A UAV-based intelligent inspection system for power transmission networks

Abstract

The present invention provides an intelligent inspection system for a power transmission network based on an unmanned aerial vehicle, comprising an unmanned aerial vehicle main body, an image acquisition unit, a data processing unit, a wireless communication unit and a ground control station. The unmanned aerial vehicle main body is equipped with a flight control module and a power device to realize autonomous flight and fixed-point hovering. The image acquisition unit is installed on the unmanned aerial vehicle main body and comprises a plurality of modules for acquiring multispectral images, thermal imaging images and three-dimensional point cloud data of the power transmission line. The data processing unit is integrated in the unmanned aerial vehicle main body and is responsible for receiving and processing data from the image acquisition unit. The wireless communication unit is also integrated in the unmanned aerial vehicle main body and is used to transmit the processed data to the ground control station. The ground control station is used to receive and store data from the unmanned aerial vehicle and perform fault analysis and diagnosis. The present invention realizes efficient and accurate inspection of the power transmission network through the integration of multiple sensors and the automation of data processing, and significantly improves the safety and efficiency of the power transmission network maintenance.

Description

Technical Field

[0001] The present invention relates to the technical field of unmanned aerial vehicles (UAVs), and in particular to an intelligent inspection system for power transmission networks based on UAVs. Background Art

[0002] Traditional power transmission network inspection methods mainly rely on manual inspection and ground equipment. With the advancement of technology, more and more inspection systems are beginning to introduce drones. Existing drone inspection systems are usually equipped with basic flight control modules and camera equipment, which can fly within a predetermined route and obtain image data of transmission lines. These systems have improved inspection efficiency to a certain extent and reduced the danger and work intensity of manual inspections.

[0003] However, the existing drone inspection system has some technical limitations. First, traditional systems are often equipped with only a single type of camera, such as a visible light camera, which cannot fully capture all kinds of potential problems in the transmission line. Secondly, the data processing capacity is limited, and the image data usually needs to be transmitted back to the ground for subsequent processing and analysis, resulting in poor real-time performance and failure to detect and handle faults in a timely manner. In addition, the data transmission process is easily affected by environmental interference, resulting in unstable or lost data transmission, affecting the accuracy of the inspection results.

[0004] Therefore, it is urgent to develop a new type of intelligent inspection system for power transmission networks based on drones. Summary of the invention

[0005] The present application provides a transmission network intelligent inspection system based on drones to improve the safety and efficiency of transmission network maintenance.

[0006] The present application provides a transmission network intelligent inspection system based on a drone, comprising:

[0007] The main body of the drone has a flight control module and a power unit for achieving autonomous flight and fixed-point hovering;

[0008] The image acquisition unit includes a multispectral imaging module, an infrared imaging module and a laser radar module installed on the main body of the UAV; wherein the multispectral imaging module is used to obtain a multispectral image of the transmission line; the infrared imaging module is used to obtain a thermal imaging image of the transmission line; and the laser radar module is used to obtain three-dimensional point cloud data of the transmission line;

[0009] a data processing unit, integrated in the drone body, for receiving and processing data from the image acquisition unit;

[0010] A wireless communication unit, integrated in the drone body, for transmitting the data processed by the data processing module to a ground control station;

[0011] The ground control station is used to receive and store data from the UAV body and perform fault analysis and diagnosis.

[0012] Furthermore, the flight control module includes a GPS positioning system and an inertial measurement unit for providing precise flight navigation and attitude control.

[0013] Furthermore, the data processing unit includes a multispectral image processing module, which executes a first image processing algorithm for identifying wire wear and corrosion in the multispectral image, wherein the first image processing algorithm includes the following steps:

[0014] According to the following formula 1, the multispectral image is subjected to noise removal:

[0015]

[0016] Where I(x, y, λ) represents the original multispectral image; I median (x, y, λ) is the denoised image; x and y are image coordinates; λ is the wavelength; median represents the median filter operation;

[0017] According to the following formula 2-4, for the image I after noise removal using the gradient operator median (x, y, λ) extracts the edge features of the image and obtains the edge intensity image E(x, y, λ):

[0018]

[0019] Among them, G x (x,y,λ) and G y (x, y, λ) represents the gradient of the image in the x and y directions respectively;

[0020] According to the following formula 5, the edge images of different wavelengths are weighted fused to generate the fused feature image F(x,y):

[0021]

[0022] Among them, ω λ is the weighting coefficient of different wavelengths;

[0023] According to the following formula 6, the fused feature image F(x, y) is binarized to obtain the damaged area image D(x, y):

[0024]

[0025] Among them, τ is the set threshold;

[0026] According to the following formulas 7 and 8, the damaged area is refined to obtain the damaged area Adamage and injury severity S damage :

[0027]

[0028] Wherein, {(x′, y′)|D(x′, y′)=1} represents the set of pixel coordinates belonging to the damaged area;

[0029] The damaged area A damage , Severity S damage The damage inspection report includes the damage location, area and severity.

[0030] Furthermore, the data processing unit includes a thermal imaging image processing module, which executes a second image processing algorithm for identifying temperature anomalies in thermal imaging images, and the second image processing algorithm includes the following steps:

[0031] According to the following formula 9, the temperature of the thermal imaging image is calibrated to remove the influence of the ambient temperature:

[0032] T cal (x,y)=T raw (x,y)-T ambient (9)

[0033] Among them, T cal (x, y) is the calibrated temperature image, T raw (x, y) is the original thermal imaging temperature image, T ambient is the ambient temperature;

[0034] The temperature gradient is calculated according to the following formula 10:

[0035]

[0036] Among them, G T (x, y) is the temperature gradient image;

[0037] According to the following formula 11, the temperature gradient image is analyzed to identify the temperature abnormality area:

[0038]

[0039] Among them, θ is the threshold for temperature anomaly detection; A T (x, y) is the binary image of the temperature anomaly area;

[0040] According to the following formulas 12 and 13, the temperature anomaly area is quantified to obtain the temperature anomaly area A anom and the temperature anomaly S damage:

[0041]

[0042] According to the following formula 14, the perimeter P of the temperature anomaly area is obtained: anom :

[0043]

[0044] Among them, Boundary (A T ) represents the boundary pixel set of the binary temperature anomaly area; Δx and Δy are the distances between boundary pixels, both of which are 1 pixel unit;

[0045] According to the following formula 15, the shape compactness C of the temperature anomaly area is calculated: shape :

[0046]

[0047] The temperature abnormal area A anom , temperature abnormality S damage , shape compactness of the temperature anomaly area C shape Combined with the geographic coordinate data, a temperature inspection report is generated, wherein the temperature anomaly report includes the temperature anomaly location, area, temperature value and shape characteristics.

[0048] Furthermore, the data processing unit includes a point cloud data processing module, which executes a point cloud processing algorithm for detecting deformation of transmission towers and conductors, and the point cloud processing algorithm includes the following steps:

[0049] According to the following formula 16, the collected three-dimensional point cloud data P is denoised:

[0050] P clean ={p i ∣p i ∈P,||p i -median(P)||<σ} (16)

[0051] Among them, P clean is the 3D point cloud data set after denoising; p i represents the i-th point in the 3D point cloud data P; median(P) is the median point of the point cloud data; σ is the preset noise threshold;

[0052] The denoised three-dimensional point cloud data set is plane-fitted to extract the reference planes of the transmission towers and conductors; wherein the plane-fitting uses the least squares method, and the fitting formula is as follows: Formula 17:

[0053]

[0054] Among them, a, b, c, d are the parameters of the fitting plane equation; (x i ,y i ,z i ) is point p i The coordinates of i is the adaptive weight, calculated according to the following formula 18:

[0055]

[0056] Among them, ρ i For point p i The local density of

[0057] According to the following formula 19, calculate point p i Distance to the fitted plane:

[0058]

[0059] Where a, b, c, and d are the parameters of the fitting plane; (x i ,y i ,z i ) is point p i The coordinates of

[0060] According to the following formula 20, the calculated distance is analyzed to identify and mark the deformation area:

[0061]

[0062] Among them, M i For point p i The deformation mark, 1 indicates abnormal deformation, 0 indicates normal; v is the adaptive threshold of deformation detection;

[0063] According to the following formulas 21 and 22, calculate the deformation degree D deform and deformation area A deform :

[0064]

[0065] Among them, N deform is the number of points in the deformation area; A i For point p i Area represented;

[0066] The deformation degree D deform , deformation area A deform Combined with the geographic coordinate data, a deformation inspection report is generated, wherein the deformation inspection report includes the deformation location, degree and range.

[0067] Furthermore, click p i The local density ρ i The calculation formula steps include:

[0068] Select radius r as the neighborhood range;

[0069] Statistics at point p i The number of points N within the neighborhood radius r i :

[0070] The local density ρ is calculated according to the following formula 23: i :

[0071]

[0072] in, is the volume of the spherical neighborhood.

[0073] Furthermore, the multi-spectral imaging module includes a plurality of filters for capturing spectral images within different wavelength ranges.

[0074] Furthermore, the laser radar module has a rotating scanning device for acquiring 360-degree three-dimensional point cloud data.

[0075] Furthermore, the ground control station includes a user interface module for displaying inspection results and fault analysis reports, and allowing operators to perform interactive control and data query.

[0076] Furthermore, the ground control station has a data storage unit for long-term storage of historical inspection data and analysis results to facilitate trend analysis and predictive maintenance.

[0077] The beneficial effects of this application mainly include: (1) The flight control module and power unit equipped on the main body of the drone enable it to achieve autonomous flight and fixed-point hovering, which improves the flexibility and coverage of inspections and reduces the complexity and uncertainty of manual operations. (2) The laser radar module is used to obtain three-dimensional point cloud data of the transmission line, which can accurately measure the shape and position of the wires, identify the deformation of transmission towers and wires, and improve the efficiency of equipment management and maintenance. (3) The data processing unit is integrated in the main body of the drone and can receive and process data from the image acquisition unit in real time, realizing real-time analysis and processing of data, and improving inspection efficiency and response speed. BRIEF DESCRIPTION OF THE DRAWINGS

[0078] Figure 1 It is a schematic diagram of a UAV-based intelligent inspection system for power transmission networks provided in the first embodiment of the present application. DETAILED DESCRIPTION

[0079] Many specific details are described in the following description to facilitate a full understanding of the present application. However, the present application can be implemented in many other ways than those described herein, and those skilled in the art can make similar generalizations without violating the connotation of the present application, so the present application is not limited by the specific implementation disclosed below.

[0080] The first embodiment of the present application provides a transmission network intelligent inspection system based on drones. Figure 1 , which is a schematic diagram of the first embodiment of the present application. Figure 1 The first embodiment of the present application provides a detailed description of a power transmission network intelligent inspection system based on a drone.

[0081] The UAV-based power transmission network intelligent inspection system includes a UAV body 101 , an image acquisition unit 102 , a data processing unit 103 , a wireless communication unit 104 and a ground control station 105 .

[0082] The drone body 101 has a flight control module and a power device for realizing autonomous flight and fixed-point hovering.

[0083] The drone body 101 is one of the core components, which has a flight control module and a power unit for realizing autonomous flight and fixed-point hovering. The flight control module contains multiple submodules, including a GPS positioning system, an inertial measurement unit (IMU), and an attitude control system. The GPS positioning system provides high-precision geographic location information, enabling the drone to navigate autonomously according to a predetermined flight path. The inertial measurement unit includes an accelerometer and a gyroscope, which monitors the attitude and motion state of the drone in real time and provides data to the attitude control system. The attitude control system adjusts the flight attitude of the drone based on these data to ensure that it remains stable during flight.

[0084] The power unit includes multiple rotors, electric motors and battery packs. Each rotor is driven by a high-efficiency electric motor to provide the necessary lift and thrust. The electric motor is connected to the flight control module through the electronic speed controller (ESC), and receives control signals to adjust the speed of the rotors to achieve the lifting, forward, backward and steering of the drone. The battery pack provides power for all electronic devices of the drone, with high energy density and fast charging function, ensuring that the drone can operate stably for a long time.

[0085] The main body of the drone also includes an obstacle avoidance module, which is used to detect and avoid obstacles during flight. The obstacle avoidance module consists of multiple ultrasonic sensors and infrared sensors, which are distributed in the front, back, left, right and bottom of the drone. The ultrasonic sensor measures the distance to the obstacle by transmitting and receiving ultrasonic signals, while the infrared sensor determines the location of the obstacle by detecting the reflection of infrared light. The obstacle avoidance module transmits the detected information to the flight control module, which adjusts the flight path accordingly to avoid collision with obstacles.

[0086] The frame of the drone body is made of lightweight and high-strength materials such as carbon fiber and aluminum alloy to ensure sufficient structural strength while reducing weight, thereby improving flight performance and flight time. In order to ensure stable flight in adverse weather conditions, the frame design also has windproof and shock-resistant functions. The shell design of the drone body takes into account aerodynamic characteristics to reduce flight resistance.

[0087] In addition, the flight control module also includes an automatic return function, which can automatically return to the take-off point when the drone is low on power or the signal is lost, ensuring the safety of the drone and the collected data.

[0088] The drone body 101 realizes highly automated and intelligent autonomous flight and fixed-point hovering functions through the integration and coordinated work of the above-mentioned multiple modules and devices, meeting various requirements of intelligent inspection of the power transmission network. The drone body 101 not only has high-precision flight control capabilities, but also has good obstacle avoidance capabilities and long-term stable working characteristics, providing a solid foundation for the overall performance and reliability of the system.

[0089] Furthermore, the flight control module includes a GPS positioning system and an inertial measurement unit for providing precise flight navigation and attitude control.

[0090] The flight control module is the core component of the drone body, responsible for realizing the autonomous flight and precise hovering of the drone. First of all, the GPS positioning system is an important component of the flight control module, which provides real-time location information of the drone by receiving signals from global positioning system satellites. The GPS system includes a high-sensitivity receiver and an antenna. The receiver analyzes satellite signals, calculates the latitude, longitude and altitude of the drone, and ensures that the drone can navigate accurately during flight. In order to improve positioning accuracy, the system can also integrate differential GPS technology, and further reduce positioning errors through correction data provided by ground base stations to achieve sub-meter accuracy.

[0091] The inertial measurement unit (IMU) is another key component of the flight control module, which is used to measure the acceleration and angular velocity of the drone. The IMU usually includes a three-axis accelerometer and a three-axis gyroscope. The accelerometer measures the linear acceleration of the drone in all directions, and the gyroscope measures the rotation speed of each axis. The IMU monitors the motion state and attitude changes of the drone in real time and transmits this data to the flight control system. The flight control system calculates the current attitude and position of the drone by fusing IMU data and GPS data, and performs attitude control and heading adjustment. The high-frequency data update of the IMU enables the system to respond quickly to attitude changes in flight and maintain flight stability and trajectory accuracy.

[0092] To achieve these functions, a high-performance processor is integrated into the flight control module to execute the navigation algorithm and attitude control algorithm. The processor receives data from the GPS positioning system and IMU, and eliminates noise and errors through data fusion technologies such as Kalman filtering to improve the estimation accuracy of position and attitude. Based on the real-time estimated attitude and position data, the processor calculates the required control instructions and adjusts the various power units of the drone to ensure that the drone performs inspections according to the predetermined flight path.

[0093] In addition, the flight control module also has an automatic return function. When the drone is low on power or the communication signal with the ground control station is lost, the system will automatically start the return program and use the GPS positioning system to guide the drone back to the take-off point, ensuring the safety of the drone and the collected data.

[0094] Through the above design, the flight control module can provide precise flight navigation and attitude control, ensuring that the UAV can complete its tasks stably and efficiently during the power transmission network inspection process.

[0095] The image acquisition unit 102 includes a multispectral imaging module, an infrared imaging module and a laser radar module installed on the drone body; wherein the multispectral imaging module is used to obtain a multispectral image of the transmission line; the infrared imaging module is used to obtain a thermal imaging image of the transmission line; and the laser radar module is used to obtain three-dimensional point cloud data of the transmission line.

[0096] The image acquisition unit 102 is an important part of the intelligent inspection system for power transmission lines. It is installed on the main body of the drone and is designed to obtain detailed images and three-dimensional data of the power transmission lines. The unit includes a multi-spectral imaging module, an infrared imaging module, and a lidar module, which work together to provide comprehensive inspection information.

[0097] The multispectral imaging module is used to obtain multispectral images of transmission lines. The module is equipped with multiple filters and high-sensitivity sensors, which can capture spectral information in different wavelength ranges. Specifically, the multispectral imaging module includes a high-resolution imaging sensor and a set of switchable filter wheels, each filter corresponding to a specific wavelength range, such as red light, green light, blue light, near infrared, etc. By switching the filters sequentially, the sensor can continuously capture image data in multiple bands, which can reveal the physical and chemical properties of transmission line materials and help identify wear, corrosion and other potential problems of the conductors.

[0098] The infrared imaging module is used to obtain thermal imaging images of the transmission line. The module contains a highly sensitive infrared detector that can detect infrared radiation emitted by objects and convert it into thermal images. The infrared imaging module is equipped with a high-precision uncooled microbolometer, which can provide high-resolution thermal imaging data over a wide temperature range. By analyzing the thermal image, overheating of the transmission line and equipment can be detected, thereby warning of potential failures and safety hazards. The infrared imaging module also integrates a temperature calibration device to ensure the accuracy of the temperature data obtained under different environmental conditions.

[0099] The LiDAR module is used to obtain three-dimensional point cloud data of transmission lines and provide accurate spatial information. The LiDAR module consists of a rotating laser transmitter and multiple receivers. It can emit laser pulses at a frequency of hundreds of thousands of times per second and measure the time it takes for the laser pulse to reflect back from the surface of an object. Based on the flight time and speed of the laser pulse, the system can calculate the precise distance of each reflection point and generate a high-density three-dimensional point cloud map. These point cloud data can be used to build a three-dimensional model of transmission towers and wires and detect their deformation and displacement. The LiDAR module can also be equipped with an optical stabilization system to reduce the impact of vibration and displacement generated by the drone during flight on data collection.

[0100] Each module of the image acquisition unit 102 is connected to the data processing unit in the drone body through a high-speed data interface to ensure that the collected images and three-dimensional data can be transmitted and processed in real time. The installation positions of these modules are carefully designed and calibrated to ensure that the best observation angle and data quality can be obtained during the flight.

[0101] By integrating multi-spectral imaging, infrared imaging and lidar technology, the image acquisition unit 102 can provide comprehensive information of the transmission line, thereby improving the accuracy and efficiency of inspection.

[0102] Furthermore, the multi-spectral imaging module includes a plurality of filters for capturing spectral images within different wavelength ranges.

[0103] The multispectral imaging module is an important part of this system, responsible for acquiring multispectral images of the transmission line in order to identify and analyze the status and potential problems of the conductors. The module is equipped with a set of precisely designed filters that selectively allow light within a specific wavelength range to pass, allowing the imaging sensor to capture spectral information at different wavelengths.

[0104] In a specific implementation, a multispectral imaging module usually includes a high-resolution imaging sensor that can capture high-quality image data. A filter wheel is installed in front of the imaging sensor. The filter wheel can accommodate multiple filters, each of which corresponds to a specific wavelength range, such as red light, green light, blue light, and near-infrared light in visible light. The filter wheel can be rotated by a motor or other mechanical device to move different filters in front of the sensor at different shooting times, so that the sensor can capture image data in each wavelength range in turn.

[0105] During operation, the UAV flight control system controls the operation of the multispectral imaging module, automatically switches the filter and captures multispectral images according to the predetermined time interval or the change of the flight path. These image data are then transmitted to the data processing unit through a high-speed interface for processing and analysis.

[0106] The selection and arrangement of filters are carefully designed to cover the critical spectral ranges required for transmission line inspection. For example, red, green, and blue filters are used to capture visible light images to help identify physical wear and external contamination; near-infrared filters are used to detect the health of vegetation and the degree of corrosion on conductors; and ultraviolet filters can be used to detect corona discharge phenomena.

[0107] To improve the accuracy and consistency of the image, the multispectral imaging module also includes an automatic calibration system that can adjust the imaging parameters in real time during the shooting process to ensure the stability of the image quality captured under different lighting conditions. The automatic calibration system includes a light intensity sensor and a color correction device, which monitors the ambient light conditions in real time, adjusts the exposure time and aperture size, and optimizes the quality of each frame.

[0108] These multispectral images are fused and analyzed in the data processing unit. By comparing and calculating images of different wavelengths, more feature information can be extracted, so as to more accurately identify and evaluate the status of the transmission line. For example, the spectral reflectance characteristics can be used to distinguish different materials, identify corrosion and wear areas, and detect potential threats to the transmission line caused by vegetation.

[0109] Through the above detailed design, the multispectral imaging module can provide rich spectral information and provide high-precision data support for power transmission network inspection.

[0110] Furthermore, the laser radar module has a rotating scanning device for acquiring 360-degree three-dimensional point cloud data.

[0111] The LiDAR module is an important component of the system, specially designed to capture detailed 3D structural information of transmission lines. Its core components include laser transmitters, receivers and rotating scanning devices, which work together to generate high-resolution 3D point cloud data.

[0112] The laser transmitter quickly emits laser pulses, which are reflected back after encountering the surface of an object and received by the receiver. By measuring the time interval from the emission to the reception of the laser pulse, the system can calculate the precise distance between each reflection point and the laser radar. These distance data, combined with the emission angle information, constitute the basic unit of 3D point cloud data.

[0113] The rotating scanning device is a key component of the LiDAR module, which enables the laser transmitter and receiver to perform omnidirectional scanning. The rotating scanning device usually includes a high-precision rotating motor that can drive the laser transmitter and receiver to rotate 360 ​​degrees around the vertical axis. During each rotation, the laser transmitter and receiver can capture three-dimensional data of the surrounding environment. Through continuous rotation scanning, the system can generate complete three-dimensional point cloud data covering the entire surrounding environment.

[0114] To ensure the accuracy and reliability of the data, the LiDAR module is also equipped with a high-precision angle encoder that records the angular position of the rotating scanning device in real time. By combining the data from the angle encoder and the time measurement of the laser pulse, the system can accurately calculate the three-dimensional coordinates of each point.

[0115] In addition, the LiDAR module uses multi-echo technology, which can simultaneously process echo signals from multiple reflecting surfaces. This technology enables the system to penetrate some obstacles, such as leaves or raindrops, and obtain deeper structural information, thereby improving the integrity and accuracy of the data.

[0116] The 3D point cloud data generated by the LiDAR module is transmitted to the data processing unit via a high-speed data interface. In the data processing unit, the point cloud data undergoes further processing, including denoising, filtering, and registration steps, to generate accurate 3D models. These 3D models can be used to detect deformation, displacement, and other structural problems of transmission towers and conductors, providing important data support for maintenance and repair.

[0117] The design of the rotating scanning device of the LiDAR module also takes into account the vibration and movement generated by the drone during flight. In order to reduce these effects, a vibration isolation device and a dynamic compensation algorithm are integrated into the module. The vibration isolation device reduces the transmission of external vibration to the LiDAR module through the physical structure, while the dynamic compensation algorithm corrects the point cloud data by analyzing the motion state of the drone in real time to ensure the accuracy of the data.

[0118] Through the above detailed design, the LiDAR module can achieve 360-degree all-round scanning, provide high-resolution three-dimensional point cloud data, and provide comprehensive structural information for intelligent inspection of the power transmission network.

[0119] The data processing unit 103 is integrated in the drone body and is used to receive and process the data from the image acquisition unit.

[0120] The data processing unit 103 is integrated in the drone body and is the data processing core of the entire system, responsible for receiving and processing data from the image acquisition unit 102. The data processing unit includes a high-performance processor, memory, and various interface modules to ensure fast and efficient processing of large amounts of image and point cloud data.

[0121] High-performance processors are the core of the data processing unit, usually using multi-core CPU or GPU architecture to meet the needs of real-time data processing. These processors are able to process multispectral images, infrared images and 3D point cloud data in parallel, ensuring real-time analysis and processing of data during the flight of the drone. To optimize processing efficiency, the processor also integrates dedicated image processing and computing acceleration modules, which can quickly execute complex image processing algorithms such as edge detection, image fusion and point cloud registration.

[0122] The memory section includes cache and mass storage. The cache is used to temporarily store intermediate data to ensure that the processor can quickly access and process the data, while the mass storage is used to store processed image and point cloud data. After processing, these data will be compressed and stored to occupy the minimum bandwidth before being transmitted to the ground control station.

[0123] The data processing unit also includes a plurality of interface modules, which are used to connect the image acquisition unit and the wireless communication unit. The interface modules include high-speed serial interfaces and parallel interfaces to ensure the transmission speed and reliability of data from the image acquisition unit to the processing unit. The data processing unit receives raw data from the multispectral imaging module, infrared imaging module and lidar module through these interface modules, and then performs pre-processing such as denoising, correction and enhancement.

[0124] In multispectral image processing, the data processing unit performs operations such as image correction, denoising, and feature extraction. Infrared image data is calibrated by temperature and thermal image analysis is used to identify temperature anomalies. For lidar data, the processing unit generates an accurate 3D model through point cloud registration and 3D modeling to identify and evaluate structural deformation and displacement.

[0125] The data processing unit is also equipped with adaptive algorithms that can adjust processing parameters according to different environments and task requirements. These algorithms automatically optimize the processing flow by analyzing the characteristics of the data in real time to ensure the accuracy and efficiency of data processing.

[0126] The processed data is transmitted to the ground control station 105 via the wireless communication unit 104. In order to ensure the reliability of data transmission, the data processing unit compresses and encrypts the processed data to reduce the transmission bandwidth and ensure data security. The compression algorithm selects the optimal compression strategy according to the data type to minimize the amount of data, while the encryption algorithm uses advanced encryption technology to prevent the data from being stolen or tampered with during transmission.

[0127] In summary, the data processing unit 103 can process the data from the image acquisition unit efficiently and in real time through the collaborative work of high-performance processors, memory and multiple interface modules, and optimize the processing flow through adaptive algorithms to ensure that the drone can play the maximum role in the intelligent inspection of the power transmission network.

[0128] Furthermore, the data processing unit includes a multispectral image processing module, which executes a first image processing algorithm for identifying wire wear and corrosion in the multispectral image, wherein the first image processing algorithm includes the following steps:

[0129] According to the following formula 1, the multispectral image is subjected to noise removal:

[0130]

[0131] Where I(x, y, λ) represents the original multispectral image; I median (x, y, λ) is the denoised image; x and y are image coordinates; λ is the wavelength; median represents the median filter operation;

[0132] According to the following formula 2-4, for the image I after noise removal using the gradient operator median (x, y, λ) extracts the edge features of the image and obtains the edge intensity image E(x, y, λ):

[0133]

[0134] Among them, G x (x,y,λ) and Gy (x, y, λ) represents the gradient of the image in the x and y directions respectively;

[0135] According to the following formula 5, the edge images of different wavelengths are weighted fused to generate the fused feature image F(x,y):

[0136]

[0137] Among them, ω λ is the weighting coefficient of different wavelengths;

[0138] According to the following formula 6, the fused feature image F(x, y) is binarized to obtain the damaged area image D(x, y):

[0139]

[0140] Among them, τ is the set threshold;

[0141] According to the following formulas 7 and 8, the damaged area is refined to obtain the damaged area A damage and injury severity S damage :

[0142]

[0143] Wherein, {(x′, y′)|D(x′, y′)=1} represents the set of pixel coordinates belonging to the damaged area;

[0144] The damaged area A damage , Severity S damage The damage inspection report includes the damage location, area and severity.

[0145] The data processing unit includes a multispectral image processing module that executes a first image processing algorithm for identifying wire wear and corrosion in a multispectral image. The detailed steps of the algorithm are as follows:

[0146] First, the multispectral image is denoised using the following formula:

[0147]

[0148] Among them, I(x,y,λ) represents the original multispectral image, I median (x, y, λ) is the denoised image, y is the spatial coordinate of the image, λ is the wavelength, and median represents the median filter operation. The median filter removes noise by calculating the median of a pixel and its surrounding pixels, effectively removing random noise in the image while retaining the edge details of the image.

[0149] Next, the denoised image I median (x, y, λ) is used to extract edge features. First, the gradient of the image in the x and y directions is calculated. The formula is as follows:

[0150]

[0151] Among them, G x (x,y,λ) and G y (x, y, λ) represents the gradient of the image in the x and y directions respectively. The gradient represents the rate of change of the image brightness and is a method for detecting the edge of the image. By performing differential operations on the image, the gradient information in each direction can be obtained.

[0152] Then, use this gradient information to calculate the edge intensity image E(x,y,λ):

[0153]

[0154] The edge intensity image (x, y, λ) reflects the strength of the edge in the image by combining the gradient magnitudes in the x and y directions. The greater the edge intensity, the more dramatic the brightness change at the point, which may be an edge or boundary in the image.

[0155] In order to integrate the information of different wavelengths, the edge images of different wavelengths are weighted fused to generate the fused feature image F(x,y):

[0156]

[0157] Among them, ω λ is the weighting coefficient of different wavelengths, which can be obtained through experimental data or set according to expert knowledge. Weighted fusion combines the edge information of each wavelength to more comprehensively reflect the structural features in the image. λ The choice of wavelength depends on the specific application and can be adjusted according to the sensitivity of the target features at different wavelengths.

[0158]

[0159] Among them, τ is the set threshold, which can be obtained through experimental data or set according to expert knowledge. Binarization processing divides the image into damaged areas and non-damaged areas, which is helpful for subsequent analysis and processing.

[0160] Then, the damaged area is refined. First, the damaged area A is calculated. damage :

[0161]

[0162] The damage area is obtained by summing all the damage points in the binary image D(x,y), reflecting the total area of ​​the damaged region.

[0163] Next, calculate the injury severity S damage :

[0164]

[0165] Among them, {(x′, y′)|D(x′, y′)=1} represents the set of pixel coordinates belonging to the damaged area. The severity of the damage is reflected by averaging the values ​​of the fused feature image F(x, y) in the damaged area.

[0166] The damaged area A damage , Severity S damage The data is combined with geographic coordinate data to generate a damage inspection report that includes the location, area and severity of the damage.

[0167] Furthermore, the data processing unit includes a thermal imaging image processing module, which executes a second image processing algorithm for identifying temperature anomalies in thermal imaging images, and the second image processing algorithm includes the following steps:

[0168] According to the following formula 9, the temperature of the thermal imaging image is calibrated to remove the influence of the ambient temperature:

[0169] T cal (x,y)=T raw (x,y)-T ambient (9)

[0170] Among them, T cal (x, y) is the calibrated temperature image, T raw (x, y) is the original thermal imaging temperature image, T ambient is the ambient temperature;

[0171] The temperature gradient is calculated according to the following formula 10:

[0172]

[0173] Among them, G T (x, y) is the temperature gradient image;

[0174] According to the following formula 11, the temperature gradient image is analyzed to identify the temperature abnormality area:

[0175]

[0176] Among them, θ is the threshold for temperature anomaly detection; A T(x, y) is the binary image of the temperature anomaly area;

[0177] According to the following formulas 12 and 13, the temperature anomaly area is quantified to obtain the temperature anomaly area A anom and the temperature anomaly S damage :

[0178]

[0179] According to the following formula 14, the perimeter P of the temperature anomaly area is obtained: anom :

[0180]

[0181] Among them, Boundary (A T ) represents the boundary pixel set of the binary temperature anomaly area; Δx and Δy are the distances between boundary pixels, both of which are 1 pixel unit;

[0182] According to the following formula 15, the shape compactness C of the temperature anomaly area is calculated: shape :

[0183]

[0184] The temperature abnormal area A anom , temperature abnormality S damage , shape compactness of the temperature anomaly area C shape Combined with the geographic coordinate data, a temperature inspection report is generated, wherein the temperature anomaly report includes the temperature anomaly location, area, temperature value and shape characteristics.

[0185] The data processing unit includes a thermal imaging image processing module, which executes a second image processing algorithm for identifying temperature anomalies in thermal imaging images. The following is a detailed description of the algorithm.

[0186] First, the thermal imaging image is temperature calibrated to remove the influence of ambient temperature. The specific formula is:

[0187] T cal (x,y)=T raw (x,y)-T ambient (9)

[0188] Among them, T cal (x,y) represents the calibrated temperature image, T raw (x, y) is the original thermal imaging temperature image, T ambientis the ambient temperature. x and y are the image coordinates. The calibration process eliminates the influence of external factors on the thermal imaging data by subtracting the ambient temperature, ensuring that only the actual temperature of the wires and equipment is reflected.

[0189] Next, calculate the temperature gradient using the following formula:

[0190]

[0191] Among them, G T (x, y) represents the temperature gradient image. The temperature gradient reflects the rate of temperature change. By calculating the derivative of the image in the x and y directions, it can be detected.

[0192] Then, the temperature gradient image is analyzed to identify the temperature abnormality area. The formula is:

[0193]

[0194] Among them, θ is the threshold for temperature anomaly detection, A T (x, y) is a binary image of the temperature anomaly area. By setting an appropriate threshold, the area with a temperature gradient higher than the threshold can be marked as an abnormal area.

[0195] Next, the temperature anomaly area is quantified to obtain the area and severity of the temperature anomaly area. First, the area A of the temperature anomaly area is calculated. anom :

[0196]

[0197] This formula is obtained by binarizing the image A T The sum of all abnormal points in (x, y) is used to obtain the total area of ​​the abnormal region.

[0198] Then, calculate the severity S of the temperature anomaly damage :

[0199]

[0200] Among them, {(x′,y′)|A T (x′, y′)=1} represents the set of pixel coordinates belonging to the abnormal area. This formula reflects the temperature severity of the abnormal area by averaging the calibration temperature in the abnormal area.

[0201] Further, the perimeter P of the temperature anomaly area is obtained anom , the formula is:

[0202]

[0203] Among them, Boundary (A T) represents the set of boundary pixels of the binary temperature anomaly area, Δx and Δy are the distances between boundary pixels, both in one pixel unit. This formula calculates the sum of the distances of boundary pixels to obtain the perimeter of the anomaly area.

[0204] Finally, the shape compactness C of the temperature anomaly area is calculated shape , the formula is:

[0205]

[0206] This formula measures the shape compactness of the abnormal area through the ratio of perimeter to area. The smaller the value, the compacter the shape.

[0207] Through the above steps, the area A of the temperature anomaly area is obtained. anom , Severity S damage and shape compactness C shape The information is combined with geographic coordinate data to generate an inspection report containing the temperature anomaly location, area, temperature value and shape characteristics.

[0208] Furthermore, the data processing unit includes a point cloud data processing module, which executes a point cloud processing algorithm for detecting deformation of transmission towers and conductors, and the point cloud processing algorithm includes the following steps:

[0209] According to the following formula 16, the collected three-dimensional point cloud data P is denoised:

[0210] P clean ={p i ∣p i ∈P,||p i -median(P)||<σ}(16)

[0211] Among them, P clean is the 3D point cloud data set after denoising; p i represents the i-th point in the 3D point cloud data P; median(P) is the median point of the point cloud data; σ is the preset noise threshold;

[0212] The denoised three-dimensional point cloud data set is plane-fitted to extract the reference planes of the transmission towers and conductors; wherein the plane-fitting uses the least squares method, and the fitting formula is as follows: Formula 17:

[0213]

[0214] Among them, a, b, c, d are the parameters of the fitting plane equation; (x i ,y i ,z i ) is point p i The coordinates of iis the adaptive weight, calculated according to the following formula 18:

[0215]

[0216] Among them, ρ i For point p i The local density of

[0217] According to the following formula 19, calculate point p i Distance to base plane:

[0218]

[0219] Where a, b, c, and d are the parameters of the reference plane; (x i ,y i ,z i ) is point p i The coordinates of

[0220] According to the following formula 20, the calculated distance is analyzed to identify and mark the deformation area:

[0221]

[0222] Among them, M i For point p i The deformation mark, 1 indicates abnormal deformation, 0 indicates normal; ε is the adaptive threshold of deformation detection;

[0223] According to the following formulas 21 and 22, calculate the deformation degree D deform and deformation area A deform :

[0224]

[0225] Among them, N deform is the number of points in the deformation area; A i For point p i Area represented;

[0226] The deformation degree D deform , deformation area A deform Combined with the geographic coordinate data, a deformation inspection report is generated, wherein the deformation inspection report includes the deformation location, degree and range.

[0227] The data processing unit includes a point cloud data processing module, which executes a point cloud processing algorithm for detecting deformation of transmission towers and conductors. The following is a detailed description of the algorithm.

[0228] First, the collected 3D point cloud data P is denoised using the following formula:

[0229] P clean ={p i ∣p i ∈P,||p i -median(P)||<σ} (16)

[0230] Among them, P clean is the 3D point cloud data set after denoising, p i represents the i-th point in the 3D point cloud data P, median(P) is the median point of the point cloud data, and σ is the preset noise threshold. Through median filtering denoising, isolated points and noise points in the point cloud data can be effectively removed, retaining the actual structural features.

[0231] Next, plane fitting is performed on the denoised 3D point cloud data set to extract the reference planes of the transmission towers and conductors. The plane fitting uses the least squares method, and the formula is as follows:

[0232]

[0233] Among them, a, b, c, d are the parameters of the fitting plane equation, (x i ,y i ,z i ) is point p i The coordinates of w i is the adaptive weight, calculated according to the following formula:

[0234]

[0235] Among them, ρ i For point p i The adaptive weight adjusts the weight value by the local density to enhance the robustness of the fitting plane and reduce the influence of noise points.

[0236] Then, calculate the point p i The distance to the reference plane is given by:

[0237]

[0238] Among them, a, b, c, and d are the parameters of the fitting plane, (x i ,y i ,z i ) is point p i The distance represents the coordinates of point p i The perpendicular distance to the fitted plane, used to assess the degree of positional deviation of the point.

[0239] Next, the calculated distance is analyzed to identify and mark the deformation area. The formula is as follows:

[0240]

[0241] Among them, M i For point p i The deformation mark of , 1 indicates abnormal deformation, 0 indicates normal, and ε is the adaptive threshold of deformation detection. By setting the threshold v, the normal area and the abnormal area can be effectively distinguished.

[0242] Then, calculate the deformation degree D deform and deformation area A deform The deformation degree calculation formula is as follows:

[0243]

[0244] Among them, N deform is the number of points in the deformation area, indicating the average degree of deviation of the deformation area.

[0245] The calculation formula of deformation area is as follows:

[0246]

[0247] Among them, A i For point p i The area represented by the deformation region is obtained by summing the areas of all points marked as abnormal deformation, reflecting the total area of ​​the deformation region.

[0248] Finally, the deformation degree D deform , deformation area A deform Combined with geographic coordinate data, a deformation inspection report is generated, which includes the location, degree and scope of the deformation.

[0249] Furthermore, click p i The local density ρ i The calculation formula steps include:

[0250] Select radius r as the neighborhood range;

[0251] Statistics at point p i The number of points N within the neighborhood radius r i :

[0252] The local density ρ is calculated according to the following formula 23: i :

[0253]

[0254] in, is the volume of the spherical neighborhood.

[0255] Point p i The local density ρ iThe calculation steps include: selecting radius r as the neighborhood range, counting the number of i The number of points N within the neighborhood radius r i , and then calculate the local density ρ according to the following formula i :

[0256]

[0257] in, is the volume of the spherical neighborhood.

[0258] First, choose a suitable radius r that defines the point p i The neighborhood range around. In three-dimensional space, this neighborhood range is a point p i The size of r should be selected based on the specific application scenario and the density of the point cloud data, ensuring that the number of points in the neighborhood is sufficient to represent the local density, but not too many to increase the computational complexity.

[0259] Next, statistics at point p i The number of points N within the neighborhood radius r i This process traverses all points in the point cloud data and calculates the relationship between each point and p i The distance is determined to determine whether the distance is less than r. j With p i The distance d ij Satisfy ij ≤r, then p j Belongs to point p i Neighborhood range, statistical value N i Increase by 1. The method of counting the number of points in the neighborhood can use a spatial index structure such as a KD tree or a ball tree to speed up the search process.

[0260] Then, use the above formula to calculate the point p i The local density ρ i The numerator N in the formula i Represents point p i The number of points in the neighborhood, denominator is the volume of the spherical neighborhood. By dividing the number of points in the neighborhood by N i Divide by the volume of the spherical neighborhood Get some p i The local density ρ i , that is, the number of points per unit volume. This density value reflects the point p i The density of the point cloud in the surrounding area. The larger the value, the denser the point distribution in the local area.

[0261] The wireless communication unit 104 is integrated in the UAV body and is used to transmit the data processed by the data processing module to the ground control station.

[0262] The wireless communication unit 104 is an important component of the power transmission network intelligent inspection system, which is integrated in the main body of the UAV and is used to transmit the data processed by the data processing unit 103 to the ground control station 105. The wireless communication unit adopts efficient and stable communication technology to ensure the integrity and real-time performance of the data during the transmission process.

[0263] The wireless communication unit contains multiple submodules, including a communication control module, a data transmission module, and a signal processing module. The communication control module is responsible for managing and coordinating the work of each submodule to ensure the smooth progress of the data transmission process. The data transmission module uses advanced communication protocols such as 4G / 5G and Wi-Fi to provide high-speed and stable data transmission channels. In order to meet the communication needs in different environments, the data transmission module has an adaptive function that can dynamically adjust the transmission parameters according to the signal strength and network conditions to ensure the best transmission performance.

[0264] The signal processing module processes the transmitted and received signals, including signal modulation, encoding and decoding. Modulation technology converts the processed data into signals suitable for wireless transmission, while encoding technology compresses and encrypts the data to improve transmission efficiency and ensure data security. The decoding technology at the receiving end can accurately restore the original data, ensuring that the data information received by the ground control station is complete and correct.

[0265] The wireless communication unit is also equipped with a multi-antenna system (MIMO), which significantly improves data transmission rate and anti-interference ability through the collaboration between multiple antennas. The multi-antenna system uses spatial multiplexing technology to transmit multiple data streams simultaneously, thereby improving transmission efficiency without increasing bandwidth. In addition, the wireless communication unit also has channel coding and error correction functions, which effectively reduce data errors during transmission and improve communication reliability.

[0266] In order to ensure the stability of data transmission during long-distance inspections, the wireless communication unit also integrates signal boosters and noise suppression technologies. Signal boosters are used to amplify weak signals to ensure that signals are still clear and receivable during long-distance transmission. Noise suppression technology improves the signal-to-noise ratio of data transmission by filtering out environmental noise and interference signals, further improving communication quality.

[0267] The wireless communication unit is connected to the data processing unit and the ground control station through a high-speed interface to ensure that the processed data can be quickly transmitted to the communication module. In actual applications, when the drone collects and processes multispectral images, infrared images, and three-dimensional point cloud data, the data processing unit will transmit the processed data to the wireless communication unit through a high-speed interface. The wireless communication unit then modulates and encodes the data and sends it to the ground control station through an appropriate communication protocol.

[0268] After the ground control station receives the data, the signal processing module will decode and demodulate the data, restore it to usable images and three-dimensional data, and store it in the storage unit of the control station. Subsequently, the analysis module of the ground control station will further analyze and diagnose the data and generate a detailed inspection report.

[0269] Through the above design, the wireless communication unit 104 can efficiently and stably complete the data transmission task between the UAV and the ground control station, ensure the real-time and accuracy of the inspection data, and provide reliable communication guarantee for the intelligent inspection system of the power transmission network.

[0270] The ground control station 105 is used to receive and store data from the UAV body and perform fault analysis and diagnosis.

[0271] The ground control station 105 is an important part of the intelligent inspection system for the power transmission network, and is designed to receive and store data from the drone body and perform fault analysis and diagnosis. The ground control station is equipped with a high-performance computing unit, a large-capacity storage unit, a data processing and analysis module, and a user interaction interface to ensure that it can efficiently and accurately process and analyze the data transmitted by the drone.

[0272] The high-performance computing unit is the core of the ground control station, which usually adopts a multi-core CPU or GPU architecture and can process large amounts of data in parallel. The computing unit is responsible for executing complex image processing algorithms, 3D point cloud data analysis, and fault detection of thermal imaging data. In order to improve processing efficiency, the computing unit is also equipped with dedicated hardware accelerators, such as image processing acceleration cards and AI accelerators, which can significantly speed up data processing and ensure the realization of real-time analysis.

[0273] The large-capacity storage unit is used to store data transmitted from the drone, including multispectral images, infrared images, and 3D point cloud data. The storage unit uses a high-speed solid-state drive (SSD) or hard disk array (RAID), which provides high data read and write speeds and reliable data storage capabilities. At the same time, the storage unit has data backup and recovery functions to ensure that data is not lost in the event of an unexpected situation, thereby ensuring system reliability and data security.

[0274] The data processing and analysis module is a key part of the ground control station, responsible for further processing and analysis of the received data. This module contains a variety of professional analysis tools and algorithm libraries, which can perform operations such as image stitching, 3D modeling, and thermal map analysis. Through these analysis tools, the system can identify various defects and faults in the transmission line, such as conductor wear, overheating, and insulation damage. The data processing and analysis module also supports machine learning and artificial intelligence technologies, which can automatically identify and classify faults through training models, improving the accuracy and efficiency of analysis.

[0275] The user interface is the human-computer interaction part of the ground control station. It is user-friendly and feature-rich. The interface provides data visualization tools that can display the data collected by the drone and the analysis results in real time. Users can view detailed information on multispectral images, thermal images, and 3D point cloud data through the interface, and can also browse the fault analysis reports generated by the system. The interface also supports user interaction with the system, such as setting inspection tasks, adjusting analysis parameters, and exporting reports, ensuring convenience and flexibility of operation.

[0276] The ground control station also has a remote access function, allowing users to log in to the system remotely through the network to view inspection data and analysis results. This function allows maintenance personnel to understand the operating status of the transmission line without having to visit the site in person, improving work efficiency and response speed.

[0277] Through the above design, the ground control station 105 can efficiently and accurately receive, store and analyze the data transmitted by the drone, generate detailed fault analysis and diagnosis reports, and ensure the safe operation and efficient maintenance of the power transmission network.

[0278] Furthermore, the ground control station includes a user interface module for displaying inspection results and fault analysis reports, and allowing operators to perform interactive control and data query.

[0279] The user interface module is an important part of the ground control station. It is designed to provide an intuitive and easy-to-use platform that allows operators to easily view and manage various types of data during the drone inspection process. The user interface module includes a display screen, input devices (such as keyboard and mouse), and a set of graphical user interface (GUI) software.

[0280] The display screen is used to display multispectral images, thermal images, and 3D point cloud data collected by the drone in real time. These images and data are presented in high resolution, ensuring that operators can clearly observe and analyze the status of the transmission lines. The display screen can also be split into multiple windows, allowing different types of data to be viewed simultaneously, for example, multispectral images on one side and thermal images on the other, further improving the operator's work efficiency.

[0281] The software part of the user interface module is based on modern graphical user interface technology, with a user-friendly design and rich functions. The main interface includes multiple function menus and toolbars, through which the operator can perform various operations, such as starting or stopping drone inspections, adjusting inspection parameters, switching display modes, etc. The main interface also integrates data visualization tools, which can display analysis results in the form of charts, heat maps, 3D models, etc., making complex data easy to understand and interpret.

[0282] Interactive control is a major feature of the user interface module. Operators can interact with the system through the keyboard, mouse or touch screen. For example, the operator can use the mouse to zoom in, zoom out and rotate the view on the 3D point cloud to examine the condition of a specific area in more detail. Keyboard shortcuts provide a way to quickly access commonly used functions, further improving the convenience of operation.

[0283] The data query function allows operators to quickly retrieve and view historical inspection data and fault analysis reports. The user interface module provides a powerful search engine and filtering tool, allowing operators to query based on time, geographic location, fault type and other conditions to quickly locate the required information. Query results can be exported to report format for easy archiving and sharing.

[0284] In addition, the user interface module also has a customization function, and operators can adjust the layout and display content of the interface according to personal preferences and work needs. For example, you can customize the size and position of the display window, select the image and data type to be displayed, set the alarm threshold and notification method, etc. The customization function ensures that each operator can work in the most suitable way for him / her, improving overall work efficiency and user experience.

[0285] To ensure data security and system stability, the user interface module also integrates user rights management and system log recording functions. The user rights management function ensures that only authorized personnel can access and operate sensitive data and key functions through hierarchical authority control. The system log recording function records all operator operations and system operating status in real time, which is convenient for post-audit and problem troubleshooting.

[0286] Through the above design, the user interface module of the ground control station can provide operators with a powerful and easy-to-use operating platform to achieve comprehensive management and analysis of UAV inspection data.

[0287] Furthermore, the ground control station has a data storage unit for long-term storage of historical inspection data and analysis results to facilitate trend analysis and predictive maintenance.

[0288] The data storage unit is an important part of the ground control station, designed to reliably and efficiently store the large amount of data collected by the drone during the inspection process. The unit includes multiple storage devices such as high-speed solid-state drives (SSDs), hard disk drives (HDDs), and network-attached storage (NAS) systems. The selection and configuration of storage devices are optimized to ensure high-capacity, high-speed, and high-reliability storage services in different usage scenarios.

[0289] In actual operation, when the drone completes an inspection mission and transmits data to the ground control station, the data storage unit will store the data according to the predetermined classification and label. The stored content includes multispectral images, infrared images, 3D point cloud data, and various results analyzed from these raw data, such as fault detection reports and equipment health assessments. In order to ensure the integrity and security of the data, the storage unit uses redundant storage technology, such as RAID (Redundant Array of Independent Disks), to achieve automatic data backup and recovery by distributing and storing data on multiple hard disks.

[0290] The data storage unit also supports advanced data management functions, including data compression, deduplication and encryption. Data compression technology can significantly reduce storage space usage and improve storage efficiency; deduplication technology can further optimize the use of storage space by identifying and deleting duplicate data; data encryption technology protects sensitive information by encrypting stored data and preventing unauthorized access and data leakage.

[0291] The long-term storage of historical inspection data and analysis results can not only be kept as records, but also provide a rich data foundation for subsequent trend analysis and predictive maintenance. The ground control station uses the built-in data analysis module to perform multi-dimensional analysis and mining using the data in the storage unit. For example, by comparing and analyzing multiple inspection data, the aging trend and potential failures of transmission lines and equipment can be identified, and possible problems in the future can be predicted, so as to formulate preventive maintenance plans and improve the reliability and service life of equipment.

[0292] The design of the data storage unit also takes into account scalability and maintainability. As the inspection tasks increase and the amount of data continues to grow, the system can easily add new storage devices to expand storage capacity. The modular design of the storage unit makes maintenance easier, and technicians can quickly replace or upgrade storage devices to ensure that the system is always in the best working condition.

[0293] Through the above detailed design, the data storage unit of the ground control station can not only store and manage a large amount of inspection data efficiently and reliably, but also support complex trend analysis and predictive maintenance functions.

[0294] Although the present application is disclosed as above in the form of a preferred embodiment, it is not intended to limit the present application. Any technical personnel in this field may make possible changes and modifications without departing from the spirit and scope of the present application. Therefore, the scope of protection of the present application shall be based on the scope defined by the claims of the present application.

Claims

1. A UAV-based intelligent inspection system for power transmission networks, characterized in that: include: The main body of the drone has a flight control module and a power unit for achieving autonomous flight and fixed-point hovering; The image acquisition unit includes a multispectral imaging module, an infrared imaging module and a laser radar module installed on the main body of the UAV; wherein the multispectral imaging module is used to obtain a multispectral image of the transmission line; the infrared imaging module is used to obtain a thermal imaging image of the transmission line; and the laser radar module is used to obtain three-dimensional point cloud data of the transmission line; a data processing unit, integrated in the drone body, for receiving and processing data from the image acquisition unit; A wireless communication unit, integrated in the drone body, for transmitting the data processed by the data processing module to a ground control station; A ground control station for receiving and storing data from the UAV body and performing fault analysis and diagnosis; The data processing unit includes a multispectral image processing module, which executes a first image processing algorithm for identifying wire wear and corrosion in a multispectral image, wherein the first image processing algorithm includes the following steps: According to the following formula 1, the multispectral image is subjected to noise removal: Where I(x, y, λ) represents the original multispectral image; I median (x, y, λ) is the denoised image; x and y are image coordinates; λ is the wavelength; median represents the median filter operation; According to the following formula 2-4, for the image I after noise removal using the gradient operator median (x, y, λ) extracts the edge features of the image and obtains the edge intensity image E(x, y, λ): Among them, G x (x,y,λ) and G y (x, y, λ) represents the gradient of the image in the x and y directions respectively; According to the following formula 5, the edge images of different wavelengths are weighted fused to generate the fused feature image F(x,y): Among them, ω λ is the weighting coefficient of different wavelengths; According to the following formula 6, the fused feature image F(x, y) is binarized to obtain the damaged area image D(x, y): Among them, τ is the set threshold; According to the following formulas 7 and 8, the damaged area is refined to obtain the damaged area A damage and injury severity S damage : Among them, {(x ′ ,y ′ )|D(x ′ ,y ′ )=1} represents the set of pixel coordinates belonging to the damaged area; The damaged area A damage , Severity S damage The damage inspection report includes the damage location, area and severity.

2. The intelligent inspection system for power transmission network according to claim 1, characterized in that: The flight control module includes a GPS positioning system and an inertial measurement unit to provide precise flight navigation and attitude control.

3. The power transmission network intelligent inspection system according to claim 1, characterized in that: The data processing unit includes a thermal imaging image processing module, which executes a second image processing algorithm for identifying temperature anomalies in thermal imaging images, wherein the second image processing algorithm includes the following steps: According to the following formula 9, the temperature of the thermal imaging image is calibrated to remove the influence of the ambient temperature: T cal (x,y)= T raw (x,y)- T ambient (9) Among them, T cal (x, y) is the calibrated temperature image, T raw (x, y) is the original thermal imaging temperature image, T ambient is the ambient temperature; The temperature gradient is calculated according to the following formula 10: Among them, G T (x, y) is the temperature gradient image; According to the following formula 11, the temperature gradient image is analyzed to identify the temperature abnormality area: Among them, θ is the threshold for temperature anomaly detection; A T (x, y) is the binary image of the temperature anomaly area; According to the following formulas 12 and 13, the temperature anomaly area is quantified to obtain the temperature anomaly area A anom and the temperature anomaly S damage : According to the following formula 14, the perimeter P of the temperature anomaly area is obtained: anom : Among them, Boundary (A T ) represents the boundary pixel set of the binary temperature anomaly area; Δx and Δy are the distances between boundary pixels, both of which are 1 pixel unit; According to the following formula 15, the shape compactness C of the temperature anomaly area is calculated: shape : The temperature abnormal area A anom , temperature abnormality S damage , shape compactness of the temperature anomaly area C shape Combined with the geographic coordinate data, a temperature inspection report is generated, wherein the temperature anomaly report includes the temperature anomaly location, area, temperature value and shape characteristics.

4. The intelligent inspection system for power transmission network according to claim 1, characterized in that: The data processing unit includes a point cloud data processing module, which executes a point cloud processing algorithm for detecting deformation of transmission towers and conductors. The point cloud processing algorithm includes the following steps: According to the following formula 16, the collected three-dimensional point cloud data P is denoised: P clean = {p i ∣p i ∈P,||p i - median(P)||< σ} (16) Among them, P clean is the 3D point cloud data set after denoising; p i represents the i-th point in the 3D point cloud data P; median(P) is the median point of the point cloud data; σ is the preset noise threshold; The denoised three-dimensional point cloud data set is plane-fitted to extract the reference planes of the transmission towers and conductors; wherein the plane-fitting uses the least squares method, and the fitting formula is as follows: Formula 17: Among them, a, b, c, d are the parameters of the fitting plane equation; (x i ,y i ,z i ) is point p i The coordinates of i is the adaptive weight, calculated according to the following formula 18: Among them, ρ i For point p i The local density of According to the following formula 19, calculate point p i Distance to the fitted plane: Where a, b, c, and d are the parameters of the fitting plane; (x i ,y i ,z i ) is point p i The coordinates of According to the following formula 20, the calculated distance is analyzed to identify and mark the deformation area: Among them, M i For point p i The deformation mark, 1 indicates abnormal deformation, 0 indicates normal; ε is the adaptive threshold of deformation detection; According to the following formulas 21 and 22, calculate the deformation degree D deform and deformation area A deform : Among them, N deform is the number of points in the deformation area; A i For point p i Area represented; The deformation degree D deform , deformation area A deform Combined with the geographic coordinate data, a deformation inspection report is generated, wherein the deformation inspection report includes the deformation location, degree and range.

5. The intelligent inspection system for power transmission network according to claim 4, characterized in that: Point p i The local density ρ i The calculation formula steps include: Select radius r as the neighborhood range; Statistics at point p i The number of points N within the neighborhood radius r i : The local density ρ is calculated according to the following formula 23: i : in, is the volume of the spherical neighborhood.

6. The intelligent inspection system for power transmission network according to claim 1, characterized in that: The multi-spectral imaging module includes a plurality of filters for capturing spectral images within different wavelength ranges.

7. The intelligent inspection system for power transmission network according to claim 1, characterized in that: The laser radar module has a rotating scanning device for acquiring 360-degree three-dimensional point cloud data.

8. The intelligent inspection system for power transmission network according to claim 1, characterized in that: The ground control station includes a user interface module for displaying inspection results and fault analysis reports, and allowing operators to perform interactive control and data query.

9. The intelligent inspection system for power transmission network according to claim 1, characterized in that: The ground control station has a data storage unit for long-term storage of historical inspection data and analysis results to facilitate trend analysis and predictive maintenance.

Images