Electric power equipment flood disaster monitoring method based on satellite remote sensing

Abstract

The invention belongs to the technical field of power transmission line intrinsic safety and intelligent operation inspection, and particularly relates to a satellite remote sensing-based power equipment flood disaster monitoring method, which comprises the following steps of: extracting elements such as vegetation, roads, buildings and water bodies around power equipment by utilizing a visible light remote sensing image and combining with a power equipment ledger, and constructing a dynamic three-dimensional power equipment scene; extracting a flood inundation range of the SAR image by using a single-time-phase threshold method and a change detection threshold method, and calculating a flood inundation depth by using water level monitoring station data near the power equipment; carrying out overlay analysis on the SAR image, the visible light image and the power equipment, and carrying out disaster assessment; the flood disaster condition of the area where the power equipment is located is monitored in real time through the satellite remote sensing technology, the monitoring efficiency and accuracy of the power equipment in the flood disaster are improved, high risks and high cost in manual inspection are avoided, and the monitoring efficiency and precision are improved.

Description

Technical Field

[0001] This invention belongs to the field of intrinsic safety and intelligent operation and maintenance technology of power transmission lines, and specifically relates to a method for monitoring power equipment for flood disasters based on satellite remote sensing. Background Technology

[0002] Floods are one of the major natural disasters threatening the safe operation of power systems. Floods can soak, wash away, and even damage power equipment (such as substations, towers, and lines), seriously affecting power supply and grid security. Traditional flood monitoring methods for power equipment typically rely on manual inspections or ground-based sensors. These methods are often inefficient, have limited coverage, and even pose safety risks during or after floods. Manual inspections are limited by factors such as transportation, terrain, and water levels, making it difficult to quickly and comprehensively obtain information about the disaster area. While ground-based sensors can provide real-time data, they are costly to deploy, have limited coverage, and are susceptible to flood damage.

[0003] Therefore, there is a need for a monitoring method that can quickly, widely, and non-contactly acquire flood information about power equipment in order to improve emergency response efficiency, reduce personnel risks, and provide decision support for post-disaster recovery. Summary of the Invention

[0004] The technical problem to be solved by the present invention is: in order to overcome the above-mentioned technical problems, the present invention provides a method for monitoring flood disasters of power equipment based on satellite remote sensing.

[0005] The technical solution adopted by this invention to solve its technical problem is: a method for monitoring flood disasters in power equipment based on satellite remote sensing, comprising the following steps:

[0006] Step S1: Determine the monitoring area and obtain information on various types of power equipment within the area based on the power equipment ledger. The power equipment information includes the spatial location data of the power equipment.

[0007] Step S2: Obtain visible light images based on the location and extent of the monitoring area;

[0008] Step S3: Preprocess the visible light image obtained in step S2;

[0009] Step S4: Dynamic 3D power equipment scene construction and updating;

[0010] Step S5: When a flood disaster occurs, determine the affected area and scope, and take SAR images;

[0011] Step S6: Preprocess the SAR image obtained in step S5;

[0012] Step S7: By analyzing the backscatter histogram of the SAR image, determine the initial value of the water body, calculate the change threshold, generate a flood binary map, and extract the flood inundation range: Utilizing the principle that calm water bodies present extremely low backscatter intensity in SAR images, the SAR image is initially binarized and the water body is extracted through threshold segmentation.

[0013] Step S8: Using the flood inundation range extracted from the SAR image, and the DEM data or water level elevation of the monitoring points in the area, generate a water depth distribution map of the disaster area.

[0014] Step S9: In the constructed dynamic 3D power equipment scene, the water depth distribution map generated in step 8 is spatially overlaid with the visible light image and the spatial location data layer of the power equipment, and an overlay analysis is performed to determine the list of power equipment affected by flooding and their locations.

[0015] In step S1, the power equipment information includes spatial location data of the power equipment. The acquired spatial location data is stored and displayed in the form of geographic coordinates. The spatial location data of various types of power equipment in the region includes latitude and longitude coordinates, height, and area. The data acquisition process covers all key power assets in the region, including but not limited to substations, transmission lines and towers, transmission channels, and other ancillary power facilities. For discrete power equipment nodes, such as towers, geographic coordinates are used. For linear power transmission networks, such as transmission lines, lines are used. For power facility areas with large land areas, such as substations, areas are represented as surfaces.

[0016] In step S2, the geographical scope of the monitoring area, the required spatial and spectral resolution, the time window, and the acceptable cloud cover are clearly defined, and a suitable satellite platform and onboard sensors are selected accordingly. A detailed imaging request is submitted, and the satellite center schedules the process based on orbital characteristics and sensor capabilities, and uploads imaging commands to the satellite. When the satellite flies over the monitoring area, the onboard sensors collect data and observe the Earth, focusing the surface light signals through the optical system and converting them into digital image data. Finally, the collected raw data is transmitted to the ground receiving station for real-time reception, demodulation, and preliminary encapsulation, and undergoes preliminary system-level processing and data archiving on the ground.

[0017] In step S2, the collected raw data is transmitted to the ground receiving station via a high-speed link for real-time reception, demodulation, and preliminary encapsulation. The data is then processed and archived at the system level on the ground, including format conversion, system radiometric correction, and coarse geometric correction, to form basic-level optical remote sensing images that can be further processed and applied.

[0018] In step S3, radiometric correction and calibration are first performed. Atmospheric correction and other methods are used to convert the original digital quantization values ​​into true surface reflectance, eliminating the influence of sensors and the atmosphere and ensuring the physical accuracy and comparability of the image data. Then, geometric and orthorectification corrections are performed. High-precision ground control points and digital elevation models are used to eliminate geometric distortion and terrain deformation of the image, and the remote sensing image and spatial location data of power equipment are registered to a unified geographic coordinate system to ensure the accuracy of subsequent spatial overlay analysis. Based on this, image enhancement and denoising can be selectively performed to optimize visual effects and improve information interpretation accuracy.

[0019] Step S4 includes the following steps:

[0020] Step S41: Using the YOLOv8 target detection algorithm, the image is scanned through a sliding window or region proposal mechanism to extract the bounding box features of discrete ground objects. After deep features are extracted by a convolutional neural network, the target category and location coordinates are output through a classifier to achieve accurate positioning of power equipment such as poles and substations.

[0021] Simultaneously, U-Net and other series of models are used for semantic segmentation and recognition. The encoder extracts multi-scale features from the image, and the decoder fuses high-level semantic information with low-level detail features. At the pixel level, the surrounding environmental elements in the monitoring area are identified and classified, and the category label of each pixel is output, including elements such as vegetation, roads, buildings, and water bodies. The area, quantity, and distance of each element to power equipment are calculated.

[0022] Step S42: Create a 3D model of the power equipment that includes the location, attributes, and surrounding classified environmental elements of the power equipment. Attach the quantitative indicators extracted in step S41, such as the vegetation coverage rate and the distance to the nearest tree, as attributes to the equipment or environmental element layer of the 3D model of the power equipment to enrich the model information.

[0023] To maintain the timeliness of the model, step S4 also includes establishing an update mechanism, acquiring new visible light images quarterly, performing the preprocessing of the new visible light images as described in step S3, and identifying and extracting environmental elements in steps S41 and S42 to obtain the latest environmental status data.

[0024] By comparing and analyzing environmental data from both old and new periods, changes in the environment surrounding power equipment are identified, such as water bodies drying up, new buildings being constructed, and vegetation growth or removal. Finally, this information is updated in the model to create a dynamic 3D power equipment scene, generating change reports or early warning information to provide a basis for dynamic management and risk prevention of power facilities. The established dynamic 3D power equipment scene allows for the visualization of the 3D model of the power equipment for intuitive understanding and analysis.

[0025] In step S5, the requirements analysis, mission planning, required spatial and spectral resolution, time window, acceptable cloud cover, revisit period, and SAR-specific bands, polarization, incident angle, and imaging mode are clearly defined. A suitable satellite platform and onboard sensor are selected, and imaging commands are uploaded. When the satellite flies over the monitoring area, the onboard sensor actively transmits microwave pulses and receives echo signals from the ground, recording their intensity, time delay, and phase information. Subsequently, complex signal processing is performed on the ground, and the images from the onboard sensor are combined to form a SAR image that can be used for subsequent analysis. The SAR image includes range compression, azimuth compression, high-precision geometric correction, and despeculiar filtering.

[0026] In step S6, radiometric correction is first performed, converting the original digital quantization values ​​into physically meaningful backscattering coefficients to eliminate sensor differences, distance attenuation, and terrain effects, ensuring the comparability of image data. Then, geometric and orthorectification corrections are performed. By combining a high-precision digital elevation model and a rigorous imaging geometry model, the inherent side-view geometric distortion and terrain deformation of SAR images are eliminated, accurately mapping the images to a standard geographic coordinate system. Simultaneously, to suppress the coherent speckle noise unique to SAR images, despecimen filtering must be performed to improve image quality and information extraction accuracy. After processing individual images, all flood-related images and spatial location data of power equipment are precisely registered to a unified geographic coordinate system, ensuring data consistency and overlayability in spatial dimensions, thus providing an accurate and reliable data foundation for subsequent spatial overlay analysis and disaster impact assessment. The original digital quantization value (DN value) is a discrete integer value obtained by analog-to-digital conversion (A / D conversion) after remote sensing sensors, such as synthetic aperture radar (SAR) and optical satellite sensors, receive electromagnetic signals reflected or scattered by ground objects. A Digital Elevation Model (DEM) is a digital model that expresses surface elevation information using raster or vector data; in other words, it records terrain height data in digital form. An Imaging Geometric Model (IEM) is a mathematical model that describes the spatial relationship between ground features, the sensor's projection center, and image points during the imaging process of a remote sensing sensor. In orthorectification, the IEM provides the sensor's imaging patterns, while the DEM provides surface elevation information. Combining the two allows for the accurate calculation of the actual geographic coordinates of ground features, ultimately mapping the image to a unified standard coordinate system.

[0027] In step S7, examine the frequency distribution histogram of the backscattering coefficients in the SAR image. If water and land are clearly distinguishable, the histogram may exhibit a bimodal or multimodal shape. One peak corresponds to a low scattering value, while another or more peaks correspond to higher scattering values. The valley area between two main peaks is usually an ideal location for threshold selection. Combining the histogram, imagery, and 3D model of power equipment, observe the pixel values ​​of known water and land areas, and select an initial threshold near the valley of the histogram. Alternatively, based on samples of unaffected permanent water and land areas identified in the imagery, calculate the mean and standard deviation of these areas, and set the threshold between them. Examine the histogram of the difference image or logarithmic ratio image before and after the flood. It can be assumed that the difference or logarithmic ratio of the unchanged areas follows a certain distribution, such as a Gaussian distribution, and calculate its mean and standard deviation. The threshold formula is:

[0028] mean - k * standard deviation

[0029] Where k is a coefficient, usually taken as 2 or 3.

[0030] Apply a threshold to create a new binary raster layer and generate a binary flood map:

[0031] Pixel values ​​below or equal to the threshold: marked as 1, indicating flooding;

[0032] Pixel values ​​above the threshold: marked as 0, indicating no flooding.

[0033] By comparing the backscattering changes of SAR images before and after flooding, especially the significant intensity decrease caused by the transformation of land into water, newly flooded areas can be identified more accurately.

[0034] The generated binary water body map is cleaned to remove noise and misclassifications. This includes morphological operations to remove noise, fill holes, and remove small patches. Topographic correction and misclassification corrections are performed using auxiliary data, including a high-precision digital elevation model, to ensure the accuracy and continuity of the extracted results. Finally, the processed flood raster data is converted to vector format to identify flooded areas and calculate their areas, providing accurate geospatial information for subsequent spatial overlay analysis with power equipment and disaster impact assessment.

[0035] In step S8, the contact points between the flood boundary and the known elevation surface are found using the flood inundation range and DEM elevation data extracted from the SAR image, and the elevations of these points are used as the water surface elevation; or the water level elevation measured by the real-time water level monitoring station in the flood area is obtained and applied to the entire connected water body area, or an inclined water surface is obtained by interpolation based on data from multiple stations.

[0036] Using data from multiple water level stations:

[0037]

[0038] in:

[0039] WSE(x,y): The water surface elevation at coordinates (x,y) to be estimated;

[0040] G_i: The reading at the i-th water level station;

[0041] d_i: The distance from point (x,y) to the i-th water level station;

[0042] p: a power exponent.

[0043] Within the flood area extracted from SAR images, the generated flood water surface elevation grid is subtracted from the ground elevation grid to generate a water depth distribution map of the disaster area, showing the water depth at different locations within the flood area.

[0044] In step S9, for point-shaped devices, the device points that fall inside the flooded polygon or within its set buffer zone are determined by the point-polygon overlay analysis; for line-shaped devices, it is analyzed whether the line partially or completely intersects with or is contained within the flooded polygon; for area-shaped devices, the affected area and range of the device, as well as the impact of the flood depth on the device are analyzed.

[0045] In step S9, based on the results of spatial overlay analysis, a list of power equipment affected by flooding and their locations are determined. Remote sensing images from multiple flood events are used to analyze the duration of flooding and update the information into the dynamic 3D power equipment scene. Finally, the monitoring results are visualized and output to generate a thematic map. Combined with historical flood disaster information, different colors or symbols are used to mark the affected power equipment and their risk levels, and a detailed report or data file is generated, including the type, location, impact of the affected equipment, and possible risk assessment information, so as to facilitate emergency command, resource allocation, and repair work by the power sector.

[0046] It also includes step S10: After the disaster, optical images are used to capture, process, identify and analyze the flooded area, assess the scope, duration, affected equipment and changes in the surrounding environment of the flood disaster, and integrate the data in a dynamic three-dimensional power equipment scene to support post-disaster reconstruction work.

[0047] The beneficial effects of this invention are as follows: This invention provides a satellite remote sensing-based method for monitoring flood disasters affecting power equipment. It utilizes visible light remote sensing imagery, combined with power equipment inventory records, to extract elements such as vegetation, roads, buildings, and water bodies surrounding the power equipment. The imagery and extraction results are updated quarterly to construct a dynamic three-dimensional power equipment scene. The method uses a single-phase threshold method and a change detection threshold method to extract the flood inundation range from SAR images, and uses data from water level monitoring stations near the power equipment to calculate the flood inundation depth. Within the dynamic three-dimensional power equipment scene, SAR images, visible light images, and power equipment data are integrated. Overlay analysis is performed to conduct disaster assessment; real-time monitoring of flood conditions in areas where power equipment is located is conducted using satellite remote sensing technology, timely acquisition of key information such as the extent and depth of flood inundation, improving the efficiency and accuracy of power equipment monitoring during floods; remote monitoring using satellite remote sensing data avoids the high risks and costs of manual inspections, improving monitoring efficiency and accuracy; through the processing and analysis of satellite remote sensing images, the specific location and extent of damage to affected power equipment can be identified in a timely manner, providing power companies with scientific decision-making basis, supporting post-flood recovery work, and ensuring the stability and reliability of power supply. Attached Figure Description

[0048] Figure 1 This is a flowchart illustrating the overall process of a satellite remote sensing-based method for monitoring flood disasters affecting power equipment, as described in this invention. Detailed Implementation

[0049] The invention will now be described in further detail with reference to the accompanying drawings. It should be emphasized that the following description is merely exemplary and not intended to limit the scope or application of the invention.

[0050] like Figure 1 As shown, the present invention provides a method for monitoring power equipment for flood disasters based on satellite remote sensing, comprising the following steps:

[0051] Step S1: Determine the monitoring area and obtain information on various types of power equipment within the area based on the power equipment ledger. The power equipment information includes the spatial location data of the power equipment.

[0052] Step S2: Based on the location and extent of the monitoring area, acquire high-resolution visible light images;

[0053] Step S3: Preprocess the visible light image obtained in step S2;

[0054] Step S4: Dynamic 3D power equipment scene construction and updating;

[0055] Step S5: When floods occur, release rainfall and flood information through the meteorological department, determine the affected areas and extent, and take SAR images;

[0056] Step S6: Preprocess the SAR image obtained in step S5;

[0057] Step S7: By analyzing the backscatter histogram of the SAR image, determine the initial value of the water body, calculate the change threshold, generate a flood binary map, and extract the flood inundation range: Utilizing the principle that calm water bodies present extremely low backscatter intensity in SAR images, the SAR image is initially binarized and the water body is extracted through threshold segmentation.

[0058] Step S8: Using the flood inundation range extracted from the SAR image, and the DEM data or water level elevation of the monitoring points in the area, generate a water depth distribution map of the disaster area.

[0059] Step S9: In the constructed dynamic 3D power equipment scene, the water depth distribution map generated in step 8 is spatially overlaid with the visible light image and the spatial location data layer of the power equipment. Combined with the changes in the surrounding environment, an overlay analysis is performed to determine the list of power equipment affected by flooding and their locations.

[0060] In step S1, the power equipment information includes spatial location data of the power equipment. The acquired spatial location data is stored and displayed in a standardized geographic coordinate format. The spatial location data of various types of power equipment within the region includes latitude and longitude coordinates, height, and area. The data acquisition process covers all key power assets within the region, including but not limited to substations, transmission lines and towers, transmission channels, and other ancillary power facilities. For discrete power equipment nodes, such as towers, geographic coordinate points are used. For linear power transmission networks, such as transmission lines, lines are used. For power facility areas with large land areas, such as substations, areas are represented as surfaces.

[0061] In step S2, the geographical scope of the monitoring area, the required spatial and spectral resolution, the time window, and the acceptable cloud cover are clearly defined, and a suitable satellite platform and onboard sensors are selected accordingly. A detailed imaging request is submitted, and the satellite center schedules the process based on orbital characteristics and sensor capabilities, and uploads imaging commands to the satellite. When the satellite flies over the monitoring area, the onboard sensors collect data and observe the Earth, focusing the surface light signals through the optical system and converting them into digital image data. Finally, the collected raw data is transmitted to the ground receiving station for real-time reception, demodulation, and preliminary encapsulation, and undergoes preliminary system-level processing and data archiving on the ground.

[0062] In step S2, the massive amount of raw data collected is transmitted to the ground receiving station via a high-speed link for real-time reception, demodulation, and preliminary encapsulation. The data is then processed and archived at the system level on the ground, including format conversion, system radiometric correction, and coarse geometric correction, to form basic-level optical remote sensing images that can be further processed and applied.

[0063] In step S3, radiometric correction and calibration are first performed. Atmospheric correction and other methods are used to convert the original digital quantization values ​​into true surface reflectance to eliminate the influence of sensors and the atmosphere, ensuring the physical accuracy and comparability of the image data. Then, geometric and orthorectification corrections are performed. High-precision ground control points and digital elevation models are used to eliminate geometric distortion and terrain deformation of the image, and the remote sensing image and spatial location data of power equipment are accurately registered to a unified geographic coordinate system to ensure the accuracy of subsequent spatial overlay analysis. Based on this, image enhancement and denoising can be selectively performed to optimize visual effects and improve information interpretation accuracy.

[0064] Step S4 includes the following steps:

[0065] Step S41: Using the YOLOv8 target detection algorithm, the image is scanned through a sliding window or region proposal mechanism to extract the bounding box features of discrete ground objects. After deep features are extracted by a convolutional neural network, the target category and location coordinates are output through a classifier to achieve accurate positioning of power equipment such as poles and substations.

[0066] Simultaneously, U-Net and other series of models are used for semantic segmentation and recognition. The encoder extracts multi-scale features from the image, and the decoder fuses high-level semantic information with low-level detail features. At the pixel level, the surrounding environmental elements in the monitoring area are identified and classified, and the category label of each pixel is output, including elements such as vegetation, roads, buildings, and water bodies. The area, quantity, and distance of each element to power equipment are calculated.

[0067] Step S42: Create a 3D model of the power equipment that includes its location, attributes, and surrounding classified environmental elements. Quantitative indicators extracted in Step S41, such as vegetation cover and distance to the nearest tree, are added as attributes to the equipment or environmental element layer of the 3D model to enrich the model information. The above steps correspond to... Figure 1 The surrounding environment identification and feature information calculation in the process refers to the surrounding environmental elements.

[0068] To maintain the timeliness of the model, step S4 also includes establishing an update mechanism, acquiring new visible light images quarterly, performing the preprocessing of the new visible light images as described in step S3, and identifying and extracting environmental elements in steps S41 and S42 to obtain the latest environmental status data.

[0069] By comparing and analyzing environmental data from both old and new periods, changes in the environment surrounding power equipment are identified, such as water bodies drying up, new buildings being constructed, and vegetation growth or removal. Finally, this information is updated in the model to create a dynamic 3D power equipment scene, generating change reports or early warning information to provide a basis for dynamic management and risk prevention of power facilities. The established dynamic 3D power equipment scene allows for the visualization of the 3D model of the power equipment for intuitive understanding and analysis.

[0070] In step S5, the requirements analysis, mission planning, required spatial and spectral resolution, time window, acceptable cloud cover, revisit period, and SAR-specific bands, polarization, incident angle, and imaging mode are clearly defined. A suitable satellite platform and onboard sensor are selected, and imaging commands are uploaded. When the satellite flies over the monitoring area, the onboard sensor actively transmits microwave pulses and receives echo signals from the ground, recording their intensity, time delay, and phase information. Subsequently, complex signal processing is performed on the ground, and the images from the onboard sensor are combined to form a SAR image that can be used for subsequent analysis. The SAR image includes range compression, azimuth compression, high-precision geometric correction, and despeculiar filtering.

[0071] In step S6, radiometric correction is first performed, converting the original digital quantization values ​​into physically meaningful backscattering coefficients to eliminate sensor differences, distance attenuation, and terrain effects, ensuring the comparability of image data. Then, geometric and orthorectification corrections are performed. By combining a high-precision digital elevation model and a rigorous imaging geometry model, the inherent side-view geometric distortion and terrain deformation of SAR images are eliminated, accurately mapping the images to a standard geographic coordinate system. Simultaneously, to suppress the coherence speckle noise unique to SAR images, despecimen filtering is necessary to improve image quality and information extraction accuracy. After processing individual images, all flood-related images and spatial location data of power equipment are precisely registered to a unified geographic coordinate system, ensuring spatial consistency and overlay capability, thus providing an accurate and reliable data foundation for subsequent spatial overlay analysis and disaster impact assessment.

[0072] In step S7, examine the frequency distribution histogram of the backscattering coefficients in the SAR image. If water and land are clearly distinguishable, the histogram may exhibit a bimodal or multimodal shape. One peak corresponds to a low scattering value, while another or more peaks correspond to higher scattering values. The valley area between two main peaks is usually an ideal location for threshold selection. Combining the histogram, imagery, and 3D model of power equipment, observe the pixel values ​​of known water and land areas, and select an initial threshold near the valley of the histogram. Alternatively, based on samples of unaffected permanent water and land areas identified in the imagery, calculate the mean and standard deviation of these areas, and set the threshold between them. Examine the histogram of the difference image or logarithmic ratio image before and after the flood. It can be assumed that the difference or logarithmic ratio of the unchanged areas follows a certain distribution, such as a Gaussian distribution, and calculate its mean and standard deviation. The threshold formula is:

[0073] mean - k * standard deviation

[0074] Where k is a coefficient, usually taken as 2 or 3.

[0075] Apply a threshold to create a new binary raster layer and generate a binary flood map:

[0076] Pixel values ​​below or equal to the threshold: marked as 1, indicating flooding;

[0077] Pixel values ​​above the threshold: marked as 0, indicating no flooding.

[0078] By comparing the backscattering changes of SAR images before and after flooding, especially the significant intensity decrease caused by the transformation of land into water, newly flooded areas can be identified more accurately.

[0079] The generated binary water body map is cleaned to remove noise and misclassifications. This includes morphological operations to remove noise, fill holes, and remove small patches. Topographic correction and misclassification corrections are performed using auxiliary data, including a high-precision digital elevation model, to ensure the accuracy and continuity of the extracted results. Finally, the processed flood raster data is converted to vector format to identify flooded areas and calculate their areas, providing accurate geospatial information for subsequent spatial overlay analysis with power equipment and disaster impact assessment.

[0080] In step S8, the contact points between the flood inundation range and the DEM elevation data within the area extracted from the SAR image are found. The elevations of these points are used as the water surface elevations. Within the flood range extracted from the SAR image, the generated flood water surface elevation grid is subtracted from the ground elevation grid to generate a water depth distribution map of the disaster area, showing the water depth at different locations within the flood area.

[0081] In step S9, for point-shaped devices, the device points that fall inside the flooded polygon or within its set buffer zone are determined by the point-polygon overlay analysis; for line-shaped devices, it is analyzed whether the line partially or completely intersects with or is contained within the flooded polygon; for area-shaped devices, the affected area and range of the device, as well as the impact of the flood depth on the device are analyzed.

[0082] In step S9, based on the results of spatial overlay analysis, a list of power equipment affected by flooding and their locations are determined. Remote sensing images from multiple flood events are used to analyze the duration of flooding and update the information into the dynamic 3D power equipment scene. Finally, the monitoring results are visualized and output to generate a thematic map. Combined with historical flood disaster information, the affected power equipment and its risk level are marked with different colors or symbols, and a detailed report or data file is generated, including the type, location, and impact status of the affected equipment (such as confirmed flooding or impending flooding), as well as possible risk assessment information, to facilitate emergency command, resource allocation, and repair work by the power sector.

[0083] It also includes step S10: After the disaster, optical images are used to capture, process, identify and analyze the flooded area, assess the scope, duration, affected equipment and changes in the surrounding environment of the flood disaster, and integrate the data in a dynamic three-dimensional power equipment scene to support post-disaster reconstruction work.

[0084] High-resolution optical images that are contemporaneous with or have minimal temporal differences from SAR images are selected. The flood extent is interpreted using the YOLOv8 target detection algorithm, the U-Net semantic segmentation algorithm, or manually. The flood extent extracted by SAR is superimposed with the visual interpretation results of the optical images, and a confusion matrix is ​​calculated to generate the following five indicators: Kappa coefficient, accuracy (OA), F1 score, precision (P), and recall (R), which are used to verify the accuracy of flood extent extraction.

[0085] Based on the flood extent (S) extracted by SAR and the reference extent (O) interpreted by optical imagery, a confusion matrix is ​​calculated by comparing the “flood extent extracted by SAR” with the “true extent of the reference data”, and the results are divided into four categories:

[0086] Submerged (S=1) Non-submerged (S=0) total Submerged (O=1) TP (True Positive) FN (False Negative) <![CDATA[O1]]> Non-submerged (O=0) FP (False Positive) TN (True Negative) <![CDATA[O0]]> total <![CDATA[S1]]> <![CDATA[S0]]> N

[0087] in:

[0088] TP: The area was identified as flooded by both SAR and optical imagery (correctly extracted);

[0089] FN: The optical image shows a flooded area, but the SAR failed to identify it as an unflooded area (missed detection);

[0090] FP: The SAR image misidentified the area as flooded, but the optical image showed an unflooded area (false positive).

[0091] TN: Both are determined to be non-flooded areas (correctly excluded).

[0092] Calculate overall accuracy (OA), which is the proportion of all correctly classified samples out of the total sample:

[0093]

[0094] in:

[0095] The value of OA ranges from [0,1]. The closer it is to 1, the higher the accuracy of the overall extraction results.

[0096] Calculate the Kappa coefficient, which is a metric for measuring classification accuracy:

[0097]

[0098] Wherein: Kappa ranges from [-1,1], with the [0,1] interval usually being of interest. Kappa > 0.8 indicates high consistency, 0.6-0.8 indicates moderate consistency, 0.4-0.6 indicates general consistency, and < 0.4 indicates poor consistency.

[0099] The accuracy (P) is the proportion of the "flooded area" extracted by SAR that is actually a flooded area.

[0100]

[0101] in:

[0102] The value of P ranges from [0,1]. The closer it is to 1, the higher the reliability of the flooded area extracted by SAR.

[0103] Calculate recall (R), which is the proportion of the actual flooded area that is correctly extracted by SAR:

[0104]

[0105] in:

[0106] The value of R ranges from [0,1]. The closer it is to 1, the higher the integrity of the flooded area extracted by SAR.

[0107] Calculate the F1 score, which is the harmonic mean of P and R:

[0108]

[0109] The value of F1 ranges from [0,1]. The closer it is to 1, the better the sum of P and R performs.

[0110] By utilizing real-time water level monitoring stations within the flood area, the measured water level elevations are obtained to verify the accuracy of flood depth extraction. This elevation is then applied to the entire connected water body area, or interpolated based on data from multiple stations, to obtain a sloping water surface.

[0111] Using data from multiple water level stations:

[0112]

[0113] in:

[0114] WSE(x,y): The water surface elevation at coordinates (x,y) to be estimated;

[0115] G_i: The reading at the i-th water level station;

[0116] d_i: The distance from point (x,y) to the i-th water level station;

[0117] p: a power exponent; usually taken as 2.

[0118] Calculate the water depth during the flood season: Water depth = Water surface elevation - Water bottom elevation.

[0119] Based on the above-described preferred embodiments of the present invention, and through the foregoing description, those skilled in the art can make various changes and modifications without departing from the inventive concept. The technical scope of this invention is not limited to the contents of the specification, but must be determined according to the scope of the claims.

Claims

1. A method for monitoring flood disasters affecting power equipment based on satellite remote sensing, characterized in that, Includes the following steps: Step S1: Determine the monitoring area and obtain information on various types of power equipment within the area based on the power equipment ledger. The power equipment information includes the spatial location data of the power equipment. Step S2: Obtain visible light images based on the location and extent of the monitoring area; Step S3: Preprocess the visible light image obtained in step S2; Step S4: Dynamic 3D power equipment scene construction and updating; Step S5: When a flood disaster occurs, determine the affected area and scope, and take SAR images; Step S6: Preprocess the SAR image obtained in step S5; Step S7: By analyzing the backscatter histogram of the SAR image, determine the initial value of the water body, calculate the change threshold, generate a binary flood map, and extract the flood inundation range; Step S8: Using the flood inundation range extracted from the SAR image, and the DEM data or water level elevation of the monitoring points in the area, generate a water depth distribution map of the disaster area. Step S9: In the constructed dynamic 3D power equipment scene, the water depth distribution map generated in step 8 is spatially overlaid with the visible light image and the spatial location data layer of the power equipment, and an overlay analysis is performed to determine the list of power equipment affected by flooding and their locations.

2. The method for monitoring power equipment flood disasters based on satellite remote sensing as described in claim 1, characterized in that, In step S1, the acquired spatial location data is stored and displayed in the form of geographic coordinates; the spatial location data of various power equipment in the region includes latitude and longitude coordinates, height and area; for discrete power equipment nodes, geographic coordinate points are used; for linear power transmission networks, lines are used; and for power facility areas with large land areas, areas are represented as surfaces.

3. The method for monitoring power equipment flood disasters based on satellite remote sensing as described in claim 1, characterized in that, In step S2, the geographical scope of the monitoring area, the required spatial and spectral resolution, the time window, and the acceptable cloud cover are determined, and a suitable satellite platform and onboard sensors are selected accordingly. When the satellite flies over the monitoring area, the onboard sensors collect data and observe the Earth, focusing the surface light signals through the optical system and converting them into digital image data. Finally, the collected raw data is transmitted to the ground receiving station for real-time reception, demodulation, and preliminary encapsulation, and then undergoes preliminary system-level processing and data archiving on the ground.

4. The method for monitoring power equipment flood disasters based on satellite remote sensing as described in claim 3, characterized in that, In step S2, the collected raw data is transmitted to the ground receiving station via a high-speed link for real-time reception, demodulation, and preliminary encapsulation. The data is then processed and archived at the system level on the ground, including format conversion, system radiometric correction, and coarse geometric correction, to form basic-level optical remote sensing images that can be further processed and applied.

5. The method for monitoring power equipment flood disasters based on satellite remote sensing as described in claim 1, characterized in that, In step S3, radiometric correction and calibration are first performed to convert the original digital quantization values ​​into the true surface reflectance. Then, geometric correction and orthorectification are performed. High-precision ground control points and digital elevation models are used to eliminate geometric distortion and terrain deformation of the image. Finally, the remote sensing image and the spatial location data of power equipment are registered to a unified geographic coordinate system.

6. The method for monitoring power equipment flood disasters based on satellite remote sensing as described in claim 1, characterized in that, Step S4 includes the following steps: Step S41: Using the YOLOv8 target detection algorithm, the image is scanned through a sliding window or region proposal mechanism to extract the bounding box features of discrete ground objects. After deep features are extracted by a convolutional neural network, the target category and location coordinates are output through a classifier to realize the positioning of power equipment. Simultaneously, a model is used for semantic segmentation and recognition. The encoder extracts multi-scale features from the image, and the decoder fuses high-level semantic information with low-level detail features. At the pixel level, the surrounding environmental elements in the monitoring area are identified and classified, the category label of each pixel is output, and the area, quantity, and distance of each type of element from the power equipment are calculated. Step S42: Create a 3D model of the power equipment that includes the location, attributes, and surrounding classified environmental elements of the power equipment. Attach the quantitative indicators extracted in step S41 as attributes to the equipment or environmental element layer of the 3D model of the power equipment.

7. The method for monitoring power equipment flood disasters based on satellite remote sensing as described in claim 6, characterized in that, Step S4 also includes establishing an update mechanism: acquiring new visible light images quarterly, performing the preprocessing of the new visible light images in step S3, and the environmental element identification and extraction steps in steps S41 and S42 to obtain the latest environmental status data. By comparing and analyzing the environmental data from the old and new periods, changes in the environment surrounding the power equipment are identified. Finally, this information is updated into the 3D model of the power equipment to create a dynamic 3D power equipment scene.

8. The method for monitoring power equipment flood disasters based on satellite remote sensing as described in claim 1, characterized in that, In step S5, a satellite platform and onboard sensor are selected, and imaging commands are uploaded. When the satellite flies over the monitoring area, the onboard sensor actively transmits microwave pulses and receives echo signals from the ground, recording their intensity, time delay, and phase information. Subsequently, signal processing is performed on the ground, and the images from the onboard sensor are combined to form a SAR image. The SAR image includes range compression, azimuth compression, high-precision geometric correction, and despeculiar filtering.

9. The method for monitoring power equipment flood disasters based on satellite remote sensing as described in claim 1, characterized in that, In step S6, radiometric correction is first performed to convert the original digital quantization values ​​into backscattering coefficients; then geometric and orthorectification corrections are performed, and the image is mapped to a standard geographic coordinate system by combining the digital elevation model and the imaging geometry model; at the same time, despeccary filtering is performed; after processing a single image, all flood images and spatial location data of power equipment are registered to a unified geographic coordinate system.

10. The method for monitoring power equipment flood disasters based on satellite remote sensing as described in claim 1, characterized in that, In step S7, the frequency distribution histogram of the backscattering coefficient of the SAR image is examined, and the backscattering changes of the SAR image before and after the flood are compared. The generated binary water body map is cleaned to remove noise and misclassifications. Auxiliary data, including the digital elevation model, are combined to perform terrain correction and misjudgment correction. Finally, the processed flood raster data is converted into vector format to determine the flood area and calculate its area.

11. The method for monitoring power equipment flood disasters based on satellite remote sensing as described in claim 1, characterized in that, In step S8, the contact points between the flood boundary and the known elevation surface are found, and the elevations of these points are used as the water surface elevation; or the real-time water level monitoring stations in the flood area are used to obtain the water level elevation measured by the stations, and this elevation is applied to the entire connected water body area, or an inclined water surface is obtained by interpolation based on data from multiple stations. Using data from multiple water level stations: in: WSE(x,y): The water surface elevation at coordinates (x,y) to be estimated; G_i: The reading at the i-th water level station; d_i: The distance from point (x,y) to the i-th water level station; p: a power exponent; Within the flood area extracted from SAR images, the generated flood water surface elevation grid is subtracted from the ground elevation grid to generate a water depth distribution map of the disaster area, showing the water depth at different locations within the flood area.

12. The method for monitoring power equipment flood disasters based on satellite remote sensing as described in claim 1, characterized in that, In step S9, for point-shaped devices, the device points that fall inside the flooded polygon or within its set buffer zone are determined by the point-polygon overlay analysis; for line-shaped devices, it is analyzed whether the line partially or completely intersects with or is contained within the flooded polygon; for area-shaped devices, the affected area and range of the device, as well as the impact of the flood depth on the device are analyzed.

13. The method for monitoring power equipment flood disasters based on satellite remote sensing as described in claim 12, characterized in that, In step S9, based on the results of spatial overlay analysis, the list of power equipment affected by floods and their locations are determined. Remote sensing images from multiple flood events are used to analyze the duration of flooding and update the information into the dynamic three-dimensional power equipment scene.

14. The method for monitoring power equipment flood disasters based on satellite remote sensing as described in claim 1, characterized in that, It also includes step S10: After the disaster, optical images are used to capture, process, identify and analyze the flooded area, assess the scope, duration, affected equipment and changes in the surrounding environment of the flood disaster, and integrate the data in a dynamic three-dimensional power equipment scene to support post-disaster reconstruction work.

Images