A power line inspection control method and system based on a UAV

Abstract

The application discloses a kind of power line inspection control method and system based on unmanned aerial vehicle, method includes: obtaining inspection task, control unmanned aerial vehicle gathers the tower position coordinate information in the inspection area corresponding to all inspection tasks;According to the tower position coordinate information, generate a plurality of corresponding inspection tasks with the inspection task of oblique flight route;Control the unmanned aerial vehicle flies according to multiple oblique flight route, and in the process of flight, corresponding point cloud data of oblique flight route is collected, and flight route is generated according to the point cloud data;Height test is carried out to all flight routes, and the optimal flight route corresponding to the inspection task is obtained;Obtain target inspection task, and match the optimal flight route corresponding to target inspection task, control the unmanned aerial vehicle carries out inspection according to optimal flight route;The method can effectively improve the unmanned aerial vehicle inspection efficiency.

Description

Technical Field

[0001] This invention relates to the field of unmanned aerial vehicle (UAV) technology, and in particular to a UAV-based power line inspection and control method and system. Background Technology

[0002] Power transmission lines are often located in mountainous and uninhabited areas, making manual inspections inefficient. Moreover, incidents of employees being bitten by snakes, insects, ants, and other animals are common. In addition, transmission towers, conductors, insulators, and other equipment are located at high altitudes. Using drones for inspection can avoid the safety risks of climbing towers at heights and provide a 360° view of the equipment details, thus improving the quality of the inspection.

[0003] Currently, planning the inspection path is a challenge in power line inspection using drones. Inadequate path planning can lead to collisions with transmission lines, unstable signal transmission, and incomplete image acquisition. Patent document CN109062233A discloses an autonomous drone inspection method for power transmission lines, comprising the following steps: S1. Calculating the area of ​​the laser radar scanning region based on the laser radar model, range, opening angle, and communication distance; S2. Using Qianxun's precise positioning service to locate the RTK laser radar drone base station; S3. Combining the radar scanning area and the transmission line structural dimensions, including tower height, maximum crossarm length, lowermost phase conductor height, and direction, calculating the horizontal distance between the transmission line's center and the drone, the theoretical vertical distance between the drone and the horizontal ground, and the designed height of the drone relative to the horizontal ground; S4. Manually controlling the RTK laser radar drone to inspect the power transmission line. Road modeling involves data conversion and fitting, LiDAR data classification, denoising, and vectorization to form a 3D solid model. It automatically extracts tower coordinates, tower height, nominal height, conductor sag, transmission line structural component dimensions, and distances between the line and trees, houses, and other lines. It also extracts the spatial coordinates of insulator attachment points and hardware attachment points, and plans the drone's autonomous flight path. S5. Based on the LiDAR-generated transmission line laser point cloud solid model, it extracts the coordinates of key transmission line locations, including tower center coordinates, ground wire attachment points, and insulator attachment point coordinates. It plans the drone's take-off and landing positions, flight speed, and flight trajectory, and generates refined drone inspection and LiDAR-modeled autonomous flight paths through autonomous and intelligent flight control programs.

[0004] The above method relies on lidar data to model the entire power transmission line environment, and the reliability of planning the inspection path of the UAV is not high simply by using a 3D model. Summary of the Invention

[0005] This invention provides a method and system for power line inspection control based on unmanned aerial vehicles (UAVs), which can effectively plan the power line inspection path of UAVs and ensure the reliability of UAV inspection.

[0006] In a first aspect, the present invention provides a power line inspection and control method based on unmanned aerial vehicles (UAVs), comprising:

[0007] Obtain inspection tasks and control the drone to collect the pole and tower location coordinates within the inspection area corresponding to all inspection tasks;

[0008] Multiple inclined flight paths corresponding to the inspection task are generated based on the tower position coordinate information;

[0009] The drone is controlled to fly along multiple tilted flight paths, and point cloud data corresponding to the tilted flight paths is collected during the flight. The flight path is generated based on the point cloud data.

[0010] Altitude tests were conducted on all flight paths to obtain the optimal flight path corresponding to the inspection mission;

[0011] The system acquires the target inspection task, matches the optimal flight path corresponding to the target inspection task, and controls the UAV to perform inspections according to the optimal flight path.

[0012] Furthermore, multiple inclined flight paths are generated based on the tower position coordinate information, including:

[0013] The external dimensions of the tower corresponding to the inspection task are obtained based on the tower location coordinates.

[0014] The flight direction, flight range, and flight altitude of the UAV are determined based on the external dimensions of the tower.

[0015] Multiple waypoints are determined based on the drone's flight direction, flight range, and flight altitude;

[0016] Based on multiple waypoints, an inclined flight path is generated to correspond to the inspection task.

[0017] Further, generating a flight path based on the point cloud data includes:

[0018] Multiple point cloud route models were obtained based on the tower point cloud data from the point cloud data.

[0019] Based on the point cloud data of the transmission line, multiple point cloud line models are connected in series;

[0020] Based on the inspection task, the inspection point locations and inspection feature actions are configured on the connected point cloud line model to generate the flight path.

[0021] Furthermore, altitude tests are performed on all the aforementioned flight paths to obtain the optimal flight path corresponding to the inspection mission, including:

[0022] Multiple different preset flight altitudes are set for each flight path, and the UAV is controlled to fly along the flight path at different preset flight altitudes;

[0023] During flight, at different preset flight altitudes, test signals and test images sent back by the drone are received;

[0024] The test signal is subjected to signal strength analysis and signal quality analysis, and the test image is subjected to sharpness analysis. Based on the analysis results, the optimal altitude for each flight path is determined.

[0025] The optimal altitude is matched with the flight path to obtain the optimal flight path.

[0026] Furthermore, signal strength and quality analysis are performed on the test signal, and sharpness analysis is performed on the test image. Based on the analysis results, the optimal altitude for each flight path is determined, including:

[0027] The test signal is analyzed for signal strength and signal quality to obtain a first height region, and the test image is analyzed for sharpness to obtain a second height region;

[0028] Based on the first evaluation condition and the second evaluation condition, the first height region and the second height region are screened respectively to obtain the first optimal height and the second optimal height;

[0029] Obtain the weights for analyzing the signal strength and signal quality of the test signal, and for analyzing the sharpness of the test image;

[0030] Based on the obtained weights, as well as the first and second optimal heights, the optimal height is obtained.

[0031] Further, signal strength and signal quality analysis are performed on the test signal to obtain a first height region, including:

[0032] Statistical analysis was performed on the signal strength of the test signals at different preset flight altitudes to obtain the average, maximum and minimum signal strength values.

[0033] Based on the average, maximum, and minimum values ​​of signal strength, the trend of signal strength variation at different preset flight altitudes and the strength difference at different times at the same preset flight altitude are obtained.

[0034] Based on the trend and differences in signal strength, the coverage area and intensity distribution of the test signal are obtained;

[0035] Perform statistical analysis on the signal quality of the test signal to obtain its mean, variance, and standard deviation;

[0036] Plot the quality variation based on the mean, variance, and standard deviation of the signal quality.

[0037] Based on the quality change map, the fluctuation range of signal quality at different preset flight altitudes and the rate of change within a predetermined time period are obtained;

[0038] The first height region is obtained based on the coverage area, intensity distribution, fluctuation range, rate of change, and first preset parameter threshold of the test signal.

[0039] Furthermore, a sharpness analysis is performed on the test image to obtain a second height region, including:

[0040] Based on image processing algorithms, relevant features of the target are extracted from test images at different preset flight altitudes.

[0041] Based on the extracted relevant features, the target in the test image is identified and located using a target detection algorithm;

[0042] Obtain predetermined evaluation indicators, and obtain the evaluation parameters corresponding to the predetermined evaluation indicators based on the identified and located test images;

[0043] The evaluation parameters of the test images at different preset flight altitudes are filtered based on the second preset parameter threshold to obtain the test images that meet the conditions.

[0044] The second altitude region is obtained based on the preset flight altitude corresponding to the test image that meets the conditions.

[0045] Furthermore, before extracting relevant features of the target, the test image can be preprocessed, including denoising, enhancement, and geometric correction, which helps to improve image quality and accuracy.

[0046] Furthermore, the target inspection task is obtained, and the optimal flight path corresponding to the target inspection task is matched, including:

[0047] Feature extraction is performed on the inspection task to obtain the corresponding inspection points and inspection feature actions;

[0048] The locations of the inspection points and the characteristic actions of the inspection are matched with the inspection tasks to obtain the matching inspection tasks.

[0049] The optimal flight path is obtained based on the matched inspection task.

[0050] Furthermore, after controlling the drone to perform inspections along the optimal flight path, the process also includes:

[0051] Control the drone to collect inspection images;

[0052] The inspection image is divided into grids to obtain grid cell images;

[0053] The grid cell image is denoised using a median filtering algorithm to obtain a denoised grid cell image.

[0054] The denoised grid cell image is enhanced using a generative adversarial network to obtain an enhanced image.

[0055] Furthermore, the process of controlling the drone to perform inspections along the optimal flight path also includes:

[0056] Get the drone's remaining battery power;

[0057] If the remaining battery level of the drone is lower than the warning value, a return-to-home battery swap command is generated and sent to the corresponding drone to control the drone to return to home and swap batteries.

[0058] Furthermore, after the drone completes its return-to-base battery swap and inspection, it also includes:

[0059] Scene images of the landing area were collected by drones;

[0060] Obtain the location of the landing area from the scene image;

[0061] Obtain the drone's current position, current attitude, and current flight speed;

[0062] Based on a pre-built prediction model, the flight parameters that the UAV needs to adjust are obtained according to the location of the UAV landing area, combined with the UAV's current position, current attitude information and current flight speed; among which, the flight parameters include target altitude, target flight speed and target attitude parameters.

[0063] The drone is controlled to adjust its flight status in real time according to flight parameters until it completes landing.

[0064] Furthermore, the location of the landing area is obtained based on the scene image, including:

[0065] The scene image is sequentially processed, target detection is performed, and feature extraction is performed to obtain the relevant features of the landing area;

[0066] The relevant features of the landing area are reduced in dimension, normalized, and key features are extracted to obtain key information about the landing area in the scene image;

[0067] The location of the landing area in the image coordinate system is obtained based on key information of the landing area;

[0068] Based on the location of the landing area in the image coordinate system and the camera parameters carried by the UAV, the location of the landing area in the UAV coordinate system is obtained.

[0069] The pre-built prediction models include:

[0070] An adaptive Kalman filter prediction model is constructed based on the dynamic characteristics and motion parameters of the UAV; the motion parameters include the UAV's position, flight speed, and flight attitude.

[0071] Acquire training data for the drone, including the drone's actual training flight speed and attitude at different locations, as well as the actual training target location.

[0072] Based on the adaptive Kalman filter prediction model and combined with the UAV training data, the prediction data of the UAV when moving from different positions to the next position is obtained. The prediction data includes the predicted altitude, predicted flight speed and predicted attitude information.

[0073] Based on the predicted data and the training data, the observation residuals of the predicted data and the training data are obtained;

[0074] The covariance matrix is ​​obtained from the observed residuals;

[0075] The weights of the predicted and actual values ​​are obtained using Kalman gain.

[0076] The adaptive Kalman filter prediction model is updated based on the observation residuals, covariance matrix, and weights.

[0077] Repeat the above prediction and update steps until the predetermined update threshold is reached to obtain the final adaptive Kalman filter prediction model.

[0078] By repeating the prediction and update steps, the uncertainty of the prediction can be reduced, and the prediction accuracy of the UAV's position, attitude and flight speed can be improved. The predetermined update threshold is to ensure that the adaptive Kalman filter prediction model meets the preset uncertainty requirements.

[0079] Furthermore, controlling the corresponding drone to return to base for battery swapping also includes:

[0080] The drone's return destination is obtained based on the charging station coordinates of the capsule-shaped drone nest;

[0081] The return flight path of the drone is generated based on the return destination;

[0082] The drone was controlled to return to its origin based on the return flight path.

[0083] When the drone is docked at the charging station, the battery is replaced by the robotic arm built into the capsule.

[0084] Secondly, the present invention provides a power line inspection and control device based on a drone, comprising:

[0085] The tower coordinate acquisition module is used to acquire inspection tasks and control the drone to collect the tower position coordinate information within the inspection area corresponding to all inspection tasks.

[0086] The tilt flight path generation module is used to generate multiple tilt flight paths corresponding to the inspection task based on the tower position coordinate information.

[0087] The flight path generation module is used to control the UAV to fly along multiple tilted flight paths, and to collect point cloud data corresponding to the tilted flight paths during the flight, and to generate flight paths based on the point cloud data.

[0088] The altitude testing module is used to perform altitude tests on all the flight paths to obtain the optimal flight path corresponding to the inspection task.

[0089] The task matching module is used to obtain the target inspection task, match the optimal flight path corresponding to the target inspection task, and control the UAV to perform inspection according to the optimal flight path.

[0090] Thirdly, the present invention also provides a power line inspection and control system based on unmanned aerial vehicles (UAVs), including a control platform, a UAV, and a capsule-shaped storage device. The capsule-shaped storage device is used to store the UAV and charge the UAV. The control platform includes a processor and a storage device. The storage device stores multiple instructions, and the processor is used to read the instructions and execute the above-described method.

[0091] The power line inspection and control method and system based on unmanned aerial vehicles (UAVs) provided by this invention have at least the following beneficial effects:

[0092] (1) By establishing a line model of laser point cloud, the flight path is planned, and the optimal flight path is obtained based on altitude test, so as to ensure the safety of UAV during inspection, the stability of signal transmission and the reliability of image acquisition, and improve the inspection efficiency of UAV.

[0093] (2) During the process of obtaining the optimal flight route based on altitude testing, the optimal flight route obtained through signal quality, signal strength and image clarity makes the control platform have a lower delay in the control command of the UAV and obtain smoother high-definition video, which further improves the efficiency of UAV inspection.

[0094] (3) By replacing the battery of the drone through the capsule machine nest, the drone can fly continuously, which further improves the working efficiency of the drone. Attached Figure Description

[0095] Figure 1 This is a flowchart of one embodiment of the UAV-based power line inspection and control method provided by the present invention.

[0096] Figure 2 This is a flowchart of an embodiment of the UAV-based power line inspection and control method provided by the present invention.

[0097] Figure 3 This is a schematic diagram of one embodiment of the power line inspection and control device based on unmanned aerial vehicles (UAVs) provided by the present invention.

[0098] Figure 4 This is a schematic diagram of one embodiment of the UAV-based power line inspection and control system provided by the present invention. Detailed Implementation

[0099] To better understand the above technical solutions, the following will provide a detailed explanation of the technical solutions in conjunction with the accompanying drawings and specific implementation methods.

[0100] refer to Figure 1 In some embodiments, a power line inspection and control method based on unmanned aerial vehicles (UAVs) is provided, including:

[0101] S1. Obtain inspection tasks and control the drone to collect the pole and tower location coordinate information within the inspection area corresponding to all inspection tasks;

[0102] S2. Generate multiple inclined flight paths corresponding to the inspection task based on the tower position coordinate information;

[0103] S3. Control the drone to fly along multiple inclined flight paths, and collect point cloud data corresponding to the inclined flight paths during the flight, and generate flight paths based on the point cloud data;

[0104] S4. Perform altitude tests on all flight paths to obtain the optimal flight path corresponding to the inspection mission.

[0105] S5. Obtain the target inspection task, match the optimal flight path corresponding to the target inspection task, and control the UAV to carry out the inspection according to the optimal flight path.

[0106] Specifically, the execution entity of the above method is the control platform, which also includes a capsule-shaped drone pod connected to the control platform. The control platform is equipped with a communication module that can transmit signals to the drone's communication module, sending operation commands to the drone to control its specific actions, such as pausing, hovering, changing routes, and temporary return-to-home. The drone is equipped with an image acquisition device, which can be a high-definition zoom camera, infrared camera, night vision camera, lidar, or other sensor devices. The drone can transmit the acquired real-time image information to the control platform. The capsule-shaped drone pod refers to a mobile drone base that can store drones, replace drone batteries, and perform drone repairs. Examples include drone capsule-shaped drone pod pickups and drone chassis. The control platform can also implement the functions of the capsule-shaped drone pod by sending control commands to it.

[0107] Specifically, in step S1, the location coordinates and height information of a certain power transmission tower in the power transmission channel within the inspection area are obtained by using a drone to mark points above the tower.

[0108] Furthermore, in step S2, multiple inclined flight paths corresponding to the inspection task are generated based on the tower position coordinate information, including:

[0109] Based on the tower's location coordinates and height information, the tower's external dimensions corresponding to the inspection task are obtained;

[0110] The flight direction, flight range, and flight altitude of the UAV are determined based on the external dimensions of the tower.

[0111] Multiple waypoints are determined based on the drone's flight direction, flight range, and flight altitude;

[0112] Based on multiple waypoints, an inclined flight path is generated to correspond to the inspection task.

[0113] Furthermore, in step S3, the drone is controlled to fly along an inclined flight path, and the point cloud data of the towers along the inclined flight path is collected by the 3D laser scanner on the drone. This method obtains the point cloud data of all towers in the inspection area, and further obtains the point cloud data of the transmission line between two towers. Since the towers have a certain angle from bottom to top, the drone's flight path is not vertical. By using the drone's oblique photography, the real scene can be quickly reconstructed, which can then be used to quickly perform 3D real scene modeling. That is, the inclined flight path is the path of the drone's oblique photography.

[0114] The flight path is generated based on point cloud data, including:

[0115] Multiple point cloud line models are constructed using collected tower point cloud data. These models are then linked together in power distribution 3D flight path planning software using point cloud data of the transmission lines. Multiple patrol point locations and inspection actions are set according to the transmission channel inspection task to generate a flight path for transmission line inspection. This flight path provides a basis for controlling drones to autonomously inspect transmission lines. Each tower corresponds to one point cloud line model. Flight paths are generated based on the point cloud data of the transmission lines between two towers, and the point cloud line models are then linked together.

[0116] Further, refer to Figure 2 In step S4, altitude tests are performed on all flight paths to obtain the optimal flight path corresponding to the inspection task, including:

[0117] S41. Set multiple different preset flight altitudes for each flight path, and control the UAV to fly along the flight path at different preset flight altitudes;

[0118] S42. During flight, at different preset flight altitudes, receive test signals and test images sent back by the UAV;

[0119] S43. Perform signal strength and signal quality analysis on the test signal, perform sharpness analysis on the test image, and determine the optimal altitude for each flight path based on the analysis results;

[0120] S44. Match the optimal altitude with the flight path to obtain the optimal flight path.

[0121] In step S43, signal strength and signal quality analysis are performed on the test signal, and sharpness analysis is performed on the test image. Based on the analysis results, the optimal height is determined, including:

[0122] A flight altitude analysis model is established, which includes an input layer, a weight calculation layer, a weighted calculation layer, and an output layer.

[0123] The test signals are analyzed for signal strength and quality, and the test images are analyzed for sharpness. Based on the analysis results, the optimal altitude for each flight path is determined, including:

[0124] The test signal is analyzed for signal strength and signal quality to obtain the first height region, and the test image is analyzed for sharpness to obtain the second height region.

[0125] Based on the first evaluation condition and the second evaluation condition, the first height region and the second height region are screened respectively to obtain the first optimal height and the second optimal height;

[0126] Obtain the weights for analyzing the signal strength and signal quality of the test signal, and for analyzing the sharpness of the test image;

[0127] Based on the obtained weights, as well as the first and second optimal heights, the optimal height is obtained.

[0128] Among these steps, signal strength and signal quality analysis of the test signal is performed to obtain the first altitude region, which includes:

[0129] Statistical analysis is performed on the signal strength of the test signal to obtain its average, maximum, and minimum values;

[0130] The average, maximum, and minimum signal strength values ​​are calculated based on different preset flight altitudes.

[0131] Based on the average, maximum, and minimum values ​​of signal strength, the trend of signal strength variation at different preset flight altitudes and the strength difference at different times at the same preset flight altitude are obtained.

[0132] Based on the trend and differences in signal strength, the coverage area and intensity distribution of the test signal are obtained;

[0133] Perform statistical analysis on the signal quality of the test signal to obtain its mean, variance, and standard deviation;

[0134] Plot the quality variation based on the mean, variance, and standard deviation of the signal quality.

[0135] Based on the quality change map, the fluctuation range of signal quality at different preset flight altitudes and the rate of change within a predetermined time period are obtained;

[0136] Based on the coverage area, intensity distribution, fluctuation range, rate of change, and first preset parameter threshold of the test signal, the first height region is obtained;

[0137] A sharpness analysis of the test image revealed the second height region, which includes:

[0138] Based on image processing algorithms, relevant features of the target are extracted from test images at different preset flight altitudes.

[0139] Based on the extracted relevant features, the target in the test image is identified and located using a target detection algorithm;

[0140] Obtain predetermined evaluation indicators, and obtain the evaluation parameters corresponding to the predetermined evaluation indicators based on the identified and located test images;

[0141] The evaluation parameters of the test images at different preset flight altitudes are filtered based on the second preset parameter threshold to obtain the test images that meet the conditions.

[0142] The second altitude region is obtained based on the preset flight altitude corresponding to the test image that meets the conditions.

[0143] The average signal strength (signal quality) is the sum of signal strength (signal quality) at each preset flight altitude / the number of signal strengths (signal quality) at the corresponding preset flight altitude; the variance of signal quality is the sum of the squares of the average signal quality at all preset flight altitudes / the number of signal quality; and the standard deviation of signal quality is the square root of the variance of signal quality.

[0144] The signal strength trend indicates the direction of increase in signal strength; for example, signal strength increases with increasing flight altitude, or signal strength decreases with increasing flight altitude. Intensity difference represents the difference in signal strength between consecutive moments; a larger difference indicates a greater intensity disparity. By comparing intensity differences at different preset flight altitudes, it can be determined whether the signal strength distribution is uniform.

[0145] The coverage area of ​​the test signal represents the region where a test signal of predetermined strength exists, while the intensity distribution represents the signal strength distribution within the coverage area. The coverage area and intensity distribution of the test signal are crucial for UAV flight path planning, communication equipment deployment, and signal optimization.

[0146] The quality variation graph includes line graphs and box plots. The box plot includes information such as the statistical median and upper and lower quartiles. The rate of change represents the change in signal quality at a preset flight altitude. The smaller the rate of change, the smaller the fluctuation and the more stable the change.

[0147] The first preset parameter thresholds include a preset coverage area threshold (preset height range), a preset intensity threshold, a preset fluctuation threshold, and a preset rate of change threshold.

[0148] Image processing algorithms include Histogram of Oriented Gradients (HOG) and Convolutional Neural Networks (CNN). Relevant features of a target can include color, texture, shape, and edges. Target detection algorithms can employ deep learning methods, including convolutional neural networks.

[0149] The predetermined evaluation metrics include accuracy, recall, precision, and F1 score. The accuracy of the evaluation parameters can be improved by comparing them with real labels or human annotations.

[0150] Image brightness:

[0151] Gray value (single-channel image): B = (R + G + B) / 3;

[0152] Brightness value (single-channel image): Y = 0.299*R + 0.587*G + 0.114*B;

[0153] Image contrast:

[0154] Contrast value (single-channel image): C = (Imax - Imin) / (Imax + Imin);

[0155] Image sharpness:

[0156] Variance (single-channel image): Var = (1 / N)*Σ[(I(x,y)-mean)^2], where N represents the number of pixels, I(x,y) represents the gray value of the pixel, and mean represents the mean gray value.

[0157] Image texture:

[0158] Co-occurrence matrix (gray-level co-occurrence matrix): Calculates the relative position and frequency relationship between gray levels in a test image.

[0159] Gabor filters: Gabor filters applied to test images at multiple scales and orientations are used to extract texture features.

[0160] Predicted evaluation indicators:

[0161] Accuracy: P = TP / (TP + FP)

[0162] Recall: R = TP / (TP + FN)

[0163] F1 score: F1 = 2*(P*R) / (P+R), where TP represents true positives (the number of positive samples correctly detected), FP represents false positives (the number of negative samples incorrectly predicted as positive samples), and FN represents false negatives (the number of positive samples incorrectly predicted as negative samples).

[0164] The second preset parameter threshold is a preset evaluation parameter value. After filtering out the evaluation parameters that meet the preset evaluation parameter value, the corresponding test image is obtained through the filtered evaluation parameters. The obtained test image is the test image that meets the conditions.

[0165] The first and second evaluation conditions can be determined by analyzing the coverage area, intensity distribution, fluctuation range, rate of change, and evaluation parameters to establish the first and second optimal altitudes within the first and second altitude regions. Alternatively, the coverage area, intensity distribution, fluctuation range, rate of change, and evaluation parameters can be categorized into levels, and the first and second optimal altitudes can be determined based on the categorization results. Furthermore, weights can be assigned to the coverage area, intensity distribution, fluctuation range, and rate of change, as well as to each parameter in the evaluation parameters. The optimal preset flight altitudes can then be obtained by combining these weights with the actual values ​​at different preset flight altitudes; these optimal preset flight altitudes are the first and second optimal altitudes.

[0166] The weight 1 for analyzing the signal strength and signal quality of the test signal, and the weight 2 for analyzing the sharpness of the test image, are summed to 1. Optimal height = First optimal height * Weight 1 + Second optimal height * Weight 2.

[0167] (1) Signal strength analysis:

[0168] Collect test signal strength data: The drone is equipped with a corresponding receiving device to acquire the strength data of the test signal. This can be achieved through the API or interface of the receiving device.

[0169] Processing the test signal strength data: Based on the received test signal strength data, statistical analysis such as average value, maximum value, and minimum value can be performed to obtain the overall signal strength situation.

[0170] Analyze the changes in test signal intensity: By comparing the test signal intensity data at different locations or time points, analyze the changing trend and intensity differences of the test signal to determine the area covered by the test signal and the intensity distribution.

[0171] (2) Signal quality analysis:

[0172] Acquire test signal quality data: Use test equipment or sensors to obtain quality-related parameters of the test signal, such as signal-to-noise ratio (SNR) and bit error rate (BER).

[0173] Analyze test signal quality data: Based on the collected test signal quality data, perform statistical analysis and comparison to evaluate the reliability and stability of the test signal.

[0174] Determining the optimal altitude based on test signal quality assessment: Based on the analysis results of test signal quality, the altitude area with better test signal strength and quality is determined, thereby determining the first altitude area for the UAV during the inspection process.

[0175] (3) Sharpness analysis:

[0176] Acquire test image data: Use the camera or sensor mounted on the drone to acquire test image data.

[0177] Image sharpness assessment: Applying image processing algorithms or metrics, such as sharpness, contrast, and blur, to analyze and evaluate the sharpness of the test image.

[0178] Determining the optimal altitude based on image sharpness assessment: Based on the analysis results of image sharpness, the altitude area with higher image quality is determined, thereby determining the second altitude area for the drone during the inspection process.

[0179] Specifically, multiple flight altitudes are set, such as 30 meters, 50 meters, 70 meters, and 90 meters. The drone is controlled to conduct autonomous inspection tests at different altitudes according to the flight path, obtaining the signal strength and quality of the drone at different altitudes. Further, the image acquisition device on the drone acquires test images at different flight altitudes. Clarity analysis is performed on the test images to obtain the clarity analysis results, which are represented by an image clarity index. The clearer the image and the higher the resolution, the higher the image clarity index. A flight altitude analysis model is constructed to obtain the optimal flight altitude. This model consists of an input layer, a weight calculation layer, a weighted calculation layer, and an output layer. Weights are allocated to the signal strength, signal quality, and image clarity analysis results. The output result, the optimal flight altitude, is obtained through weighted calculation. Then, the optimal flight path, i.e., the flight path at the optimal flight altitude, is obtained using the obtained optimal flight altitude and flight path. Obtaining this optimal flight path results in lower latency in the control platform's control commands to the drone and smoother high-definition video, further improving the efficiency of drone inspections.

[0180] In general and specifically, inspection tasks refer to specific inspection missions performed by drones, such as inspections of power transmission towers, power transmission channels, substations, power distribution systems, and emergency inspections. The method for generating flight paths involves extracting features for each inspection task to obtain multiple waypoints and inspection characteristic actions unique to that task. Multiple flight paths are obtained using power distribution 3D flight path planning software, and altitude tests are performed on each path to obtain multiple optimal flight paths. Each inspection task corresponds to one optimal flight path. By generating multiple optimal flight paths, the system can control the drone to complete different inspection tasks, improving the drone's work efficiency.

[0181] Further, in step S5, the target inspection task is obtained, and the optimal flight path corresponding to the target inspection task is matched, including:

[0182] Feature extraction is performed on the target inspection task to obtain the corresponding inspection point locations and inspection characteristic actions;

[0183] The locations of inspection points and characteristic actions of inspections are matched with inspection tasks to obtain matching inspection tasks.

[0184] The optimal flight path is obtained based on the matched inspection task.

[0185] Furthermore, in step S5, the optimal flight path is determined, and the drone is controlled to perform inspections according to the optimal flight path. The control platform generates flight path instructions and sends the flight path instructions to the drone. After receiving the instructions, the drone is controlled to take off automatically to perform the task and complete the autonomous inspection. By remotely controlling the drone to perform autonomous inspections through the control platform, the use of human resources can be reduced and the working efficiency of the drone can be improved.

[0186] When the UAV of this invention performs inspections according to the optimal flight path, it uses a PID controller for closed-loop control to adjust the UAV's position, speed, and acceleration, including:

[0187] Determine the target position and target speed of the UAV during flight based on the inspection task and the optimal classification route.

[0188] Real-time acquisition of the drone's position, speed, and acceleration during flight;

[0189] The target position is compared with the real-time position, and the target velocity is compared with the real-time velocity to obtain the error signal;

[0190] The position, velocity, and acceleration of the drone are adjusted by combining the error signal and the preset acceleration range through a PID controller;

[0191] Repeat the above steps until the drone reaches the target location.

[0192] Specifically, the closed-loop control process includes:

[0193] (1) Determine the target position and speed limit: In the UAV control system, the required target position and speed limit are determined according to the inspection task and the optimal flight path, that is, the position and speed that the UAV is expected to reach.

[0194] (2) Real-time acquisition of position, speed and acceleration information: Real-time acquisition of current position, speed and acceleration information through sensors built into the drone, such as GPS, gyroscope and accelerometer.

[0195] (3) Calculate the error signal: Compare the determined target position and velocity with the actual acquired position and velocity, and calculate the error signal, that is, the difference between the expected value and the actual value.

[0196] (4) Applying a PID controller: Input the error signal into the PID (proportional-integral-derivative) controller, and calculate the control signal according to the predetermined parameters and algorithm.

[0197] Proportional (P) controller: Based on the magnitude of the error signal, it generates an output proportional to the error, used to quickly respond to changes in the error.

[0198] Integral (I) controller: Based on the accumulated amount of the error signal, it generates an integral term output to eliminate persistent errors and stabilize the system.

[0199] Differential (D) controller: Based on the rate of change of the error signal, it generates a differential term output to suppress system oscillations and improve response speed.

[0200] (5) Adjust according to control signal: Apply the control signal output by the PID controller to the actuators of the UAV, such as motors or servos, to adjust the position, speed and acceleration of the UAV.

[0201] (6) Continuous iteration: Through the feedback mechanism, the position and speed of the UAV are continuously monitored and adjusted so that it gradually approaches the set target position and speed limit, thereby achieving closed-loop control.

[0202] During closed-loop control, the drone's acceleration can be adjusted via control signals according to actual needs to achieve precise acceleration control. Specific methods may include:

[0203] Adjusting parameters in a PID controller: In a PID controller, the control signal can be made more sensitive or stable by adjusting the proportional (P), integral (I), and derivative (D) parameters, thereby affecting changes in acceleration.

[0204] Setting acceleration limits: In a closed-loop control system, an upper limit or range of acceleration can be set to limit the rate of change of the drone's acceleration, thereby ensuring safety and stability.

[0205] Feedback control based on real-time acceleration information: By monitoring the acceleration of the UAV in real time, feedback control can be performed according to the preset acceleration range, and the control signal can be adjusted in a timely manner to achieve precise acceleration control.

[0206] Before the UAV performs inspections along the optimal flight path and implements closed-loop control, the accuracy of target position estimation can be improved by repeatedly updating the prediction and estimation method of the target position. This prediction and estimation method is the same as the one used when adjusting flight parameters during UAV landing.

[0207] Furthermore, the prediction and estimation process is as follows:

[0208] (1) Adaptive Kalman Filter: Adaptive Kalman filtering is a commonly used state estimation algorithm used to make optimal state estimations based on sensor measurements and system dynamics models. In this case, adaptive Kalman filtering can be used to estimate parameters such as the altitude of the UAV.

[0209] (2) Set state variables: The altitude of the UAV is used as the state variable, that is, the physical state that needs to be estimated.

[0210] (3) System dynamics model: Based on the motion characteristics and environmental conditions of the UAV, a dynamics model is established to describe the variation law of the UAV altitude.

[0211] (4) Sensor measurements: The acceleration and position data provided by the guidance system are used as measurements to correct and adjust the estimated values.

[0212] (5) Kalman filter iteration process: Based on the Kalman filter iteration process, the optimal height is calculated using prior estimates and measurements.

[0213] Prediction phase: Using system dynamics models and prior estimates, the change in drone altitude is predicted through a prediction model.

[0214] Update phase: The actual measured value is compared with the predicted value. Based on the measurement error and the weight of the covariance matrix, the predicted value is corrected to obtain a more accurate height estimate.

[0215] Iteration: Repeat the prediction and update phases, continuously refining the estimated value so that it gradually approaches the true height.

[0216] (6) Optimal initial height estimate: After multiple iterations, the adaptive Kalman filter can provide an optimal initial height estimate for subsequent measurement and calculation processes.

[0217] By analyzing computations and using adaptive Kalman filtering, an optimal estimate of the UAV's initial altitude can be obtained. This allows for determination of whether the initial altitude meets a preset value, and if so, subsequent measurement, calculation, and control operations can be performed. This improves the UAV's positioning accuracy and flight stability.

[0218] Furthermore, after controlling the drone to perform inspections along the optimal flight path, it also includes:

[0219] Control the drone to collect inspection images;

[0220] The inspection images are divided into grids to obtain grid cell images;

[0221] Denoising is performed on the grid cell image based on the median filtering algorithm to obtain a denoised grid cell image.

[0222] An enhanced image is obtained by enhancing the denoised grid cell image using a generative adversarial network.

[0223] The denoising process includes sequentially removing Gaussian noise, salt-and-pepper noise, and motion blur from the grid cell image; specifically,

[0224] Remove Gaussian noise: Use smoothing filters such as Gaussian filters to reduce Gaussian noise in the grid cell image and improve the clarity of the grid cell image.

[0225] Salt-and-pepper noise removal: Nonlinear filters such as median filters are used to reduce salt-and-pepper noise in the grid cell image, restoring the detail and quality of the grid cell image.

[0226] Removing motion blur: For blurred images caused by camera shake or target motion, deblurring algorithms such as inverse filtering or blind deblurring methods can be applied.

[0227] The enhancement process for the denoised grid cell image includes sequential contrast enhancement, sharpening, color correction, and artifact removal; specifically,

[0228] Contrast Enhancement: By using algorithms such as histogram equalization or adaptive contrast enhancement, the pixel value distribution of the denoised grid cell image is adjusted to increase the contrast and visual effect of the denoised grid cell image.

[0229] Sharpening: Image sharpening filters (such as Laplacian filters or high-boost filters) are used to enhance the edges and details of the denoised grid cell image.

[0230] Color correction: Based on the color characteristics of the denoised grid cell image, color correction and color balance are performed to make the denoised grid cell image present a realistic color representation.

[0231] Image artifact removal: Image inpainting or background illumination correction algorithms can be used to address image artifacts or uneven lighting issues that may occur in specific application scenarios.

[0232] Specifically, the process involves obtaining inspection image information transmitted back by the UAV, dividing the image information into a grid, which means dividing the inspection image information into many small units. Then, a median filtering algorithm is used to denoise multiple grid units of the inspection image. The median filtering algorithm is a nonlinear signal processing technique based on ranking statistics theory that effectively suppresses noise. It mainly replaces the value of a point in a digital image with the median value of all points in its neighborhood, making the surrounding pixel values ​​closer to the true value, thereby eliminating isolated noise points and obtaining denoised inspection image information. Finally, a generative adversarial network (GAN) is used to enhance the denoised inspection image information. GANs are an important generative model in the field of deep learning, where the generator and discriminator are trained simultaneously and compete in a minimax algorithm to obtain the processed inspection image information. By obtaining the processed inspection image information, clearer and more intuitive inspection image information can be obtained, improving the inspection quality of the UAV.

[0233] Furthermore, controlling the drone to perform inspections along the optimal flight path also includes:

[0234] Get the drone's remaining battery power;

[0235] If the remaining battery level of the drone is lower than the warning threshold, a return-to-home battery swap command is generated and sent to the corresponding drone, controlling the drone to return to home for battery swapping.

[0236] Specifically, the system obtains the real-time battery level (remaining power) of the drone during flight and sets an alarm battery level (warning value). The alarm battery level is set based on the number and density of capsule drone nests within the inspection area, and reasonable alarm battery levels are obtained through big data testing. When the drone's real-time battery level falls below the alarm battery level, the control platform automatically generates a return-to-home battery swap command. After generating the command, the control platform immediately sends it to both the drone and the capsule drone nest. Based on the coordinates of the capsule drone nest's charging station, the platform obtains the drone's return-to-home destination coordinates and sends these coordinates back to the drone, controlling it to return. When the drone lands at the capsule drone nest's charging station, the capsule drone nest uses sensors to determine the drone's battery location and replaces the battery using its built-in robotic arm. This battery swapping via the capsule drone nest allows for uninterrupted flight, further improving the drone's operational efficiency.

[0237] After the UAV completes its return-to-base battery swap and inspection, the invention also includes:

[0238] Scene images of the landing area were collected by drones;

[0239] Obtain the location of the landing area from the scene image;

[0240] Obtain the drone's current position, current attitude, and current flight speed;

[0241] Based on a pre-built prediction model, the flight parameters that the UAV needs to adjust are obtained according to the location of the UAV landing area, combined with the UAV's current position, current attitude information and current flight speed; among which, the flight parameters include target altitude, target flight speed and target attitude parameters.

[0242] The drone is controlled to adjust its flight status in real time according to flight parameters until it completes landing.

[0243] The pre-built adaptive Kalman filter prediction model includes:

[0244] An adaptive Kalman filter prediction model is constructed based on the dynamic characteristics and motion parameters of the UAV; the motion parameters include the UAV's position, flight speed, and flight attitude.

[0245] Acquire training data for the drone, including the drone's actual training flight speed and attitude at different locations, as well as the actual training target location.

[0246] Based on the adaptive Kalman filter prediction model and combined with the UAV training data, the prediction data of the UAV when moving from different positions to the next position is obtained. The prediction data includes the predicted altitude, predicted flight speed and predicted attitude information.

[0247] Based on the predicted data and the training data, the observation residuals of the predicted data and the training data are obtained;

[0248] The covariance matrix is ​​obtained from the observed residuals;

[0249] The weights of the predicted and actual values ​​are obtained using Kalman gain.

[0250] The adaptive Kalman filter prediction model is updated based on the observation residuals, covariance matrix, and weights.

[0251] Repeat the above prediction and update steps until the predetermined update threshold is reached to obtain the final adaptive Kalman filter prediction model.

[0252] The adaptive Kalman filter prediction model is a model that uses the adaptive Kalman filter algorithm, which is an existing algorithm and will not be described in detail again.

[0253] Furthermore, the location of the landing area is obtained based on the scene image, including:

[0254] The scene image is sequentially processed, target detection is performed, and feature extraction is performed to obtain the relevant features of the landing area;

[0255] The relevant features of the landing area are reduced in dimension, normalized, and key features are extracted to obtain key information about the landing area in the scene image;

[0256] The location of the landing area in the image coordinate system is obtained based on key information of the landing area;

[0257] Based on the location of the landing area in the image coordinate system and the camera parameters carried by the UAV, the location of the landing area in the UAV coordinate system is obtained;

[0258] Image processing includes cropping operations on scene images, such as denoising, resizing, and restoration. Object detection involves processing scene images using object detection algorithms to identify landing areas. These algorithms can employ deep learning-based methods to detect target targets (landing areas) in scene images and label their positions and bounding boxes, such as YOLO (You Only Look Once) and Faster R-CNN (Region-based Convolutional Neural Networks). Feature extraction includes extracting color histograms, texture features, and shape features. Feature extraction can describe the attributes of the landing area for subsequent decision-making and control.

[0259] When associating and transforming positional information in the image coordinate system with positional information in the UAV coordinate system, technologies such as floating shape measurement, sensor fusion, or landmark recognition are typically used.

[0260] Specifically, based on the location of the landing area in the image coordinate system and the camera parameters on the UAV, the location of the landing area in the UAV coordinate system is obtained, including:

[0261] Obtain the intrinsic and extrinsic parameter matrices of the camera mounted on the drone;

[0262] Based on the location of the landing area in the image coordinate system, as well as the camera's intrinsic and extrinsic parameter matrices, the location of the landing area in the UAV coordinate system is obtained.

[0263] The method involves converting the location of the landing area in the image coordinate system to a conventional coordinate system using the camera's intrinsic and extrinsic parameter matrices. The UAV coordinate system is established with the UAV as the reference point, typically with the UAV's onboard location as the origin and the UAV's flight direction as the reference axis. For example, the X-axis might point forward, the Y-axis to the right, and the Z-axis downward. The UAV coordinate system describes the UAV's position, attitude, and motion. The image coordinate system is established with the image's pixels as the reference. The origin is the top-left corner of the image, the horizontal direction is the X-axis, and the vertical direction is the Y-axis. Pixel coordinates in the image coordinate system can be used to describe the position and size of objects.

[0264] Furthermore, the current position, current attitude information, and current flight speed of the drone are obtained, including:

[0265] The current location of the drone is obtained through GPS positioning;

[0266] The current flight speed of the drone is obtained through GPS, inertial sensors, or ground speed sensors;

[0267] The current attitude information of the drone is obtained through gyroscopes, speedometers, or magnetometers.

[0268] This invention achieves precise landing of drones by combining visual recognition and GPS positioning. Specifically,

[0269] (1) Visual recognition:

[0270] Using cameras or sensors mounted on a drone, scene images or videos are acquired, and the acquired scene videos contain continuous scene images.

[0271] Computer vision techniques, such as image processing, object detection, and feature extraction, are used to process and analyze images to identify relevant features of the landing area.

[0272] By extracting features such as edges, corners, and colors, target detection and tracking are performed to determine the location and orientation of the landing area.

[0273] Optimization of visual recognition: By using advanced computer vision algorithms and deep learning models, the accuracy and robustness of target detection and feature extraction can be improved, thereby more accurately identifying the location and attitude of the landing area.

[0274] Target position estimation formula: Estimate the position of the target in the image based on the target feature points detected in the camera image.

[0275] Distance estimation formula: Estimate the distance between the target and the drone based on the target's size in the image and known physical parameters.

[0276] One common method for target position estimation based on geometric relationships is to use a camera projection model of a monocular camera. Below is a simplified example of a geometric relationship-based target position estimation formula:

[0277] The parameters of the monocular camera carried by the drone are as follows:

[0278] Camera intrinsic parameter matrix: K = [[f_x,0,c_x],[0,f_y,c_y],[0,0,1]]

[0279] Camera extrinsic matrix (transformation matrix of the UAV relative to the world coordinate system): T_wd=[[R_wd,t_wd],[0,1]]

[0280] Where f_x and f_y are the camera's focal lengths, and c_x and c_y are the camera's optical center coordinates. R_wd is the rotation matrix, and t_wd is the translation vector.

[0281] The target's position in the UAV coordinate system is P_td = [X_t, Y_t, Z_t, 1], which means the target's position in the camera coordinate system is P_cd = [X_c, Y_c, Z_c, 1].

[0282] The target's position in the camera coordinate system can be transformed to the image coordinate system using the camera projection model:

[0283] P_id=K*T_wd*P_td

[0284] Where P_id = [u,v,1] is the position of the target in the image coordinate system, and u and v are the projected positions of the target on the image.

[0285] By using the above formula and combining the UAV's attitude and position information, the target's position in the camera coordinate system can be converted into its position in the image coordinate system, thereby estimating the target's position in the image.

[0286] It should be noted that in practical applications, factors such as distortion correction, camera nonlinearity, and optimization methods also need to be considered. The variables and matrix representations in the formula examples of this invention are for illustrative purposes only, and practical applications require appropriate adjustments and extensions based on specific camera parameters and scenarios.

[0287] (2) GPS positioning:

[0288] Use a GPS receiver to obtain the drone's current location and attitude information.

[0289] By combining sensor data such as inertial measurement units (IMUs), position and attitude data are filtered and fused to improve the accuracy and stability of positioning.

[0290] GPS positioning optimization: By combining data from sensors such as IMUs and employing filtering and fusion algorithms, noise and uncertainty in GPS signals can be reduced, thereby improving the accuracy and stability of positioning.

[0291] Position filtering formula: Using methods such as Kalman filtering or extended Kalman filtering, combined with GPS and IMU data, the position of the UAV is filtered and fused.

[0292] Location correction formula: The location is corrected based on GPS positioning error by matching with pre-stored landmarks or maps.

[0293] (3) Identification-Calculation-Landing Process:

[0294] Identification phase: Based on the results of visual recognition and GPS positioning, determine the location, size, and attitude of the landing area.

[0295] Calculation phase: Based on the location and attitude of the landing area, and combined with the current state of the UAV, calculate the flight parameters that the UAV needs to adjust, such as altitude, speed, and attitude control.

[0296] Landing phase: Based on the calculated flight parameters, the UAV performs a precise landing operation through the automatic control system, including altitude control, horizontal position adjustment and attitude stabilization.

[0297] The optimized algorithm is able to achieve precise landing for the following reasons:

[0298] Altitude control formula: Calculate altitude control commands based on the error between the current altitude of the UAV and the target landing altitude.

[0299] Lateral position control formula: Calculate the lateral position control command based on the error between the current position of the UAV and the target landing position.

[0300] Attitude control formula: Calculate attitude control commands based on the error between the current attitude of the UAV and the landing attitude of the target.

[0301] Algorithm optimization: By optimizing the algorithms in the identification, calculation, and landing processes, such as using optimization algorithms to solve the optimal control problem, it is possible to improve landing speed and stability while ensuring landing accuracy.

[0302] In addition to achieving precise landing, the optimized algorithm may also have the following effects:

[0303] Improved anti-interference capability: Through optimization of the recognition stage, the algorithm can better handle the complex environment of the landing area, such as avoiding misidentification of obstacles or handling changes in lighting.

[0304] Improved real-time performance: Through algorithm optimization and performance enhancement, the latency of identification, calculation and control can be reduced, thereby improving the real-time performance and responsiveness of the landing process.

[0305] Enhanced robustness: The optimized algorithm has better adaptability and robustness to changes in different landing areas and unknown environments, improving the UAV's landing capability under various conditions.

[0306] refer to Figure 3 In some embodiments, a power line inspection and control device based on a drone is also provided, comprising:

[0307] The tower coordinate acquisition module 201 is used to acquire inspection tasks and control the drone to collect tower position coordinate information within the inspection area corresponding to all inspection tasks.

[0308] The tilt flight path generation module 202 is used to generate multiple tilt flight paths corresponding to the inspection task based on the tower position coordinate information.

[0309] The flight path generation module 203 is used to control the UAV to fly along multiple tilted flight paths, and to collect point cloud data corresponding to the tilted flight paths during the flight, and to generate flight paths based on the point cloud data.

[0310] Altitude test module 204 is used to perform altitude tests on all flight paths to obtain the optimal flight path corresponding to the inspection task.

[0311] The task matching module 205 is used to acquire the target inspection task, match the optimal flight path corresponding to the target inspection task, and control the UAV to carry out inspection according to the optimal flight path.

[0312] This invention solves the technical problems of low efficiency and poor inspection quality in traditional manual drone inspection modes. By enabling autonomous drone inspection and intelligent battery replacement via a capsule-like pod, the efficiency of drone inspection can be improved. Optimal flight altitude testing ensures smoother transmission of inspection image information. Noise reduction and enhancement processing of the inspection image information improves the quality of drone inspections and further enhances the reliability of regional power supply. Furthermore, obtaining the optimal flight path reduces the latency of control commands from the control platform and provides smoother, higher-definition video. Remote control of the drone for autonomous inspection reduces the need for manpower. Battery replacement via the capsule-like pod enables uninterrupted flight, significantly improving the efficiency of drone inspections.

[0313] Furthermore, the altitude testing module 204 performs altitude tests on all flight paths to obtain the optimal flight path corresponding to the inspection mission, including:

[0314] Multiple preset flight altitudes are set for each flight path, and the drone is controlled to fly along the flight path at different preset flight altitudes;

[0315] During flight, at different preset flight altitudes, test signals and test images sent back by the drone are received;

[0316] The test signals are analyzed for signal strength and signal quality, and the test images are analyzed for sharpness. Based on the analysis results, the optimal altitude for each flight path is determined.

[0317] The optimal flight path is obtained by matching the optimal altitude with the flight path.

[0318] refer to Figure 4 In some embodiments, a power line inspection and control system based on unmanned aerial vehicles (UAVs) is also provided, including a control platform 1, a UAV 2, and a capsule-shaped storage device 3. The capsule-shaped storage device 3 is used to store the UAV and charge the UAV. The control platform 1 includes a processor and a storage device. The storage device stores multiple instructions, and the processor is used to read the instructions and execute the above-described methods.

[0319] 5G data transmission is used between the capsule machine nest and the control platform.

[0320] After receiving inspection images from the drone, the capsule-shaped sensor will process the images and send them to the control platform. The specific processing procedure is as follows:

[0321] (1) Image compression: Use appropriate compression algorithms (such as JPEG, WebP, etc.) to compress the inspection images to reduce the amount of data, thereby reducing transmission delay and bandwidth consumption.

[0322] (2) Image slicing: The inspection image is sliced ​​into multiple small pieces or tiles so that they can be sent in parallel during transmission, thereby improving transmission efficiency.

[0323] (3) Error correction coding: Introduce error correction codes, such as Reed-Solomon coding, to detect and repair errors during data transmission, thereby improving data integrity and reliability.

[0324] After receiving the inspection image data sent by the capsule machine, the control platform will process it accordingly. The specific processing procedure is as follows:

[0325] (1) Image stitching: The tiles of the received inspection images are stitched together in the prescribed order to restore the complete original inspection images.

[0326] (2) Image decompression: Decompress the received compressed image to restore the original high-quality image.

[0327] (3) Image cropping: Cropping the received inspection images according to requirements, extracting the areas of interest, and reducing data processing overhead.

[0328] In improving transmission quality through the capsule terminal and control platform using the above-mentioned operations, the following operations can be employed to achieve the desired effect. Specifically,

[0329] (1) Image compression ratio control: Select appropriate compression parameters and algorithms to balance image quality and data compression ratio. A lower compression ratio can retain more details and image quality, but will increase the amount of data transmitted and the transmission latency.

[0330] (2) Lossless compression algorithm: Using lossless compression algorithms (such as PNG) can avoid the loss of image quality caused by compression, but usually results in a higher data transmission volume.

[0331] (3) High-efficiency coding algorithm: Select advanced coding algorithms (such as HEVC, AV1) to improve image compression rate and transmission efficiency while maintaining high image quality.

[0332] (4) Image enhancement algorithm: Before transmission, the image is enhanced, such as noise reduction, contrast enhancement, sharpening, etc., to improve the visual quality of the image.

[0333] (5) Error correction coding: Error correction coding (such as RS code) can be used to detect and repair errors during transmission, ensuring the integrity of image data and thus improving image quality.

[0334] In summary, the UAV-based power line inspection and control method and system provided in the above embodiments have at least the following beneficial effects:

[0335] (1) By establishing a line model of laser point cloud, the flight path is planned, and the optimal flight path is obtained based on altitude test, so as to ensure the safety of UAV during inspection, the stability of signal transmission and the reliability of image acquisition, and improve the inspection efficiency of UAV.

[0336] (2) During the process of obtaining the optimal flight route based on altitude testing, the optimal flight route obtained through signal quality, signal strength and image clarity makes the control platform have a lower delay in the control command of the UAV and obtain smoother high-definition video, which further improves the efficiency of UAV inspection.

[0337] (3) By replacing the battery of the drone through the capsule machine nest, the drone can fly continuously, which further improves the working efficiency of the drone.

[0338] Although preferred embodiments of the invention have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including both the preferred embodiments and all changes and modifications falling within the scope of the invention. Clearly, those skilled in the art can make various alterations and modifications to the invention without departing from its spirit and scope. Thus, if these modifications and modifications of the invention fall within the scope of the claims and their equivalents, the invention is also intended to include these modifications and modifications.

Claims

1. A power line inspection and control method based on unmanned aerial vehicles (UAVs), characterized in that, include: Obtain inspection tasks and control the drone to collect the pole and tower location coordinates within the inspection area corresponding to all inspection tasks; Multiple inclined flight paths corresponding to the inspection task are generated based on the tower position coordinate information; The drone is controlled to fly along multiple tilted flight paths, and point cloud data corresponding to the tilted flight paths is collected during the flight. The flight path is generated based on the point cloud data. Multiple different preset flight altitudes are set for each flight path, and the UAV is controlled to fly along the flight path at different preset flight altitudes; During flight, at different preset flight altitudes, test signals and test images sent back by the drone are received; The test signal is analyzed for signal strength and signal quality to obtain a first height region, and the test image is analyzed for sharpness to obtain a second height region. The first and second height regions are then filtered based on a first evaluation condition and a second evaluation condition to obtain a first optimal height and a second optimal height. Weights are assigned to the signal strength and signal quality analysis of the test signal and the sharpness analysis of the test image. Based on the assigned weights, and the first and second optimal heights, the optimal height is obtained. The optimal altitude is matched with the flight path to obtain the optimal flight path; The system acquires the target inspection task, matches the optimal flight path corresponding to the target inspection task, and controls the UAV to perform inspections according to the optimal flight path.

2. The method according to claim 1, characterized in that, Based on the tower position coordinate information, multiple inclined flight paths corresponding to the inspection task are generated, including: The external dimensions of the tower corresponding to the inspection task are obtained based on the tower location coordinates. The flight direction, flight range, and flight altitude of the UAV are determined based on the external dimensions of the tower. Multiple waypoints are determined based on the drone's flight direction, flight range, and flight altitude; Based on multiple waypoints, an inclined flight path is generated to correspond to the inspection task.

3. The method according to claim 1, characterized in that, Generate a flight path based on the point cloud data, including: Multiple point cloud route models were obtained based on the tower point cloud data from the point cloud data. Based on the point cloud data of the transmission line, multiple point cloud line models are connected in series; Based on the inspection task, the inspection point locations and inspection feature actions are configured on the connected point cloud line model to generate the flight path.

4. The method according to claim 1, characterized in that, The process of controlling the drone to perform inspections along the optimal flight path also includes: Get the drone's remaining battery power; If the remaining battery level of the drone is lower than the warning value, a return-to-home battery swap command is generated and sent to the corresponding drone to control the drone to return to home and swap batteries.

5. The method according to claim 4, characterized in that, After the drone completes its return-to-base battery swap and inspection, the process also includes: Scene images of the landing area were collected by drones; Obtain the location of the landing area from the scene image; Obtain the drone's current position, current attitude, and current flight speed; Based on the location of the drone's landing area, combined with the drone's current position, current attitude information, and current flight speed, the flight parameters that the drone needs to adjust are obtained; among them, the flight parameters include target altitude, target flight speed, and target attitude parameters; The drone is controlled to adjust its flight status in real time according to flight parameters until it completes landing.

6. The method according to claim 1, characterized in that, The test signal is analyzed for signal strength and signal quality to obtain a first height region, including: Statistical analysis was performed on the signal strength of the test signals at different preset flight altitudes to obtain the average, maximum and minimum signal strength values. Based on the average, maximum, and minimum values ​​of signal strength, the trend of signal strength variation at different preset flight altitudes and the strength difference at different times at the same preset flight altitude are obtained. Based on the trend and differences in signal strength, the coverage area and intensity distribution of the test signal are obtained; Perform statistical analysis on the signal quality of the test signal to obtain its mean, variance, and standard deviation; Plot the quality variation based on the mean, variance, and standard deviation of the signal quality. Based on the quality change map, the fluctuation range of signal quality at different preset flight altitudes and the rate of change within a predetermined time period are obtained; The first height region is obtained based on the coverage area, intensity distribution, fluctuation range, rate of change, and first preset parameter threshold of the test signal.

7. The method according to claim 1 or 6, characterized in that, A sharpness analysis was performed on the test image to obtain the second height region, including: Based on image processing algorithms, relevant features of the target are extracted from test images at different preset flight altitudes. Based on the extracted relevant features, the target in the test image is identified and located using a target detection algorithm; Obtain predetermined evaluation indicators, and obtain the evaluation parameters corresponding to the predetermined evaluation indicators based on the identified and located test images; The evaluation parameters of the test images at different preset flight altitudes are filtered based on the second preset parameter threshold to obtain the test images that meet the conditions. The second altitude region is obtained based on the preset flight altitude corresponding to the test image that meets the conditions.

8. A power line inspection and control system based on unmanned aerial vehicles (UAVs), characterized in that, The system includes a control platform, a drone, and a capsule-shaped storage device for storing and charging the drone. The control platform includes a processor and a storage device, the storage device storing multiple instructions, and the processor for reading the instructions and executing the method as described in any one of claims 1-7.

Images