Transmission line settlement risk assessment method, storage medium, and computer equipment

Abstract

This application discloses a method for assessing transmission line settlement risk, as well as a storage medium and computer equipment. The method includes: calculating pore water pressure by considering rainfall and dynamic processes, and then calculating the slope sliding velocity in the target area at different times. Based on a rainfall-driven creeping landslide collapse model, a dynamic equilibrium differential equation for the landslide is constructed. The equation and the slope sliding velocity are used to predict the time required for a potential landslide in the target area to progress from slow creep to collapse, thereby obtaining the slope collapse rate. Deformation data is obtained based on an interferometric synthetic aperture radar deformation method. The deformation data is used as one of the landslide driving factors and is input into three machine learning methods along with other driving factors to obtain landslide susceptibility, i.e., landslide probability. Combining landslide probability with slope collapse rate to assess landslide risk along transmission towers improves the interpretability and accuracy of landslide risk assessment, providing strong support for landslide disaster prevention and post-disaster repair of transmission lines.

Description

Technical Field

[0001] The present application relates to the field of electric power technology, and in particular to a method for evaluating the risk of transmission line subsidence, a storage medium, and a computer device. Background Art

[0002] Landslides are the most widespread geological hazards worldwide. Due to climate change, earthquakes, human engineering activities, and other factors, they frequently occur, posing a serious threat to human life, property, and infrastructure. Landslide movement can generally be defined as extremely slow, relatively slow, moderate, relatively fast, and extremely fast, depending on the scale of displacement rate or velocity. Unlike rapidly collapsing landslides, slow-moving landslides typically evolve at speeds of millimeters to several meters per year over years to decades and are dominated by frictional sliding and / or viscoplastic flow along discrete sliding zones. While slow-moving landslides rarely claim lives, they can cause severe damage to terrain, housing, infrastructure, and agricultural production. It is worth noting that slow-moving landslides can also accelerate to rapid movement and collapse due to pore water pressure disturbances caused by rainfall, resulting in loss of life.

[0003] Predicting the velocity and displacement of reactivated slow-moving landslides is a key issue in understanding landslide kinematics and implementing early warning systems. Various data-based numerical models have been used to predict landslide displacement; however, dynamics-based models are relatively limited, as they consider the shear behavior of the soil within the sliding zone, which fundamentally controls landslide motion. The ability to reliably predict landslide velocity and displacement is crucial for predicting the movement and state of landslides, as well as for providing early warning of slow-moving landslides. However, such predictions of slip rate and state are often based on potential landslide areas, limiting predictions to a specific area and preventing holistic risk assessment across a large study area.

[0004] Landslide susceptibility assessment is the process of comprehensively studying factors such as the geology, hydrology, and geomorphology of a landslide region to assess the probability of a landslide occurring in a target area. With the development and application of technologies such as remote sensing and GIS, the research methods for landslide susceptibility assessment have been greatly expanded and improved. Among susceptibility assessment technologies, machine learning methods can adaptively adjust to different data types, effectively balancing assessment accuracy and operational efficiency, and discovering hidden patterns and relationships in the data. However, landslide susceptibility assessments based on machine learning rely on static evaluation factors, often lacking temporal and dynamic information, and cannot account for the continuous changes in surface conditions over time. Summary of the Invention

[0005] In view of this, the present application provides a transmission line settlement risk assessment method, storage medium, and computer equipment. The method calculates pore water pressure by considering rainfall and dynamic processes, and then calculates the slope sliding velocity in the target area at different times. Based on the rainfall-driven creeping landslide collapse model, a dynamic equilibrium differential equation for the landslide is constructed. The equation and the slope sliding velocity are used to predict the time required for a potential landslide in the target area to develop from slow creep to collapse, thereby obtaining the slope collapse rate. Deformation data is obtained based on the interferometric synthetic aperture radar deformation method. The deformation data is used as one of the landslide driving factors and is input into three machine learning methods along with other driving factors to obtain landslide susceptibility, that is, landslide probability. The landslide risk along the transmission tower is assessed by combining the landslide probability with the slope collapse rate, which improves the interpretability and accuracy of the landslide risk assessment and provides strong support for landslide disaster prevention and post-disaster repair of transmission lines.

[0006] According to one aspect of the present application, a method for assessing the risk of transmission line subsidence is provided, the method comprising:

[0007] Evaluate the subsidence risk of transmission lines based on demand, determine target areas, and obtain original elevation information of the target areas;

[0008] Acquiring single-view complex images of the target area at different times within a preset evaluation period, pairing the single-view complex images to obtain a plurality of candidate interference image pairs, and optimizing the candidate interference image pairs to obtain a plurality of target interference image pairs;

[0009] For any target interference image pair, an interference fringe pattern is generated based on the target interference image pair, and the interference phase in the interference fringe pattern is converted into elevation information using a phase-to-elevation conversion method. The terrain phase is simulated and removed from the interference fringe pattern using the converted elevation information and the original elevation information to obtain a deformation phase pattern. The deformation phase patterns of each target interference image pair are combined to obtain deformation data of the target area within a preset evaluation period, wherein the interference fringe pattern includes an interference phase, which is composed of a terrain phase and a deformation phase.

[0010] Calculating the pore water pressure of the target area within a preset evaluation period; calculating the slope sliding velocity of the target area at different times within the preset evaluation period based on the pore water pressure and the landslide dynamic equilibrium differential equation; reclassifying the slope sliding velocity at different times based on the natural discontinuity method to obtain the slope collapse rate of the target area at different times within the preset evaluation period, wherein the slope collapse rate includes extremely fast collapse, relatively fast collapse, moderate collapse, relatively slow collapse, and extremely slow collapse;

[0011] Acquiring settlement driving factor data of the target area within a preset evaluation period, and predicting the probability of a landslide occurring in the target area within the preset evaluation period based on the deformation data, the settlement driving factor data, and a preset landslide probability prediction model to obtain a landslide probability prediction result, wherein the settlement driving factor data includes data of multiple preset settlement driving factors;

[0012] The landslide probability prediction results of the target area within the preset evaluation period and the slope collapse rates at different times are combined to obtain the settlement risk assessment results of the target area within the preset evaluation period.

[0013] According to another aspect of the present application, a storage medium is provided, on which a computer program is stored. When the program is executed by a processor, the above-mentioned transmission line settlement risk assessment method is implemented.

[0014] According to another aspect of the present application, a computer device is provided, including a storage medium, a processor, and a computer program stored on the storage medium and executable on the processor, wherein the processor implements the above-mentioned transmission line settlement risk assessment method when executing the program.

[0015] Through the above-mentioned technical solution, this application provides a transmission line settlement risk assessment method, storage medium, and computer equipment. This method calculates pore water pressure by considering rainfall and dynamic processes, and then calculates the slope sliding velocity in the target area at different times. Based on a rainfall-driven creeping landslide collapse model, a dynamic equilibrium differential equation for the landslide is constructed. This equation and the slope sliding velocity are used to predict the time it may take for a potential landslide in the target area to progress from slow creep to collapse, thereby obtaining the slope collapse rate. Deformation data is obtained based on an interferometric synthetic aperture radar deformation method. This deformation data is used as one of the landslide driving factors and is input into three machine learning methods along with other driving factors to determine the landslide susceptibility, or landslide probability. Combining landslide probability with slope collapse rate to assess landslide risk along transmission towers improves the interpretability and accuracy of landslide risk assessment, providing strong support for landslide disaster prevention and post-disaster repair of transmission lines.

[0016] The above description is only an overview of the technical solution of the present application. In order to more clearly understand the technical means of the present application, it can be implemented in accordance with the contents of the specification. In order to make the above and other purposes, features and advantages of the present application more obvious and easy to understand, the specific implementation methods of the present application are listed below. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] The drawings described herein are used to provide a further understanding of the present application and constitute a part of the present application. The illustrative embodiments of the present application and their descriptions are used to explain the present application and do not constitute an improper limitation on the present application. In the drawings:

[0018] Figure 1 A schematic diagram of a flow chart of a transmission line settlement risk assessment method provided in an embodiment of the present application is shown;

[0019] Figure 2 A schematic diagram showing a flow chart of another transmission line settlement risk assessment method provided in an embodiment of the present application is shown;

[0020] Figure 3 A schematic diagram of an SBAS deformation inversion process provided in an embodiment of the present application is shown;

[0021] Figure 4 A schematic diagram showing a flow chart of another transmission line settlement risk assessment method provided in an embodiment of the present application is shown;

[0022] Figure 5 A schematic diagram of a transmission line slope collapse time estimation process provided by an embodiment of the present application is shown;

[0023] Figure 6 A schematic diagram of a landslide susceptibility assessment process along a transmission line provided in an embodiment of the present application is shown. DETAILED DESCRIPTION

[0024] The present application will be described in detail below with reference to the accompanying drawings and in combination with embodiments. It should be noted that, unless there is a conflict, the embodiments and features in the embodiments of the present application can be combined with each other.

[0025] In this embodiment, a method for assessing the risk of transmission line subsidence is provided. Figure 1 As shown, the method includes:

[0026] Step 101: Evaluate the subsidence risk of transmission lines based on demand, determine a target area, and obtain original elevation information of the target area.

[0027] In the above embodiment of the present application, based on the demand evaluation of the transmission line with the risk of settlement, the target area is determined, and then the original elevation information of the target area is obtained. For example, the SR TMDEM data of the target area can be obtained. The SR TMDEM data is the digital elevation model (Digital Elevation Model) data collected by the Shuttle Radar Topography Mission (ShuttleRad arTopography Mission). It is obtained by cooperation between NASA and the National Bureau of Surveying, Mapping and other institutions of the Ministry of National Defense using radar equipment on the space shuttle to perform high-precision remote sensing measurements of the earth's surface. These data can be used to create a global digital elevation model and provide accurate elevation information of the earth's surface.

[0028] Step 102 : acquiring single-view complex images of the target area at different times within a preset evaluation period, pairing the single-view complex images to obtain a plurality of candidate interference image pairs, and optimizing the candidate interference image pairs to obtain a plurality of target interference image pairs.

[0029] Step 103: for any target interference image pair, generate an interference fringe pattern based on the target interference image pair, use a phase-elevation conversion method to convert the interference phase in the interference fringe pattern into elevation information, use the converted elevation information and the original elevation information to simulate and remove the terrain phase in the interference fringe pattern to obtain a deformation phase pattern, and combine the deformation phase patterns of each target interference image pair to obtain deformation data of the target area within a preset evaluation period, wherein the interference fringe pattern includes an interference phase, and the interference phase is composed of a terrain phase and a deformation phase.

[0030] Next, single-look complex images of the target area are acquired at different times within a preset evaluation period (e.g., two months). These images are paired to obtain multiple candidate interferometric image pairs. These candidate interferometric image pairs are then optimized to obtain multiple target interferometric image pairs. Single-look complex images, such as Sentinel-1A (SLC) data, are used. Sentinel-1A, or Sentinel-1A, is an Earth observation satellite in the European Space Agency's Copernicus program. It carries a C-band synthetic aperture radar and provides all-weather, all-time Earth surface imaging services. SLC data is a data product from Sentinel-1A that contains amplitude and phase information. In particular, satellite orbit data covering the target area can also be downloaded to assist in monitoring and analyzing the target area's topography. The acquired Sentinel-1A images are then subjected to interferometric pair combination, interferometric processing, orbit refinement and re-flattening, deformation inversion, and geocoding to obtain the final deformation data.

[0031] Step 104, calculate the pore water pressure of the target area within the preset evaluation period, calculate the slope sliding speed of the target area at different times within the preset evaluation period based on the pore water pressure and the landslide dynamic equilibrium differential equation, reclassify the slope sliding speed at different times based on the natural discontinuity method, and obtain the slope collapse rate of the target area at different times within the preset evaluation period, wherein the slope collapse rate includes extremely fast collapse, relatively fast collapse, medium-speed collapse, relatively slow collapse and extremely slow collapse.

[0032] Next, rainfall data such as the groundwater level of the target area soil, the distance of groundwater below the surface from the surface, the steady-state friction coefficient, and the inclination of the sliding surface are collected, and the pore water pressure in the soil is calculated using the rainfall data. Combined with the steady-state equation of the landslide (the dynamic equilibrium differential equation of the landslide), the slope sliding velocity of the target area at different times within the preset evaluation period is calculated. The slope sliding velocity at different times is reclassified based on the natural break method to obtain the slope collapse rate of the target area at different times within the preset evaluation period. The Natural Breaks Method is a statistical method for data classification and grouping, which is particularly suitable for geographic information systems (GIS) and spatial data analysis. It is based on the natural distribution characteristics of the data and identifies the natural break points in the data by analyzing the numerical distribution in the data set. These break points are places where the data values change significantly. The specific classification method is as follows:

[0033] 1. Collect, organize, and quantify the data to be classified (i.e., slope sliding velocity at different times);

[0034] 2. Select the number of categories (k) into which the data should be divided. This is usually determined based on the characteristics of the data and the purpose of analysis. In the above embodiment of the present application, the categories can be set to 5, namely, very fast collapse, relatively fast collapse, medium collapse, relatively slow collapse, and very slow collapse;

[0035] 3. For each possible grouping scheme, calculate the within-group variance (Within-group variance), the formula is:

[0036]

[0037] k is the number of categories, n j is the number of data of the jth category, x ij is the i-th data point in the j-th class, is the mean of the j-th class of data.

[0038] 4. Calculate the between-group variance (Between-group variance), the formula is:

[0039]

[0040] is the overall mean of all data.

[0041] 5. The goal is to minimize the within-group variance W or maximize the between-group variance B. The optimal grouping can be found by iterating different grouping schemes.

[0042] Step 105: Acquire settlement driving factor data of the target area within a preset evaluation period; predict the probability of landslide occurring in the target area within the preset evaluation period based on the deformation data, the settlement driving factor data, and a preset landslide probability prediction model to obtain a landslide probability prediction result, wherein the settlement driving factor data includes data of multiple preset settlement driving factors.

[0043] Step 106 , combining the landslide probability prediction results of the target area within the preset evaluation period and the slope collapse rates at different times, to obtain the settlement risk assessment results of the target area within the preset evaluation period.

[0044] Next, the system obtains settlement driving factor data for the target area. Based on the deformation data, settlement driving factor data, and a pre-set landslide probability prediction model, it predicts the probability of a landslide in the target area, resulting in a landslide probability prediction result for the target area. The settlement driving factor data includes data on multiple pre-set settlement driving factors. Finally, by combining the landslide probability prediction results for the target area within the pre-set evaluation period with the slope collapse rates at different times, a settlement risk assessment result for the target area within the pre-set evaluation period is obtained, providing a more reliable assessment result.

[0045] By applying the technical solution of this embodiment, pore water pressure is calculated by considering rainfall and dynamic processes, and then the slope sliding velocity in the target area at different times is calculated. Based on the rainfall-driven creeping landslide collapse model, a dynamic equilibrium differential equation for the landslide is constructed. This equation and the slope sliding velocity are used to predict the time it may take for a potential landslide in the target area to progress from slow creep to collapse, thereby obtaining the slope collapse rate. Deformation data is obtained based on the interferometric synthetic aperture radar deformation method. This deformation data is used as one of the landslide driving factors and is input into three machine learning methods along with other driving factors to determine the landslide susceptibility, namely the landslide probability. Combining landslide probability with slope collapse rate to assess landslide risk along transmission towers improves the interpretability and accuracy of landslide risk assessment, providing strong support for landslide disaster prevention and post-disaster repair of transmission lines.

[0046] Furthermore, as a refinement and extension of the specific implementation of the above embodiment, in order to fully illustrate the specific implementation process of this embodiment, another transmission line settlement risk assessment method is provided, such as Figure 2 As shown, the method includes:

[0047] Step 201: determine the transmission lines whose settlement risk needs to be evaluated, expand the transmission lines whose settlement risk needs to be evaluated by a preset width to obtain a target area, and obtain the original elevation information of the target area.

[0048] Step 202 is to obtain single-view complex images of the target area at different times within a preset evaluation period, and determine two single-view complex images whose time interval does not exceed the preset time and whose spatial distance does not exceed the preset distance as a candidate interference image pair, until multiple candidate interference image pairs are determined.

[0049] In the above implementation of this application, the transmission lines that need to be evaluated for settlement risk are determined, such as Figure 3 As shown, transmission line vector data can be obtained. A preset width (e.g., 10 km) can be expanded around the transmission line vector data to obtain the target area. That is, within the data coverage, a buffer zone of a certain width around the transmission line is selected as the target area, and the original elevation information (SRTMDEM data) of the target area is obtained. Simultaneously, Sentinel-1A single-look complex images (Sentinel-1A SLC data) of the target area at different times within a preset evaluation period are obtained, and satellite orbit data is downloaded. Two single-look complex images acquired with a time interval not exceeding a preset time and a spatial distance not exceeding a preset distance are identified as a candidate interferometric image pair, and this is continued until multiple candidate interferometric image pairs are identified. For example, the time baseline threshold of the SBAS-InSAR process is set to 60 days, and the spatial baseline threshold is set to 150 meters. Candidate interferometric pairs that meet these thresholds are generated so that the subsidence risk of the target area can be subsequently predicted in combination with InSAR deformation. SBAS, or Satellite-Based Augmentation System, can enhance the positioning accuracy of satellite navigation systems such as GPS. InSAR deformation is the abbreviation of "Interferometric Synthetic Aperture Radar deformation" or "Synthetic Aperture Radar Interferometry Deformation". "InSAR" is the abbreviation of "Interferometric Synthetic Aperture Radar", which means the interferometric synthetic aperture radar technology, and "deformation" refers to the small deformation or displacement of the surface or objects monitored by this technology. After the interference combination, such as Figure 3 As shown, interference processing, track refinement and re-leveling, and deformation inversion geocoding can be performed to facilitate data clipping and transmission line settlement analysis. Finally, transmission line settlement analysis, Kriging interpolation, and deformation rate result verification can be performed.

[0050] Step 203: For any candidate interference image pair, calculate the similarity between the two single-view complex images in the candidate interference image pair to obtain a coherence coefficient map, perform conjugate multiplication on the two single-view complex images in the candidate interference image pair to obtain an interference fringe map, and use the minimum cost flow algorithm of the Delaunay triangulation network to phase unwrap the interference fringe map to obtain an unwrapped map.

[0051] In step 204, among the multiple candidate interference image pairs, the candidate interference image pairs corresponding to the coherence coefficient maps with low correlation, the candidate interference image pairs corresponding to the interference fringe maps with low interference quality, and the candidate interference image pairs corresponding to the detangling maps with low detangling quality are removed to obtain multiple target interference image pairs.

[0052] Next, phase unwrapping can be performed using the minimum cost flow (MCF) algorithm of the Delaunay triangulation. After the interferometric processing is completed, the quality of the interferometric pairs is determined based on the coherence coefficient graph, the interferogram, and the unwrapping graph. Interferometric pairs with low coherence and poor unwrapping quality are removed to avoid affecting the inversion effect.

[0053] Step 205: for any target interference image pair, generate an interference fringe pattern based on the target interference image pair, use a phase-elevation conversion method to convert the interference phase in the interference fringe pattern into elevation information, use the converted elevation information and the original elevation information to simulate and remove the terrain phase in the interference fringe pattern to obtain a deformation phase pattern, and combine the deformation phase patterns of each target interference image pair to obtain deformation data of the target area within a preset evaluation period, wherein the interference fringe pattern includes an interference phase, and the interference phase is composed of a terrain phase and a deformation phase.

[0054] Next, the acquired raw SRTM data can be used as an external DEM (digital elevation model) to eliminate the terrain phase of the interferometric pair. For example, during the interferometric processing, a multi-look operation of 8 pixels in range and 2 pixels in azimuth can be performed to suppress noise and improve the signal-to-noise ratio of the interferogram. In particular, after obtaining the interferometric fringe pattern, GACOS data can also be used for data processing and interpretation to eliminate or reduce the phase changes caused by atmospheric delay in the Yellow River source region, reduce the impact of atmospheric delay, and use Kriging interpolation to interpolate and supplement the incoherence in the deformation rate results. At the same time, the deformation rate results are verified in combination with measured slip data along the transmission towers to verify the reliability of the deformation rate results. GACOS data, or Generic Atmospheric Correction Online Service for InSAR, is specifically used for atmospheric correction. Based on the iterative decomposition model of the troposphere, it can separate the stratification and turbulence signals from the total tropospheric delay, and then generate a high-spatial-resolution zenith total delay map, which enables atmospheric correction for applications such as InSAR measurements.

[0055] Step 206, calculate the pore water pressure of the target area within the preset evaluation period, calculate the slope sliding speed of the target area at different times within the preset evaluation period based on the pore water pressure and the landslide dynamic equilibrium differential equation, reclassify the slope sliding speed at different times based on the natural discontinuity method, and obtain the slope collapse rate of the target area at different times within the preset evaluation period, wherein the slope collapse rate includes extremely fast collapse, relatively fast collapse, medium-speed collapse, relatively slow collapse and extremely slow collapse.

[0056] Step 207: Acquire settlement driving factor data of the target area within a preset evaluation period; predict the probability of landslide occurring in the target area within the preset evaluation period based on the deformation data, the settlement driving factor data, and a preset landslide probability prediction model to obtain a landslide probability prediction result, wherein the settlement driving factor data includes data of multiple preset settlement driving factors.

[0057] Next, the pore water pressure of the target area within the preset evaluation period is calculated. Based on the pore water pressure and the landslide dynamic equilibrium differential equation, the slope sliding velocity of the target area at different times within the preset evaluation period is calculated. The slope sliding velocity at different times is reclassified based on the natural discontinuity method to obtain the slope collapse rate of the target area at different times within the preset evaluation period. The settlement driving factor data of the target area within the preset evaluation period is obtained. Based on the deformation data, the settlement driving factor data and the preset landslide probability prediction model, the probability of landslide in the target area within the preset evaluation period is predicted to obtain the landslide probability prediction result, in preparation for the subsequent settlement risk assessment.

[0058] Step 208: If the landslide probability prediction result shows that the probability of landslide occurring in the target area falls within a preset low probability range, the settlement risk assessment result of the target area is low settlement risk.

[0059] Step 209: If the landslide probability prediction result shows that the probability of landslide in the target area does not fall within the preset low probability range, the settlement risk assessment result of the target area is that there is a settlement risk, and based on the slope collapse rate of the target area at different times within the preset evaluation period, a settlement risk warning information is generated and sent to the preset receiving terminal.

[0060] Since the rainfall and dynamics-based susceptibility evaluation (the calculated slope collapse rate of the target area at different times within the preset evaluation period) is based on the premise that the target is a potential landslide body, and the result of the machine learning-based susceptibility evaluation (landslide probability prediction result) is the possibility of a landslide occurring in the target, when comprehensively optimizing the two risks, we cannot simply compare the two and take the maximum value. When the probability of a landslide occurring in the target is extremely small (belonging to the preset small probability range), it is meaningless to discuss the possible sliding speed after the landslide occurs. Therefore, when fusion is performed, the landslide risk prediction results based on machine learning need to be given priority. Therefore, it can solve the problem that the current landslide risk assessment method based on machine learning only considers a single static index and cannot obtain time series and dynamic information.

[0061] By applying the technical solution of this embodiment, it is possible to take into account the continuous changes in surface conditions over a period of time, as well as the dependence of rainfall and dynamics-based methods on information about potential landslide areas. Compared with traditional methods, more complete and more accurate risk assessment results with a larger buffer range can be obtained. That is, based on rainfall and dynamics and a variety of machine learning, combined with nine surface settlement driving factors, a spatial simulation of terrain change rate can be performed on the transmission line corridor, such as a 10km buffer zone. By comparing a variety of machine learning methods, the algorithm with the highest regression accuracy is determined. At the same time, the rainfall and dynamics methods are combined to assess the landslide risk. This optimizes the landslide susceptibility results of machine learning and results in a more accurate risk assessment.

[0062] Furthermore, as a refinement and extension of the specific implementation of the above embodiment, in order to fully illustrate the specific implementation process of this embodiment, another transmission line settlement risk assessment method is provided, such as Figure 4 As shown, the method includes:

[0063] Step 301: Based on the transmission lines with subsidence risk evaluated on demand, a target area is determined, and original elevation information of the target area is obtained.

[0064] Step 302 : acquiring single-view complex images of the target area at different times within a preset evaluation period, pairing the single-view complex images to obtain a plurality of candidate interference image pairs, and optimizing the candidate interference image pairs to obtain a plurality of target interference image pairs.

[0065] Step 303: for any target interference image pair, generate an interference fringe pattern based on the target interference image pair, use a phase-elevation conversion method to convert the interference phase in the interference fringe pattern into elevation information, use the converted elevation information and the original elevation information to simulate and remove the terrain phase in the interference fringe pattern to obtain a deformation phase pattern, and combine the deformation phase patterns of each target interference image pair to obtain deformation data of the target area within a preset evaluation period, wherein the interference fringe pattern includes an interference phase, and the interference phase is composed of a terrain phase and a deformation phase.

[0066] In the above-described embodiment of the present application, based on the need to evaluate the subsidence risk of a transmission line, a target area is determined and the original elevation information of the target area is obtained. Single-view complex images of the target area at different times within a preset evaluation period are obtained, and the single-view complex images are paired to obtain multiple candidate interference image pairs. The candidate interference image pairs are then optimized to obtain multiple target interference image pairs. SBAS deformation inversion is performed based on the target interference image pairs to obtain deformation data of the target area within the preset evaluation period, preparing for the subsequent landslide probability, i.e., landslide susceptibility assessment.

[0067] Step 304: Calculate the pore water pressure of the target area within a preset evaluation period based on a pore water pressure calculation formula. Calculate the slope sliding velocity of the target area at different times within the preset evaluation period based on the pore water pressure and the landslide dynamic equilibrium differential equation. The pore water pressure calculation formula is:

[0068] U=γ w h w ,

[0069] The landslide dynamic equilibrium differential equation formula is:

[0070]

[0071] U is the pore water pressure, γ w is the unit weight of pore water, h w is the height of pore water in the soil, α is the inclination of the sliding surface of the landslide, γ* is the equivalent unit weight of the landslide, H is the thickness of the landslide, μ ss is the steady-state friction coefficient of the sliding surface of the landslide, g is the acceleration of gravity, v is the sliding velocity of the slope relative to the land surface, and t is time.

[0072] Step 305: reclassify the slope sliding speed at different times based on the natural discontinuity method to obtain the slope collapse rate of the target area at different times within the preset evaluation period, wherein the slope collapse rate includes extremely fast collapse, relatively fast collapse, medium-speed collapse, relatively slow collapse and extremely slow collapse.

[0073] Then, if Figure 5 As shown in the figure, time-series rainfall data covering the transmission line is obtained and combined with the soil moisture content of the transmission line slope to calculate the pore water pressure of the transmission line slope. For example, daily precipitation in the target area over the course of a year can be used as a basis to calculate the pore water pressure within the target area, providing the input data for the steady-state equation. Next, a rainfall intrusion model (a landslide dynamic equilibrium differential equation) is established. The slope sliding velocity at different times is derived using this landslide dynamic equilibrium differential equation. This differential equation can be used to calculate the final sliding velocity of the landslide after the stress perturbation. In particular, the sliding velocity can be predicted using a landslide sliding rate prediction model (based on the stochastic Senli model). Specifically, after determining the dynamic parameters of the landslide, the landslide sliding rate prediction model is used to estimate the landslide sliding rate. Finally, the collapse rate of the transmission line slope is reclassified into five categories: extremely rapid collapse, relatively rapid collapse, moderate collapse, relatively slow collapse, and extremely slow collapse.

[0074] Step 306: Based on the historical settlement driving factor data and historical deformation data of the target area, a random forest model, a support vector machine model, and an extreme gradient boosting model are trained respectively, and the model with the highest prediction accuracy is used as the landslide probability prediction model.

[0075] Preset surface subsidence driving factors such as Figure 6 As shown in the figure, including slope, lithology, profile curvature, plane curvature, distance from road, vegetation coverage, intensity of human activities, distance from river and precipitation (amount), Figure 6 In this method, a random forest (RF) model, a support vector machine (SVM) model, an eXtreme Gradient Boosting (XGBoost) model, or a backpropagation neural network (BPNN) model are trained based on historical subsidence driving factors and deformation data in the target area. The model with the highest prediction accuracy is used as the landslide probability prediction model. Combined with machine learning model optimization, a landslide hazard assessment index (slope collapse rate) is output. The index is then reclassified into extremely low risk, relatively low risk, moderate risk, relatively high risk, and very high risk, corresponding to extremely slow collapse, relatively slow collapse, moderate collapse, relatively fast collapse, and extremely fast collapse, respectively.

[0076] Specifically, before training the three models, collinearity analysis can be performed on the data of the nine surface subsidence driving factors and the deformation data to ensure that there is no multicollinearity between the factors during modeling. After the collinearity analysis, 70% of the data is used as the training set to train the random forest, support vector machine, and extreme gradient boosting models and simulate the landslide susceptibility risk of the target area, and the remaining 30% of the data is used as the test set to test the accuracy of the model prediction. During the training process, the parameters of the three models are optimized, and based on the training results of the models under the optimal parameter conditions, the ROC (Receiver Operating Characteristic Curve) accuracy curves of the three models are calculated. The model with the highest accuracy is selected and reclassified according to the results of its landslide susceptibility evaluation to obtain the landslide risk prediction results based on machine learning. In particular, when using the data of the nine surface subsidence driving factors and deformation data to train the model, the nine factor data and soil data can be cropped to the same range as the target area to improve the accuracy of model training.

[0077] When performing collinearity analysis, the tolerance (TOL) and variance inflation factor (VIF) of the 10 factors are calculated. The specific calculation method is as follows:

[0078]

[0079] TOL is the tolerance, VIF is the variance inflation factor, and the tolerance and variance inflation factor are a pair of reciprocals. is the coefficient of determination of the residual factor regression model.

[0080] For the factors for which TOL and VIF are calculated, if their variance inflation factor value is greater than 5 or the tolerance is less than 0.2, it proves that there is collinearity between the input factors and some factors need to be eliminated. After elimination, the data of the nine factors are re-obtained and re-analyzed until all factors meet the variance inflation factor value < 5 and the tolerance > 0.2.

[0081] The process of training a random forest model is as follows: N estimators are extracted from the original data set as training sets using the bootstrap sampling technique. The size of each training set is approximately two-thirds of the original data set. During the training process, approximately one-third of the data in each bootstrap sample of RF will not be drawn, and this part of the data is called out-of-bag data. Next, a regression tree is created for each training set, forming a total of N estimator regression trees, and finally forming a "forest", but these regression trees are not "pruned". During the growth of each tree, not all optimal attributes are selected as internal nodes of the branch, but the optimal attributes are selected from the randomly selected maximum depth (Maxdepth) attributes for branching. Therefore, the random algorithm increases the differences between regression models by constructing different training sets, thereby improving the extrapolation prediction ability of the combined regression model.

[0082] After n times of model training, the regression model sequence {t1(x), t2(x), t3(x), ..., t k (x)}, used to form a multivariate regression model system. Then, the prediction results of the regression trees of N estimators are collected and the value of the new sample is calculated using a simple averaging strategy. The final regression decision formula is:

[0083]

[0084] represents the combined regression model, t i is a single decision tree regression model, and K is the number of regression trees (N estimators).

[0085] The process of training the support vector machine model is as follows: First, define a set of data points P = (x i ,a i ), i = 1, ... n, where x i Input vector for data point i, a i is the actual value, and n is the number of data points. For a linear function f, the hyperplane constructed by SVR (Support Vector Regression) is determined as: f(x) = wx + b, where the predicted value f(x) depends on the slope w and the intercept b.

[0086] In general, one wants to strike a balance between learning the relationship between input and output while maintaining good generalization behavior. Excessive focus on minimizing training error can lead to overfitting. Low-complexity models are restricted in terms of decision boundaries, but are less likely to overfit. In Cortes and Vapnik (the two contributors to the Support Vector Machine, SVM), it was shown that the probability of test error depends on two factors, the frequency of training errors and the confidence interval, where these two factors form a trade-off. The confidence interval is related to the dimensionality of the SVM (Support Vector Machine model), which can be thought of as the complexity of the learned model. Therefore, improved generalization can be obtained by improving the confidence interval, at the cost of additional training error. In the following formula, R represents the compound risk caused by training error and model complexity. Of course, the risk R needs to be kept as low as possible.

[0087]

[0088] The formula that produces estimates of w and b consists of two main parts:

[0089] Part 1 It consists of training risk or empirical risk and contains the loss function, L ∈ This function (a,y) means that if the difference between the predicted value f(x) and the actual value a is less than ∈, the prediction error will be ignored. The loss function is formally defined as follows:

[0090]

[0091] The second part of the formula 1 / 2W 2 is a regularization term, which is related to the complexity of the model.

[0092] C controls the trade-off between the regularization term and training accuracy. A larger value of C means that more weight is placed on correctly predicting training points, at the expense of higher generalization error. The problem of finding the optimal hyperplane is a convex optimization problem. For nonlinear relationships between input vectors and outputs, it is necessary to define a mapping The training point x i Convert to high-dimensional feature space. Calculate Φ(x i ) and Φ(x). Function Φ(x i )Φ(x) is usually defined as K(x i ,x), and is called a kernel function, which attempts to achieve linear separability between training points in a high-dimensional feature space.

[0093] The process of training the XGBoost model is as follows:

[0094] The output of XGBoost can be expressed as the sum of the prediction values of all trees f k (x i ):

[0095]

[0096] Where Γ is the space of regression trees, K is the number of regression trees, and x i Represents the feature corresponding to sample i.

[0097] The process of improving the machine learning algorithm needs to continue until the objective function is reduced to a certain limit. In order to approximate the set of functions used in the model, the function set is defined as follows:

[0098]

[0099] Where n is the data sample, It is the training loss function, which is used to describe how well the model fits the training data.

[0100]

[0101] Is a regularization term used to penalize model complexity. In the regularization term, γ is the complexity cost of introducing additional leaves, λ is a regularization hyperparameter, is the L2 norm of the weight of leaf node j.

[0102] In additive learning, all trees are built sequentially, and each newly added tree learns from the previous trees and updates the residuals in the predictions. has already included the iteration results of all trees. Therefore, for the kth iteration, Can be expressed Objective function Φ (k) is written as:

[0103] To effectively optimize the general setting objective of the first loss training function, a second-order Taylor expansion is used to approximate it:

[0104]

[0105] Where: and are the first-order gradient statistics and second-order gradient statistics of the loss function respectively. In step k, the constant term can be removed to obtain the following approximate target:

[0106]

[0107] Here we use a vector of leaf scores to define the tree, and use a leaf index mapping function to map an instance to a leaf j. This process can be expressed as:

[0108]

[0109] Given a fixed tree structure, the optimal leaf weight on each leaf node is solved by simple quadratic programming as well as The extreme value of:

[0110]

[0111] The mean absolute error (MAE), root mean square error (RMSE), and coefficient of determination (R²) of the three methods were calculated and used as a basis to compare the accuracy of the three algorithms and determine the optimal one. Based on the characteristics of the three models and the characteristics of the study area, several sets of parameter values were preset for the three models. For machine learning models, there are often more than one adjustable parameter, which requires permutations and combinations, and comprehensive testing of all parameter combinations to find the optimal parameter settings.

[0112] The training set was used to train machine learning models based on various parameter groups. The evaluation accuracy of each model was verified using the data of the validation set, and the optimal parameter combination of the three models was selected.

[0113] The data of the training set are then placed into three machine learning models based on the optimal parameter combination for modeling. The data of the training set are used for modeling, and then the data of the validation set are input into the model. The output data is used to calculate the ROC accuracy curve and AUC of the three models. The AUC (Area Under Curve) indicator is a super important indicator for evaluating the performance of the binary classification model.

[0114] The ROC curve is a common tool for evaluating the performance of classification models and is widely used in landslide susceptibility assessment. It plots the relationship between the true positive rate (TPR) and the false positive rate (FPR) to illustrate the model's performance at different thresholds. The horizontal axis typically represents the FPR, and the vertical axis represents the TPR. The calculation formula is as follows:

[0115] FPR=FP / (FP+TN), TPR=TP / (TP+FN),

[0116] In the formula, FP stands for False Positives (false positives); TP stands for True Positives (true positives); FN stands for False Negatives (false negatives); and TN stands for True Negatives (true negatives). The closer the ROC curve is to the upper left corner, the better the model's classification ability. The closer it is to the diagonal, the model has no classification ability and its classification performance is equivalent to random guessing.

[0117] The significance of the AUC indicator is to quantify the performance of the ROC curve. The value of AUC ranges from 0 to 1, and the closer it is to 1, the better the performance of the classifier. Specifically, the size of the AUC value can be used to evaluate the classification ability of the classifier: when AUC is equal to 0.5, it means that the performance of the classifier is equivalent to random guessing; when AUC is greater than 0.5, the performance of the classifier is better than random guessing;

[0118] Based on the AUC indicator, the model with the highest evaluation accuracy (the model with the highest AUC value) was selected as the optimal machine learning model. The landslide susceptibility of the entire study area was evaluated. The evaluation results were reclassified based on the natural break method to obtain the landslide risk prediction results based on machine learning.

[0119] Step 307: Acquire settlement driving factor data of the target area within a preset evaluation period; predict the probability of a landslide occurring in the target area within the preset evaluation period based on the deformation data, the settlement driving factor data, and a preset landslide probability prediction model to obtain a landslide probability prediction result; wherein the settlement driving factor data includes data of a plurality of preset settlement driving factors, and the preset surface settlement driving factors include slope, lithology, profile curvature, plane curvature, distance from a road, vegetation coverage, human activity intensity, distance from a river, and precipitation.

[0120] Step 308 : Combining the landslide probability prediction results of the target area within the preset evaluation period and the slope collapse rates at different times, obtain the settlement risk assessment results of the target area within the preset evaluation period.

[0121] Finally, the settlement driving factor data of the target area within the preset evaluation period is obtained. Based on the deformation data, settlement driving factor data and the preset landslide probability prediction model, the probability of landslide in the target area within the preset evaluation period is predicted to obtain the landslide probability prediction result. The landslide probability prediction result of the target area within the preset evaluation period and the slope collapse rate at different times are combined to obtain the settlement risk evaluation result of the target area within the preset evaluation period.

[0122] By applying the technical solution of this embodiment, pore water pressure is calculated by considering rainfall and dynamic processes, and then the slope sliding velocity in the target area at different times is calculated. Based on the rainfall-driven creeping landslide collapse model, a dynamic equilibrium differential equation for the landslide is constructed. This equation and the slope sliding velocity are used to predict the time it may take for a potential landslide in the target area to progress from slow creep to collapse, thereby obtaining the slope collapse rate. Deformation data is obtained based on the interferometric synthetic aperture radar deformation method. This deformation data is used as one of the landslide driving factors and is input into three machine learning methods along with other driving factors to determine the landslide susceptibility, namely the landslide probability. Combining landslide probability with slope collapse rate to assess landslide risk along transmission towers improves the interpretability and accuracy of landslide risk assessment, providing strong support for landslide disaster prevention and post-disaster repair of transmission lines.

[0123] It should be noted that for other corresponding descriptions of the functional units involved in the transmission line settlement risk assessment device provided in the embodiment of the present application, reference can be made to Figure 1 、 Figure 2 and Figure 4 The corresponding description in the method will not be repeated here.

[0124] Based on the above Figure 1 、 Figure 2 and Figure 4 The method shown in FIG. 1 is a method for performing the above-mentioned operation. Accordingly, the embodiment of the present application further provides a storage medium on which a computer program is stored. When the computer program is executed by a processor, the above-mentioned operation is performed. Figure 1 、 Figure 2 and Figure 4 The transmission line settlement risk assessment method shown in Figure 2 is shown in Figure 2.

[0125] Based on this understanding, the technical solution of the present application can be embodied in the form of a software product, which can be stored in a non-volatile storage medium (which can be a CD-ROM, USB flash drive, mobile hard disk, etc.), including a number of instructions for enabling a computer device (which can be a personal computer, server, or network device, etc.) to execute the methods described in each implementation scenario of the present application.

[0126] Based on the above Figure 1 、 Figure 2 and Figure 4 In order to achieve the above purpose, the embodiment of the present application further provides a computer device, which can be a personal computer, a server, a network device, etc., and the computer device includes a storage medium and a processor; the storage medium is used to store a computer program; the processor is used to execute the computer program to achieve the above purpose. Figure 1 、 Figure 2 and Figure 4 The transmission line settlement risk assessment method shown in Figure 2 is shown in Figure 2.

[0127] Optionally, the computer device may further include a user interface, a network interface, a camera, a radio frequency (RF) circuit, a sensor, an audio circuit, a Wi-Fi module, etc. The user interface may include a display, an input unit such as a keyboard, etc., and the optional user interface may also include a USB interface, a card reader interface, etc. The network interface may optionally include a standard wired interface, a wireless interface (such as a Bluetooth interface, a Wi-Fi interface), etc.

[0128] Those skilled in the art will understand that the computer device structure provided in this embodiment does not constitute a limitation on the computer device, and may include more or fewer components, or a combination of certain components, or different component arrangements.

[0129] The storage medium may also include an operating system and a network communication module. An operating system is a program that manages and stores the hardware and software resources of a computer device, supporting the execution of information processing programs and other software and / or programs. The network communication module facilitates communication between components within the storage medium, as well as with other hardware and software within the physical device.

[0130] Through the description of the above implementation methods, those skilled in the art can clearly understand that the present application can be implemented by means of software plus the necessary general hardware platform, or by hardware implementation, taking into account rainfall and dynamic processes to calculate pore water pressure, and then calculating the slope sliding speed in the target area at different times. Based on the rainfall-driven creeping landslide collapse model, a landslide dynamic equilibrium differential equation is constructed, and the equation and slope sliding speed are used to predict the time required for potential landslides in the target area to develop from slow creep to collapse, thereby obtaining the slope collapse rate. Deformation data is obtained based on the interferometric synthetic aperture radar deformation method, and the deformation data is used as one of the landslide driving factors. Together with other driving factors, it is input into three machine learning methods to obtain landslide susceptibility, i.e., landslide probability. Combining landslide probability with slope collapse rate to assess landslide risk along transmission towers improves the interpretability and accuracy of landslide risk assessment, providing strong support for landslide disaster prevention and post-disaster repair of transmission lines.

[0131] Those skilled in the art will understand that the accompanying drawings are only schematic diagrams of a preferred implementation scenario, and the modules or processes in the accompanying drawings are not necessarily required to implement the present application. Those skilled in the art will understand that the modules in the devices in the implementation scenario can be distributed in the devices of the implementation scenario according to the implementation scenario description, or can be changed accordingly and located in one or more devices different from the implementation scenario. The modules of the above-mentioned implementation scenario can be combined into one module, or can be further split into multiple sub-modules.

[0132] The serial numbers of the above application are for descriptive purposes only and do not represent the advantages or disadvantages of the implementation scenarios. The above disclosure only discloses several specific implementation scenarios of the present application, but the present application is not limited thereto. Any changes that can be conceived by those skilled in the art should fall within the scope of protection of the present application.

Claims

1. A method for assessing the risk of transmission line settlement, characterized in that: The method comprises: Evaluate the subsidence risk of transmission lines based on demand, determine target areas, and obtain original elevation information of the target areas; Acquiring single-view complex images of the target area at different times within a preset evaluation period, pairing the single-view complex images to obtain a plurality of candidate interference image pairs, and optimizing the candidate interference image pairs to obtain a plurality of target interference image pairs; For any target interference image pair, an interference fringe pattern is generated based on the target interference image pair, and the interference phase in the interference fringe pattern is converted into elevation information using a phase-to-elevation conversion method. The terrain phase is simulated and removed from the interference fringe pattern using the converted elevation information and the original elevation information to obtain a deformation phase pattern. The deformation phase patterns of each target interference image pair are combined to obtain deformation data of the target area within a preset evaluation period, wherein the interference fringe pattern includes an interference phase, which is composed of a terrain phase and a deformation phase. Calculating the pore water pressure of the target area within a preset evaluation period; calculating the slope sliding velocity of the target area at different times within the preset evaluation period based on the pore water pressure and the landslide dynamic equilibrium differential equation; reclassifying the slope sliding velocity at different times based on the natural discontinuity method to obtain the slope collapse rate of the target area at different times within the preset evaluation period, wherein the slope collapse rate includes extremely fast collapse, relatively fast collapse, moderate collapse, relatively slow collapse, and extremely slow collapse; Acquiring settlement driving factor data of the target area within a preset evaluation period, and predicting the probability of a landslide occurring in the target area within the preset evaluation period based on the deformation data, the settlement driving factor data, and a preset landslide probability prediction model to obtain a landslide probability prediction result, wherein the settlement driving factor data includes data of multiple preset settlement driving factors; The landslide probability prediction results of the target area within the preset evaluation period and the slope collapse rates at different times are combined to obtain the settlement risk assessment results of the target area within the preset evaluation period.

2. The method according to claim 1, characterized in that The landslide probability prediction results of the target area within the preset evaluation period and the slope collapse rate at different times are integrated to obtain the settlement risk assessment results of the target area within the preset evaluation period, including: If the landslide probability prediction result shows that the probability of landslide occurring in the target area falls within the preset low probability range, the settlement risk assessment result of the target area is low settlement risk; If the landslide probability prediction result shows that the probability of landslide in the target area does not fall within the preset small probability range, the settlement risk assessment result of the target area is that there is a settlement risk, and based on the slope collapse rate of the target area at different times within the preset evaluation period, a settlement risk warning information is generated and sent to the preset receiving terminal.

3. The method according to claim 1, characterized in that Before predicting the probability of a landslide occurring in the target area within a preset evaluation period based on the deformation data, the settlement driving factor data, and a preset landslide probability prediction model, the method further includes: Based on the historical settlement driving factor data and historical deformation data of the target area, a random forest model, a support vector machine model and an extreme gradient boosting model are trained respectively, and the model with the highest prediction accuracy is used as the landslide probability prediction model.

4. The method according to claim 1, wherein The optimizing process on the candidate interference image pairs to obtain a plurality of target interference image pairs includes: For any candidate interference image pair, calculating the similarity between two single-view complex images in the candidate interference image pair to obtain a coherence coefficient map, performing conjugate multiplication on the two single-view complex images in the candidate interference image pair to obtain an interference fringe map, and performing phase unwrapping processing on the interference fringe map using a minimum cost flow algorithm of a Delaunay triangulation to obtain an unwrapped map; Among multiple candidate interference image pairs, candidate interference image pairs corresponding to coherence coefficient maps with low correlation, candidate interference image pairs corresponding to interference fringe maps with low interference quality, and candidate interference image pairs corresponding to detangling maps with low detangling quality are removed to obtain multiple target interference image pairs.

5. The method according to claim 1, wherein The pairing of the single-view complex images to obtain a plurality of candidate interference image pairs includes: Two single-view complex images acquired with a time interval not exceeding a preset time and a spatial distance not exceeding a preset distance are determined as a candidate interference image pair, until a plurality of candidate interference image pairs are determined.

6. The method according to claim 1, characterized in that The demand-based assessment of subsidence risk of transmission lines and determination of target areas include: The transmission lines for which the subsidence risk needs to be evaluated are determined, and a target area is obtained by expanding a preset width around the transmission lines for which the subsidence risk needs to be evaluated.

7. The method according to claim 1, characterized in that The calculating of the pore water pressure of the target area within a preset evaluation period, and the calculating of the slope sliding velocity of the target area at different times within the preset evaluation period based on the pore water pressure and the landslide dynamic equilibrium differential equation, include: The pore water pressure of the target area within a preset evaluation period is calculated based on the pore water pressure calculation formula. The slope sliding velocity of the target area at different times within the preset evaluation period is calculated based on the pore water pressure and the landslide dynamic equilibrium differential equation. The pore water pressure calculation formula is: U=γ w h w , The landslide dynamic equilibrium differential equation formula is: U is the pore water pressure, γ w is the unit weight of pore water, h w is the height of pore water in the soil, α is the inclination of the sliding surface of the landslide, γ* is the equivalent unit weight of the landslide, H is the thickness of the landslide, μ ss is the steady-state friction coefficient of the sliding surface of the landslide, g is the acceleration of gravity, v is the sliding velocity of the slope relative to the land surface, and t is time.

8. The method according to claim 1, characterized in that The preset settlement driving factors include slope, lithology, profile curvature, plane curvature, distance from roads, vegetation coverage, intensity of human activities, distance from rivers and precipitation.

9. A storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the method for evaluating transmission line settlement risk according to any one of claims 1 to 8 is implemented.

10. A computer device comprising a storage medium, a processor, and a computer program stored in the storage medium and executable on the processor, wherein: When the processor executes the computer program, the method for evaluating transmission line subsidence risk according to any one of claims 1 to 8 is implemented.

Images