topographic water storage index and the building vulnerability index are reduced proportionally.

[0024] After each adjustment, the three weights are renormalized.

[0025] S4. Compare the comprehensive risk value with the preset conditions to trigger the early warning mechanism, and optimize the weight of each index by incorporating residents' feedback.

[0026] Furthermore, the optimization of the weights of each index by incorporating resident feedback includes the following steps:

[0027] Based on images of the leakage site reported by residents through the system, the YOLOv5 model was used to segment the leakage area and extract the area ratio of leakage traces and the leakage pattern classification.

[0028] If the area of ​​leakage marks is greater than the predetermined threshold, but the house does not trigger an alert, it is considered a missed report; if the area of ​​leakage marks is less than the predetermined threshold, but a red alert is triggered, it is considered a false alarm. For missed and false alarm cases, the index weights are backtracked and optimized. In the optimization process, the learning rate is introduced to control the adjustment range, and the difference between the actual leakage severity and the original risk value, as well as the dynamic adjustment of each index value, are taken into account.

[0029] The advantages of this invention are:

[0030] This invention collects meteorological, topographic, soil, and building data. Meteorological data is combined with topographic data to calculate slope enhancement factors to correct for building rainfall, capturing the dynamic changes in the impact of topography on rainfall. Soil data is used to compensate for the water absorption delay effect through a model to dynamically update soil saturation, and integrated into fine-grained building-level unit data to ensure real-time and accurate data input. Based on the integrated data, rainfall threat, topographic water storage, and building vulnerability indices are calculated, and an adaptive weighting mechanism is used to dynamically adjust the weights of each index. The influence of each factor is flexibly adjusted according to real-time data, allowing for real-time tracking of the dynamic evolution of risks and solving the problem that static models with fixed weights cannot reflect the dynamic contribution of risk factors. Finally, early warnings are triggered based on real-time comprehensive risk values, and resident feedback is incorporated to continuously optimize weights. Actual feedback corrects model biases and reduces false alarm rates. Simultaneously, real-time updates of multi-source data and dynamic weighted fusion ensure timely response to risk changes, effectively solving the technical problems of poor timeliness and high false alarm rates in static models. Attached Figure Description

[0031] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used together with the embodiments of the invention to explain the invention and do not constitute a limitation thereof.

[0032] Figure 1 This is a flowchart illustrating the overall steps of the present invention. Detailed Implementation

[0033] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. 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 are within the scope of protection of the present invention.

[0034] Example 1

[0035] In this embodiment, as Figure 1 As shown, this invention provides a method for dynamic assessment of rainwater infiltration risk in mountainous houses based on multi-source data fusion, the specific steps of which include:

[0036] S1. Collect meteorological data, topographic data, soil data, and building data;

[0037] Meteorological data acquisition: Real-time rainfall intensity and cumulative rainfall duration are obtained from local meteorological stations, while rainfall trend forecasts for the next 2 hours are obtained from meteorological satellites (represented by a probability distribution from 0 to 100%). The data is updated every 5 minutes, with a spatial resolution of 200-meter grid.

[0038] Terrain data acquisition: High-precision digital elevation models (DEMs) with a resolution of 5 meters are generated using LiDAR aerial photography. Slope and aspect are calculated based on DEM data, and backflow path analysis is performed, covering a 50-meter radius around each building.

[0039] Soil data acquisition: A distributed wireless sensor network is installed in the monitoring area. Each sensor node has two burial depths to monitor the volumetric water content and electrical conductivity of the soil. The distance between sensor nodes does not exceed 100 meters, and data is collected every 15 minutes.

[0040] Housing data collection: Obtain housing archive information from the GIS database of the housing and construction department, and at the same time use drones to conduct oblique photography to build a 3D model and extract the building age (years), roof material (divided into tiles, asphalt, concrete, etc., represented by codes) and foundation type (divided into rubble, concrete, pile foundation, etc., represented by categories).

[0041] S2. Combine meteorological data with topographic data, calculate the slope enhancement factor to correct the rainfall received by the houses, and obtain the corrected meteorological data; at the same time, when processing soil data, consider the water absorption delay effect, use the model to compensate for soil saturation based on soil type, and obtain the compensated soil data; then integrate all data into a unified house-level unit to obtain the corrected dataset.

[0042] In this embodiment, rainfall data from a 200-meter grid at a weather station is mapped to the coordinates of a building. The percentage of the building's center slope is extracted, and a correction value is added proportionally to this slope percentage to obtain a terrain enhancement coefficient. The measured rainfall from the weather station is multiplied by the terrain enhancement coefficient to obtain the actual rainfall the building can withstand. This is then overlaid with the rainfall increase predicted by meteorological satellites to generate a time series updated every five minutes. This results in corrected meteorological data that includes the actual rainfall the building can withstand and the predicted increase. The expression for calculating the actual rainfall the building can withstand is as follows:

[0043] ,

[0044] ,

[0045] In the formula: This indicates the most recent rainfall measured at the weather station; Indicates the terrain enhancement coefficient; This indicates the rainfall amount after the house location has been corrected. This indicates the slope of the house's center point extracted from the DEM.

[0046] In natural terrain, slope angle is a key factor affecting rainfall confluence and localized rainfall enhancement, as detailed below:

[0047] When the slope is less than or equal to 45 degrees, surface runoff is mainly laminar or slow-moving. Rainfall stays on the slope for a longer time. The enhancement effect of topography on airflow increases linearly with the increase of slope, which is consistent with the law of positive correlation between slope and flow velocity in fluid mechanics.

[0048] When the slope is greater than 45 degrees, surface runoff turns into rapids or waterfalls, and slope erosion intensifies. This exceeds the actual slope range of most houses. According to the requirement of collecting topographic data within 50m around houses in this embodiment, steep slopes greater than 45 degrees are rarely seen in urban and suburban areas. The maximum slope around mountain residents is usually controlled within 45 degrees to ensure safety.

[0049] If you directly use the slope angle (e.g.) Calculating the enhancement ratio using a slope of 60 degrees may result in an excessively large K value (e.g., 60 / 45 = 1.33, which, when multiplied by 0.1, results in an enhancement of 13.3%, exceeding the lower limit of the reasonable range of "10%-30%" in "Mountain Meteorology"). Using 45 degrees as the slope with the maximum impact can convert any slope into a dimensionless coefficient, ensuring that the enhancement ratio is controllable.

[0050] Secondly, 0.1 is the maximum proportionality coefficient for the terrain enhancement effect. When the slope is 45 degrees, the rainfall is enhanced by 10%. According to the theory of mountain meteorology, the rainfall enhancement rate on the windward slope can reach 10% to 30%. In the case of building risk assessment, it is necessary to balance the enhancement effect and safety redundancy. If the upper limit of 30% (i.e., coefficient 0.3) is taken, it may lead to excessive enhancement for the slope, which is inconsistent with the actual observation result that the enhancement of a 15-degree slope is about 3%-5%. If the lower limit of 10% (i.e. coefficient 0.1) is taken, when the slope is 15 degrees, the enhancement ratio is exactly 10%, which is highly consistent with the result that the actual rainfall of a 45-degree steep slope house in a 23-year rainstorm event in a certain area was 9.8% higher than that of the meteorological station.

[0051] The process of compensating for soil saturation includes the following steps:

[0052] Based on the soil type, a lag time constant is set, and combined with the current rate of change in soil moisture, dynamic attenuation compensation is applied to the real-time monitored saturation. That is, if an upward trend in moisture content is detected, the compensation amount is increased based on the real-time value; if the moisture content decreases, the current value is maintained until the compensation period ends. The specific expression is as follows:

[0053] ,

[0054] In the formula: Indicates the soil saturation after compensation; This indicates the real-time monitoring value of the sensor; Represents the lag time constant. Sandy soil: Soil: Clay: ;

[0055] According to the soil water and water characteristics database, as shown in Table 1:

[0056] Table 1 Soil Irrigation Properties Database

[0057]

[0058] In this embodiment, The value is the median measured value of the aforementioned moisture characteristic time constant.

[0059] The 3D model generated by the drone's oblique photography is reconstructed, and then the roof material is classified (e.g., tile, asphalt, concrete, etc.) using image recognition technology (semantic segmentation). The building age is extracted from the GIS attribute database and grouped (e.g., 0-10 years, 10-20 years, 20 years and above). The roof material and building age information are integrated to generate a basic parameter table for the building vulnerability index.

[0060] Based on topographic data, corrected meteorological data, compensated soil data, and structured building data, all data timestamps are unified to a predetermined time interval. All data are bound by building IDs, and soil sensor data is matched to the nearest building. Topographic data is converted into a polar coordinate grid centered on the building. Corrected meteorological data is converted into a time series matrix, topographic data is converted into a raster file, and soil and building data are encapsulated into JSON and relational tables, respectively, to obtain a standardized dataset with spatiotemporal alignment.

[0061] S3. Based on the corrected dataset, calculate the rainfall threat index, topographic water storage index, and building vulnerability index. Then, use an adaptive weighting mechanism to dynamically adjust the weights of each index and perform weighted fusion to obtain a comprehensive risk value; the specific expression is as follows:

[0062] ,

[0063] In the formula: Indicates the building's water penetration risk value; Indicates the rainfall threat index; Indicates the topographic water storage index; Indicates the building's vulnerability index; , , This represents the dynamic weighting coefficients of each index, which sum to 1;

[0064] ,

[0065] In the formula: This represents the critical rainfall intensity threshold; This indicates the increase in rainfall over the next 2 hours (satellite forecast value, range 0~1, e.g., if the predicted rainfall increases by 30%, then take 0.3). The prediction reliability coefficient is used to adjust the contribution weight of the rainfall increase predicted by meteorological satellites in the next 2 hours to the rainfall threat index. Essentially, it quantifies the "error level" and "reliability" of prediction data sources such as satellites. Its value ranges from 0 to 1: k=1: indicates that the prediction data is completely reliable (error approaches 0, with sufficient experimental or correction support); k=0: indicates that the prediction data is completely unreliable (error is extremely large, with no verification basis).

[0066] ,

[0067] In the formula: This indicates the density of runoff paths (the number of water flow paths calculated using a 200m×200m grid DEM around the building). Indicates the real-time soil saturation; Indicates drainage adequacy (range 0~1; concrete drainage ditch = 0.2, natural soil ditch = 0.5, no drainage = 1.0, identified through drone imagery).

[0068] ,

[0069] In the formula: This represents the historical probability of roof infiltration (range 0-1, based on the number of infiltrations recorded in the GIS database within 5 years). calculate, (Take the upper limit of 1.0). This indicates the foundation's impermeability grade (discrete score: reinforced concrete = 0.1, brick-concrete = 0.3, rubble foundation = 0.8, old rammed earth = 1.0).

[0070] The dynamic weighting coefficients are adaptively adjusted as follows:

[0071] ,

[0072] in, This indicates rainfall intensity, in millimeters per hour; after each adjustment, it needs to be re-normalized to ensure that the sum of the three weights is 1.

[0073] S4. Compare the comprehensive risk value with the preset conditions to trigger the early warning mechanism, and optimize the weight of each index by incorporating residents' feedback.

[0074] The early warning mechanism is as follows:

[0075] Yellow alert trigger conditions: and The combination of medium-to-high risk and high soil saturation may trigger slow infiltration.

[0076] Orange alert trigger conditions: and The combination of high risk and fragile foundations is prone to rapid failure; among them, The values ​​represent the foundation's impermeability grade, with the following rules: 0.1: rubble foundation without waterproof layer (old houses), 0.3: simple concrete foundation, 0.6: reinforced concrete foundation with waterproof coating, 1.0: pile foundation + waterproof raft foundation (modern structure).

[0077] Red alert trigger conditions: or There is an extremely high risk of surface flooding, posing an immediate threat; among them, Indicates the depth of water accumulation on the ground.

[0078] In this embodiment, the yellow alert emphasizes the soil's water storage potential to avoid delayed infiltration caused by soil saturation at low risk levels; the orange alert targets building defect sensitivity and immediately escalates the response when the risk level is high and the foundation is fragile; the red alert is set to dual-channel triggering, covering both the risk model's limit value and actual observed hazards.

[0079] In this embodiment, the YOLOv5 model is used to segment the leakage area based on the images reported by residents, and the area ratio of water seepage traces is extracted. Penetration mode classification (0: Roof cracks; 1: Damp walls; 2: Water seepage in the foundation)

[0080] Early warning and failure judgment based on features extracted from the model: when the area of ​​water seepage traces is greater than... Furthermore, the house did not trigger an early warning, indicating that the model missed a report; when the area of ​​water seepage marks is greater than... However, the house triggered a red alert, indicating a false alarm from the model;

[0081] For cases of underreporting or false reporting, retrospective adjustments will be made according to the following formula. , , :

[0082] ,

[0083] In the formula: Indicates the updated weights; Indicates the weights before the update; Indicates the learning rate; Indicates the actual severity of leakage. ; This represents the index values ​​of each factor (building, topography, rainfall); This represents the maximum value of the corresponding index among the three types of indices; similarly, the adjusted value is calculated. , :

[0084] By continuously comparing the model's predicted risk value R with the actual leakage severity, the weight allocation strategy is dynamically calibrated to gradually adapt the model to the characteristics of the local microenvironment.

[0085] Finally, it should be noted that the above descriptions are merely preferred embodiments of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments or make equivalent substitutions for some of the technical features. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A dynamic assessment method for rainwater infiltration risk of houses in mountainous areas based on multi-source data fusion, characterized in that, Includes the following steps: S1. Collect meteorological data, topographic data, soil data, and building data; S2. Combine meteorological data with topographic data, calculate the slope enhancement factor to correct the rainfall received by the houses, and obtain the corrected meteorological data; use a model to compensate for soil saturation based on soil type, and obtain the compensated soil data. All data is integrated into a unified house-level unit to obtain the corrected dataset; The method for correcting the amount of rain received by a house includes the following steps: The weather station's grid rainfall data is located at the house coordinates, the house's center slope percentage is extracted, and a correction value is added proportionally to the slope percentage to obtain the terrain enhancement coefficient; the actual rainfall measured by the weather station is multiplied by the terrain enhancement coefficient to obtain the actual rainfall that the house can withstand, and then the rainfall increase predicted by the meteorological satellite is added to generate a time series updated every five minutes to obtain the corrected meteorological data; S3. Based on the corrected dataset, calculate the rainfall threat index, topographic water storage index, and building vulnerability index. Use an adaptive weighting mechanism to dynamically adjust the weights of each index, and then perform a weighted fusion of the indices to obtain a comprehensive risk value. The adaptive weighting mechanism for dynamically adjusting the weights of each index includes the following steps: When soil saturation exceeds a predetermined threshold, the weight of the topographic water storage index is increased, while the weights of the rainfall threat index and the building vulnerability index are reduced proportionally. When rainfall intensity exceeds a predetermined threshold, the weight of the rainfall threat index is increased, while the weights of the topographic water storage index and the building vulnerability index are reduced proportionally. After each adjustment, the three weights are renormalized; S4. Compare the comprehensive risk value with the preset conditions to trigger the early warning mechanism, and incorporate resident feedback to optimize the weight of each index; The optimization of the weights of each index by incorporating resident feedback includes the following steps: Based on images of the leakage site reported by residents through the system, the YOLOv5 model was used to segment the leakage area and extract the area ratio of leakage traces and the leakage pattern classification. If the area of ​​leakage marks is greater than the predetermined threshold, but the house does not trigger an alert, it is considered a missed report; if the area of ​​leakage marks is less than the predetermined threshold, but a red alert is triggered, it is considered a false alarm. For missed and false alarm cases, the index weights are backtracked and optimized. In the optimization process, the learning rate is introduced to control the adjustment range, and the difference between the actual leakage severity and the original risk value, as well as the dynamic adjustment of each index value, are taken into account.

2. The method for dynamic assessment of rainwater infiltration risk in mountainous houses based on multi-source data fusion according to claim 1, characterized in that, The process of compensating for soil saturation includes the following steps: Based on the soil type, a lag time constant is set, and combined with the rate of change of current soil moisture, dynamic attenuation compensation is performed on the real-time monitored saturation. If an upward trend in moisture content is detected, the compensation amount is increased based on the real-time value. If the moisture content decreases, the current value is maintained until the compensation period ends.

3. The method for dynamic assessment of rainwater infiltration risk in mountainous houses based on multi-source data fusion according to claim 2, characterized in that, The corrected dataset includes the following steps: The 3D model of the drone was analyzed, the roof material was identified through texture features, the address type and risk level were associated, and the houses were divided into age groups according to the year to obtain structured housing data. Based on topographic data, corrected meteorological data, compensated soil data, and structured house data, all data timestamps are unified to a predetermined time interval. All data are bound by house IDs, and soil data is matched to the nearest house. Topographic data is converted into a polar coordinate grid centered on the house. The corrected meteorological data was converted into a time series matrix, the topographic data into a raster file, and the soil data and building data were encapsulated into JSON and relational tables, respectively, to obtain a standardized dataset with spatiotemporal alignment.

4. The method for dynamic assessment of rainwater infiltration risk in mountainous houses based on multi-source data fusion according to claim 3, characterized in that, The rainfall threat index is calculated by dividing the real-time rainfall intensity by the critical rainfall intensity threshold and multiplying it by a factor that takes into account future rainfall increases.

5. The method for dynamic assessment of rainwater infiltration risk in mountainous houses based on multi-source data fusion according to claim 4, characterized in that, The topographic water storage index is calculated from the return path density, soil saturation, and drainage smoothness; the drainage smoothness is assigned a value according to the type of drainage facility.

6. The method for dynamic assessment of rainwater infiltration risk of houses in mountainous areas based on multi-source data fusion according to claim 5, characterized in that, The building vulnerability index is the product of the historical probability of roof permeability and the foundation impermeability grade.

Images