Risk evaluation method, system and equipment for power transmission and distribution tower in mining subsidence area of mountain coal mine and storage medium

Abstract

The invention relates to the technical field of geological disaster risk evaluation, in particular to a mountain coal mine goaf subsidence area power transmission and distribution tower risk evaluation method, system and device and a storage medium. Obtaining multi-dimensional geological environment data of the evaluation area, extracting the multi-dimensional geological environment data corresponding to the sample points and the marked collapse sample data as training features, generating a risk prediction model through training, and carrying out risk level identification on each space unit in the evaluation area through the risk prediction model, forming a regional risk distribution result; the method comprises the following steps: acquiring an accumulated vertical displacement value at a tower position, establishing a probability distribution relationship between the accumulated vertical displacement value and a tower damage grade by combining a tower damage state sample investigated on site, calculating tower equivalent damage probabilities under different accumulated vertical displacement values, and fitting according to the equivalent damage probabilities to generate a damage probability curve; and substituting the accumulated vertical displacement value of each tower into the damage probability curve to calculate a damage probability value.

Description

Technical Field

[0001] This invention relates to the field of geological disaster risk assessment technology, and in particular to a method, system, equipment and storage medium for risk assessment of power transmission and distribution towers in mountainous coal mine subsidence areas. Background Technology

[0002] Underground goafs formed by coal mining can cause surface subsidence, posing a serious threat to surface infrastructure. Surface subsidence in goafs leads to uneven settlement of power transmission tower foundations, causing safety accidents such as tower tilting and conductor breakage, severely impacting the stable operation of the power system. Therefore, a scientific assessment of the risk status of power transmission and distribution towers in goaf subsidence areas is of great significance.

[0003] In existing technologies, synthetic aperture radar interferometry has been widely used for monitoring surface deformation in mining areas. Temporal interferometry can acquire large-scale, high-precision surface subsidence data, effectively identifying areas of surface deformation. However, existing monitoring technologies primarily focus on the quantitative analysis of surface deformation itself, lacking evaluation methods that combine deformation information with damage to specific disaster-bearing structures. The risk of mining subsidence is influenced by multiple factors, and relying solely on deformation data makes it difficult to systematically assess the risk of tower failure.

[0004] The geological environment of mining subsidence areas in mountainous regions is complex, with discontinuous subsidence patterns. The location of the subsidence center is affected by factors such as topography, lithology, and rock strata dip angle, resulting in irregularly shaped subsidence basins. Evaluation methods for mining subsidence in plains areas are difficult to apply directly to mountainous environments. Existing evaluation methods struggle to quantify the complex relationship between dynamic geological activity in mining subsidence areas and tower failure, and lack the technical means to effectively couple geological environmental hazards with the vulnerability of tower structures.

[0005] Existing risk assessment methods largely focus on qualitative analysis or simple semi-quantitative evaluation, with inadequate evaluation index systems that fail to fully consider the comprehensive impact of multi-dimensional factors such as topography, geological conditions, vegetation cover, and mining parameters. The accuracy and reliability of the evaluation results need improvement, making it difficult to provide accurate basis for the formulation of engineering protection measures. Existing methods lack the ability to identify refined risks at the individual tower level, failing to achieve tower-level risk classification management and targeted protection. Summary of the Invention

[0006] In view of the problems existing in the prior art, the present invention is proposed.

[0007] Therefore, the problem to be solved by this invention is how to address the following issues in the existing technology: First, it is impossible to effectively integrate multi-dimensional geological environmental factors and mining-related factors for systematic risk assessment; second, there is a lack of vulnerability assessment methods that quantitatively correlate surface deformation monitoring data with the damage status of tower structures; third, it fails to achieve a scientific coupling of hazard and vulnerability to form a comprehensive risk assessment system; and fourth, the assessment accuracy is insufficient, making it difficult to support refined risk management and graded early warning for individual towers.

[0008] To solve the above-mentioned technical problems, the present invention provides the following technical solution: In a first aspect, embodiments of the present invention provide a method for risk assessment of power transmission and distribution towers in mountainous coal mine subsidence areas, which includes: acquiring multidimensional geological environment data of the assessment area, extracting multidimensional geological environment data corresponding to each sample point and labeled subsidence sample data as training features, generating a hazard prediction model through training, and using the hazard prediction model to identify the hazard level of each spatial unit in the assessment area to form a regional hazard distribution result. The cumulative vertical displacement value at the tower location is obtained. Combined with the tower damage status sample from the field investigation, the probability distribution relationship between the cumulative vertical displacement value and the tower damage level is established. The equivalent damage probability of the tower under different cumulative vertical displacement values ​​is calculated. A damage probability curve is generated by fitting the equivalent damage probability. The cumulative vertical displacement value of each tower is substituted into the damage probability curve to calculate the damage probability value. The vulnerability level is classified according to the damage probability value. Construct a coupling matrix between hazard level and vulnerability level, query the hazard level of the spatial grid cell where each tower is located and the vulnerability level of the corresponding tower, determine the comprehensive risk level of each tower based on the coupling matrix, and generate a tower risk distribution map based on the comprehensive risk level of each tower.

[0009] As a preferred embodiment of the risk assessment method for power transmission and distribution towers in mountainous coal mine subsidence areas described in this invention, the method for obtaining multidimensional geological environment data of the assessment area includes topographic parameters, geological condition parameters, surface cover parameters, coal seam mining parameters, tower spatial location parameters, and temporal surface deformation parameters. The labeled collapse sample data includes collapsed sample points labeled according to time-series surface deformation parameters in areas where mining-induced collapses have occurred, and non-collapse sample points labeled in stable areas.

[0010] As a preferred embodiment of the risk assessment method for power transmission and distribution towers in the subsidence area of ​​mountainous coal mines described in this invention, the topographic parameters include elevation, slope, aspect and topographic relief; the surface cover parameters include vegetation cover and land use type; and the coal seam mining parameters include the number of mining layers, mining thickness, mining depth and mining depth-to-thickness ratio.

[0011] As a preferred embodiment of the risk assessment method for power transmission and distribution towers in mountainous coal mine subsidence areas described in this invention, the method for establishing the probability distribution relationship between cumulative vertical displacement values ​​and tower damage levels includes: determining the tower damage levels corresponding to different cumulative vertical displacement value intervals based on field investigations, using a log-normal distribution function to describe the probability of occurrence of each damage level, and weighted summing of the probability of occurrence of each damage level to obtain the equivalent damage probability.

[0012] The beneficial effects of this technical solution are as follows: By obtaining real tower damage data through on-site investigation, a correspondence between cumulative vertical displacement values ​​and damage levels is established. The use of a log-normal distribution function accurately characterizes the non-uniform distribution of tower damage probability during surface subsidence, consistent with the actual physical process of tower failure caused by mining subsidence. Through weighted summation of the probability of occurrence for each damage level, multiple discrete damage levels are transformed into continuous equivalent damage probability values, preserving the differences between different damage states while achieving a quantitative expression of vulnerability.

[0013] As a preferred embodiment of the risk assessment method for power transmission and distribution towers in the subsidence area of ​​mountainous coal mines described in this invention, the damage probability curve is fitted with a hyperbolic tangent function to fit the nonlinear relationship between the cumulative vertical displacement value and the equivalent damage probability; a nonlinear fitting method is used to establish a continuous functional relationship between the cumulative deformation and the equivalent damage probability, which serves as a quantitative basis for vulnerability assessment.

[0014] As a preferred embodiment of the risk assessment method for power transmission and distribution towers in mountainous coal mine subsidence areas described in this invention, the construction of the coupling matrix between hazard level and vulnerability level includes setting classification standards for hazard level and vulnerability level. Based on the combination relationship between hazard level and vulnerability level, the rules for determining the comprehensive risk level are defined; Using the hazard level and vulnerability level of each tower as input, the corresponding comprehensive risk level is determined through matrix lookup.

[0015] The beneficial effects of this technical solution are as follows: By establishing a coupling matrix of hazard and vulnerability, the organic integration of geological environmental risk and the vulnerability of disaster-bearing bodies is achieved. Setting grading standards transforms continuous hazard and vulnerability indices into operable level classifications, facilitating intuitive risk identification and tiered management. The definition of judgment rules fully considers the combined effects of hazard and vulnerability, reflecting the objective law that "high hazard + high vulnerability" leads to extremely high risk, and "low hazard + low vulnerability" results in low risk. A matrix lookup method is used to determine risk levels, avoiding complex calculation processes, improving evaluation efficiency, and ensuring the consistency and repeatability of evaluation results.

[0016] As a preferred embodiment of the risk assessment method for power transmission and distribution towers in mountainous coal mine subsidence areas described in this invention, the method further includes: setting graded early warning thresholds based on the cumulative vertical displacement values ​​of each tower, and triggering a risk early warning signal of the corresponding level when the cumulative vertical displacement value of the tower reaches the early warning threshold.

[0017] Secondly, embodiments of the present invention provide a risk assessment system for power transmission and distribution towers in mountainous coal mine subsidence areas, which includes a hazard assessment module, which acquires multidimensional geological environment data of the assessment area, extracts multidimensional geological environment data corresponding to each sample point and labeled subsidence sample data as training features, generates a hazard prediction model through training, and identifies the hazard level of each spatial unit in the assessment area using the hazard prediction model to form a regional hazard distribution result. The vulnerability assessment module obtains the cumulative vertical displacement value at the tower location, combines it with the tower damage status samples from the field survey, establishes the probability distribution relationship between the cumulative vertical displacement value and the tower damage level, calculates the equivalent damage probability of the tower under different cumulative vertical displacement values, generates a damage probability curve based on the equivalent damage probability, substitutes the cumulative vertical displacement value of each tower into the damage probability curve to calculate the damage probability value, and classifies the vulnerability level based on the damage probability value. The results output module constructs a coupling matrix between hazard level and vulnerability level, queries the hazard level of the spatial grid cell where each tower is located and the vulnerability level of the corresponding tower, determines the comprehensive risk level of each tower based on the coupling matrix, and generates a tower risk distribution map based on the comprehensive risk level of each tower.

[0018] Thirdly, embodiments of the present invention provide a computer device, including a memory and a processor, wherein the memory stores a computer program, wherein: when the computer program instructions are executed by the processor, they implement the steps of the risk assessment method for power transmission and distribution towers in mountainous coal mine subsidence areas as described in the first aspect of the present invention.

[0019] Fourthly, embodiments of the present invention provide a computer-readable storage medium storing a computer program thereon, wherein: when the computer program instructions are executed by a processor, they implement the steps of the risk assessment method for power transmission and distribution towers in mountainous coal mine subsidence areas as described in the first aspect of the present invention.

[0020] The beneficial effects of this invention are as follows: By acquiring multidimensional geological environment data of the evaluation area and constructing a hazard prediction model based on machine learning, this invention systematically integrates multidimensional factors such as topography, geological conditions, surface cover, and coal seam mining, solving the problem that traditional methods cannot comprehensively consider multi-source heterogeneous data, and realizing a spatial and quantitative evaluation of the risk of mining subsidence in complex geological environments in mountainous areas. By establishing the probability distribution relationship between cumulative vertical displacement values ​​and tower damage levels and fitting damage probability curves, the surface subsidence data obtained from InSAR time-series monitoring is transformed into tower damage probability, achieving a quantitative transformation from surface deformation to structural damage. By constructing a coupling matrix of hazard and vulnerability, the possibility of geological subsidence is organically combined with the disaster-bearing vulnerability of the tower structure, providing a comprehensive risk level judgment and graded early warning threshold setting for each tower, solving the engineering problem of lacking refined risk identification and targeted protection basis for transmission towers in mountainous mining subsidence areas. Attached Figure Description

[0021] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0022] Figure 1 A flowchart for risk assessment of power transmission and distribution towers in mountainous coal mine subsidence areas; Figure 2 Computer equipment diagram for risk assessment of power transmission and distribution towers in coal mining subsidence areas in mountainous regions; Figure 3 A schematic diagram of the confusion matrix of various evaluation models for risk assessment of power transmission and distribution towers in coal mining subsidence areas in mountainous regions; Figure 4 A schematic diagram of ROC curves for various evaluation models of power transmission and distribution towers in mountainous coal mine subsidence areas; Figure 5 Damage probability diagrams of power transmission and distribution towers at various damage scales in the risk assessment method for power transmission and distribution towers in coal mining subsidence areas in mountainous regions. Figure 6 The fitting curve of the average damage probability of power transmission and distribution towers in the coal mine subsidence area is used for the risk assessment method of power transmission and distribution towers in the coal mine subsidence area in mountainous areas. Detailed Implementation

[0023] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0024] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the spirit of the invention. Therefore, the invention is not limited to the specific embodiments disclosed below.

[0025] Secondly, the term "an embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in different places throughout this specification does not necessarily refer to the same embodiment, nor is it a single embodiment or an embodiment selectively excluded from other embodiments.

[0026] Example 1 Reference Figure 1 - Figure 2 This is the first embodiment of the present invention, which provides a method for risk assessment of power transmission and distribution towers in mountainous coal mine subsidence areas, including: S100: Obtain multidimensional geological environment data of the evaluation area, extract multidimensional geological environment data corresponding to each sample point and labeled collapse sample data as training features, generate a hazard prediction model through training, identify the hazard level of each spatial unit in the evaluation area using the hazard prediction model, and form the regional hazard distribution results.

[0027] S200: Obtain the cumulative vertical displacement value at the tower location, combine it with the tower damage status sample from the on-site investigation, establish the probability distribution relationship between the cumulative vertical displacement value and the tower damage level, calculate the equivalent damage probability of the tower under different cumulative vertical displacement values, generate a damage probability curve based on the equivalent damage probability, substitute the cumulative vertical displacement value of each tower into the damage probability curve to calculate the damage probability value, and classify the vulnerability level based on the damage probability value.

[0028] S300: Construct a coupling matrix between hazard level and vulnerability level, query the hazard level of the spatial grid cell where each tower is located and the vulnerability level of the corresponding tower, determine the comprehensive risk level of each tower based on the coupling matrix, and generate a tower risk distribution map based on the comprehensive risk level of each tower.

[0029] It should be noted that during the sample labeling process, areas that have collapsed were identified and labeled as positive samples using time-series surface deformation parameters, while negative samples were labeled in stable areas, forming a training dataset for comparing positive and negative samples. This allows the machine learning model to learn the characteristic differences between collapsed and stable areas. In the vulnerability assessment stage, the cumulative vertical displacement value, as the core variable, directly reflects the intensity of the effect of surface subsidence on the tower foundation. Combined with damage state samples from field investigations, a quantitative relationship between deformation and damage level is established, realizing the transformation from monitoring data to damage state.

[0030] This technical solution constructs a hazard assessment system based on multi-source data fusion through step S100. It integrates multi-dimensional factors such as topography, geological conditions, surface cover, and mining parameters. Using machine learning models, it uncovers the complex nonlinear relationships between these factors and subsidence occurrence. After generating a hazard prediction model, it identifies hazard at the spatial unit level for the assessment area, forming a regional hazard distribution result reflecting the probability of mining subsidence. This solves the problem of traditional methods' difficulty in systematically integrating multi-dimensional factors. Step S200 establishes a vulnerability assessment method, obtaining the cumulative vertical displacement values ​​of tower locations. Combined with on-site damage survey data, it uses a probability distribution function to establish a quantitative relationship between displacement and damage. Through equivalent damage probability calculation and damage probability curve fitting, continuous displacement monitoring data is converted into vulnerability levels, achieving a quantitative correlation between surface deformation monitoring results and tower structural damage status. This solves the problem of existing technologies lacking a quantitative relationship between deformation data and damage to the disaster-bearing body. The S300 process achieves a scientific coupling of hazard and vulnerability, constructing a coupling matrix that organically combines the hazards of the geological environment with the vulnerability of the tower structure. Through matrix queries, a comprehensive risk level is determined for each tower, generating an intuitive tower risk distribution map that reflects both the possibility of mining subsidence and the tower's vulnerability to disasters. This enables refined risk assessment of individual towers, addressing the issues of insufficient assessment accuracy and lack of refined risk identification capabilities.

[0031] Example 2 Reference Figure 2 - Figure 6 This is the second embodiment of the present invention.

[0032] In this embodiment, step S100 involves acquiring multidimensional geological environment data of the evaluation area, extracting multidimensional geological environment data corresponding to each sample point and labeled collapse sample data as training features, generating a hazard prediction model through training, and using the hazard prediction model to identify the hazard level of each spatial unit within the evaluation area to form a regional hazard distribution result, including the following steps A1-A2: A1: Obtain multidimensional geological environment data for the evaluation area, including topographic parameters, geological condition parameters, surface cover parameters, coal seam mining parameters, tower spatial location parameters, and temporal surface deformation parameters; Specifically, topographic parameters include elevation, slope, aspect, and topographic relief; surface cover parameters include vegetation cover and land use type; and coal seam mining parameters include the number of mining layers, mining thickness, mining depth, and mining depth-to-thickness ratio.

[0033] Specifically, the topographic parameters are obtained using a Digital Elevation Model (DEM) as the basic data source. In this embodiment, a 30-meter resolution DEM provided by the ALOS satellite is selected. The DEM stores the elevation information of each location on the Earth's surface in raster form, with each raster cell representing the average elevation value within a 30-meter × 30-meter area. Four topographic parameters are extracted based on the DEM: elevation is directly read from the elevation values ​​of each raster cell in the DEM. The elevation range of the evaluation area is from 800 meters to 2800 meters, divided into seven levels in 300-meter intervals: 800–1100 meters, 1100–1400 meters, 1400–1700 meters, 1700–2000 meters, 2000–2300 meters, 2300–2600 meters, and 2600–2800 meters. The topographic features and surface processes differ between different elevation zones. The slope parameter is calculated using a slope analysis algorithm from a geographic information system. This algorithm calculates the slope angle based on the elevation difference between the central grid cell and its eight adjacent grid cells. The calculation uses the ratio of the elevation difference in the direction of maximum descent to the horizontal distance, and then converts it into a slope angle using an arctangent function. The slope range is from 0 to 90 degrees, divided into nine levels in 10-degree intervals. The slope in gentle areas is in the range of 0-10 degrees, in mild areas it is 10-20 degrees, in sloping areas it is 20-30 degrees, in steep areas it is 30-40 degrees, in very steep areas it is 40-50 degrees, and in dangerous areas it exceeds 50 degrees. Aspect parameters indicate the orientation of the land surface. They are determined by calculating the maximum downward direction of the land surface, which is the projection direction of the slope normal vector onto the horizontal plane. This direction is expressed as an azimuth angle, measured clockwise from due north, ranging from 0 to 360 degrees. The 360-degree azimuth angle is divided into 9 directional categories: the horizontal category corresponds to flat areas with a slope of less than 1 degree and no clear slope aspect; the due north category corresponds to azimuth angles of 337.5 degrees to 22.5 degrees; the northeast category corresponds to 22.5 degrees to 67.5 degrees; the due east category corresponds to 67.5 degrees to 112.5 degrees; the southeast category corresponds to 112.5 degrees to 157.5 degrees; the due south category corresponds to 157.5 degrees to 202.5 degrees; the southwest category corresponds to 202.5 degrees to 247.5 degrees; the due west category corresponds to 247.5 degrees to 292.5 degrees; and the northwest category corresponds to 292.5 degrees to 337.5 degrees. The terrain relief parameter reflects the complexity of the local terrain. It is calculated using the moving window statistical method. A 3×3 grid window centered on the target grid is selected, and the window contains 9 grid cells. The maximum and minimum elevation values ​​within the window are calculated, and the difference between the two is the terrain relief value of the central grid. The terrain relief value is graded in 10-meter intervals: 0-10 meters is slight relief, 10-20 meters is small relief, 20-30 meters is medium relief, 30-50 meters is large relief, and more than 50 meters is strong relief.

[0034] Land cover parameters include vegetation cover and land use type. Vegetation cover is characterized by the Normalized Difference Vegetation Index (NDVI), which is calculated based on the strong absorption of red light by plant chlorophyll and the strong reflectance of near-infrared light by leaf structure, using the reflectance in the near-infrared and red light bands. The formula for calculating NDVI is: In the formula, NIR represents the reflectance in the near-infrared band, and Red represents the reflectance in the red band.

[0035] This embodiment uses imagery from the European Space Agency's Sentinel-2 multispectral satellite. This satellite is equipped with 13 spectral bands and a spatial resolution of 10–60 meters. The center wavelength of the red band B4 is 665 nm, and the center wavelength of the near-infrared band B8 is 842 nm, both with a spatial resolution of 10 meters. The NDVI calculation formula is: NDVI = near-infrared reflectance minus red reflectance, then divided by the sum of near-infrared and red reflectance. The NDVI value ranges from -1 to +1. NDVI values ​​for water bodies and bare land are close to or less than 0; NDVI values ​​for sparse vegetation are between 0.1 and 0.3; NDVI values ​​for medium-density vegetation are between 0.3 and 0.5; and NDVI values ​​for dense vegetation are greater than 0.5. In this embodiment, NDVI is divided into six levels according to five thresholds: 0.1, 0.2, 0.3, 0.4, and 0.5. A NDVI value less than 0.1 indicates no vegetation or water body; 0.1 to 0.2 indicates sparse vegetation; 0.2 to 0.3 indicates relatively sparse vegetation; 0.3 to 0.4 indicates moderate vegetation; 0.4 to 0.5 indicates good vegetation; and greater than 0.5 indicates dense vegetation. Land use type data uses the 2023 version of the land cover dataset CLCD. This dataset is generated based on Landsat satellite imagery using a machine learning classification algorithm, with a spatial resolution of 30 meters and a temporal resolution of annual updates. The classification system includes nine primary categories: cultivated land, forest land, grassland, shrubland, wetland, water body, permafrost and long-term snow cover, artificial impermeable surfaces, and bare land. Each land use type has different surface characteristics: cultivated land is used for agricultural planting, with vegetation cover that changes with the seasons and the surface is disturbed by human farming activities; forest land is covered by arbor forests, with deep root systems that help stabilize the soil; grassland is covered by herbaceous plants, with relatively stable vegetation cover; shrubland is covered by shrub vegetation, falling between grassland and forest land; wetland is a water-saturated or seasonally waterlogged area, with high soil moisture content and low carrying capacity; water bodies are rivers, lakes, and other bodies of water; artificial impermeable surfaces are artificial surfaces such as building sites and roads; bare land is a natural surface without vegetation cover.

[0036] Coal seam mining parameters are obtained from the mining engineering data of coal mining enterprises, including the number of mining layers, mining thickness, mining depth, and the depth-to-thickness ratio. The mining engineering data comes from the coal mine's mining engineering plan and mining design drawings, which indicate the location, thickness, mining range, and mining progress of each coal seam. The number of mining layers is used to evaluate the number of coal seams that have been mined or are currently being mined within the evaluation area. Mountainous coal mines typically have multiple coal seams, separated by rock strata. Mining proceeds sequentially from bottom to top or top to bottom. A higher number of mining layers means a wider vertical distribution of underground goaf areas and a greater disturbance to the overlying strata. The mining thickness M is directly taken for single coal seam mining, measured as the vertical distance from the top to the bottom of the coal seam. For multi-coal seam mining, the goaf areas of each coal seam are superimposed vertically, and the accumulated mining height affects the overall deformation of the overlying strata. Therefore, the cumulative value of the thicknesses of all mined coal seams is used as the total mining thickness. The mining depth H is the burial depth of the coal seam, the vertical distance from the surface to the roof of the coal seam. For multi-seam mining, the burial depth of the shallowest coal seam is used as the representative value of the mining depth. The mining depth reflects the distance that the mining impact travels to the surface. In deep mining, the mining impact gradually attenuates during the upward transmission process, and the surface deformation is relatively small. In shallow mining, the attenuation of the mining impact is less, and the surface deformation is larger. The mining depth-to-thickness ratio is calculated by dividing the mining depth H by the mining thickness M. This ratio is a key parameter in mining subsidence prediction. According to the probability integral method theory, when the mining depth-to-thickness ratio is less than the critical mining depth-to-thickness ratio, the surface subsidence reaches its maximum, indicating a state of full mining activity. The risk of surface collapse is highest during full mining activity. The value of the critical mining depth-to-thickness ratio is related to the lithology and structure of the overlying strata. The critical mining depth-to-thickness ratio is larger for hard rock strata and smaller for soft rock strata, generally between 30 and 50.

[0037] The spatial location parameters of the power transmission and distribution lines are extracted from the engineering archives, which record basic information about the lines and detailed data about the towers. The geographic coordinates of the towers are obtained through GPS measurements using the WGS84 geographic coordinate system, recording the longitude and latitude of each tower with meter-level accuracy. The tower coordinates are then converted to a projected coordinate system consistent with other spatial data. In this embodiment, the UTM projected coordinate system is used. After conversion, the tower coordinates are expressed in east and north coordinates, in meters, facilitating distance and area calculations within the geographic information system. Using the spatial analysis function of the geographic information system, the tower point features are overlaid with the goaf area features. The Euclidean distance from each tower to the nearest goaf boundary is calculated. The distance calculation uses the shortest distance algorithm from a point to a surface to find the point on the goaf boundary closest to the tower. The straight-line distance between the tower point and this boundary point is then calculated as the horizontal distance between the tower and the goaf boundary. This distance parameter reflects the spatial position of the tower relative to the goaf. According to the mining impact zoning theory, the area directly above and near the boundary of the goaf is the area of ​​strong mining impact, with large surface deformation. The mining impact gradually weakens in areas far from the goaf, with small surface deformation. Therefore, the distance between the tower and the boundary of the goaf is an important indicator for evaluating the risk level of the tower.

[0038] Temporal surface deformation parameters were obtained using interferometric synthetic aperture radar (InSAR) time-series analysis. Synthetic aperture radar (SAR) is an active microwave remote sensing sensor that transmits microwave signals to the Earth's surface and receives the backscattered echoes. It generates radar images by recording the amplitude and phase information of the echoes. Interferometric InSAR technology uses two radar images of the same area acquired at different times for interferometric processing. By calculating the phase difference between corresponding pixels in the two images, the line-of-sight (LOS) phase displacement of that pixel within the time interval between the two imaging events is obtained. The relationship between the phase difference and the LOS displacement is: LOS displacement equals radar wavelength multiplied by the phase difference and then divided by 4π. Based on traditional two-image interferometry, temporal InSAR technology utilizes multi-temporal radar image sequences to construct a time-series analysis model, enabling the acquisition of the temporal deformation history of each pixel and improving the accuracy and reliability of deformation monitoring. This embodiment uses the Small Baseline Set Temporal InSAR (SBAS-InSAR) technique. The basic idea of ​​this technique is to combine multi-temporal radar images into multiple short-time interval, small-space-baseline interferometric pairs according to temporal and spatial baseline constraints. Each interferometric pair generates an interferogram, which records the surface deformation information within the time interval. By establishing a linear equation set between the phase of the interferogram and the deformation of the time series, the cumulative deformation at each moment in the time series is solved using the least squares method or singular value decomposition method.

[0039] This embodiment uses C-band synthetic aperture radar data from the ESA's Sentinel-1 satellites. Sentinel-1 is a dual-satellite network consisting of Sentinel-1A and Sentinel-1B, with a revisit period of 6 days and a single-satellite revisit period of 12 days. The radar band is C-band, with a center frequency of 5.405 GHz and a corresponding wavelength of 5.55 cm. The imaging mode is interferometric wide swath (IW), which has a swath width of 250 km and a spatial resolution of 5 m × 20 m. After multi-look processing, the spatial resolution for deformation monitoring is approximately 30 m × 30 m. The radar data is in single-look complex SLC format, containing amplitude and phase information. Sentinel-1 radar data from January 2021 to December 2023, spanning three years, was selected for the evaluation area. A total of 60 images were acquired in the descent direction, with a sampling interval of approximately 18 days. The SBAS-InSAR processing workflow includes the following steps: First, radar image registration is performed. An image from the midpoint of the time series is selected as the master image, and other images are registered to the master image's coordinate system with a registration accuracy of 0.1 pixels, ensuring that pixels with the same name in images from different times correspond to the same location on the ground. Second, interferometric pair selection is performed. Based on constraints of a time baseline of less than 180 days and a spatial baseline of less than 200 meters, 150 interferometric pairs are generated from 60 images. The time interval between each interferometric pair is short to reduce the impact of temporal uncorrelation, and the spatial baseline is small to reduce the impact of spatial uncorrelation and terrain phase. Third, interferogram generation is performed. The master and slave images of each interferometric pair are multiplied by conjugate to obtain the interferometric phase. The interferometric phase includes four parts: terrain phase, deformation phase, atmospheric phase, and noise phase. The terrain phase is removed using a digital elevation model, and the atmospheric and noise phases are suppressed using spatiotemporal filtering methods. The deformation phase is extracted. The fourth step is phase unwrapping. The interferometric phase is the principal phase, ranging from negative π to positive π. The true phase value needs to be recovered using a phase unwrapping algorithm; this embodiment uses the minimum cost flow phase unwrapping algorithm. The fifth step is to establish a time-series inversion model. The cumulative deformation of each pixel at each time step is set as an unknown, and the unwrapped phase of each interferogram is set as the observed value. A linear relationship between the observed value and the unknown is established based on the master-slave image times of the interferometric pair, forming an overdetermined system of equations. The singular value decomposition (SVD) method is used to solve this system of equations, obtaining the cumulative deformation of each pixel at each time step relative to the first time step in the time series. The sixth step is to calculate the annual average deformation rate. A linear fit is performed on the cumulative deformation of each pixel in the time series. The slope of the fitted line is the annual average deformation rate, expressed in millimeters per year.

[0040] A2: The labeled collapse sample data includes collapsed sample points labeled according to time-series surface deformation parameters in areas where mining-induced collapse has occurred, and non-collapse sample points labeled in stable areas.

[0041] Specifically, see Table 1 for data retrieval details:

[0042] The annotation of subsidence sample points, i.e., positive samples, comprehensively utilizes three information sources: time-series surface deformation monitoring results, coal mine boundary boundaries, and on-site geological hazard investigation results. The time-series surface deformation monitoring results are derived from the cumulative vertical displacement field and annual average vertical displacement rate field generated by SBAS-InSAR technology. These two products visually demonstrate the spatial distribution and deformation rate of surface subsidence. The deformation threshold for identifying subsidence areas is set as follows: the absolute value of the annual average vertical displacement rate is greater than 60 mm per year, or the absolute value of the cumulative vertical displacement over 3 years is greater than 180 mm. This threshold is based on the following considerations: the surface deformation rate caused by normal tectonic geological activity is generally less than 10 mm per year, while the surface subsidence rate caused by mining subsidence is usually from tens to hundreds of millimeters per year. Setting a threshold of 60 mm per year can distinguish between mining subsidence deformation and background deformation. In the geographic information system, the annual average vertical displacement rate field is segmented by threshold, and raster cells with an absolute rate value greater than 60 mm per year are extracted. These raster cells constitute preliminary suspected subsidence areas. The boundary of a coal mine is obtained from the attached map of the mining license of the coal mining enterprise. The boundary is the boundary of the legal mining area of ​​the coal mine. Goaf areas are generally located within or near the boundary. By spatially superimposing the suspected subsidence area with the boundary, suspected subsidence areas located within or within a 500-meter buffer zone outside the boundary are screened out, and non-subsidence surface deformations far away from the mining area are further excluded.

[0043] The investigation included: observing the presence of subsidence pits on the surface. Subsidence pits are closed depressions formed by the sudden subsidence of the ground above the mined-out area. They are mostly elliptical or irregular in shape, ranging in depth from a few meters to tens of meters, with steep edges. The investigation also included observing the presence of ground fissures on the surface. Ground fissures are tensile cracks caused by surface stretching deformation. The width of these fissures ranges from a few centimeters to a few meters, and their length from a few meters to hundreds of meters. The direction of these fissures is related to the boundary of the mined-out area or the slope of the terrain. The investigation also included observing the presence of stepped elevations on the surface. Steps are steep slopes formed by differential surface subsidence, with height differences ranging from tens of centimeters to several meters. The investigation also included observing whether surface buildings showed signs of deformation or damage, such as cracked walls, foundation subsidence, or tilting. Finally, the investigation included observing whether power transmission and distribution towers were tilted, using laser rangefinders and levels to measure the tilt, and observing whether conductors exhibited abnormal sag or loose joints. Areas within suspected subsidence zones where one or more of the above geological hazards were found were confirmed as areas where mined-out subsidence had occurred and were designated as positive sample areas. This embodiment investigated and evaluated 12 coal mines within the region. Through on-site investigation and verification, it was confirmed that some areas of 8 coal mines had experienced mining subsidence, with a total confirmed subsidence area of ​​approximately 15 square kilometers.

[0044] Sample points were extracted within the positive sample annotation area using random sampling to ensure the representativeness of the spatial distribution of samples within the collapse area. First, the positive sample annotation area was rasterized, with a raster size consistent with the preprocessed data layer set (30m x 30m). The total number of raster cells within the positive sample annotation area was approximately 16,667. From these raster cells, 1200 raster cells were randomly selected as positive sample points using a random number generator, representing a sampling ratio of approximately 7.2%. This sampling ratio ensured sufficient sample size for training the machine learning model while avoiding excessive computational burden due to an excessive number of samples. Each extracted positive sample point was assigned a hazard category label of "1," indicating that a mining subsidence had occurred at that location, classifying it as a high-risk area. The attribute values ​​of the raster cell containing the positive sample point were extracted from the 12 evaluation factor layers to form the feature vector of that sample point. The feature vector is a 12-dimensional vector. The first four dimensions are the standardized values ​​of topographic factors, the fifth dimension is the uniquely encoded vector of stratigraphic lithology factors, the sixth and seventh dimensions are the values ​​and codes of external environmental factors, the eighth to eleventh dimensions are the standardized values ​​of underground mining factors, and the twelfth dimension is the standardized value of tower location relationship factors. The feature vectors and class labels of 1200 positive sample points are used to form a positive sample dataset.

[0045] Non-collapse sample points, i.e., negative samples, were labeled within stable regions, which are areas where no significant goaf collapse has occurred. The identification of stable regions was based on time-series surface deformation monitoring results and coal mine distribution information. The identification criteria were: an absolute value of the annual average vertical displacement rate less than 10 mm per year, an absolute value of the cumulative vertical displacement over 3 years less than 30 mm, and a distance greater than 500 meters from the nearest goaf boundary. An absolute value of the annual average vertical displacement rate less than 10 mm per year indicates that surface deformation is at the background deformation level; an absolute value of the cumulative vertical displacement less than 30 mm indicates that the overall deformation over 3 years is very small; and areas more than 500 meters from the goaf boundary are unaffected by mining. These three conditions together ensure the stability of the negative sample areas. In the geographic information system, the annual average vertical displacement rate field is segmented by thresholding, and grid cells with an absolute rate value of less than 10 mm per year are extracted. The surface features of the goaf area are buffered by an extension of 500 meters, and the grid cells in the buffer are excluded. The remaining grid cells constitute the stable region. The total number of grid cells contained in the stable region is approximately 400,000. The area of ​​the stable region is much larger than the area of ​​the collapse region, which is consistent with the actual situation.

[0046] Negative sample points were randomly generated within the stable region using a uniformly distributed random number generator. 4000 grid cells were randomly selected from the grid cell set of the stable region as negative sample points. The number of negative samples was set to 4000, with a positive-to-negative sample ratio of 1:3.3. This ratio was set considering two factors: first, stable regions dominate in actual spatial distribution, and a larger number of negative samples than positive samples reflects reality; second, machine learning models trained on imbalanced datasets are prone to biased predictions towards the majority class, and controlling the number of negative samples prevents the positive-to-negative sample ratio from becoming too disparate, which is beneficial for the model to learn the feature differences between the two types of regions. Each negative sample point was assigned a hazard category label of "0," indicating that no mining subsidence had occurred at that location, classifying it as a low-risk area. The attribute values ​​of the grid cells containing the negative sample points were extracted on 12 evaluation factor layers to form a 12-dimensional feature vector for the negative sample points. The feature vectors and category labels of the 4000 negative sample points were combined to form the negative sample dataset.

[0047] The positive and negative sample datasets were merged to form a complete labeled sample dataset with a total of 5200 samples, of which 1200 are positive (23.1%) and 4000 are negative (76.9%). Each sample in the dataset contains a 12-dimensional feature vector and a class label. The feature vector reflects the geological environment characteristics of the sample's spatial location, and the class label indicates whether mining subsidence has occurred at that location. The dataset was randomly divided into training and test sets in a 7:3 ratio using stratified random sampling to ensure that the ratio of positive to negative samples in the training and test sets remained consistent with the overall sample ratio. Specifically, the training set contains 3640 samples, of which 840 are positive (23.1%) and 2800 are negative (76.9%); the test set contains 1560 samples, of which 360 are positive (23.1%) and 1200 are negative (76.9%). The training set is used for parameter learning and optimization of the machine learning model, while the test set is used for independent verification of the model's generalization ability. The test set is not involved in the model training process at all, to avoid optimistic bias in the model evaluation results.

[0048] like Figure 3 The number of misclassified sample points in the GWO-RF model is 60+88, and the number of misclassified sample points in the MLP-RF model is 76+48, which is significantly smaller than that in the MLP model and the RF model. The confusion matrix of the MLP model (top left) and the confusion matrix of the RF model (top right) are shown. The confusion matrix of the GWO-RF model (bottom left) and the confusion matrix of the MLP-RF model (bottom right) are shown. Figure 4 The AUC values ​​of GWO-RF and MLP-RF were 0.95 and 0.96, respectively, exceeding those of the MLP model and the RF model. The AUC value of the MLP-RF model reached 0.96, significantly outperforming the single model.

[0049] In this embodiment, step S200 involves obtaining the cumulative vertical displacement value at the tower location, combining it with the tower damage status samples from the on-site investigation, establishing a probability distribution relationship between the cumulative vertical displacement value and the tower damage level, calculating the equivalent damage probability of the tower under different cumulative vertical displacement values, generating a damage probability curve based on the equivalent damage probability, substituting the cumulative vertical displacement value of each tower into the damage probability curve to calculate the damage probability value, and classifying the vulnerability level based on the damage probability value, including the following steps B1-B2: B1: Establishing the probability distribution relationship between cumulative vertical displacement value and tower damage level includes: determining the tower damage level corresponding to different cumulative vertical displacement value intervals based on field investigation, using a log-normal distribution function to describe the probability of occurrence of each damage level, and weighted summing of the probability of occurrence of each damage level to obtain the equivalent damage probability.

[0050] Specifically, the classification of tower damage levels is based on the relevant provisions regarding tower deformation in the State Grid's industry standards, "Design Code for Overhead Transmission Lines in Mining-Affected Areas" and "Operation Regulations for Overhead Transmission Lines." Mining subsidence leads to uneven settlement of the tower foundation, which in turn causes tower tilting. Tower tilt is the core indicator for judging the degree of tower damage. Tower tilt is defined as the ratio of the vertical displacement difference at the tower's center point to the horizontal distance from the tower foundation; a larger tilt indicates a more severe tilt. The formula for calculating tower tilt is: In the formula, I is the tower tilt, Δd is the vertical settlement difference at the tower center point, and l is the horizontal distance from the tower foundation.

[0051] This embodiment classifies tower damage into four levels, from C0 to C3, corresponding to no significant damage, minor damage, moderate damage, and severe damage, respectively. Level C0 (no significant damage) corresponds to a tower tilt of less than or equal to 1%. At this level, the tower remains essentially vertical, with no significant deformation or cracking on the ground around the base, and the tower is safe to operate. Level C1 (minor damage) corresponds to a tower tilt between 1% and 1.5%. The tower tilts slightly, and a few small cracks and shallow subsidence appear on the ground around the base. The impact on tower stability is minimal and can be monitored through routine inspections. Level C2 (moderate damage) corresponds to a tower tilt between 1.5% and 2.0%. The tower tilts more significantly, with a large area of ​​subsidence and multiple tensile cracks around the base. However, the cracks do not directly threaten the tower foundation, and the tower can still maintain operation. Enhanced monitoring and the development of emergency plans are necessary. C3 level severe damage corresponds to a tower tilt greater than 2.0%. The tower is severely tilted, and large-scale landslides, sinkholes, and wide tensile cracks appear around the tower base, indicating surface deformation. The surface deformation directly threatens the stability of the tower foundation and poses a risk of tower collapse. Immediate reinforcement measures or relocation of the line are required.

[0052] The tower tilt is calculated using cumulative vertical displacement values. This calculation requires obtaining the cumulative vertical displacement values ​​at the tower location and the horizontal distance from the tower foundation. The cumulative vertical displacement values ​​at the tower location are extracted from the cumulative vertical displacement field generated by SBAS-InSAR time-series analysis. Based on the tower's coordinates, the corresponding grid cell is located within the cumulative vertical displacement field, and the cumulative vertical displacement value of that grid cell is read as the cumulative ground settlement at the tower location. It should be noted that in this embodiment, the cumulative vertical displacement field grid size is 30m × 30m. The grid cell corresponding to the tower location actually covers an area of ​​approximately 30m × 30m centered on the tower. The cumulative vertical displacement value of the grid cell is the average deformation of multiple radar-scattering targets within this area, representing the settlement status of the tower foundation and its surrounding surface. The horizontal distance from the tower foundation refers to the span of the tower foundation base. For steel pipe towers, the foundation is a single pile foundation, and the horizontal distance is taken as 0. For lattice-type iron towers, the foundation is a separate foundation, with each tower leg corresponding to an independent foundation. The four foundations are arranged in a square or rectangle, and the horizontal distance is the distance between the diagonal foundations. The towers evaluated in this embodiment are mainly lattice-type iron towers. The horizontal distance between the foundations varies from 4 to 8 meters depending on the tower type. To simplify the calculation, a uniform horizontal distance of 30 meters (the size of a lattice unit) is used. This means that the foundations at both ends of the tower are assumed to be located at the centers of two adjacent lattice units. The cumulative vertical displacement difference between the two lattice units represents the uneven settlement of the tower foundation. The tower tilt is calculated by dividing the cumulative vertical displacement difference by 30 meters and then multiplying by 100% to convert it to a percentage. For example, if the cumulative vertical displacement at a certain tower location is -60 mm and the cumulative vertical displacement of an adjacent lattice unit is -30 mm, the displacement difference is 30 mm (0.03 meters). The tilt is 0.03 meters divided by 30 meters, which equals 0.001 (0.1%), corresponding to a C0 level of no significant damage.

[0053] A field survey was conducted to obtain accurate data on the damage status of power transmission and distribution lines within the goaf areas of 12 coal mines in Liupanshui and Bijie, Guizhou Province. The survey covered 93 power transmission and distribution lines that traversed or were located near goaf areas. The survey period was from October to December 2023, corresponding to the end of the SBAS-InSAR time series analysis window. Survey methods included: measuring the tilt angle of the power transmission and distribution lines using a total station with an accuracy down to the arcsecond level; measuring the relative height difference between the foundations of the power transmission and distribution lines using a laser rangefinder; using drones to take aerial photographs of the power transmission and distribution lines and surrounding ground surfaces, and identifying ground deformation phenomena through image interpretation; and interviewing line maintenance personnel to understand the historical deformation of the power transmission and distribution lines and their inspection records. For each power transmission and distribution line, the damage level was determined based on the combined tilt measurement results and ground deformation phenomena, according to the aforementioned four damage level standards. Information such as the power transmission and distribution line number, coordinates, measured tilt, determined damage level, and description of surrounding ground deformation was recorded. Simultaneously, the SBAS-InSAR cumulative vertical displacement value of each tower location was extracted, and a dataset corresponding to the cumulative vertical displacement value and damage level was established. The survey results showed that among the 93 towers, 64 were classified as C0 with no obvious damage, 13 as C1 with slight damage, 12 as C2 with moderate damage, and 4 as C3 with severe damage. The distribution of damage levels showed a trend of increasing level with increasing cumulative vertical displacement value.

[0054] The log-normal distribution function is used to describe the probability of each damage level. The log-normal distribution function is suitable for describing the probability distribution of results generated by the multiplication of multiple random factors. The process of tower damage caused by surface subsidence is affected by various factors such as geological conditions, mining intensity, tower foundation type, and construction quality. The combined effect of these factors makes the tower damage probability exhibit a log-normal distribution characteristic. The probability density function of the log-normal distribution is: probability density equals 1 divided by the cumulative vertical displacement multiplied by the standard deviation parameter multiplied by the square root of 2π, then multiplied by an exponential function, where the exponent is the negative log-cumulative vertical displacement minus the square of the log-median parameter, divided by twice the square of the standard deviation parameter. Integrating the probability density function yields the cumulative distribution function, which represents the probability that the tower damage reaches a certain level when the cumulative vertical displacement is less than or equal to a certain value. In this embodiment, the cumulative distribution function is used to describe the probability that the tower damage level reaches or exceeds a certain level, denoted as P. The expression for P is the cumulative distribution function of the standard normal distribution, with the independent variable being the log-cumulative vertical displacement minus the log-median parameter, divided by the standard deviation parameter. The cumulative distribution function of the log-normal distribution is expressed as: In the formula, P represents the damage level that reaches or exceeds the cumulative settlement Δ. The probability of , where Φ is the standard normal cumulative distribution function. Damage level The corresponding median cumulative settlement, where β is the standard deviation parameter.

[0055] The cumulative distribution function of the standard normal distribution is obtained by looking up a table or by numerical calculation.

[0056] The log-median parameter and standard deviation parameter are two parameters of the log-normal distribution function, which need to be estimated based on field survey data. Maximum likelihood estimation is used for parameter estimation, the basic idea of ​​which is to find the parameter value that maximizes the probability of the observed data. For C1 level minor damage, the cumulative vertical displacement values ​​of towers classified as C1 in the statistical survey data show that the cumulative vertical displacement values ​​are mainly concentrated in the range of 15 mm to 35 mm, with a median of approximately 25 mm. The standard deviation parameter was determined to be 0.15 through a goodness-of-fit test. The log-median parameter is taken as the logarithm of 25 mm, approximately 3.22, while the standard deviation parameter remains unchanged at 0.15. The probability distribution function for C1 level damage is a standard normal cumulative distribution function, with the independent variable being the log-cumulative vertical displacement minus 3.22 and then divided by 0.15. For C2 level moderate damage, the cumulative vertical displacement of C2 level towers mainly falls within the range of 35 mm to 55 mm, with a median of approximately 44.2 mm. The logarithmic median parameter is approximately 3.79 (logarithm 44.2), and the standard deviation parameter is 0.10. For C3 level severe damage, the cumulative vertical displacement of C3 level towers mainly falls within the range of 60 mm to 90 mm, with a median of approximately 72 mm. The logarithmic median parameter is approximately 4.28 (logarithm 72), and the standard deviation parameter is 0.10. (Reference) Figure 5 The graph displays the probability distribution curves for three damage levels. The horizontal axis represents the cumulative vertical displacement value, and the vertical axis represents the damage probability. The three curves correspond to levels C1, C2, and C3, respectively. The curves are S-shaped, with the probability increasing most rapidly near the median value. The probability is lower when the cumulative vertical displacement value is less than the median and higher when it is greater than the median. As shown in the graph, when the cumulative vertical displacement value is 25 mm, the probability of level C1 damage is approximately 50%, and the probability of level C2 damage is close to 0%. When the cumulative vertical displacement value is 50 mm, the probability of level C1 damage is close to 100%, the probability of level C2 damage is approximately 50%, and the probability of level C3 damage is approximately 10%. When the cumulative vertical displacement value is 72 mm, the probability of level C3 damage is approximately 50%.

[0057] The equivalent damage probability is obtained by weighted summation of the probabilities of occurrence for each damage level. This equivalent damage probability comprehensively reflects the overall damage severity of the tower under different cumulative vertical displacement values. The formula for the weighted summation is: In the formula, The equivalent damage probability, Let i be the probability of the tower experiencing Class i damage. This represents the median cumulative settlement corresponding to damage level i.

[0058] The equivalent damage probability is equal to the C1 damage probability multiplied by the median cumulative vertical displacement corresponding to C1, plus the C2 damage probability multiplied by the median cumulative vertical displacement corresponding to C2, plus the C3 damage probability multiplied by the median cumulative vertical displacement corresponding to C3, and then divided by the sum of the three median cumulative vertical displacements. The weighting coefficients in the formula use the median cumulative vertical displacement corresponding to each level, reflecting the magnitude of settlement corresponding to different levels of damage. The larger the settlement, the greater the weight in the equivalent damage probability. The median cumulative vertical displacement is 25 mm for C1, 44.2 mm for C2, and 72 mm for C3. After normalization, the weighting coefficients are approximately 0.177 for 25 divided by 141.2, approximately 0.313 for 44.2 divided by 141.2, and approximately 0.510 for 72 divided by 141.2. Calculation Example: For a tower with a cumulative vertical displacement of 34.95 mm, substituting the probability distribution functions of the three damage levels, we calculate the probability of occurrence for each level. The probability of damage at level C1 is approximately 0.987 (logarithm 34.95 minus 3.22, divided by 0.15). The probability of damage at level C2 is approximately 0.009 (logarithm 34.95 minus 3.79, divided by 0.10). The probability of damage at level C3 is approximately 0 (logarithm 34.95 minus 4.28, divided by 0.10). The equivalent damage probability is calculated as 0.987 multiplied by 25, plus 0.009 multiplied by 44.2, plus 0 multiplied by 72, resulting in 24.675. Adding 0.398 gives 25.073, which, divided by 141.2, yields 0.1776. The equivalent damage probability ranges from 0 to 1, where 0 represents no damage, 1 represents complete damage, and 0.1776 indicates that the tower is at a low to moderate level of damage, which is basically consistent with the on-site investigation that determined the tower to be a C1 level minor damage.

[0059] B2: The damage probability curve is fitted with a hyperbolic tangent function to establish the nonlinear relationship between the cumulative vertical displacement value and the equivalent damage probability; a nonlinear fitting method is used to establish a continuous functional relationship between the cumulative deformation and the equivalent damage probability, which serves as a quantitative basis for vulnerability evaluation.

[0060] Specifically, the equivalent damage probability was calculated for each of the 93 surveyed towers, forming a data pair set of cumulative vertical displacement values ​​and equivalent damage probabilities. Each data pair contained 93 sample points, with the x-axis representing the absolute value of the cumulative vertical displacement and the y-axis representing the equivalent damage probability. Observation of the data point distribution revealed that as the cumulative vertical displacement value increased, the equivalent damage probability showed a trend of first slowly increasing, then rapidly increasing, and finally gradually saturating. This trend conforms to the morphological characteristics of the hyperbolic tangent function. The hyperbolic tangent function tanh is an S-shaped curve. As the independent variable changes from negative infinity to positive infinity, the function value monotonically increases from -1 to positive 1, with the fastest growth near the independent variable of 0. By performing a linear transformation on the hyperbolic tangent function, different S-shaped curves can be fitted. The fitting function form used in this embodiment is: the equivalent damage probability equals parameter a plus parameter b multiplied by the hyperbolic tangent function, and the independent variable of the hyperbolic tangent function is parameter c multiplied by the cumulative vertical displacement plus parameter d. The expression of the hyperbolic tangent fitting function is: In the formula, Let Δ be the equivalent damage probability under the cumulative settlement, a, b, c, and d be the fitting parameters, and tanh be the hyperbolic tangent function.

[0061] The function contains four undetermined parameters: parameter a controls the vertical offset of the function, parameter b controls the vertical amplitude of the function, parameter c controls the horizontal scaling (i.e., the growth rate) of the function, and parameter d controls the horizontal offset (i.e., the inflection point position) of the function.

[0062] The fitting method employs nonlinear least squares, which determines parameter values ​​by minimizing the sum of squared residuals between the fitted function values ​​and the observed values. The sum of squared residuals is defined as the sum of the squares of the fitted function values ​​minus the squares of the observed values ​​for all sample points, denoted as RSS. Minimizing the sum of squared residuals is a nonlinear optimization problem solved using an iterative algorithm. This embodiment uses the Levenberg-Marquardt algorithm, which combines the advantages of the Gauss-Newton method and gradient descent, exhibiting fast convergence and good stability. During algorithm initialization, initial values ​​for the parameters are given: parameter a is set to the average value of the observed equivalent damage probability (approximately 0.5); parameter b is set to the standard deviation of the observed equivalent damage probability (approximately 0.3); parameter c is set to 0.03; and parameter d is set to -2. During algorithm iteration, the partial derivatives of the sum of squared residuals with respect to each parameter are calculated, a Jacobian matrix is ​​constructed, and the parameter values ​​are updated based on the Jacobian matrix. This iteration is repeated until the sum of squared residuals converges or the maximum number of iterations is reached. This embodiment converges after 100 iterations, and the optimal parameter values ​​are: parameter a = 0.53, parameter b = 0.919, parameter c = 0.036, and parameter d = -1.95.

[0063] Goodness of fit was evaluated using the coefficient of determination (R²), which is defined as 1 minus the sum of squared residuals divided by the sum of squared total deviations. The formula for calculating the coefficient of determination is: In the formula, The coefficient of determination is RSS, the sum of squared residuals is TSS, the sum of squared total deviations is TSS, and yi is the observed value. These are the fitted values. is the mean of the observed values, and n is the sample size.

[0064] The total sum of squares of deviations is the sum of the squares of the observed values ​​minus the mean of all observed values, reflecting the overall dispersion of the observed data. The coefficient of determination ranges from 0 to 1; the closer to 1, the better the fit. In this embodiment, the coefficient of determination R² is 0.989, indicating that the fitting function can explain 98.9% of the variation in the observed data, and the fitting effect is very good. Substituting the parameter values ​​into the fitting function, the explicit expression of the damage probability curve is obtained as follows: the equivalent damage probability equals 0.53 plus 0.919 multiplied by the hyperbolic tangent function (0.036 multiplied within parentheses) minus the cumulative vertical displacement minus 1.95. (Refer to...) Figure 6 The damage probability curve and 93 observation data points are displayed. The horizontal axis represents the cumulative vertical displacement value in millimeters, and the vertical axis represents the equivalent damage probability, ranging from 0 to 1. The blue dots represent the observation data points, and the red solid line represents the fitted damage probability curve. The observation points are evenly distributed on both sides of the fitted curve, and the fitted curve describes the overall trend of the observation data well.

[0065] Analysis of the damage probability curve revealed the following characteristics: When the cumulative vertical displacement is less than 20 mm, the equivalent damage probability is less than 0.1, the curve is relatively flat, and the damage probability increases slowly; when the cumulative vertical displacement is between 30 mm and 70 mm, the curve enters a rapid growth phase, with the damage probability increasing from approximately 0.2 to approximately 0.9. This range is the critical stage for the rapid deterioration of the tower's damage condition; when the cumulative vertical displacement is greater than 80 mm, the equivalent damage probability approaches 1, the curve tends to saturate, indicating that almost all towers have entered a state of severe damage. The warning thresholds are determined based on the damage probability curve: a blue warning is issued when the cumulative vertical displacement value is between 10 mm and 25 mm, with an equivalent damage probability of approximately 0.05 to 0.15. The tower is in the transition stage from no obvious damage to slight damage, requiring enhanced monitoring. A yellow warning is issued when the cumulative vertical displacement value is between 25 mm and 45 mm, with an equivalent damage probability of approximately 0.15 to 0.45. The tower is progressing from slight damage to moderate damage, requiring the development of an emergency plan. An orange warning is issued when the cumulative vertical displacement value is between 45 mm and 72 mm, with an equivalent damage probability of approximately 0.45 to 0.7. The tower is in the stage of moderate to severe damage, requiring reinforcement measures. A red warning is issued when the cumulative vertical displacement value is greater than or equal to 72 mm, with an equivalent damage probability greater than 0.7. The tower has entered a state of severe damage, posing a risk of collapse, requiring immediate emergency response measures.

[0066] Damage probability curves were applied to assess the vulnerability of all towers within the evaluation area. For each tower, the cumulative vertical displacement value of the tower location was extracted from the SBAS-InSAR cumulative vertical displacement field. This cumulative vertical displacement value was then substituted into the expression of the damage probability curve to calculate the equivalent damage probability value of the tower. The equivalent damage probability value serves as a quantitative indicator of tower vulnerability; a higher value indicates a greater susceptibility to damage under the current surface subsidence conditions. Based on the equivalent damage probability value, tower vulnerability was classified into five levels: extremely low vulnerability (equivalent damage probability less than 0.1), low vulnerability (equivalent damage probability 0.1 to 0.3), moderate vulnerability (equivalent damage probability 0.3 to 0.5), high vulnerability (equivalent damage probability 0.5 to 0.7), and extremely high vulnerability (equivalent damage probability greater than 0.7). The vulnerability level classification employed the natural breakpoint method, which determines the classification threshold based on the natural grouping characteristics of the data, minimizing intra-group differences and maximizing inter-group differences. In this embodiment, a total of 93 towers were evaluated within the area. The distribution of vulnerability levels for each tower was as follows: extremely low vulnerability (52 towers, 55.9%), low vulnerability (18 towers, 19.4%), medium vulnerability (15 towers, 16.1%), high vulnerability (6 towers, 6.5%), and extremely high vulnerability (2 towers, 2.2%). The vulnerability level distribution resembled a pyramid, with most towers at extremely low and low vulnerability, and a few at high and extremely high vulnerability. This is consistent with the actual situation, indicating that the vulnerability evaluation method is reasonable.

[0067] In this embodiment, step S300 involves constructing a coupling matrix between hazard level and vulnerability level, querying the hazard level of the spatial grid cell where each tower is located and the vulnerability level of the corresponding tower, determining the comprehensive risk level of each tower based on the coupling matrix, and generating a tower risk distribution map based on the comprehensive risk level of each tower. This includes the following steps C1-C2: C1: Constructing the coupling matrix between hazard level and vulnerability level includes setting the classification criteria for hazard level and vulnerability level; Based on the combination relationship between hazard level and vulnerability level, the rules for determining the comprehensive risk level are defined; Using the hazard level and vulnerability level of each tower as input, the corresponding comprehensive risk level is determined through matrix lookup.

[0068] Specifically, the hazard level classification criteria are determined based on the hazard index output by the hazard prediction model. The multilayer perceptron-random forest hybrid model trained in step S100 is applied to the entire evaluation area to predict the hazard index of each 30m × 30m grid cell within the evaluation area. The hazard index ranges from 0 to 1, with a higher value indicating a higher probability of mining subsidence at that location. The hazard indices of all grid cells within the evaluation area are divided into five levels according to the natural breakpoint method: extremely low hazard corresponds to a hazard index less than 0.2, low hazard corresponds to a hazard index of 0.2 to 0.4, medium hazard corresponds to a hazard index of 0.4 to 0.6, high hazard corresponds to a hazard index of 0.6 to 0.8, and extremely high hazard corresponds to a hazard index greater than 0.8. The hazard zoning map is generated in the geographic information system. Each raster cell is assigned a different color according to its hazard level: dark green for extremely low hazard, light green for low hazard, yellow for medium hazard, orange for high hazard, and red for extremely high hazard. The shade of the color directly reflects the level of hazard. The hazard level distribution of the statistical evaluation area is as follows: extremely low hazard areas account for 56.04%, low hazard areas account for 21.99%, medium hazard areas account for 11.45%, high hazard areas account for 7.40%, and extremely high hazard areas account for 3.06%. This hazard level distribution conforms to the spatial distribution pattern of actual mining subsidence, which is mainly concentrated in and around coal mining areas, occupying a relatively small area.

[0069] The grading criteria for vulnerability levels have been determined in step S200. Based on the equivalent damage probability value of the tower, the vulnerability of the tower is divided into five levels: extremely low vulnerability, low vulnerability, moderate vulnerability, high vulnerability, and extremely high vulnerability. The equivalent damage probability thresholds corresponding to each level are 0.1, 0.3, 0.5, and 0.7, respectively. The vulnerability level reflects the damage risk of the tower under the current surface subsidence state; the higher the level, the more fragile the tower is and the more easily it is damaged.

[0070] The rules for determining the comprehensive risk level are based on the definition of risk, which is defined as the product of the probability of a disaster occurring and the severity of its consequences. The basic relationship for risk assessment is: In the formula, R represents the comprehensive risk level, D represents the hazard level (the probability of mining subsidence), and S represents the vulnerability level (the severity of tower damage).

[0071] In this technical solution, hazard represents the probability of a mining subsidence collapse, and vulnerability represents the severity of tower damage. The combination of these two factors determines the overall risk faced by the tower. A 5×5 coupling matrix is ​​constructed, with rows corresponding to five hazard levels from extremely low to extremely high, and columns corresponding to five vulnerability levels from extremely low to extremely high. Each cell in the matrix represents a combination of hazard and vulnerability, and the corresponding overall risk level is entered within the cell. The overall risk level is also divided into five levels: extremely low risk, low risk, medium risk, high risk, and extremely high risk. The judgment rules follow these principles: when both hazard and vulnerability are extremely low, the overall risk is extremely low; when either hazard or vulnerability is extremely high, the overall risk is at least high; when both hazard and vulnerability are extremely high, the overall risk is extremely high; for other combinations, the overall risk is determined by combining the levels of both hazard and vulnerability. Generally, the higher the hazard and vulnerability levels, the higher the overall risk level. The specific coupling matrix is ​​shown in Table 2 below. Table 2 Risk Matrix

[0072] When both the hazard level and vulnerability level are extremely high, both are at the highest level, and the overall risk assessment is extremely high, indicating that the tower faces a very serious risk and requires immediate emergency measures. When the hazard level is extremely high but the vulnerability level is high, medium, low, or extremely low, the overall risk assessment is extremely high, high, medium, and low respectively. Extremely high hazard means that the possibility of mining subsidence is very high, and even if the tower's vulnerability level is low, there is still a high risk. When the hazard level is high, the combination with extremely high vulnerability level results in extremely high risk; the combination with high vulnerability level results in high risk; the combination with medium vulnerability level results in medium risk; the combination with low vulnerability level results in low risk; and the combination with extremely low vulnerability level results in extremely low risk. When the hazard level is medium, the combination with extremely high vulnerability level results in extremely high risk; the combination with high vulnerability level results in high risk; the combination with medium vulnerability level results in medium risk; the combination with low vulnerability level results in low risk; and the combination with extremely low vulnerability level results in extremely low risk. When the risk level is low, combining it with extremely high vulnerability results in high risk; combining it with high vulnerability results in medium risk; combining it with medium vulnerability results in low risk; and combining it with low vulnerability and extremely low vulnerability results in extremely low risk. When the risk level is extremely low, combining it with extremely high vulnerability results in high risk; combining it with high vulnerability results in medium risk; combining it with medium vulnerability results in low risk; and combining it with low vulnerability and extremely low vulnerability results in extremely low risk.

[0073] C2: Also includes: setting graded early warning settings based on the cumulative vertical displacement value of each tower, and triggering the corresponding level of risk warning signal when the cumulative vertical displacement value of the tower reaches the early warning setting value.

[0074] Specifically, a comprehensive risk assessment is conducted on each tower within the evaluation area. The assessment process includes three steps: First, the hazard level of the spatial grid cell containing the tower is determined. Based on the tower's coordinates, the corresponding grid cell is located on the hazard zoning map, and its hazard level is read. The hazard level ranges from one of five levels, from extremely low to extremely high. Second, the vulnerability level of the tower is determined. The equivalent damage probability value and corresponding vulnerability level of the tower are retrieved from the vulnerability assessment results in step S200. The vulnerability level is also one of five levels, from extremely low to extremely high. Third, the comprehensive risk level is determined by querying the coupling matrix. Using the hazard level as the row index and the vulnerability level as the column index, the corresponding cell in the coupling matrix is ​​located, and the comprehensive risk level within that cell is read. This level represents the comprehensive risk assessment result for the tower.

[0075] This embodiment conducted a comprehensive risk assessment on each of the 93 towers within the evaluation area. The statistical assessment results were as follows: 59 towers (63.4%) were classified as extremely low risk, 17 towers (18.3%) as low risk, 11 towers (11.8%) as medium risk, 4 towers (4.3%) as high risk, and 2 towers (2.2%) as extremely high risk. The distribution of the comprehensive risk level resembled a pyramid, with most towers in extremely low or low risk status and a few in high or extremely high risk status. This aligns with the tower damage distribution observed during the field investigation, validating the effectiveness of the risk assessment method. A focused analysis was conducted on the 6 towers classified as high or extremely high risk. These towers are mainly located in areas with severe mining subsidence, such as the Jinhe Coal Mine, Xiejiahegou Coal Mine, and Hongxing Coal Mine. Their hazard level is high or extremely high, and their vulnerability level is medium or higher. Field investigations revealed that these towers indeed exhibited significant tilting or large-scale subsidence cracks in the surrounding ground surface, demonstrating a high degree of consistency between the risk assessment results and the actual situation.

[0076] Based on the cumulative vertical displacement values ​​of each tower, a tiered early warning threshold is set. This threshold is based on the early warning threshold determined in step S200, dividing the early warning into four levels: a blue warning corresponds to a cumulative vertical displacement value of 10 mm to 25 mm, indicating that the tower is in a transitional stage from no significant damage to minor damage, requiring strengthened daily inspections at least once a month, focusing on checking tower tilt and surrounding ground deformation; a yellow warning corresponds to a cumulative vertical displacement value of 25 mm to 45 mm, indicating that the tower is progressing from minor damage to moderate damage, requiring an increase in inspection frequency to once every two weeks. Establish emergency plans and prepare reinforcement materials and equipment; an orange alert corresponds to a cumulative vertical displacement value of 45 mm to 72 mm, indicating that the tower is in a stage of moderate to severe damage, requiring weekly inspections, activation of the emergency plan, organization of professional teams to reinforce the tower, and, if necessary, implementation of load-limiting measures; a red alert corresponds to a cumulative vertical displacement value greater than or equal to 72 mm, indicating that the tower has entered a state of severe damage and is at risk of collapse, requiring daily inspections, immediate emergency measures, including tower reinforcement, reduction of conductor tension, and line shutdown, and, if necessary, relocation of the line to avoid the collapse area.

[0077] The early warning signal is triggered using a combination of real-time monitoring and periodic assessment. Real-time monitoring utilizes the periodic overpasses of synthetic aperture radar (SAR) satellites to acquire the latest surface deformation data. The Sentinel-1 satellite has a revisit cycle of 12 days, updating the cumulative vertical displacement data every 12 days. The latest data is compared with previous data to calculate the deformation rate and determine whether surface subsidence is accelerating. Periodic assessments are conducted quarterly, comprehensively analyzing information such as cumulative vertical displacement values, deformation rates, on-site inspection records, and tower monitoring data to assess the changing trends of tower risk levels and dynamically adjust the early warning level. When the cumulative vertical displacement value at a tower location reaches the early warning setpoint, the system automatically generates an early warning signal. The early warning signal includes the early warning level, tower number, tower location, cumulative vertical displacement value, vulnerability level, overall risk level, and recommended measures. The early warning signal is promptly sent to line maintenance and management personnel via SMS, telephone, and email to ensure that the early warning information is delivered to the relevant responsible persons as soon as possible.

[0078] A tower risk distribution map is generated based on the comprehensive risk level of each tower. This map is produced in a geographic information system (GIS), using remote sensing imagery or topographic maps as a base map, overlaid with elements such as coal mine boundaries, goaf boundaries, transmission line routes, and tower locations. Tower locations are marked with symbols of different colors and shapes: extremely low-risk towers are marked with dark green triangles, low-risk towers with light green triangles, medium-risk towers with yellow triangles, high-risk towers with orange triangles, and extremely high-risk towers with red triangles. The color and size of the symbols visually reflect the tower's risk level. The risk distribution map clearly displays the spatial distribution and risk status of each tower within the evaluation area, providing intuitive decision support for line operation and maintenance management. Maintenance personnel can rationally arrange inspection routes based on the risk distribution map, prioritizing inspections of high-risk and extremely high-risk towers, concentrating resources on reinforcing and protecting key towers, and improving the targeted nature and effectiveness of maintenance work.

[0079] In summary, step S100 acquires multi-dimensional geological environmental data of the evaluation area and constructs a hazard prediction model, enabling intelligent identification and spatial distribution evaluation of mining subsidence hazards, providing a scientific basis for risk assessment regarding the probability of mining subsidence. Step S200 establishes the probability distribution relationship between cumulative vertical displacement values ​​and tower damage levels and fits damage probability curves, achieving a quantitative evaluation of tower vulnerability and establishing a quantifiable correlation between surface deformation monitoring data and tower damage status. Step S300 constructs a coupling matrix of hazard and vulnerability and determines the comprehensive risk level, achieving a scientific coupling between mining subsidence hazards and tower vulnerability, providing comprehensive risk assessment results and graded early warning schemes for each tower. The entire technical solution can provide strong technical support for risk management and protection decisions for power transmission and distribution towers in mountainous coal mine mining subsidence areas, improving the safe operation level of transmission lines under the influence of mining subsidence.

[0080] Example 3 The above is a schematic scheme for a risk assessment method for power transmission and distribution towers in mountainous coal mine subsidence areas. It should be noted that the technical solution of this system for risk assessment of power transmission and distribution towers in mountainous coal mine subsidence areas is based on the same concept as the aforementioned method for risk assessment of power transmission and distribution towers in mountainous coal mine subsidence areas. Details not described in detail in this embodiment can be found in the description of the aforementioned method for risk assessment of power transmission and distribution towers in mountainous coal mine subsidence areas.

[0081] This embodiment also provides a risk assessment system for power transmission and distribution towers in mountainous coal mine subsidence areas, including: The hazard assessment module acquires multidimensional geological environment data of the assessment area, extracts multidimensional geological environment data corresponding to each sample point and labeled collapse sample data as training features, generates a hazard prediction model through training, and identifies the hazard level of each spatial unit in the assessment area using the hazard prediction model to form the regional hazard distribution results. The vulnerability assessment module obtains the cumulative vertical displacement value at the tower location, combines it with the tower damage status samples from the field survey, establishes the probability distribution relationship between the cumulative vertical displacement value and the tower damage level, calculates the equivalent damage probability of the tower under different cumulative vertical displacement values, generates a damage probability curve based on the equivalent damage probability, substitutes the cumulative vertical displacement value of each tower into the damage probability curve to calculate the damage probability value, and classifies the vulnerability level based on the damage probability value. The results output module constructs a coupling matrix between hazard level and vulnerability level, queries the hazard level of the spatial grid cell where each tower is located and the vulnerability level of the corresponding tower, determines the comprehensive risk level of each tower based on the coupling matrix, and generates a tower risk distribution map based on the comprehensive risk level of each tower.

[0082] This embodiment also provides an electronic device applicable to the risk assessment of power transmission and distribution towers in mountainous coal mine subsidence areas, including: a memory and a processor; the memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions to realize the risk assessment method for power transmission and distribution towers in mountainous coal mine subsidence areas as proposed in the above embodiment.

[0083] This embodiment also provides a storage medium storing a computer program that, when executed by a processor, implements the risk assessment method for power transmission and distribution towers in mountainous coal mine subsidence areas as proposed in the above embodiments.

[0084] The storage medium proposed in this embodiment and the method for risk assessment of power transmission and distribution towers in mountainous coal mine subsidence areas proposed in the above embodiments belong to the same inventive concept. Technical details not described in detail in this embodiment can be found in the above embodiments, and this embodiment has the same beneficial effects as the above embodiments.

[0085] Based on the above description of the implementation methods, those skilled in the art can clearly understand that the present invention can be implemented using software and necessary general-purpose hardware, and of course, it can also be implemented using hardware. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as a computer floppy disk, read-only memory (ROM), random access memory (RAM), flash memory, hard disk, or optical disk, etc., including several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods of the various embodiments of the present invention.

[0086] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A method for risk assessment of power transmission and distribution towers in mountainous coal mine subsidence areas, characterized in that: This includes acquiring multidimensional geological environment data of the evaluation area, extracting multidimensional geological environment data corresponding to each sample point and labeled collapse sample data as training features, generating a hazard prediction model through training, identifying the hazard level of each spatial unit in the evaluation area using the hazard prediction model, and forming regional hazard distribution results. The cumulative vertical displacement value at the tower location is obtained. Combined with the tower damage status sample from the field investigation, the probability distribution relationship between the cumulative vertical displacement value and the tower damage level is established. The equivalent damage probability of the tower under different cumulative vertical displacement values ​​is calculated. A damage probability curve is generated by fitting the equivalent damage probability. The cumulative vertical displacement value of each tower is substituted into the damage probability curve to calculate the damage probability value. The vulnerability level is classified according to the damage probability value. Construct a coupling matrix between hazard level and vulnerability level, query the hazard level of the spatial grid cell where each tower is located and the vulnerability level of the corresponding tower, determine the comprehensive risk level of each tower based on the coupling matrix, and generate a tower risk distribution map based on the comprehensive risk level of each tower.

2. The risk assessment method for power transmission and distribution towers in mountainous coal mine subsidence areas as described in claim 1, characterized in that: The multidimensional geological environment data of the evaluation area includes topographic parameters, geological condition parameters, surface cover parameters, coal seam mining parameters, tower spatial location parameters, and temporal surface deformation parameters. The labeled collapse sample data includes collapsed sample points labeled according to time-series surface deformation parameters in areas where mining-induced collapses have occurred, and non-collapse sample points labeled in stable areas.

3. The risk assessment method for power transmission and distribution towers in mountainous coal mine subsidence areas as described in claim 2, characterized in that: Topographic parameters include elevation, slope, aspect, and topographic relief; surface cover parameters include vegetation cover and land use type; and coal seam mining parameters include the number of mining layers, mining thickness, mining depth, and mining depth-to-thickness ratio.

4. The risk assessment method for power transmission and distribution towers in mountainous coal mine subsidence areas as described in claim 3, characterized in that: Establishing the probability distribution relationship between cumulative vertical displacement value and tower damage level includes: determining the tower damage level corresponding to different cumulative vertical displacement value intervals based on field investigation, using a log-normal distribution function to describe the probability of occurrence of each damage level, and weighting and summing the probability of occurrence of each damage level to obtain the equivalent damage probability.

5. The risk assessment method for power transmission and distribution towers in mountainous coal mine subsidence areas as described in claim 4, characterized in that: The damage probability curve is fitted with a hyperbolic tangent function to determine the nonlinear relationship between the cumulative vertical displacement and the equivalent damage probability. A nonlinear fitting method is used to establish a continuous functional relationship between the cumulative deformation and the equivalent damage probability, which serves as a quantitative basis for vulnerability evaluation.

6. The risk assessment method for power transmission and distribution towers in mountainous coal mine subsidence areas as described in claim 5, characterized in that: The construction of the coupling matrix between hazard level and vulnerability level includes setting the classification criteria for hazard level and vulnerability level; Based on the combination relationship between hazard level and vulnerability level, the rules for determining the comprehensive risk level are defined; Using the hazard level and vulnerability level of each tower as input, the corresponding comprehensive risk level is determined through matrix lookup.

7. The risk assessment method for power transmission and distribution towers in mountainous coal mine subsidence areas as described in claim 6, characterized in that: Also includes: A tiered early warning threshold is set based on the cumulative vertical displacement value of each tower. When the cumulative vertical displacement value of a tower reaches the early warning threshold, a risk warning signal of the corresponding level is triggered.

8. A risk assessment system for power transmission and distribution towers in mountainous coal mine subsidence areas, based on the risk assessment method for power transmission and distribution towers in mountainous coal mine subsidence areas as described in any one of claims 1 to 7, characterized in that: It also includes a hazard assessment module, which acquires multidimensional geological environment data of the assessment area, extracts multidimensional geological environment data corresponding to each sample point and labeled collapse sample data as training features, generates a hazard prediction model through training, and uses the hazard prediction model to identify the hazard level of each spatial unit in the assessment area to form the regional hazard distribution results. The vulnerability assessment module obtains the cumulative vertical displacement value at the tower location, combines it with the tower damage status samples from the field survey, establishes the probability distribution relationship between the cumulative vertical displacement value and the tower damage level, calculates the equivalent damage probability of the tower under different cumulative vertical displacement values, generates a damage probability curve based on the equivalent damage probability, substitutes the cumulative vertical displacement value of each tower into the damage probability curve to calculate the damage probability value, and classifies the vulnerability level based on the damage probability value. The results output module constructs a coupling matrix between hazard level and vulnerability level, queries the hazard level of the spatial grid cell where each tower is located and the vulnerability level of the corresponding tower, determines the comprehensive risk level of each tower based on the coupling matrix, and generates a tower risk distribution map based on the comprehensive risk level of each tower.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that: When the processor executes the computer program, it implements the steps of the risk assessment method for power transmission and distribution towers in the mining subsidence area of ​​mountainous coal mines as described in any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by the processor, it implements the steps of the risk assessment method for power transmission and distribution towers in the mining subsidence area of ​​mountainous coal mines as described in any one of claims 1 to 7.

Images