A power transmission tower high-resolution satellite remote sensing image acquisition and processing method and system

Abstract

The application provides a power transmission tower high-resolution satellite remote sensing image acquisition and processing method and system, and relates to the technical field of power systems.The method comprises the following steps: S1, acquiring optical remote sensing images through a satellite data interface and performing data preprocessing; S2, constructing an improved YOLOv5x deep learning model and inputting the preprocessed remote sensing image data to perform tower target detection; S3, performing multi-dimensional position correction according to the tower target detection result; S4, extracting state parameters according to the corrected data; and S5, constructing a tower health degree evaluation model through an improved random forest algorithm, inputting the extracted state parameters into the model, and outputting a tower health index, and automatically generating early warning information when the index is lower than a threshold value.The application can realize wide-range, high-precision and automatic power transmission tower state monitoring, and solves the technical problems of low traditional inspection efficiency and poor adaptability to complex environments.

Description

Technical Field

[0001] This invention relates to the field of power system technology, and more specifically, to a method and system for acquiring and processing high-resolution satellite remote sensing images of transmission towers. Background Technology

[0002] With the continuous expansion of my country's power grid, the total length of 220kV and above transmission lines has exceeded 838,800 kilometers, with more than 70% located in complex terrains such as mountains and hills. However, traditional manual inspection methods have significant limitations. While existing drone inspection technology has improved efficiency to some extent, it is limited by its endurance (the radius of a single operation is usually less than 15 kilometers) and complex weather conditions (such as strong winds and dense fog), making it difficult to achieve large-scale, routine monitoring. At the same time, the error rate of manual recording is as high as 15%, which is insufficient to meet the dynamic monitoring needs of ultra-high voltage power grids and new energy grid-connected equipment, resulting in a high rate of missed detections and failing to support preventive maintenance decisions.

[0003] Satellite remote sensing technology has advantages such as wide coverage and stable data acquisition cycle, but existing technologies have the following bottlenecks in the identification of power transmission towers: (1) High-resolution satellite image data has problems such as atmospheric interference and geometric distortion, which affect the accuracy of target detection; (2) Traditional target detection algorithms have an identification rate of less than 60% for small-scale towers (which usually occupy only 5-15 pixels in the image); (3) There is a lack of position correction mechanism for complex environments such as mountainous terrain and vegetation cover, which often results in positioning errors exceeding 3 meters. Summary of the Invention

[0004] The purpose of this invention is to provide a method and system for acquiring and processing high-resolution satellite remote sensing images of power transmission towers, which can realize large-scale, high-precision, and automated monitoring of the status of power transmission towers, and solve the technical problems of low efficiency and poor adaptability to complex environments in traditional inspections.

[0005] The technical solution of this invention is as follows:

[0006] In a first aspect, this application provides a method and system for acquiring and processing high-resolution satellite remote sensing images of power transmission towers, which includes the following steps:

[0007] S1. Acquire optical remote sensing images through the satellite data interface and perform data preprocessing;

[0008] S2. Construct an improved YOLOv5x deep learning model and input preprocessed remote sensing image data to perform pole target detection;

[0009] S3. Perform multi-dimensional position correction based on the tower target detection results;

[0010] S4. Extract state parameters based on the corrected data;

[0011] S5. Construct a pole health assessment model, input the extracted state parameters into an improved random forest algorithm, output the pole health index, and automatically generate early warning information when the index is lower than the threshold.

[0012] Furthermore, in step S1, the aforementioned satellite data interface supports access to multi-source satellite data, including Gaofen-7, WorldView-3, and Sentinel-2, and has the ability to automatically select the best image. The selection criteria include cloud coverage <10%, shooting angle <30°, and data acquisition time within the last 30 days.

[0013] Further, in step S1, the aforementioned data preprocessing includes sequentially performing radiometric correction, geometric correction, cloud removal, and resolution fusion processing. Radiometric correction employs the 6S atmospheric correction model, utilizing the FLAASH module in ENVI. By inputting solar elevation angle and aerosol optical thickness parameters, errors caused by atmospheric scattering and absorption during image capture are eliminated. Then, a quadratic polynomial transformation model is used for geometric fine correction to ensure the authenticity of ground feature reflectance. Geometric correction uses geographic reference data and high-precision control points, such as Google Earth landmarks, to register the remote sensing image, ensuring the image accurately corresponds to the actual geographical location and guaranteeing the spatial accuracy of subsequent classification results. Cloud removal employs a multispectral band threshold segmentation and inpainting restoration algorithm. Resolution fusion processing uses a pan-sharpening algorithm to improve image clarity while preserving spectral information. The image of the study area is cropped according to the administrative boundaries of the study area, reducing data volume and improving processing efficiency.

[0014] Furthermore, in step S2, the improved YOLOv5x deep learning model includes: adding a coordinate attention mechanism after the third CSP module of the backbone network CSPDarknet to enhance the ability to extract structural features of transmission towers; introducing a BiFPN bidirectional feature pyramid network in the Neck part; using the CIoU loss function to optimize bounding box regression in the output layer; and using Mosaic data augmentation and label smoothing techniques during model training.

[0015] Furthermore, in step S3, the aforementioned multi-dimensional position correction includes image registration based on SIFT feature point matching, elevation correction combined with the DSM digital surface model, comparison of the pixel elevation value at the bottom of the tower with the design elevation, and spatial coordinate calibration using the tower's GIS coordinates. That is, the pixel coordinates are converted into WGS84 latitude and longitude coordinates and compared with the GIS ledger coordinates through the RPC model.

[0016] Furthermore, in step S4, the aforementioned state parameters include: tower inclination, number of insulator strings, conductor sag, and tower foundation settlement.

[0017] The tower tilt is calculated using the Hough transform to detect the tilt along the tower's centerline. The calculation formula includes:

[0018] ρ=xcosθ+ysinθ

[0019] Inclination = |θ−90°|

[0020] In the formula, ρ is the perpendicular distance from the origin to the line, θ is the angle between the perpendicular line and the positive x-axis, and (x,y) are the pixel coordinates on the line in the image.

[0021] The number of insulator strings was identified by combining HSV threshold segmentation with morphological operations.

[0022] The conductor sag is calculated based on parabolic fitting, and its calculation formula includes:

[0023] f=wL 2 / 8T

[0024] In the formula, f is the conductor sag, w is the weight per unit length of the conductor, L is the horizontal distance between the two towers, and T is the horizontal tension of the conductor.

[0025] The settlement of the tower foundation was detected by analyzing the grayscale changes in images.

[0026] In step S5, the calculation process for constructing the tower health assessment model includes:

[0027] Weighted calculation of pole health index:

[0028]

[0029] An improved random forest calculates the health index through ensemble predictions from 100 decision trees. The output of a single decision tree is as follows:

[0030]

[0031] Integrated prediction:

[0032]

[0033] Feature importance weight calculation:

[0034]

[0035] In the formula, The pole health index ranges from 0 to 100, with higher values ​​indicating better health. The weight of the i-th state parameter; is the standardized score of the i-th state parameter; n is the total number of state parameters in the input model; denoted as , where is the prediction score of the k-th decision tree for the input feature vector x; c is the number of labels. Let represent the feature vector of the j-th training sample; Let this represent the label corresponding to the j-th training sample; This is an indicator function; it returns 1 when the condition is true and 0 otherwise. Let be the training sample set for the k-th tree; Let $\frac{i}{i}$ be the total decrease in Gini impurity caused by the i-th state parameter during the splitting of all decision trees in the random forest. Let Gini impurity be the sum of all state parameters.

[0036] Secondly, this application provides a high-resolution satellite remote sensing image acquisition and processing system for power transmission towers, characterized in that it includes:

[0037] The image acquisition and processing module is used to acquire optical remote sensing images through the satellite data interface and perform data preprocessing.

[0038] The target detection module is used to build an improved YOLOv5x deep learning model and input preprocessed remote sensing image data to detect tower targets.

[0039] The position correction module is used to perform multi-dimensional position correction based on the tower target detection results;

[0040] The parameter extraction module is used to extract state parameters based on the corrected data.

[0041] The health assessment module is used to build a pole health assessment model using an improved random forest algorithm. The extracted state parameters are input into the model, and the pole health index is output. When the index is lower than the threshold, an early warning message is automatically generated.

[0042] Thirdly, this application provides an electronic device, comprising:

[0043] Memory, used to store one or more programs;

[0044] processor;

[0045] When one or more of the above programs are executed by the above processor, a method and system for acquiring and processing high-resolution satellite remote sensing images of power transmission towers as described in any of the first aspects above is implemented.

[0046] Fourthly, this application provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements a high-resolution satellite remote sensing image acquisition and processing method and system for power transmission towers as described in any of the first aspects above.

[0047] Compared with the prior art, the present invention has at least the following advantages or beneficial effects:

[0048] (1) The present invention provides a high-resolution satellite remote sensing image acquisition and processing method and system for power transmission towers. It adopts an improved YOLOv5x model and significantly improves the target detection accuracy of towers by adding a coordinate attention mechanism, BiFPN feature fusion and CIoU loss function optimization. It provides high-precision data support for power inspection, thereby improving inspection efficiency, reducing costs, and solving the technical bottleneck of low efficiency due to terrain limitations in manual inspection.

[0049] (2) This invention can work stably under cloudy and complex terrain conditions by using the 6S atmospheric correction model, Pan-sharpening resolution fusion and cloud removal technology, which makes up for the meteorological limitations of UAV inspection, realizes year-round routine monitoring, and significantly enhances environmental adaptability.

[0050] (3) This invention constructs a tower health assessment model based on an improved random forest algorithm, integrates multi-dimensional parameters such as tilt and sag, and outputs a quantitative health index to provide data-driven status warning and maintenance decision-making basis for power grid operation and maintenance. Attached Figure Description

[0051] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of the present invention and should not be regarded as a limitation on the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0052] Figure 1 This is a step diagram of a high-resolution satellite remote sensing image acquisition and processing method for power transmission towers according to the present invention;

[0053] Figure 2 This is a schematic structural block diagram of an electronic device according to an embodiment of the present invention.

[0054] Icons: 101, memory; 102, processor; 103, communication interface. Detailed Implementation

[0055] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. The components of the embodiments of this application described and shown in the accompanying drawings can generally be arranged and designed in various different configurations.

[0056] Therefore, the following detailed description of the embodiments of this application provided in the accompanying drawings is not intended to limit the scope of the claimed application, but merely to illustrate selected embodiments of the application. All other embodiments obtained by those skilled in the art based on the embodiments of this application without inventive effort are within the scope of protection of this application.

[0057] It should be noted that similar labels and letters in the following figures indicate similar items. Therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures.

[0058] It should be noted that, in this document, the term "comprising" or any other variation thereof is intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0059] The following detailed description of some embodiments of this application is provided in conjunction with the accompanying drawings. Unless otherwise specified, the various embodiments and features described below can be combined with each other.

[0060] Example 1

[0061] Please see Figure 1 , Figure 1 The diagram shows the steps of a high-resolution satellite remote sensing image acquisition and processing method and system for power transmission towers provided in an embodiment of this application.

[0062] In a first aspect, this application provides a method and system for acquiring and processing high-resolution satellite remote sensing images of power transmission towers, which includes the following steps:

[0063] S1. Acquire optical remote sensing images through the satellite data interface and perform data preprocessing;

[0064] S2. Construct an improved YOLOv5x deep learning model and input preprocessed remote sensing image data to perform pole target detection;

[0065] S3. Perform multi-dimensional position correction based on the tower target detection results;

[0066] S4. Extract state parameters based on the corrected data;

[0067] S5. Construct a pole health assessment model, input the extracted state parameters into an improved random forest algorithm, output the pole health index, and automatically generate early warning information when the index is lower than the threshold.

[0068] It should be noted that the range of the pole health index is generally 0-100, and the threshold is generally set to 70.

[0069] As a preferred implementation, in step S1, the satellite data interface supports access to multi-source satellite data, including Gaofen-7, WorldView-3 and Sentinel-2, and has the ability to automatically select the best image. The selection criteria include cloud coverage <10%, shooting angle <30° and data acquisition time within the last 30 days.

[0070] Among them, Gaofen-7 has a resolution of 0.5m, WorldView-3 has a resolution of 0.3m, and Sentinel-2 has a resolution of 10m.

[0071] In a preferred implementation, step S1, data preprocessing includes sequentially performing radiometric correction, geometric correction, cloud removal, and resolution fusion processing. Radiometric correction employs a 6S atmospheric correction model, utilizing the FLAASH module in ENVI. By inputting solar elevation angle and aerosol optical thickness parameters, errors caused by atmospheric scattering and absorption during image capture are eliminated. Then, geometric fine correction is performed using a quadratic polynomial transformation model, selecting no fewer than 20 ground control points, with the root mean square error controlled within 0.5 pixels to ensure the authenticity of ground feature reflectivity. Geometric correction uses georeferenced data and high-precision control points, such as Google... Earth landmarks were used to register remote sensing images, ensuring accurate correspondence between the images and actual geographical locations. This guaranteed the spatial accuracy of subsequent classification results, keeping the geometric correction error within 0.5 pixels. Cloud removal employed a multispectral band threshold segmentation and inpainting restoration algorithm, achieving a texture similarity of ≥95% between the restored area and its surroundings. Resolution fusion was performed using a pan-sharpening algorithm, fusing 15m multispectral images with 0.5m panchromatic images. The spectral distortion was <2%, improving image clarity while preserving spectral information. Furthermore, the study area image was cropped based on the administrative boundaries of the study area, reducing data volume and improving processing efficiency.

[0072] As a preferred implementation, in step S2, the improved YOLOv5x deep learning model includes: adding a coordinate attention mechanism after the third CSP module of the backbone network CSPDarknet to enhance the ability to extract structural features of transmission towers; introducing a BiFPN bidirectional feature pyramid network in the Neck part; using the CIoU loss function to optimize bounding box regression in the output layer; employing Mosaic data augmentation and label smoothing techniques during model training; and, based on the test set, in the object detection model evaluation, when the IoU threshold is 0.5, the average accuracy of the model on the test set is not less than 96%.

[0073] As a preferred implementation, step S3 includes multi-dimensional position correction, which includes image registration based on SIFT feature point matching, with ≥80 effective matching points and a registration error ≤0.8 pixels; elevation correction combined with the DSM digital surface model, using the pixel elevation value at the bottom of the tower to compare with the design elevation, achieving a correction accuracy of ±0.5m; and spatial coordinate calibration using the tower's GIS coordinates, i.e., converting the pixel coordinates to WGS84 latitude and longitude coordinates through the RPC model and comparing them with the GIS ledger coordinates, with an average positioning error ≤1.2m, meaning that the tower positioning error after correction does not exceed 1.2 meters.

[0074] In a preferred embodiment, the state parameters in step S4 include: tower inclination, number of insulator strings, conductor sag, and tower foundation settlement.

[0075] The tower tilt is calculated using the Hough transform to detect the tilt along the tower's centerline, with a measurement accuracy of ±0.5°. The calculation formula includes:

[0076] ρ=xcosθ+ysinθ

[0077] Inclination = |θ−90°|

[0078] In the formula, ρ is the perpendicular distance from the origin to the line, θ is the angle between the perpendicular line and the positive x-axis, and (x,y) are the pixel coordinates on the line in the image.

[0079] The number of insulator strings was identified by combining HSV threshold segmentation with morphological operations, achieving an accuracy of ≥98%.

[0080] The conductor sag is calculated based on parabolic fitting, with an error ≤0.3m. The calculation formula includes:

[0081] f=wL 2 / 8T

[0082] In the formula, f is the conductor sag, w is the weight per unit length of the conductor, L is the horizontal distance between the two towers, and T is the horizontal tension of the conductor.

[0083] The settlement of the tower foundation was detected by analyzing the grayscale changes in images, with a sensitivity of 0.01m.

[0084] Step S5, the calculation process for constructing the tower health assessment model includes:

[0085] Weighted calculation of pole health index:

[0086]

[0087] An improved random forest calculates the health index through ensemble predictions from 100 decision trees. The output of a single decision tree is as follows:

[0088]

[0089] Integrated prediction:

[0090]

[0091] Feature importance weight calculation:

[0092]

[0093] In the formula, The pole health index ranges from 0 to 100, with higher values ​​indicating better health. The weight of the i-th state parameter; is the standardized score of the i-th state parameter; n is the total number of state parameters in the input model; denoted as , where is the prediction score of the k-th decision tree for the input feature vector x; c is the number of labels. Let represent the feature vector of the j-th training sample; Let this represent the label corresponding to the j-th training sample; This is an indicator function; it returns 1 when the condition is true and 0 otherwise. Let be the training sample set for the k-th tree; Let $\frac{i}{i}$ be the total decrease in Gini impurity caused by the i-th state parameter during the splitting of all decision trees in the random forest. Let Gini impurity be the sum of all state parameters.

[0094] Example 2

[0095] Secondly, this application provides a high-resolution satellite remote sensing image acquisition and processing system for power transmission towers, characterized in that it includes:

[0096] The image acquisition and processing module is used to acquire optical remote sensing images through the satellite data interface and perform data preprocessing.

[0097] The target detection module is used to build an improved YOLOv5x deep learning model and input preprocessed remote sensing image data to detect tower targets.

[0098] The position correction module is used to perform multi-dimensional position correction based on the tower target detection results;

[0099] The parameter extraction module is used to extract state parameters based on the corrected data.

[0100] The health assessment module is used to build a pole health assessment model using an improved random forest algorithm. The extracted state parameters are input into the model, and the pole health index is output. When the index is lower than the threshold, an early warning message is automatically generated.

[0101] It should be noted that the range of the pole health index is generally 0-100, and the threshold is generally set to 70.

[0102] As a preferred implementation, the satellite data interface supports access to multi-source satellite data, including Gaofen-7, WorldView-3, and Sentinel-2, and has the ability to automatically select the best image. The selection criteria include cloud coverage <10%, shooting angle <30°, and data acquisition time within the last 30 days.

[0103] Among them, Gaofen-7 has a resolution of 0.5m, WorldView-3 has a resolution of 0.3m, and Sentinel-2 has a resolution of 10m.

[0104] As a preferred implementation, data preprocessing includes sequentially performed radiometric correction, geometric correction, cloud removal, and resolution fusion. Radiometric correction employs a 6S atmospheric correction model, utilizing the FLAASH module in ENVI. By inputting solar elevation angle and aerosol optical thickness parameters, errors caused by atmospheric scattering and absorption during image capture are eliminated. Then, geometric fine correction is performed using a quadratic polynomial transformation model, selecting no fewer than 20 ground control points, with the root mean square error controlled within 0.5 pixels to ensure the accuracy of ground feature reflectivity. Geometric correction uses georeferenced data and high-precision control points, such as Google... Earth landmarks were used to register remote sensing images, ensuring accurate correspondence between the images and actual geographical locations. This guaranteed the spatial accuracy of subsequent classification results, keeping the geometric correction error within 0.5 pixels. Cloud removal employed a multispectral band threshold segmentation and inpainting restoration algorithm, achieving a texture similarity of ≥95% between the restored area and its surroundings. Resolution fusion was performed using a pan-sharpening algorithm, fusing 15m multispectral images with 0.5m panchromatic images. The spectral distortion was <2%, improving image clarity while preserving spectral information. Furthermore, the study area image was cropped based on the administrative boundaries of the study area, reducing data volume and improving processing efficiency.

[0105] As a preferred implementation, the improved YOLOv5x deep learning model includes: adding a coordinate attention mechanism after the third CSP module of the backbone network CSPDarknet to enhance the ability to extract structural features of transmission towers; introducing a BiFPN bidirectional feature pyramid network in the Neck part; using the CIoU loss function to optimize bounding box regression in the output layer; employing Mosaic data augmentation and label smoothing techniques during model training; and demonstrating through the test set that, in the object detection model evaluation, when the IoU threshold is 0.5, the model's average accuracy on the test set is no less than 96%.

[0106] As a preferred implementation, multi-dimensional position correction includes image registration based on SIFT feature point matching, with ≥80 effective matching points and a registration error ≤0.8 pixels; elevation correction combined with the DSM digital surface model, using the pixel elevation value at the bottom of the tower to compare with the design elevation, achieving a correction accuracy of ±0.5m; and spatial coordinate calibration using the tower's GIS coordinates, i.e., converting pixel coordinates to WGS84 latitude and longitude coordinates through the RPC model and comparing them with the GIS ledger coordinates, with an average positioning error ≤1.2m, meaning that the tower positioning error after correction does not exceed 1.2 meters.

[0107] As a preferred implementation, the state parameters include: tower inclination, number of insulator strings, conductor sag, and tower foundation settlement.

[0108] The tower tilt is calculated using the Hough transform to detect the tilt along the tower's centerline, with a measurement accuracy of ±0.5°. The calculation formula includes:

[0109] ρ=xcosθ+ysinθ

[0110] Inclination = |θ−90°|

[0111] In the formula, ρ is the perpendicular distance from the origin to the line, θ is the angle between the perpendicular line and the positive x-axis, and (x,y) are the pixel coordinates on the line in the image.

[0112] The number of insulator strings was identified by combining HSV threshold segmentation with morphological operations, achieving an accuracy of ≥98%.

[0113] The conductor sag is calculated based on parabolic fitting, with an error ≤0.3m. The calculation formula includes:

[0114] f=wL 2 / 8T

[0115] In the formula, f is the conductor sag, w is the weight per unit length of the conductor, L is the horizontal distance between the two towers, and T is the horizontal tension of the conductor.

[0116] The settlement of the tower foundation was detected by analyzing the grayscale changes in images, with a sensitivity of 0.01m.

[0117] The calculation process for constructing the pole health assessment model includes:

[0118] Weighted calculation of pole health index:

[0119]

[0120] An improved random forest calculates the health index through ensemble predictions from 100 decision trees. The output of a single decision tree is as follows:

[0121]

[0122] Integrated prediction:

[0123]

[0124] Feature importance weight calculation:

[0125]

[0126] In the formula, The pole health index ranges from 0 to 100, with higher values ​​indicating better health. The weight of the i-th state parameter; is the standardized score of the i-th state parameter; n is the total number of state parameters in the input model; denoted as , where is the prediction score of the k-th decision tree for the input feature vector x; c is the number of labels. Let represent the feature vector of the j-th training sample; Let this represent the label corresponding to the j-th training sample; This is an indicator function; it returns 1 when the condition is true and 0 otherwise. Let be the training sample set for the k-th tree; Let $\frac{i}{i}$ be the total decrease in Gini impurity caused by the i-th state parameter during the splitting of all decision trees in the random forest. Let Gini impurity be the sum of all state parameters.

[0127] Example 3

[0128] Please see Figure 2 , Figure 2 This is a schematic structural block diagram of an electronic device provided in an embodiment of this application.

[0129] An electronic device includes a memory 101, a processor 102, and a communication interface 103. The memory 101, processor 102, and communication interface 103 are electrically connected directly or indirectly to enable data transmission or interaction. For example, these components can be electrically connected to each other via one or more communication buses or signal lines. The memory 101 can be used to store software programs and modules. The processor 102 executes the software programs and modules stored in the memory 101 to perform various functional applications and data processing. The communication interface 103 can be used for signaling or data communication with other node devices.

[0130] The memory 101 may be, but is not limited to, random access memory (RAM), read-only memory (ROM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), etc.

[0131] The processor 102 can be an integrated circuit chip with signal processing capabilities. The processor 102 can be a general-purpose processor, including a central processing unit (CPU), a network processor (NP), etc.; it can also be a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components.

[0132] It is understood that the structure shown in the figure is for illustrative purposes only. A method and system for acquiring and processing high-resolution satellite remote sensing images of power transmission towers may include more or fewer components than those shown in the figure, or have a different configuration. The components shown in the figure can be implemented using hardware, software, or a combination thereof.

[0133] In the embodiments provided in this application, it should be understood that the disclosed methods can also be implemented in other ways. The embodiments described above are merely illustrative. For example, the flowcharts or block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of methods and computer program products according to various embodiments of this application. In this regard, each block in the flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the blocks may occur in a different order than those marked in the drawings. For example, two consecutive blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in the block diagram and / or flowchart, and combinations of blocks in the block diagram and / or flowchart, can be implemented using a dedicated hardware-based system that performs the specified function or action, or using a combination of dedicated hardware and computer instructions.

[0134] In addition, the functional modules in the various embodiments of this application can be integrated together to form an independent part, or each module can exist independently, or two or more modules can be integrated to form an independent part.

[0135] If the aforementioned functions are implemented as software functional modules and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0136] The above description is merely a preferred embodiment of this application and is not intended to limit this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the protection scope of this application.

[0137] It will be apparent to those skilled in the art that this application is not limited to the details of the exemplary embodiments described above, and that this application can be implemented in other specific forms without departing from the spirit or essential characteristics of this application. Therefore, the embodiments should be considered illustrative and non-limiting in all respects, and the scope of this application is defined by the appended claims rather than the foregoing description. Thus, all variations falling within the meaning and scope of equivalents of the claims are intended to be included within this application. No reference numerals in the claims should be construed as limiting the scope of the claims.

Claims

1. A method for acquiring and processing high-resolution satellite remote sensing images of power transmission towers, characterized in that, Includes the following steps: S1. Acquire optical remote sensing images through the satellite data interface and perform data preprocessing; S2. Construct an improved YOLOv5x deep learning model and input preprocessed remote sensing image data to perform pole target detection; S3. Perform multi-dimensional position correction based on the tower target detection results; S4. Extract state parameters based on the corrected data; S5. Construct a pole health assessment model using an improved random forest algorithm. Input the extracted state parameters into the model and output the pole health index. When the index is below the threshold, an early warning message is automatically generated.

2. The method for acquiring and processing high-resolution satellite remote sensing images of power transmission towers as described in claim 1, characterized in that, In step S1, the satellite data interface supports access to multi-source satellite data, including Gaofen-7, WorldView-3 and Sentinel-2, and has the ability to automatically select the best image. The selection criteria include cloud coverage <10%, shooting angle <30° and data acquisition time within the last 30 days.

3. The method for acquiring and processing high-resolution satellite remote sensing images of power transmission towers as described in claim 1, characterized in that, In step S1, the data preprocessing includes radiometric correction, geometric correction, cloud removal, and resolution fusion processing performed sequentially. Radiometric correction uses the 6S atmospheric correction model and the FLAASH module in ENVI to eliminate errors caused by atmospheric scattering and absorption during image acquisition. Geometric correction uses geographic reference data and high-precision control points to register the remote sensing image, ensuring that the image accurately corresponds to the actual geographical location and guaranteeing the spatial accuracy of subsequent classification results. Cloud removal uses a multispectral band threshold segmentation and inpainting restoration algorithm. Resolution fusion processing uses the Pan-sharpening algorithm.

4. The method for acquiring and processing high-resolution satellite remote sensing images of power transmission towers as described in claim 1, characterized in that, In step S2, the improved YOLOv5x deep learning model includes: adding a coordinate attention mechanism after the third CSP module of the backbone network CSPDarknet; introducing a BiFPN bidirectional feature pyramid network in the Neck part; using the CIoU loss function to optimize bounding box regression in the output layer; and using Mosaic data augmentation and label smoothing techniques during model training.

5. The method for acquiring and processing high-resolution satellite remote sensing images of power transmission towers as described in claim 1, characterized in that, In step S3, the multi-dimensional position correction includes image registration based on SIFT feature point matching, elevation correction combined with DSM digital surface model, and spatial coordinate calibration using tower GIS coordinates.

6. The method for acquiring and processing high-resolution satellite remote sensing images of power transmission towers as described in claim 1, characterized in that, In step S4, the state parameters include: tower inclination, number of insulator strings, conductor sag, and tower foundation settlement.

7. The method for acquiring and processing high-resolution satellite remote sensing images of power transmission towers as described in claim 1, characterized in that, In step S5, the calculation process for constructing the tower health assessment model includes: Weighted calculation of pole health index: ， An improved random forest calculates the health index through ensemble predictions from 100 decision trees. The output of a single decision tree is as follows: ， Integrated prediction: ， Feature importance weight calculation: ， In the formula, The pole health index ranges from 0 to 100, with higher values ​​indicating better health. The weight of the i-th state parameter; is the standardized score of the i-th state parameter; n is the total number of state parameters in the input model; denoted as , where is the prediction score of the k-th decision tree for the input feature vector x; c is the number of labels. Let represent the feature vector of the j-th training sample; Let this represent the label corresponding to the j-th training sample; This is an indicator function; it returns 1 when the condition is true and 0 otherwise. Let be the training sample set for the k-th tree; Let $\frac{i}{i}$ be the total decrease in Gini impurity caused by the i-th state parameter during the splitting of all decision trees in the random forest. Let Gini impurity be the sum of all state parameters.

8. A high-resolution satellite remote sensing image acquisition and processing system for power transmission towers, characterized in that, include: The image acquisition and processing module is used to acquire optical remote sensing images through the satellite data interface and perform data preprocessing. The target detection module is used to build an improved YOLOv5x deep learning model and input preprocessed remote sensing image data to detect tower targets. The position correction module is used to perform multi-dimensional position correction based on the tower target detection results; The parameter extraction module is used to extract state parameters based on the corrected data. The health assessment module is used to build a pole health assessment model using an improved random forest algorithm. The extracted state parameters are input into the model, and the pole health index is output. When the index is lower than the threshold, an early warning message is automatically generated.

9. An electronic device, characterized in that, include: Memory, used to store one or more programs; processor; When the processor executes the one or more programs, it implements a method for acquiring and processing high-resolution satellite remote sensing images of power transmission towers as described in any one of claims 1-7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements a method for acquiring and processing high-resolution satellite remote sensing images of power transmission towers as described in any one of claims 1-7.

Images