Method for intelligently monitoring geological disasters based on satellite remote sensing

Abstract

The invention relates to the technical field of geological monitoring, in particular to a method for intelligently monitoring geological disasters based on satellite remote sensing, which comprises the following steps of: obtaining corresponding earth surface coverage information through image registration and ortho-rectification processing; identifying terrain boundaries, geological structure features and geological attribute information through the land surface coverage information; recognizing surrounding hidden danger points through a topographic feature extraction method; an image fusion enhancement method is matched with a permanent scatterer interference method, a stable scatterer is extracted, a deformation monitoring network is constructed, and a phase difference value in a high-risk area is obtained through calculation; carrying out unwrapping processing through the phase difference value, and carrying out inversion to obtain earth surface three-dimensional deformation field information in the monitoring area; establishing a geological disaster risk assessment model through space-time clustering analysis; by establishing a risk assessment model, the geological disaster risk in the monitoring area is quantitatively assessed. The beneficial effect of the invention is that comprehensive monitoring and accurate early warning of geological hazard hidden dangers around the transformer substation are realized.

Description

Technical Field

[0001] The invention relates to the technical field of geological monitoring, and in particular to a method for intelligently monitoring geological disasters based on satellite remote sensing. Background Art

[0002] With the acceleration of urbanization, the surrounding environment of important infrastructure such as substations has become more complex, and the risk of geological disasters has become increasingly prominent. Geological disasters such as landslides, collapses and ground subsidence not only threaten personal safety and property, but may also lead to power supply interruptions, seriously affecting social and economic stability and development. In recent years, the rapid development of remote sensing technology has provided new means for geological disaster monitoring. The combination of high-resolution multispectral satellite images and synthetic aperture radar (SAR) images makes the extraction of surface cover changes and geological features more accurate. Through the analysis of multi-temporal images, potential geological disaster risk areas can be effectively identified, so as to conduct risk assessment and early warning. In this process, the integration of visible light image intelligent analysis and geological disaster risk assessment faces multiple technical challenges. After obtaining high-resolution multispectral satellite images, how to achieve accurate surface cover information extraction and accurate ground object classification through image registration and orthorectification. When extracting surface texture information and identifying terrain boundaries, how to effectively combine edge detection and segmentation technology to improve the identification accuracy of potential geological disaster risk points. Finally, when fusing optical images with synthetic aperture radar images for surface deformation monitoring, it is necessary to improve the interferometric processing accuracy of permanent scatterer interferometric radar to obtain more accurate three-dimensional surface deformation field information. Solving these interrelated technical problems is crucial to achieving accurate geological hazard risk assessment. Summary of the invention

[0003] In view of the above problems or problems existing in the prior art, the present invention is proposed.

[0004] Therefore, the purpose of the present invention is to provide a method for intelligently monitoring geological disasters based on satellite remote sensing, which can accurately identify, assess risks and provide early warning for potential geological disaster risk areas.

[0005] In order to solve the above technical problems, the present invention provides the following technical solutions: a method for intelligently monitoring geological disasters based on satellite remote sensing, which comprises obtaining corresponding surface coverage information through image registration and orthorectification processing;

[0006] Identify terrain boundaries and geological structure features through surface cover information;

[0007] Extract geological attribute information of lithology and stratigraphic occurrence through terrain boundaries and geological structural features;

[0008] Identify surrounding hidden danger points through terrain feature extraction method;

[0009] Through the image fusion enhancement method combined with the permanent scatterer interference method, stable scatterers are extracted and a deformation monitoring network is constructed to calculate the phase difference value in the high-risk area.

[0010] Through phase difference unwrapping processing, combined with orbital parameters and atmospheric phase, the three-dimensional surface deformation field information in the monitoring area is inverted, and the cumulative deformation is calculated using time series analysis methods.

[0011] Establish a geological disaster risk assessment model through spatiotemporal cluster analysis;

[0012] By establishing a risk assessment model, the geological disaster risks in the monitoring area are quantitatively assessed, and risk levels are divided to generate early warning information.

[0013] As a preferred solution of the method for intelligently monitoring geological disasters based on satellite remote sensing of the present invention, the image registration and orthorectification processing includes: using a high-resolution multispectral satellite to obtain multi-time period multispectral image data and synthetic aperture radar images of the area surrounding the substation;

[0014] Multispectral image data includes visible, near-infrared, and short-wave infrared bands;

[0015] The synthetic aperture radar image includes the radar echo signal of the area surrounding the substation and the corresponding radar image data.

[0016] As a preferred solution of the method for intelligently monitoring geological disasters based on satellite remote sensing of the present invention, the surface coverage information includes extracting surface texture information by a texture feature extraction method, combining edge detection and segmentation methods to identify terrain boundaries and geological structure features;

[0017] The texture feature extraction method includes calculating the texture feature by using the gray level co-occurrence matrix to obtain the surface texture information, and the surface texture information is the surface texture feature map;

[0018] The edge detection and segmentation method includes detecting the enhanced texture feature map to obtain the terrain boundary segments; according to the terrain boundary segments, the watershed algorithm is used to segment the edge detection results, the minimum area threshold is set, and the areas smaller than the threshold are merged to obtain the geological structure feature areas.

[0019] As a preferred solution of the method for intelligently monitoring geological disasters based on satellite remote sensing of the present invention, the watershed algorithm includes gradient calculation, marking minimum value, flood filling, water confluence point meeting, post-processing, minimum area threshold and morphological operation.

[0020] As a preferred solution of the method for intelligently monitoring geological disasters based on satellite remote sensing of the present invention, the geological attribute information of lithology and stratum dip includes extracting shape index and directional characteristics from the geological structure characteristic area, and calculating the circularity and main direction angle of each segmented area, combining the strike information of the terrain boundary segment, identifying the lithology category and stratum dip, judging the stratum dip, and obtaining the geological attribute information.

[0021] As a preferred solution of the method for intelligently monitoring geological disasters based on satellite remote sensing of the present invention, the terrain feature extraction method includes extracting terrain slope and slope aspect information from geological attribute information;

[0022] The surrounding hidden danger points include high-risk areas prone to geological disasters such as landslides, collapses and ground subsidence, which are identified through terrain slope and slope aspect information to form a distribution map of hidden danger points around the substation.

[0023] As a preferred solution of the method for intelligently monitoring geological disasters based on satellite remote sensing of the present invention, the image fusion enhancement method includes: using discrete wavelet transform to fuse the optical image and the synthetic aperture radar image, taking the average value of the low-frequency coefficient, selecting the high-frequency coefficient by the absolute value maximum method, extracting the high-frequency and low-frequency information by multi-scale decomposition, fusing and reconstructing by using different weight coefficients, obtaining a fused high-resolution image, and geo-registering the fused high-resolution image with the hidden danger point distribution map;

[0024] The permanent scatterer interference method includes selecting points in the high-risk area with an amplitude deviation index less than a preset threshold and a coherence greater than a preset threshold as candidate permanent scatterers based on the registered fusion image and the hidden danger point distribution map, screening the candidate permanent scatterers, and performing time series interference processing on the screened stable scatterers.

[0025] As a preferred solution of the method of satellite remote sensing intelligent monitoring of geological disasters of the present invention, the phase difference unwrapping process includes: using the least squares phase unwrapping algorithm to unwrap the phase difference to obtain a continuous phase field. According to the continuous phase field, atmospheric phase correction is performed by combining the tropospheric delay model and the ionospheric delay model to obtain the corrected phase information; the conversion matrix between the line of sight deformation and the three-dimensional deformation component is solved by the least squares method, and the corrected phase information is converted into a three-dimensional surface deformation field, and the three-dimensional surface deformation field is decomposed at multiple scales, the long-term trend term is extracted, and the cumulative deformation amount is calculated.

[0026] As a preferred solution of the method of satellite remote sensing intelligent monitoring of geological disasters of the present invention, the spatiotemporal clustering analysis includes extracting the spatial distribution characteristics and time evolution laws of abnormal deformation areas, combining historical disaster data statistics, meteorological factors and geological attribute information.

[0027] As a preferred solution of the method for satellite remote sensing intelligent monitoring of geological disasters of the present invention, the risk assessment model includes a quantitative assessment of the risks of landslides, collapses, and ground subsidence geological disasters in the monitoring area, divides the risk levels and generates multiple types of hidden danger warning information, and the multiple types of hidden danger warning information include risk type, level, impact range and recommended measures.

[0028] The beneficial effects of the invention are as follows: the invention can more accurately extract surface cover changes and geological features through the combination of high-resolution multispectral satellite images and SAR images, effectively identify potential geological disaster risk areas, and also help to conduct risk assessment and early warning of geological disasters, improve the monitoring capabilities of disasters such as landslides, collapses and ground subsidence, thereby protecting personal safety, property and power supply, and promoting social and economic stability and development. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for describing the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work. Among them:

[0030] Figure 1 The figure is a flowchart of the method for intelligent monitoring of geological disasters based on satellite remote sensing. DETAILED DESCRIPTION

[0031] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are described in detail below in conjunction with the accompanying drawings.

[0032] In the following description, many specific details are set forth to facilitate a full understanding of the present invention, but the present invention may also be implemented in other ways different from those described herein, and those skilled in the art may make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0033] Secondly, the term "one 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 term "in one embodiment" that appears in different places in this specification does not necessarily refer to the same embodiment, nor does it refer to a separate or selective embodiment that is mutually exclusive with other embodiments.

[0034] Example

[0035] Reference Figure 1, which is an embodiment of the present invention, provides a method for intelligently monitoring geological disasters based on satellite remote sensing, which can accurately identify, assess risks and provide early warning for potential geological disaster risk areas.

[0036] Specifically, the corresponding surface coverage information is obtained through image registration and orthorectification processing;

[0037] Identify terrain boundaries and geological structure features through surface cover information;

[0038] Extract geological attribute information of lithology and stratigraphic occurrence through terrain boundaries and geological structural features;

[0039] Identify surrounding hidden danger points through terrain feature extraction method;

[0040] Through the image fusion enhancement method combined with the permanent scatterer interference method, stable scatterers are extracted and a deformation monitoring network is constructed to calculate the phase difference value in the high-risk area.

[0041] Through phase difference unwrapping processing, combined with orbital parameters and atmospheric phase, the three-dimensional surface deformation field information in the monitoring area is inverted, and the cumulative deformation is calculated using time series analysis methods.

[0042] Establish a geological disaster risk assessment model through spatiotemporal cluster analysis;

[0043] By establishing a risk assessment model, the geological disaster risks in the monitoring area are quantitatively assessed, risk levels are divided and early warning information is generated.

[0044] Furthermore, the image registration and orthorectification processing includes using high-resolution multispectral satellites to obtain multi-time period multispectral image data and synthetic aperture radar images of the area surrounding the substation;

[0045] Multispectral image data includes visible, near-infrared, and short-wave infrared bands;

[0046] The synthetic aperture radar image includes the radar echo signal of the area surrounding the substation and the corresponding radar image data.

[0047] Specifically, a high-resolution multispectral satellite sensor is used to image the area around the substation in multiple time periods, and multispectral image data containing visible light, near infrared and short-wave infrared bands are obtained once a quarter. At the same time, a synthetic aperture radar system is used to collect radar echo signals in the area to generate corresponding radar images. For the acquired multispectral images and radar images, the SIFT (Scale Invariant Feature Transform) algorithm is used to extract image feature points, which includes: multi-scale analysis of the input image, creating a series of spatial images of different scales, comparing each pixel between adjacent scales, and finding the position of the local maximum in the scale space as the key point; fitting the candidate key points, removing the key points with edge response and low contrast; calculating the gradient histogram of the neighborhood around the key point, and assigning one or more main directions to each key point; calculating the gradient direction histogram in the neighborhood of each key point, and quantizing it into a vector of fixed length to form a descriptor; then using R The ANSAC (random sample consistency) algorithm removes mismatched points, establishes a geometric transformation model between images, realizes accurate registration of multi-source heterogeneous images, and eliminates geometric deviations between images, including: randomly selecting a minimum number of matching point pairs (at least 3 pairs of points are required for affine transformation) to estimate transformation parameters; using the estimated transformation parameters to check whether the remaining matching point pairs conform to the transformation, and the matching point pairs are called inliers; iterative optimization repeats the above steps multiple times, each time selecting a different matching point pair, and finding the transformation parameter set with the most inliers; the final model construction uses the largest set of inliers to re-estimate the transformation parameters and obtain the final geometric transformation model. According to the SRTM (shuttle radar topography mission) digital elevation model data and the RPCs (rational polynomial coefficients) model, the registered multispectral images and radar images are orthorectified to eliminate the geometric deformation caused by terrain undulation and imaging perspective, and to obtain orthophotos with a unified geographic coordinate system; orthorectification includes: loading SRTM DEM data to ensure that its coverage matches the multispectral images and radar images to be corrected; extracting RPC coefficients from the image metadata, which are used to describe the imaging geometry of the sensor; using DEM data to calculate the terrain height corresponding to each pixel, this step is necessary because the height of the terrain will affect the geometry of the image; using the RPC model to reversely project each pixel into the ground coordinate system, this step takes into account the influence of the terrain as well as the attitude and position of the sensor; in order to generate an orthophoto, the result of the reverse projection needs to be resampled so that it falls on a standard grid. The resampling methods include the nearest neighbor method, bilinear interpolation and cubic convolution interpolation. The final output orthophoto is terrain-corrected and perspective-corrected, eliminating the geometric deformation caused by terrain undulation and imaging perspective. All pixels are located on the same horizontal plane and have a unified geographic coordinate system.The atmospheric correction and radiation calibration are performed on the multi-spectral images after orthorectification to eliminate the atmospheric influence and sensor error, improve the spectral information quality of the images, and then the principal component analysis is used to reduce the dimension of the multi-band images and extract the main feature information. The multi-temporal orthophoto data are used to construct the surface cover classification features, and the support vector machine classification algorithm based on the radial basis function kernel is used. The optimal penalty parameter and kernel function parameter are determined through grid search. The image features after dimensionality reduction are trained and classified to identify the surface cover type in the area around the substation and generate a high-precision surface cover classification map. The high-resolution multi-spectral satellite sensor collects images within 10 kilometers around the substation, with a spatial resolution of 0.5 meters, including four bands of blue, green, red and near infrared, and is acquired once a quarter. The X-band synthetic aperture radar system is used simultaneously with an operating frequency of 9.6GHz to obtain radar images. The SIFT algorithm is applied to the acquired multi-source images, and the threshold of the number of feature points is set to 1000. After extracting the feature points, the RANSAC algorithm is used to eliminate the mismatched points, and the threshold of the internal point ratio is set to 0.8. The SRTM digital elevation model with a vertical accuracy of 16 meters and a horizontal resolution of 30 meters was used to perform orthorectification in combination with the RPC model to obtain orthophoto images with a geographic coordinate system of WGS84. The FLAASH atmospheric correction model was used for the orthophoto multispectral images, and the mid-latitude summer atmospheric model was selected. The visibility was set to 40 kilometers for atmospheric correction and radiation calibration. The principal component analysis method was used to reduce the dimension of the corrected multi-band images, and the principal components with a cumulative contribution rate of 95% were retained. The surface cover classification features including the normalized vegetation index (NDVI), the normalized water index (NDWI) and the normalized building index (NDBI) were constructed. The support vector machine with a radial basis function kernel was used for classification. The optimal penalty parameter C was 100 and the kernel function parameter γ was 0.01 through five-fold cross validation and grid search. Finally, a surface cover classification map with a spatial resolution of 0.5 meters was generated, including four types of objects: water bodies, vegetation, buildings and bare land.

[0048] Among them, SIFT algorithm is a feature extraction method used to detect and describe key points in images at different scales and directions. The main parameters of SIFT algorithm include scale threshold, edge threshold and key point threshold, which are used to determine the stability and uniqueness of key points;

[0049] The RANSAC algorithm is an iterative method for estimating the parameters of a mathematical model from a set of data containing outliers. The main parameters of RANSAC include the number of iterations, the inlier threshold, and the model fit threshold. These parameters determine the robustness and accuracy of the algorithm;

[0050] The SRTM digital elevation model provides high-resolution terrain data for the world. SRTM data is usually provided in the form of a digital elevation model (DEM), with parameters including resolution and accuracy.

[0051] The RPCs model is a set of polynomial coefficients used for geometric correction of satellite images. The RPCs parameters include coefficients of linear, quadratic, and cubic terms, which are used to convert image pixel coordinates into ground coordinates.

[0052] The NDVI formula is as follows:

[0053]

[0054] Among them, NIR is the reflectivity of the near-infrared band, and R is the reflectivity of the red light band;

[0055] The NDWI formula is as follows:

[0056]

[0057] Where G is the reflectivity of the green light band;

[0058] The NDBI formula is as follows:

[0059]

[0060] Among them, MIR is the reflectivity of the mid-wave infrared band, and SWIR1 is the reflectivity of the short-wave infrared band;

[0061] The FLAASH atmospheric correction model can remove atmospheric effects from remote sensing images to obtain more accurate surface reflectance. The FLAASH atmospheric correction model is based on the physical atmospheric radiation transfer model and takes into account the effects of gases, aerosols and clouds in the atmosphere. The formula is as follows:

[0062] I(λ)=I0(λ)·τ(λ)+(1-τ(λ))·L(λ)

[0063] Where I(λ) is the observed radiance (including atmospheric path radiation), I0(λ) is the radiance at the top of the atmosphere (i.e., surface reflected radiation), τ(λ) is the atmospheric transparency, which represents the atmospheric attenuation of radiation, and L(λ) is the atmospheric path radiation.

[0064] Further, the surface cover information includes extracting surface texture information by texture feature extraction method, combining edge detection and segmentation method to identify terrain boundaries and geological structure features;

[0065] The texture feature extraction method includes calculating the texture feature by using the gray level co-occurrence matrix to obtain the surface texture information, and the surface texture information is the surface texture feature map;

[0066] The edge detection and segmentation method includes detecting the enhanced texture feature map to obtain the terrain boundary segments; according to the terrain boundary segments, the watershed algorithm is used to segment the edge detection results, the minimum area threshold is set, and the areas smaller than the threshold are merged to obtain the geological structure feature areas.

[0067] Furthermore, the watershed algorithm includes gradient calculation, marking minimum, flood filling, watershed meeting, post-processing, minimum area threshold and morphological operations.

[0068] Furthermore, the geological attribute information of lithology and stratigraphic occurrence includes extracting shape index and directional characteristics from the geological structure characteristic area, calculating the circularity and main direction angle of each segmented area, combining the strike information of the terrain boundary segment, identifying the lithology category and stratigraphic dip, judging the stratigraphic occurrence, and obtaining the geological attribute information.

[0069] Specifically, according to the acquired surface coverage information, the gray-level co-occurrence matrix is ​​used to calculate the texture features. By setting the sliding window size to 9×9 pixels, the second-order moment, contrast, entropy and correlation statistics in the four directions of 0°, 45°, 90° and 135° are calculated to obtain the surface texture feature map. The texture feature map is processed by adaptive histogram equalization to improve the image contrast and enhance the texture details. For the enhanced texture feature map, the Canny edge detection operator is used for edge detection, in which the Canny edge detection operator is used to process the texture feature map, including: Gaussian filtering denoising, calculating gradient strength and direction, non-maximum suppression and dual threshold detection; the OTSU algorithm is used to adaptively calculate the high and low thresholds, and the terrain boundary segments are detected. The OTSU algorithm adaptively determines the high and low thresholds required by the Canny edge detection operator by maximizing the inter-class variance between the foreground and the background. The distance transform is used to calculate the marker map, which is used to indicate the seed point or initial segmentation area in the image. The distance transform can measure the distance from each pixel to the nearest edge point, and then the watershed algorithm is applied to the edge detection result for image segmentation. The image is regarded as terrain, where the pixel intensity represents the altitude. The goal of segmentation is to find the "valleys" and "peaks" in the image, that is, the regional boundaries. The minimum regional area threshold is set to 100 square pixels to filter out small and unimportant areas, and merge those areas with an area smaller than the threshold to obtain geological structural feature areas. The shape index and directional characteristics are extracted from the segmented geological structural feature areas, and the circularity and main direction angle of each segmented area are calculated. Combined with the trend information of the terrain boundary segment, the lithology category and formation dip are identified through the support vector machine classifier, the formation occurrence is judged, and the geological attribute information is obtained. Use known geological sample data to train the support vector machine classifier, use the radial basis kernel function, and determine the optimal parameters through grid search, including: ensure that the geological sample data has been preprocessed, such as cleaning, standardization / normalization, etc., so that they are suitable for training models. The data should be divided into features (X) and labels (y), where features are the data sets used for training and labels are the corresponding categories or target variables; select the support vector machine classifier and specify the use of the RBF kernel function. The RBF kernel function is very suitable for processing nonlinearly separable data sets because it can map the original feature space to a higher-dimensional space, where it is easier to find a suitable hyperplane to divide the data; SVM has two key hyperparameters that need to be adjusted: one is the regularization parameter C, and the other is the width parameter gamma of the RBF kernel function. The selection of these parameters will affect the generalization ability of the model; in order to find the optimal parameter combination, the grid search will traverse all possible parameter combinations, and cross-validation will be used to evaluate the effect of each combination. According to geological principles and combined with known regional geological background information, the classification results are post-processed to correct abnormal classification results that do not conform to geological laws and improve the accuracy of geological attribute information.The texture features of the surface cover information were calculated using the gray-level co-occurrence matrix. A sliding window of 9×9 pixels was set to calculate the angular second-order moment, contrast, entropy and correlation in the four directions of 0°, 45°, 90° and 135°, respectively, to obtain a 16-dimensional texture feature vector. Adaptive histogram equalization was applied to the texture feature map, and the image block size was set to 8×8 pixels and the contrast limit threshold was set to 0.02 to enhance the texture details. The Canny edge detection operator was used for edge detection, and the high and low thresholds were adaptively calculated by the OTSU algorithm. In typical cases, the high threshold was about 120 and the low threshold was about 60. The distance transform was used to calculate the label map, and the Euclidean distance from the pixel to the boundary was used as the transformation value. Then, the watershed algorithm was applied for image segmentation, and the minimum area threshold was set to 100 square pixels. The shape index and directional features were extracted from the segmented area, and the circularity (4π×area / perimeter2) and the main direction angle (ellipse fitting based on the regional second-order moment) were calculated. The support vector machine classifier is used to identify lithology and formation occurrence. The training samples include 5 common lithologies and 3 formation occurrences. The radial basis kernel function is used. The penalty factor C is 10 and the kernel function parameter γ is 0.01 through grid search. According to the regional geological background information, such as the known formation trend of 45° northeast, the classification results are post-processed, and the abnormal classification results that deviate from the main trend by more than 30° are corrected to improve the accuracy of geological attribute information to more than 90%.

[0070] Among them, the main parameters of the Canny edge detection operator include the standard deviation of the Gaussian filter, the gradient threshold and the suppression threshold;

[0071] The OTSU algorithm is an automatic image threshold segmentation method, and its formula is as follows:

[0072]

[0073] Among them, the main parameter of the OTSU algorithm is the threshold T, which divides the histogram of the image into foreground and background so that the intra-class variance between them is minimized.

[0074] Further, the terrain feature extraction method includes extracting terrain slope and aspect information from geological attribute information;

[0075] The surrounding hidden danger points include high-risk areas prone to geological disasters such as landslides, collapses and ground subsidence, which are identified through terrain slope and slope aspect information to form a distribution map of hidden danger points around the substation.

[0076] Specifically, the terrain slope is calculated using the finite difference method based on the elevation data in the geological attribute information. The slope value is obtained by the ratio of the elevation difference of adjacent grid cells to the horizontal distance. The slope is divided into five levels: 0-5°, 5-15°, 15-25°, 25-35°, and >35°, which are used to comprehensively assess potential risk areas. The terrain slope is calculated using the elevation data, and the slope angle is determined by the elevation gradient of the grid cells in the east-west and north-south directions. The slope value of 0 to 360 degrees is obtained and evenly divided into eight 45° fan-shaped areas, corresponding to the eight slope categories of north, northeast, east, southeast, south, southwest, west, and northwest. Combined with slope, aspect information and geological properties, a random forest algorithm is used to establish a geological hazard risk assessment model; data cleaning and feature engineering are carried out through historical disaster point data obtained from the geological department, model parameters are trained, and the probability of landslides, collapses and ground subsidence is calculated, including: obtaining location information of landslides, collapses and ground subsidence events in historical records from the geological department; obtaining DEM data of the study area, from which information such as slope and aspect is calculated; collecting geological attribute data such as soil type, rock formation properties, groundwater level, etc.; checking whether there are missing values ​​in the data set and deciding how to deal with these missing values ​​(deletion, filling, etc.); identifying and processing outliers to avoid their negative impact on model training, ensuring the consistency and reliability of all data sources. reliability; extract slope and aspect information from DEM data, these two features are very important for the risk assessment of geological hazards; extract useful features such as soil permeability and rock compressive strength based on geological attribute data; determine which features are most important for predicting geological hazards through correlation analysis or other feature selection methods; divide the data set into training set and test set, and select the random forest algorithm as the geological hazard risk assessment model; adjust the hyperparameters of the random forest, such as the number of trees, maximum depth, etc., through methods such as cross-validation, and use the training set data to train the random forest model; evaluate the performance of the model on the test set, such as accuracy, recall, F1 score, etc.; for each test sample, the model will output the probability of occurrence of landslides, collapses, and ground subsidence. According to the calculation results of the risk assessment model, the risk threshold is set, and the 90% quantile is selected as the high-risk threshold to identify high-risk areas. The kernel density estimation method is used to perform spatial clustering analysis on high-risk points. GIS software or other visualization tools are used to draw a heat map based on the results of density estimation to display the distribution of potential danger points around the substation, including: using the selected bandwidth to perform density estimation on high-risk points to obtain the density value at each point; using the density value obtained by KDE, high-density areas can be identified. These areas are potential potential danger point clusters. The density threshold can be used to determine which areas belong to the same cluster, thereby identifying the potential danger point clusters in space.Combined with the location information of the substation and the distribution of surrounding infrastructure, a secondary assessment of high-risk areas is conducted, taking into account the impact of human activities on the geological environment, such as excavation, filling and other engineering activities, and adjusting the risk assessment results through weighted overlay analysis to form the final distribution map of hidden danger points around the substation. Based on the digital elevation model with a resolution of 10 meters, the finite difference method is used to calculate the terrain slope, and the maximum slope value of the central pixel is calculated using a 3×3 pixel window. The slope is divided into five levels: 0-5°, 5-15°, 15-25°, 25-35°, and >35°, and the risk value is assigned from 1 to 5. The terrain slope is calculated using the same elevation data, and the slope angle is calculated using the Zevenbergen&Thorne algorithm to obtain the slope value of 0 to 360 degrees, which is evenly divided into eight 45° fan-shaped areas, corresponding to the eight slope categories of north, northeast, east, southeast, south, southwest, west, and northwest, and the coding values ​​​​are assigned from 1 to 8 respectively. Combined with the slope, aspect information and geological attributes such as lithology and formation dip, a random forest model was constructed, using 500 decision trees, with a maximum tree depth of 10 and a minimum number of leaf node samples of 5. The data of 100 historical disaster points in the past 20 years provided by the geological department were divided into 80% training set and 20% test set, and the probability of landslide, collapse and ground subsidence was obtained after cross-validation. The 90% quantile was set as the high-risk threshold, and the high-risk points were spatially clustered by the kernel density estimation method (bandwidth set to 1000 meters) to generate a 100×100-meter grid distribution heat map of hidden danger points. Combined with the engineering activity information within the 500-meter buffer zone of the substation, such as excavation depth, fill height, etc., an adjustment coefficient of 0.8-1.2 was set, and the risk assessment results were fine-tuned through weighted overlay analysis, and finally a 1:10000 scale distribution map of hidden danger points around the substation was formed.

[0077] Among them, the Zevenbergen & Thorne algorithm includes slope calculation, flow direction determination, water flow accumulation, threshold setting, river network extraction, and river network refinement;

[0078] The slope calculation formula is as follows:

[0079]

[0080] Among them, z i,j is the elevation of cell (i, j), Δx and Δy are the dimensions of the cell.

[0081] Furthermore, the image fusion enhancement method includes: using discrete wavelet transform to fuse the optical image and the synthetic aperture radar image, taking the average value of the low-frequency coefficients, selecting the high-frequency coefficients by the absolute value maximum method, extracting the high-frequency and low-frequency information by multi-scale decomposition, fusing and reconstructing using different weight coefficients to obtain a fused high-resolution image, and geo-registering the fused high-resolution image with the hidden danger point distribution map;

[0082] The permanent scatterer interference method includes selecting points in the high-risk area with an amplitude deviation index less than a preset threshold and a coherence greater than a preset threshold as candidate permanent scatterers based on the registered fused image and the hidden danger point distribution map, and screening the candidate permanent scatterers, and performing time series interference processing on the screened stable scatterers. Specifically, a deformation monitoring network is constructed using a spatial Delaunay triangulation network, and the network adjustment calculation is performed using the least squares adjustment method to obtain the relative phase difference between adjacent monitoring points in the high-risk area.

[0083] Specifically, the optical image and the synthetic aperture radar image are fused by discrete wavelet transform, and the db4 wavelet basis function is used for three-layer decomposition. The low-frequency coefficients are averaged, and the high-frequency coefficients are selected by the absolute maximum method. The high-frequency and low-frequency information are extracted by multi-scale decomposition, and fusion and reconstruction are performed using different weight coefficients to obtain a fused high-resolution image, which includes: performing three-layer wavelet decomposition on the two original images (optical image and SAR image); for the low-frequency coefficients of each layer, the average value of the low-frequency coefficients corresponding to the two images is taken; for the high-frequency coefficients of each layer, the absolute maximum method is used (the coefficient with the larger absolute value is selected as the final fusion coefficient); after completing the fusion of the low-frequency and high-frequency coefficients, the information extracted by multi-scale decomposition is used to reconstruct the fusion coefficients by inverse discrete wavelet transform (IDWT) to obtain a fused high-resolution image. The fused high-resolution image is geo-referenced with the distribution map of hidden danger points, and the affine transformation model and the nearest neighbor resampling method are used to ensure that the two correspond accurately in spatial position. According to the registered fusion image and the distribution map of hidden danger points, points with amplitude deviation index less than 0.3 and coherence greater than 0.7 in the high-risk area are selected as candidate permanent scatterers. The phase and baseline relationship of the scatterer is fitted by the least squares method, and the pseudo scattering points with residuals greater than the preset threshold are eliminated. The screened stable scatterers are subjected to time series interference processing, and the cumulative deformation and deformation rate of each monitoring point are obtained by constructing interference pattern pairs and using the minimum cost flow algorithm for phase unwrapping.The deformation monitoring network is constructed using the spatial Delaunay triangulation network. The network adjustment calculation is performed using the least squares adjustment method to obtain the relative phase difference between adjacent monitoring points in the high-risk area, and to determine the regional deformation trend and potential risks, including: selecting a certain number of monitoring points, which should be evenly distributed in the monitoring area, and increasing the density in the high-risk area to ensure sufficient monitoring accuracy; installing appropriate monitoring equipment at the selected monitoring points, such as GNSS receivers or other deformation monitoring instruments; using the Delaunay triangulation algorithm to connect these monitoring points to form a triangular network without crossing boundaries. Delaunay triangulation ensures that the maximum gap circle of each triangle does not contain any other monitoring points; after the triangulation network is constructed, the three vertices in each triangle constitute an interconnected monitoring unit, which is convenient for subsequent deformation analysis; regularly or continuously collecting phase measurement data from each monitoring point, which can be used to Carrier phase observations from GNSS receivers; Calculate the baseline vector of each triangle side (i.e., the line between two monitoring points); Calculate the relative phase difference of each baseline vector using the least squares method, which reflects the relative movement between two points; Perform adjustment calculations on the entire monitoring network using the least squares method to reduce measurement errors and improve the accuracy of monitoring results; Based on the adjustment calculation results, analyze the relative phase difference within each triangle to determine the deformation trend in the area; Identify possible deformation trends by observing the changes in the relative phase difference between monitoring points in different time periods; Evaluate potential risk levels based on deformation trends, for example, if the monitoring points in a certain area show a consistent moving direction, it may indicate that there is a deformation risk in the area; Summarize the analysis results and prepare a detailed deformation monitoring report to provide decision-making support for relevant departments to take necessary preventive measures or emergency response measures. Use discrete wavelet transform to fuse 10-meter resolution optical images with 30-meter resolution synthetic aperture radar images, use db4 wavelet basis function for 3-layer decomposition, average the low-frequency coefficients, and select the high-frequency coefficients using the absolute maximum method, and fuse and reconstruct to obtain a high-resolution image with a resolution of 10 meters. The fused image was geo-referenced with the 1:10000 scale distribution map of potential hazards. The 6-parameter affine transformation model and the nearest neighbor resampling method were used, and the registration accuracy was better than 0.5 pixels. Points with an amplitude deviation index less than 0.3 and a coherence greater than 0.7 were selected as candidate permanent scatterers in high-risk areas, with an average of 50-100 points per square kilometer. The phase and baseline relationship of the scatterer was fitted using the least squares method, and the residual threshold was set to 0.5 radians to remove pseudo-scattering points that exceeded the threshold. The screened stable scatterers were subjected to time series interference processing. 190 interference pattern pairs were constructed using 20 SAR images, and the minimum cost flow algorithm was used for phase unwrapping to calculate the cumulative deformation and annual average deformation rate of each monitoring point.A deformation monitoring network was constructed using spatial Delaunay triangulation with a maximum side length of 500 meters. The network adjustment was performed using the least squares adjustment method to obtain the relative phase difference between adjacent monitoring points in high-risk areas with an accuracy better than 0.1 radians, which was used to judge regional deformation trends and potential risks.

[0084] Furthermore, the phase difference unwrapping process includes: using the least squares phase unwrapping algorithm to unwrap the phase difference to obtain a continuous phase field. According to the continuous phase field, atmospheric phase correction is performed by combining the tropospheric delay model and the ionospheric delay model to obtain the corrected phase information; the least squares method is used to solve the conversion matrix between the line-of-sight deformation and the three-dimensional deformation component, and the corrected phase information is converted into a three-dimensional surface deformation field, and the three-dimensional surface deformation field is decomposed at multiple scales to extract the long-term trend term and calculate the cumulative deformation.

[0085] Specifically, the least squares phase unwrapping algorithm is used to unwrap the phase difference value, the upper limit of the number of iterations is set to 50, the convergence threshold is set to 0.1 radians, and the phase jump is eliminated through iterative calculation to obtain a continuous phase field. According to the satellite orbit parameters and the atmospheric phase model, the atmospheric phase correction is performed by combining the tropospheric delay model and the ionospheric delay model, and the baseline is refined in combination with the ground GPS control point data to improve the deformation measurement accuracy, including: obtaining accurate satellite orbit files, such as precise ephemeris SP3 files, which contain the position information of the satellite at different time points; correcting the orbital position of the satellite according to the information in the orbit file to reduce the impact of the orbit error on the deformation measurement; collecting the data of the ground GPS control point, including the position of the receiver, the time of reception, and the received satellite signal, and preprocessing the collected data, such as removing gross errors and filtering; using the tropospheric delay model, according to the meteorological parameters such as atmospheric pressure, temperature and humidity in the direction of the satellite line of sight, the influence of the tropospheric delay is estimated; by subtracting the tropospheric delay value, the carrier phase observation value is corrected; using the ionospheric delay model, according to the electron content in the direction of the satellite line of sight, the influence of the ionospheric delay is estimated, so that Use a combination of dual-frequency observations to eliminate most ionospheric effects, or use single-frequency observations combined with an external ionospheric model for correction; use the corrected carrier phase observations to calculate the baseline vector, that is, the relative position between the two monitoring sites; use the least squares method or other optimization algorithms to adjust the baseline vector in the entire monitoring network to reduce measurement errors and improve accuracy; use ground GPS control points with known positions as constraints to further improve the accuracy of the baseline vector; incorporate the true coordinates of the control points into the adjustment calculation to correct the baseline vector, thereby obtaining more accurate coordinates of the monitoring points; through the above steps, the accuracy of deformation measurement can be significantly improved, which is specifically reflected in the following aspects: by correcting the delay of the troposphere and ionosphere, the influence of atmospheric effects on the received signal is reduced; the accuracy of the baseline vector is improved through baseline refinement and the constraints of ground control points; through network adjustment calculation, the overall stability of the monitoring network is enhanced, making the monitoring results more reliable. The least square method is used to solve the conversion matrix between the line-of-sight deformation and the three-dimensional deformation component, and the corrected phase information is converted into the three-dimensional deformation field of the surface, and the displacement components of each scattering point in the monitoring area in the east-west, north-south and vertical directions are calculated. Multi-scale decomposition is performed through db4 wavelet, long-term trend items are extracted, and time series processing is performed on the three-dimensional deformation field. The cumulative deformation is calculated, and the cumulative deformation threshold is set to 30 mm / year. According to the cumulative deformation and geological background information, the fuzzy comprehensive evaluation method is used to classify the potential hidden danger points, set high, medium and low risk levels, generate a hidden danger point distribution map, and determine the high-risk level area as a potential geological disaster hidden danger point.The phase difference unwrapping process was performed on 300,000 scattering points in the monitoring area. The least squares phase unwrapping algorithm was used, the upper limit of the number of iterations was set to 50, the convergence threshold was set to 0.1 radians, and a continuous phase field was obtained after averaging 15 iterations. The tropospheric delay model was constructed using ECMWF meteorological data, and the ionospheric delay correction was performed in combination with the global ionospheric TEC map. At the same time, the baseline was refined using 10 ground GPS control point data to improve the deformation measurement accuracy to the millimeter level. By constructing a 100×100 km monitoring grid, the least squares method was used to solve the conversion matrix between the line-of-sight deformation and the three-dimensional deformation component, and the east-west, north-south and vertical displacement components of each scattering point were calculated, with an average accuracy of 3mm, 4mm and 2mm respectively. The db4 wavelet 4-layer decomposition was performed on 20 SAR images, the long-term trend term was extracted, the cumulative deformation was calculated, and the threshold was set to 30 mm / year. According to the accumulated deformation and geological background information, the fuzzy comprehensive evaluation method is used to classify the potential hidden danger points, and the membership function is set with the weight vector of [0.4, 0.3, 0.3]. Corresponding to the deformation, geological conditions and terrain factors, a 1:10000 scale hidden danger point distribution map is generated, in which the high-risk areas account for 5% of the total area and are determined to be potential geological disaster hidden danger points.

[0086] Furthermore, the spatiotemporal clustering analysis includes extracting the spatial distribution characteristics and temporal evolution laws of deformation anomaly areas, combining historical disaster data statistics, meteorological factors and geological attribute information.

[0087] Specifically, the DBSCAN algorithm is used to spatially cluster the identified geological hazard risk points, and the cluster radius is set to 500 meters, and the minimum point number threshold is 10, to obtain the spatial distribution characteristics of the deformation anomaly area, and to determine the geometric shape and area size of the abnormal area, including: preparing a data set containing the longitude and latitude coordinates of the risk points, which can be the high-risk point coordinates obtained from the previous step, and converting the longitude and latitude coordinates into points in two-dimensional space; using the DBSCAN algorithm in the Scikit-learn library to cluster the converted coordinates, setting the cluster radius (eps) to 500 meters and the minimum point number threshold (min_samples) to 10; analyzing the cluster labels output by the DBSCAN algorithm, the spatial distribution characteristics of the deformation anomaly area can be obtained, including calculating the bounding box and area information of each cluster. The STL decomposition method is used to perform time evolution analysis on the deformation anomaly area, setting the seasonal period to 12 months, and extracting trend terms, periodic terms, and random terms to obtain the long-term change trend and seasonal fluctuation law of the deformation anomaly area. According to the statistical information of historical disaster data, the Gaussian kernel function is used for kernel density estimation, and the bandwidth selection adopts the Silverman criterion to calculate the frequency distribution of disasters in the region. Combined with the data of meteorological factors such as rainfall and temperature, the disaster sensitivity index is constructed. The geological attribute information such as lithology, formation dip, and fault distribution is integrated, and the weight of each factor is determined by the hierarchical analysis method to construct the geological vulnerability index. Through the multivariate regression analysis method, the spatial distribution characteristics, time evolution law, historical disaster frequency, meteorological factor sensitivity index and geological vulnerability index of the deformation abnormal area are used as input variables to establish a geological disaster risk assessment model, and the disaster risk level in the study area is calculated. The risk level is divided into three levels: low, medium and high, with thresholds of 0.3 and 0.7 respectively. This includes obtaining the data of the deformation abnormal area in the study area, extracting its spatial distribution characteristics and time evolution law as one of the input variables of the model. Obtain historical geological disaster data in the study area, and statistically calculate the frequency of disasters as the second input variable of the model. Obtain meteorological data in the study area, and calculate the sensitivity index of meteorological factors to geological disasters as the third input variable of the model. Obtain the geological environment data in the study area and calculate the vulnerability index of the geological environment as the fourth input variable of the model. Input the above four types of variables into the pre-established multivariate regression analysis model, and calculate the geological disaster risk value of each location in the study area through the model. According to the preset risk level threshold, the calculated risk value is divided into three levels: low, medium and high, and the distribution map of the geological disaster risk level of the study area is obtained. If the risk level of a certain location is high, key monitoring and early warning will be carried out at that location; if the risk level is medium, the inspection frequency of that location will be increased; if the risk level is low, regular monitoring will be maintained.In the 10km×10km study area, the DBSCAN algorithm was used to spatially cluster 1000 identified geological disaster risk points, with a cluster radius of 500 meters and a minimum point threshold of 10. Finally, 15 deformation anomaly areas were obtained, of which the largest area was 2.5 square kilometers and the smallest area was 0.3 square kilometers. The STL decomposition method was applied to these abnormal areas for time evolution analysis. Using 36 months of monitoring data and setting the seasonal cycle to 12 months, the long-term trend of an average annual deformation rate of -15mm / year and seasonal fluctuations of ±5mm were extracted. Using the data of 200 historical disaster points in the past 50 years, the Gaussian kernel function was used for kernel density estimation. The bandwidth was calculated by the Silverman criterion to be 1.2km, and a disaster frequency distribution map of a 100m×100m grid was generated. Combined with meteorological data such as an average annual rainfall of 1200mm and an average annual temperature of 15℃, a disaster sensitivity index was constructed. The five lithology types, the formation dip angle range of 15°-45° and the distribution information of three main fault zones in the region were integrated, and the weights were determined to be 0.4, 0.3 and 0.3 respectively by using the hierarchical analysis method, and the geological vulnerability index in the range of 0-1 was calculated. After all the above indicators were standardized and input into the multivariate regression model, the regression coefficients were 0.3, 0.25, 0.2, 0.15 and 0.1 respectively, and a geological disaster risk assessment model was established. According to the calculation results of the model, the study area was divided into three risk areas of low, medium and high, with thresholds of 0.3 and 0.7 respectively, of which the high risk area accounted for 15% of the total area.

[0088] Furthermore, the risk assessment model includes a quantitative assessment of the risks of geological disasters such as landslides, collapses, and ground subsidence in the monitoring area, classifies risk levels, and generates multiple types of hidden danger warning information, which include risk type, level, impact range, and recommended measures.

[0089] Specifically, according to the established risk assessment model, a sliding window of 30m×30m and a step length of 15m are used to analyze the terrain data in the monitoring area, calculate parameters such as slope, slope aspect and terrain undulation, and combine geological attribute information such as lithology and formation dip angle to obtain the sensitivity index of landslide, collapse and ground subsidence. Multi-source remote sensing data and ground monitoring data are used to decompose the time series through the wavelet transform method, extract long-term trends and periodic changes, and combine triggering factors such as rainfall and groundwater level to calculate the hazard index of landslide, collapse and ground subsidence. The analytic hierarchy process is used to determine the weights of the sensitivity index and hazard index, construct a judgment matrix, calculate the eigenvector, and perform consistency tests. The comprehensive risk index of landslide, collapse and ground subsidence is obtained through weighted superposition, and the risk level is divided according to the preset threshold. Combined with the distribution of facilities around the substation and the importance level, the risk level is locally adjusted, and differentiated warning thresholds and response strategies are formulated. Based on the comprehensive risk index and risk level, combined with the geographic information around the substation, multiple types of hidden danger warning information are generated, including risk type, level, impact range and recommended measures. It is updated once a day and updated in real time in case of emergencies. The risk warning information is visualized and dynamically updated through the geographic information system platform. In the monitoring area of ​​10km×10km, a 30m×30m sliding window with a step length of 15m is used to analyze the DEM data, and the slope range of 0-60°, 8 azimuth categories of slope direction and terrain relief of 0-200m are calculated. Combined with 5 types of lithology and 15°-45° formation dip information, the fuzzy comprehensive evaluation method is used to obtain the landslide, collapse and ground subsidence sensitivity index in the range of 0-1. Using 12 Sentinel-1SAR data and 50 ground GPS monitoring points, db4 wavelet was used for 4-layer decomposition to extract the long-term trend of the annual average deformation rate from -20mm / year to +5mm / year and the seasonal change of ±10mm. Combined with the trigger factors such as the annual average rainfall of 1500mm and the groundwater level change of ±2m, the hazard index in the range of 0-1 was calculated. A 3×3 judgment matrix was constructed by the hierarchical analysis method, and the sensitivity index weight of 0.6 and the hazard index weight of 0.4 were calculated. The consistency ratio CR was 0.05, and the weighted superposition was used to obtain a comprehensive risk index in the range of 0-1. The risk was divided into low, medium and high levels according to the thresholds of 0.3 and 0.7. The risk level within the 500m buffer zone of the substation was increased by one level, and the warning threshold was lowered by 20%. Finally, warning information including risk type, level, impact scope and recommended measures is generated, and a 1:10,000 scale risk warning map is visualized through the WebGIS platform. It is updated every 24 hours and the real-time update mechanism is triggered when the daily rainfall exceeds 50 mm.

[0090] In summary, the present invention combines high-resolution multispectral satellite images with SAR images to more accurately extract surface cover changes and geological features, effectively identify potential geological disaster risk areas, and also help to conduct risk assessment and early warning of geological disasters, improve the monitoring capabilities of disasters such as landslides, collapses and ground subsidence, thereby protecting personal safety, property and power supply, and promoting social and economic stability and development.

[0091] Importantly, it should be noted that the construction and arrangement of the present application shown in a plurality of different exemplary embodiments are only exemplary. Although only a few embodiments are described in detail in this disclosure, it should be readily understood by those who refer to this disclosure that many modifications are possible without substantially departing from the novel teachings and advantages of the subject matter described in the application (e.g., mounting arrangements, use of materials, color, changes in orientation, etc.). For example, the element shown as integrally formed may be composed of a plurality of parts or elements, the position of the element may be inverted or otherwise changed, and the nature or number or position of the discrete element may be altered or changed. Therefore, all such modifications are intended to be included within the scope of the present invention. The order or sequence of any process or method steps may be changed or reordered according to an alternative embodiment. In the claims, any "bracket plus function" clause is intended to cover the structure of the execution function described herein, and is not only structurally equivalent but also equivalent structure. Without departing from the scope of the present invention, other replacements, modifications, changes and omissions may be made in the design, operating conditions and arrangement of the exemplary embodiments. Therefore, the present invention is not limited to a specific embodiment, but extends to a variety of modifications that still fall within the scope of the appended claims.

[0092] Additionally, in an effort to provide a concise description of example embodiments, all features of an actual implementation may not be described.

[0093] It should be understood that in the development of any actual implementation, as in any engineering or design project, numerous implementation-specific decisions may be made. Such a development effort may be complex and time-consuming, but for those of ordinary skill having the benefit of this disclosure, the development effort will be a routine task of design, fabrication, and production without undue experimentation.

[0094] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. A method for intelligent monitoring of geological disasters based on satellite remote sensing, characterized in that: include, The corresponding surface coverage information is obtained through image registration and orthorectification processing; Identify terrain boundaries and geological structure features through surface cover information; Extract geological attribute information of lithology and stratigraphic occurrence through terrain boundaries and geological structural features; Identify surrounding hidden danger points through terrain feature extraction method; Through the image fusion enhancement method combined with the permanent scatterer interference method, stable scatterers are extracted and a deformation monitoring network is constructed to calculate the phase difference value in the high-risk area. Through phase difference unwrapping processing, combined with orbital parameters and atmospheric phase, the three-dimensional surface deformation field information in the monitoring area is inverted, and the cumulative deformation is calculated using time series analysis methods. Establish a geological disaster risk assessment model through spatiotemporal cluster analysis; By establishing a risk assessment model, the geological disaster risks in the monitoring area are quantitatively assessed, risk levels are divided and early warning information is generated.

2. The method for intelligently monitoring geological disasters based on satellite remote sensing according to claim 1, characterized in that: The image registration and orthorectification processing includes using a high-resolution multispectral satellite to obtain multi-time period multispectral image data and synthetic aperture radar images of the area surrounding the substation; The multispectral image data includes visible light, near infrared and short-wave infrared bands; The synthetic aperture radar image includes radar echo signals of the area surrounding the substation and corresponding radar image data.

3. The method for intelligently monitoring geological disasters based on satellite remote sensing as claimed in claim 2, characterized in that: The surface coverage information includes extracting surface texture information by a texture feature extraction method, combining edge detection and segmentation methods to identify terrain boundaries and geological structure features; The texture feature extraction method comprises calculating texture features using a gray level co-occurrence matrix to obtain surface texture information, wherein the surface texture information is a surface texture feature map; The edge detection and segmentation method comprises detecting the enhanced texture feature map to obtain the terrain boundary line segments; applying the watershed algorithm to perform image segmentation on the edge detection result according to the terrain boundary line segments, setting the minimum area threshold, merging the areas smaller than the threshold, and obtaining the geological structure feature areas.

4. The method for intelligently monitoring geological disasters based on satellite remote sensing as claimed in claim 3, characterized in that: The watershed algorithm includes gradient calculation, minimum marking, flood filling, watershed meeting, post-processing, minimum area threshold and morphological operations.

5. The method for intelligently monitoring geological disasters based on satellite remote sensing as claimed in claim 4, characterized in that: The geological attribute information of lithology and stratigraphic dip includes extracting shape index and directional characteristics from the geological structure characteristic area, calculating the circularity and main direction angle of each segmented area, identifying the lithology category and stratigraphic dip angle, judging the stratigraphic dip, and obtaining the geological attribute information in combination with the strike information of the terrain boundary line segment.

6. The method for intelligently monitoring geological disasters based on satellite remote sensing as claimed in claim 5, characterized in that: The terrain feature extraction method includes extracting terrain slope and aspect information from geological attribute information; The surrounding hidden danger points include high-risk areas prone to geological disasters such as landslides, collapses and ground subsidence, which are identified through the terrain slope and the slope direction information to form a distribution map of hidden danger points around the substation.

7. The method for intelligently monitoring geological disasters based on satellite remote sensing as claimed in claim 6, characterized in that: The image fusion enhancement method includes: using discrete wavelet transform to fuse the optical image and the synthetic aperture radar image, taking the average value of the low-frequency coefficients, selecting the high-frequency coefficients by the absolute value maximum method, extracting the high-frequency and low-frequency information by multi-scale decomposition, fusing and reconstructing by using different weight coefficients to obtain a fused high-resolution image, and geo-registering the fused high-resolution image with the hidden danger point distribution map; The permanent scatterer interference method includes selecting points in a high-risk area whose amplitude deviation index is less than a preset threshold and whose coherence is greater than a preset threshold as candidate permanent scatterers based on the registered fusion image and the potential hazard point distribution map, screening the candidate permanent scatterers, and performing time series interference processing on the screened stable scatterers.

8. The method for intelligently monitoring geological disasters based on satellite remote sensing as claimed in claim 7, characterized in that: The phase difference unwrapping process includes: using the least squares phase unwrapping algorithm to unwrap the phase difference to obtain a continuous phase field. According to the continuous phase field, atmospheric phase correction is performed by combining the tropospheric delay model and the ionospheric delay model to obtain corrected phase information; using the least squares method to solve the conversion matrix between the line-of-sight deformation and the three-dimensional deformation component, the corrected phase information is converted into a three-dimensional surface deformation field, the three-dimensional surface deformation field is decomposed at multiple scales, the long-term trend term is extracted, and the cumulative deformation amount is calculated.

9. The method for intelligently monitoring geological disasters based on satellite remote sensing as claimed in claim 8, characterized in that: The spatiotemporal cluster analysis includes extracting the spatial distribution characteristics and temporal evolution laws of deformation anomaly areas, combining historical disaster data statistics, meteorological factors and geological attribute information.

10. The method for intelligently monitoring geological disasters based on satellite remote sensing according to claim 9, characterized in that: The risk assessment model includes a quantitative assessment of the risks of landslides, collapses, and ground subsidence in the monitoring area, classifies risk levels, and generates multiple types of hidden danger warning information, which include risk type, level, impact range, and recommended measures.

Images