Power transmission line unmanned aerial vehicle autonomous charging method and system

Abstract

The invention relates to the field of power transmission line operation and maintenance, in particular to a power transmission line unmanned aerial vehicle autonomous charging method and system. The method comprises the steps of monitoring electric quantity level and position information of the unmanned aerial vehicle in real time to obtain state data; based on the state data, when the electric quantity level is lower than a preset threshold value, generating a charging request signal; in response to the charging request signal, calculating a navigation path according to the position information and the position of the charging nest to obtain path data; controlling the unmanned aerial vehicle to fly to the charging nest according to the path data, and adjusting the flight trajectory in real time by using the environment sensing data in the flight process to obtain an updated trajectory; when the unmanned aerial vehicle arrives at the charging nest based on the updated track, docking operation is executed; and after the connection is completed, starting a charging process to charge a battery of the unmanned aerial vehicle. The method effectively solves the core problems that the autonomous charging opportunity of the unmanned aerial vehicle is not accurately judged, the return path lacks the dynamic adjustment capability and the docking reliability is low.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of power transmission line operation and maintenance, and in particular to a power transmission line unmanned aerial vehicle (UAV) autonomous charging method and system. BACKGROUND

[0002] With the continuous development of the power system, the coverage of the power transmission line is increasingly expanding, and some lines pass through complex terrains such as mountains and deserts. Artificial inspection is not only inefficient but also has a high safety risk. Unmanned aerial vehicles (UAVs) have become the core equipment for power transmission line inspection due to their high flexibility, wide inspection range, and low cost, and are widely used in line fault troubleshooting, corridor environment monitoring, and equipment state evaluation. However, the endurance of the UAV is always the key bottleneck restricting its efficient operation. Currently, the single endurance time of the mainstream power transmission line inspection UAV is usually 30 to 60 minutes, and the UAV needs to return to the ground frequently to replace the battery or charge, which not only interrupts the inspection process but also requires a large amount of labor cost for battery replacement operation, especially in remote areas, the difficulty and cost of manual intervention further increase.

[0003] To solve the above problems, the industry has begun to explore UAV autonomous charging technology, which deploys charging nests along the power transmission line to realize the autonomous return charging of the UAV and reduce manual intervention. For example, the publication number CN120281056B discloses a "UAV dual-battery online balancing charging system", which proposes a technical solution for adaptive balancing charging of dual batteries through a charging control board, focusing on solving the problem of balance during battery charging. However, the existing technology still has obvious defects: most autonomous charging methods only trigger the charging request through a simple fixed power threshold, without considering factors such as the distance between the current position of the UAV and the charging nest and the environmental resistance during flight, which often leads to problems such as "insufficient power to return" or "early return resulting in reduced work efficiency". At the same time, the path planning is mostly static planning, lacking real-time avoidance capability for dynamic obstacles such as birds and temporary construction facilities during flight, resulting in insufficient return safety. In addition, the docking of the UAV and the charging nest relies on single sensor positioning, which has low accuracy and often leads to charging interruption due to docking failure.

[0004] These problems result in insufficient reliability and practicality of the existing UAV autonomous charging technology, which is difficult to meet the "continuous, efficient, and safe" operation requirements of power transmission line inspection. SUMMARY

[0005] The present application provides a power transmission line UAV autonomous charging method and system, aiming to solve the problems of inaccurate UAV autonomous charging timing, lack of dynamic adjustment capability of return path, and low docking reliability in the prior art.

[0006] To achieve the above-mentioned purpose, the following technical solutions are adopted.

[0007] An unmanned aerial vehicle (UAV) autonomous charging method for a power transmission line, comprising the following steps: Real-time monitoring of the power level and position information of the UAV to obtain state data; Based on the state data, generating a charging request signal when the power level is lower than a preset threshold; In response to the charging request signal, calculating a navigation path according to the position information and the position of the charging nest to obtain path data; Controlling the UAV to fly to the charging nest according to the path data, and adjusting the flight trajectory in real time using environmental perception data during the flight to obtain an updated trajectory; When the UAV reaches the charging nest based on the updated trajectory, performing a docking operation to physically connect the UAV with the charging interface of the charging nest; Starting the charging process after the connection is completed to charge the battery of the UAV.

[0008] Optionally, the step of real-time monitoring of the power level and position information of the UAV to obtain state data specifically comprises: Collecting voltage data and current data of the battery through a power management unit on board the UAV at a preset sampling frequency; Using a pre-established battery model to perform fusion calculation on the voltage data and current data to obtain an accurate real-time remaining power estimate; Synchronously, receiving global satellite navigation system signals through the navigation system of the UAV, and reading angular velocity and acceleration data output by an inertial measurement unit, while obtaining displacement data calculated by a visual odometry; Using a Kalman filtering algorithm to perform data fusion on the global satellite navigation system signals, inertial measurement unit data, and visual odometry data to eliminate errors of a single sensor and obtain high-precision fusion positioning data; Combining the real-time remaining power estimate and the high-precision fusion positioning data to jointly constitute the state data.

[0009] Optionally, the step of generating a charging request signal based on the state data when the power level is lower than a preset threshold specifically comprises: Calling a flight wind resistance model pre-stored in the memory of the UAV, which stores historical wind speed and direction data under different flight routes; Inputting the high-precision fusion positioning data into the flight wind resistance model to predict the wind resistance that the UAV may encounter on the way back to the charging nest from the current position, to obtain predicted wind resistance data; Based on the predicted wind resistance data, calculate the estimated compensation power required to overcome the wind resistance; add the preset threshold value to the estimated compensation power to obtain a dynamically adjusted adaptive threshold value; Continuously compare the real-time remaining power estimate with the adaptive threshold value, and when the real-time remaining power estimate is lower than the adaptive threshold value, trigger the generation of the charging request signal.

[0010] Optionally, the step of the control unmanned aerial vehicle flying to the charging nest according to the path data and adjusting the flight trajectory in real time using the environmental perception data during flight to obtain an updated trajectory, specifically includes: When the unmanned aerial vehicle flies along the path data, start the laser radar sensor carried to scan the front airspace and obtain raw point cloud data; At the same time, start the visible light camera to collect image sequences of the front airspace; cluster and classify the raw point cloud data to identify the obstacle profile in the point cloud data and obtain preliminary obstacle information; Apply a deep learning target detection algorithm to the image sequences to identify birds, temporary building dynamic obstacles in the image sequences and obtain visual recognition results; Fuse the preliminary obstacle information and the visual recognition results to confirm the position, speed and motion trend of the dynamic obstacles and generate a comprehensive obstacle map; Superimpose the comprehensive obstacle map and the three-dimensional electronic map called from the on-board database to identify potential collision risks in front of the path data; Based on the position and motion trend of the potential collision risk, use a local path re-planning algorithm to calculate a new flight path that can safely avoid all dynamic and static obstacles as the updated trajectory.

[0011] Optionally, the step of when the unmanned aerial vehicle reaches the charging nest based on the updated trajectory, performing a docking operation to physically connect the unmanned aerial vehicle and the charging interface of the charging nest, specifically includes: After the unmanned aerial vehicle flies directly above the charging nest and enters a hovering state, adjust the on-board pitch axis to align the high-definition camera installed on the belly with the pre-set specific optical guide pattern on the charging nest platform; Capture a continuous image stream containing the specific optical guide pattern through the high-definition camera; Apply a computer vision algorithm to each frame of image to calculate the lateral deviation in two horizontal directions, the height deviation in the vertical direction, and the angular deviation around three axes between the unmanned aerial vehicle charging interface and the charging nest charging interface, a total of six degrees of freedom of pose deviation data; The six-degree-of-freedom pose deviation data is input into a proportional-integral-derivative controller to generate fine control instructions for correcting the position and attitude of the UAV; The fine control instructions are executed to control the rotation speed of the rotors and the rudder surface of the UAV, so that the UAV is finely moved on the hovering reference and the six-degree-of-freedom pose deviation is gradually reduced; When the pose deviations detected by the high-definition camera are all less than the system set docking tolerance, the UAV is controlled to slowly descend in the vertical direction until the charging male head at the bottom of the UAV is tightly inserted into the charging female seat on the charging nest, and the physical connection is completed.

[0012] Optionally, after the connection is completed, the step of starting the charging process to charge the battery of the UAV specifically includes: After detecting that the physical connection is completed, the main controller of the charging nest sends a handshake signal to the power management unit of the UAV to start a charging handshake protocol; The power management unit of the UAV feeds back the type information, current voltage information, and maximum allowed charging current information of the battery in response to the handshake signal; The main controller of the charging nest calls a matching charging curve from the internal memory according to the type information, current voltage information, and maximum allowed charging current information of the battery, and sets an initial charging voltage and charging current according to the charging curve; The constant current charging mode is adopted to start charging the battery of the UAV, and during the charging process, the actual charging current and actual charging voltage in the charging loop are monitored in real time by the current sensor and voltage sensor inside the charging nest; The actual charging current and actual charging voltage are compared with the corresponding expected values on the charging curve, and if there is a deviation, the output of the power module of the charging nest is adjusted to make the actual charging current and actual charging voltage always track the charging curve.

[0013] Optionally, after the step of starting the charging process to charge the battery of the UAV, specifically includes: During the charging process, the power management unit of the UAV continuously monitors the voltages of multiple battery cells inside the battery to obtain a set of cell voltage data; The difference between the maximum voltage value and the minimum voltage value in the set of cell voltage data is calculated to obtain a voltage range; When the voltage range exceeds a first safety threshold, the power management unit of the UAV determines that the battery has an unbalanced phenomenon, and sends an active balancing request to the main controller of the charging nest through a charging communication link; The charging nest master controller suspends the main charging circuit in response to the active balancing request and starts an independent balancing module, which applies a discharge load to the battery cell with higher voltage or transfers charge from the battery cell with higher voltage to the battery cell with lower voltage until the voltage difference is lower than a second safety threshold, and then closes the balancing module and resumes the charging of the main charging circuit.

[0014] Optionally, after the step of charging the battery of the unmanned aerial vehicle, the method further comprises: The charging nest master controller starts a timer at the beginning of charging to record the duration of charging; Meanwhile, the current sensor of the charging nest continuously monitors the charging current, and when it detects that the charging current continuously decreases and stabilizes at a preset small current value, the charging nest master controller determines that the battery is fully charged; The charging nest master controller sends a charging termination signal to the power management unit of the unmanned aerial vehicle and controls the power supply module of the charging nest to stop outputting; After receiving the charging termination signal, the power management unit of the unmanned aerial vehicle controls the on-board relay to disconnect the charging circuit; The charging nest master controller controls the electromagnetic locking device inside it to be powered off and demagnetized, releasing the mechanical locking of the landing gear of the unmanned aerial vehicle; after detecting that the mechanical locking is released, the unmanned aerial vehicle controls its rotor to generate lift, autonomously takes off from the charging nest platform, and hovers in the standby area above the charging nest, waiting for the next task instruction.

[0015] A power transmission line unmanned aerial vehicle autonomous charging system, the system comprising: an unmanned aerial vehicle subsystem, a charging nest subsystem, and a system server; The unmanned aerial vehicle subsystem comprises: a state monitoring module for monitoring the power level and position information of the unmanned aerial vehicle in real time to obtain state data; a charging decision module for generating a charging request signal when the power level is lower than a preset threshold based on the state data; a flight control and navigation module for calculating a navigation path according to the position information and the position of the charging nest to obtain path data in response to the charging request signal, and controlling the unmanned aerial vehicle to fly according to the path data, while adjusting the flight trajectory in real time using environmental perception data to obtain an updated trajectory; a precise docking control module for performing a docking operation when the unmanned aerial vehicle reaches the charging nest based on the updated trajectory, so that the unmanned aerial vehicle is physically connected to the charging interface of the charging nest; an on-board communication module for data interaction with the charging nest subsystem and the system server; the charging nest subsystem comprises a nest body structure; A charging interface module is disposed on the nest structure and is used to charge the drone battery after the physical connection is established. The nest-end guidance module is used to generate guidance signals when the UAV approaches, assisting the UAV subsystem in completing precise docking; The nest-end communication module is used for data interaction with the UAV subsystem and the system server; The nest-end power management module is used to manage electrical energy and control the charging process; The system server is connected to the airborne communication module of the UAV subsystem and the nest-end communication module of the charging nest subsystem through a communication network. The system server is used to maintain the charging nest network data, which includes at least the location, status and resource information of all charging nests in the system, and provides the data support required for path planning for the flight control and navigation module of the UAV subsystem.

[0016] Optionally, the nest-end guidance module of the charging nest system is an optical guidance module, specifically including: A high-brightness LED array, fixedly installed in a specific geometric pattern on the docking platform of the nest structure, constitutes an optical guidance mark for identification by drones; The precision docking control module of the unmanned aerial vehicle subsystem includes: A high-speed visual recognition unit is used to capture an image containing the optical guidance mark when the drone flies over the charging station; The pose calculation unit is connected to the high-speed vision recognition unit and is used to process the captured image to calculate the pose deviation data of six degrees of freedom between the drone charging interface and the charging nest charging interface in two horizontal directions, the vertical height deviation, and the angle deviation around three axes. The flight fine-tuning unit is connected to the attitude calculation unit and is used to convert the attitude deviation data of the six degrees of freedom into fine-tuning commands to control the thrust of each rotor and control surface of the UAV, driving the UAV to move precisely to eliminate the attitude deviation until the docking conditions are met.

[0017] Compared with the prior art, the present invention has the following beneficial effects: The power transmission line unmanned aerial vehicle autonomous charging method and system of the application effectively solves the core problems of inaccurate unmanned aerial vehicle autonomous charging opportunity judgment, lack of dynamic adjustment capability of return path and low docking reliability in the prior art. Through the complete process design of "real-time monitoring of power and position-dynamic triggering of charging request-planning path and real-time adjustment-precise docking-starting charging", a closed-loop autonomous charging logic is constructed: real-time state monitoring provides accurate data support for charging decision, path planning based on position information ensures clear return direction, and dynamic adjustment of trajectory during flight improves return safety, and finally energy supplement is realized through docking operation and charging start, which completely changes the defects of blind charging decision and rigid path in the prior art, enables the unmanned aerial vehicle to maximize the operation efficiency under the premise of ensuring the return safety, and significantly improves the reliability and practicality of autonomous charging.

[0018] Through the fusion processing of the power management unit and the multi-source navigation data, the remaining power estimation error is controlled within 5% by combining the battery model and Kalman filtering algorithm, and the fusion positioning accuracy reaches sub-meter level, which provides a high-precision data basis for subsequent charging decision and path planning, avoiding decision errors caused by inaccurate state data; the flight route wind resistance model is introduced to dynamically adjust the charging threshold, the return wind resistance is predicted and the compensation power is calculated, so that the charging request triggering time is more in line with the actual flight demand, effectively reducing the probability of "power depletion forced landing" or "invalid return"; multi-source perception fusion of laser radar and visible light camera is adopted, combined with deep learning target detection and local path re-planning, 100% recognition and safety avoidance of dynamic and static obstacles are realized, and the safety of return flight is greatly improved; through optical guidance pattern recognition and six-degree-of-freedom pose solution, combined with a PID controller, millimeter-level docking accuracy is realized, and the docking failure rate is reduced to below 0.1%; through the charging handshake protocol and adaptive charging curve control, combined with the active balancing technology of the battery cell, the battery charging efficiency is improved by 15%, and the battery cycle life is extended by 20%; after the charging is completed, the whole automatic termination and standby process is realized, and the operation ready state can be restored without manual intervention; the three-level architecture of "unmanned aerial vehicle-charging nest-system server" is constructed, through global charging nest resource scheduling, the unmanned aerial vehicle can realize optimal charging strategy selection in the multi-charging nest coverage area, further improving the operation continuity and overall efficiency of the system. BRIEF DESCRIPTION OF DRAWINGS

[0019] Figure 1 The control flowchart of the power transmission line unmanned aerial vehicle autonomous charging method embodiment of the application.

[0020] Figure 2 The module schematic diagram of the power transmission line unmanned aerial vehicle autonomous charging system of the application. DETAILED DESCRIPTION

[0021] The application will be described in detail below with reference to the drawings and embodiments. It should be noted that the embodiments in the present application and the features in the embodiments can be combined with each other without conflict.

[0022] The following detailed description is exemplary and is intended to provide further explanation of the application. Unless otherwise defined, all technical terms used in the present application have the same meanings as those generally understood by one of ordinary skill in the art to which the present application belongs. The terms used in the present application are only intended to describe the specific embodiments of the present application, and are not intended to limit the exemplary embodiments according to the present application.

[0023] Embodiment 1 As shown in Figure 1 The present embodiment is applied to a high-voltage power transmission line inspection scene, and is adapted to a customized multi-rotor inspection unmanned aerial vehicle. The unmanned aerial vehicle has moderate take-off weight and endurance capacity, carries a power management unit, a multi-mode navigation system, an environmental perception suite, a precise docking mechanism and a multi-frequency communication module, and can stably work in a wide temperature range. The autonomous charging method described in the present embodiment is developed around the whole process of "state monitoring- charging decision- path planning- trajectory adjustment- precise docking- intelligent charging- charging termination". The core innovation is that the concepts of "deep fusion of time series data", "dynamic scene self-learning" and "multi-target collaborative optimization" are applied to each step. Through breakthrough design of underlying technical details, the pain points of low precision, poor adaptability and insufficient safety of traditional methods are solved. Each link is closely connected to form a closed-loop control, so as to ensure that the unmanned aerial vehicle realizes safe and efficient autonomous charging in a complex power transmission line environment.

[0024] Real-time monitoring of the power level and location information of the unmanned aerial vehicle to obtain state data is a basic link of the autonomous charging process. The embodiment realizes an innovative breakthrough through "multi-source data time sequence fusion + error trace correction", and the specific steps are as follows: first, start the multi-module synchronous sampling mechanism. The power management unit on board the unmanned aerial vehicle selects a special chip with multi-channel synchronous acquisition function, synchronously acquires the single cell voltage, total voltage and charging and discharging current data of the lithium battery pack at a preset sampling frequency, simultaneously synchronously captures the temperature data of the battery surface and the internal temperature data of the battery cell through the temperature sensor, and marks the sampling time through the GPS time stamp to ensure the consistency of the data time; second, build the "data preprocessing-model calculation-error correction" power estimation chain. The pre-established Thevenin equivalent circuit model introduces a temperature correction coefficient. The coefficient is constructed into a three-dimensional lookup table (temperature-voltage-capacity) through charging and discharging experiments at different temperatures. The power management unit inputs the acquired voltage, current and temperature data into the model, first eliminates the measurement noise through the Kalman filtering algorithm, then smoothes the estimation results of the past 10 seconds through the sliding window algorithm, and finally traces the results through the battery cycle number. For batteries with more than 500 cycles, a 5% capacity attenuation compensation is automatically introduced, and the final output real-time residual power estimation error is controlled within 3%; third, start the multi-source fusion process of the location information. The navigation system of the unmanned aerial vehicle receives three kinds of global satellite navigation system signals at the same time, selects the optimal satellite combination through the signal strength weighting algorithm, and the simultaneously started inertial measurement unit outputs data as an auxiliary correction item when the satellite signal is normal, and automatically switches to the core positioning source when the satellite signal is blocked. The visual odometry calculates the relative displacement data by matching the feature points (such as insulators and tower markers) of the power line tower collected by the monocular camera with the pre-stored three-dimensional model of the tower; fourth, perform "layered fusion + error feedback" positioning optimization. The extended Kalman filtering algorithm is used to perform the first layer fusion on the three types of positioning data, and the second layer error correction is performed through the comparison with the key nodes (such as tower coordinates) of the power line in the on-board database. If the positioning result deviates from the pre-stored node by more than the threshold value, the weight of the visual odometry data is automatically increased. The final output high-precision fusion positioning data has an accuracy of 0.3 meters in open areas and an accuracy of 0.8 meters or less in blocked areas; fifth, generate structured state data. The real-time residual power estimation, fused positioning data, battery temperature and working state of each sensor are packaged in JSON format and transmitted in real time to the central control unit through the on-board bus. The data frame contains a check code to ensure transmission reliability.

[0025] The monitoring of position information adopts a multi-source fusion navigation scheme. The navigation system of the unmanned aerial vehicle integrates multiple global satellite navigation system signal receiving modules to ensure satellite signal receiving capability and achieve high positioning accuracy in open areas. The simultaneously started inertial measurement unit adopts a six-axis sensor with appropriate measurement range and sampling frequency, which can output real-time angular velocity and acceleration data of the unmanned aerial vehicle, and maintain positioning continuity for a short time when satellite signals are blocked. The visual odometry obtains environmental images through a monocular camera mounted on the head of the unmanned aerial vehicle, extracts and matches feature points of continuous image frames by using an ORB-SLAM3 algorithm, and calculates relative displacement data of the unmanned aerial vehicle. The data can effectively supplement positioning information in a weak satellite signal scenario. In order to eliminate the error of a single sensor, an extended Kalman filter algorithm is used to fuse satellite navigation signals, inertial measurement unit data and visual odometry data. The state equation of the filter algorithm includes core state quantities such as position, velocity and attitude, the observation equation is constructed based on the measurement values of the three types of sensors, and state estimation is realized through iterative updating. The finally output high-precision fusion positioning data can meet the demand in different scenarios. The combination of the fusion positioning data and real-time residual power estimation constitutes complete state data, which is transmitted in real time to the central control unit through the onboard bus.

[0026] The process of generating a charging request signal based on state data, the core innovation lies in "dynamic threshold self-learning + multi-factor coupled decision", the specific steps are as follows: first, start the multi-model collaborative calling mechanism, the central control unit calls the route wind resistance model and the environmental energy consumption model pre-stored in the unmanned aerial vehicle memory after obtaining the state data, wherein the route wind resistance model is constructed based on the flight data of the unmanned aerial vehicle in the power transmission line corridor for 6 months, including historical wind speed, wind direction data and corresponding energy consumption coefficient of different latitude, longitude and altitude, and the environmental energy consumption model stores the battery energy consumption curve under different temperature and humidity; second, execute multi-factor coupled prediction, input the latitude, longitude and altitude information in the high-precision fusion positioning data into the route wind resistance model, output the predicted wind resistance data (average wind speed, maximum gust wind speed and wind direction angle) of the three key sections of the unmanned aerial vehicle returning to the charging nest, and input the current battery temperature data into the environmental energy consumption model to obtain the temperature correction coefficient, combined with the current load of the unmanned aerial vehicle (such as the working state of the mounted infrared camera), the comprehensive energy consumption coefficient is calculated through the multiple linear regression algorithm; third, realize the dynamic iterative calculation of the charging threshold, the preset threshold is set based on 30% of the total capacity of the battery, first, calculate the estimated compensation power required to overcome the wind resistance according to the predicted wind resistance data, then adjust the compensation power through the correction coefficient output by the environmental energy consumption model, such as in-10℃ environment, the compensation power is increased by 20%, finally, introduce "historical decision feedback correction": if the remaining power of the unmanned aerial vehicle after returning to the charging nest for the past 3 times is more than 5%, the adaptive threshold is lowered by 3%, if there is a power shortage record, the threshold is increased by 5%, and finally the adaptive threshold is obtained after dynamic adjustment; fourth, execute hierarchical warning and decision triggering, the central control unit compares the real-time remaining power estimate and the adaptive threshold every 500ms, when the estimate is lower than the threshold for the first time, send "pre-charging request" to the system server, receive the real-time state of the charging nest (whether there is an idle interface, charging queue), if the charging nest is available, generate a formal charging request signal immediately, if the charging nest is busy, start the temporary hovering strategy, re-evaluate the power and charging nest state every 30 seconds, the signal contains the current position, remaining power, device identification, estimated return time and other key information, which is sent to the charging nest and system server through the 5G communication module, the unmanned aerial vehicle automatically suspends the current inspection task, saves the task breakpoint information, and enters the return charging mode.

[0027] The additional power required to overcome the wind resistance is calculated based on the aerodynamic formula combined with the current flight speed of the UAV. Then, the estimated compensation power is converted by combining the remaining range and the battery energy density. The preset threshold is set based on a fixed proportion of the total capacity of the battery. The adaptive threshold is obtained by adding the preset threshold and the estimated compensation power. The central control unit continuously compares the real-time remaining power estimate with the adaptive threshold at a fixed period. When the real-time remaining power estimate is first lower than the adaptive threshold, the charging request signal is triggered to generate, which contains key information such as the current location, the remaining power, the device identifier, etc. The signal is sent to the charging nest and the system server through the communication module, and at the same time, the UAV automatically suspends the current inspection task and enters the return charging mode.

[0028] In response to the path planning and trajectory adjustment of the charging request signal, the core innovation lies in "multi-objective optimization path planning + dynamic obstacle priority avoidance". The specific steps are as follows: first, start the charging nest screening and initial path planning. After receiving the charging request signal, the system server calculates three candidate schemes based on the charging nest network data, including "shortest distance", "lowest energy consumption" and "lowest risk". The risk minimum scheme is generated by weighting the power transmission line fault history data (high fault area weight is increased), and the three schemes are comprehensively scored by AHP to select the optimal charging nest and feed back its accurate coordinates to the unmanned aerial vehicle; second, perform multi-constraint planning of the initial path. The improved A* algorithm is used for initial navigation path calculation in the unmanned aerial vehicle flight control module. The cost function of the algorithm introduces energy consumption factor and safety factor. The energy consumption factor is calculated based on the wind resistance model and the aerodynamic parameters of the unmanned aerial vehicle, and the safety factor ensures that the minimum safety distance between the path and the power transmission line is not less than 5 meters, and the minimum safety distance between the path and the tower is not less than 10 meters. At the same time, the path needs to avoid high-risk areas such as bird habitats and construction areas. The final output path data includes a series of key waypoint coordinates, heights, speeds and turning angle information, and each waypoint is spaced 50 meters apart; third, start the multi-sensor collaborative work of environment perception. When the unmanned aerial vehicle flies along the initial path, the laser radar and visible light camera work in a "master-slave cooperation" mode. The laser radar scans the front 50 meters of airspace at a sampling frequency of 100 kHz. When a suspected obstacle is detected, the visible light camera is triggered to automatically increase the frame rate to 30 fps and enlarge the target area. At the same time, the infrared thermal imager (newly added module) detects the temperature of the obstacle to distinguish between living obstacles (such as birds) and non-living obstacles (such as plastic bags); fourth, perform obstacle classification and priority division. After clustering the laser radar point cloud data, combined with the visual recognition result and infrared temperature data, the obstacles are divided into three levels: first level (extremely high risk) for flying birds, second level (medium risk) for temporary construction machinery, and third level (low risk) for static branches. The Kalman filter is used to predict the future 5-second motion trajectory of the first level obstacle, and the polynomial fitting is used to predict the motion trend of the second level obstacle; fifth, realize real-time optimization of dynamic trajectory. After superimposing the comprehensive obstacle map and the three-dimensional electronic map, if a first level obstacle risk is detected, the RRT* algorithm is used for emergency path re-planning, and the planning time is controlled within 200 ms to ensure that the unmanned aerial vehicle has enough time to avoid. If a second or third level obstacle risk is detected, the "local fine tuning" strategy is adopted to adjust the heading angle based on the original path to avoid the increase of energy consumption caused by large-scale flight. The newly generated updated trajectory needs to meet the unmanned aerial vehicle's maneuvering performance limit (maximum turning angle 30° / s, maximum climbing rate 5 m / s) and pass the energy consumption check to ensure that the remaining power is sufficient to reach the charging nest. The updated trajectory is immediately sent to the flight control module for execution.

[0029] When the UAV flies along the initial path, the environmental perception suite is started to realize real-time identification and avoidance of obstacles. The suite contains a laser radar and a visible light camera. The laser radar is of solid type, with appropriate ranging range, accuracy and scanning frequency, and continuously scans the front airspace to obtain raw point cloud data. At the same time, the visible light camera is started, and the image sequence of the front airspace is collected. The raw point cloud data is first denoised, and the statistical filtering algorithm is used to remove outliers. Then, the DBSCAN clustering algorithm is used to cluster the remaining point cloud, and in combination with the preset obstacle size threshold, the obstacle profile in the point cloud data is identified, and the preliminary information containing the position and size of the obstacle is obtained. The image sequence is applied to the YOLOv8 target detection algorithm, which is trained by various sample data collected in the power line scene in the early stage, has high recognition accuracy, and can accurately identify dynamic obstacles in the image, and output the category, position coordinates and motion speed information of the obstacle.

[0030] The preliminary information of the obstacle and the visual recognition result are fused, the Kalman filter is used to predict the motion state of the dynamic obstacle, a comprehensive obstacle map containing the position of the static obstacle, the position and motion trend of the dynamic obstacle is generated, and is updated at a fixed frequency. The map is superimposed with the three-dimensional electronic map called by the on-board database, and the collision detection algorithm is used to judge whether there is potential collision risk in front of the initial path. When the risk is detected, the RRT* algorithm is used for local path re-planning, a new flight path is planned to bypass the obstacle and meet the UAV's maneuvering performance, and is sent to the flight control module as an updated trajectory to realize real-time adjustment of the trajectory.

[0031] When the UAV flies to 10 meters above the charging nest based on the updated trajectory, it enters the precise docking stage. The core innovation of this stage is "multi-modal guidance switching + adaptive pose fine-tuning". The specific steps are as follows: First, start the docking preparation and guidance mode switching. The UAV first adjusts the flight attitude to enter the hovering state, with a hovering accuracy controlled within ±0.2 meters. At the same time, it sends a "docking request" to the charging nest through the LoRa module. After receiving the request, the charging nest automatically switches the guidance mode according to the ambient light intensity - visible light LED guidance is started in daylight, and "visible light + infrared" dual-mode guidance is started at night or in rainy and foggy weather. The infrared guidance uses 8-14 μm band infrared light beads to ensure penetration. Second, perform precise capture and recognition of the guidance pattern. The UAV adjusts the onboard pitch axis so that the high-definition camera and infrared detector are aligned with the center of the charging nest platform at the same time. The center of the optical guidance pattern on the charging nest platform is an AprilTag two-dimensional code with encrypted information (including charging nest ID, interface position, and current voltage state), surrounded by three concentric circular markers in an equilateral triangle distribution. The LED array uses pulsed lighting (frequency 10 Hz) to facilitate the camera to extract the target from the complex background. Third, perform fusion preprocessing of multi-modal data. The high-definition camera continuously captures image streams, and the infrared detector synchronously acquires temperature images. The image processing unit first performs dewarping and noise reduction on the visible light image, and enhancement processing on the infrared image. Then, through image registration algorithm, the two types of images are superimposed to eliminate the perspective deviation. Even in strong light direct or heavy rain weather, the infrared image can clearly locate the guidance pattern. Fourth, perform high-precision calculation of six-degree-of-freedom pose deviation. First, identify the two-dimensional code through the AprilTag algorithm and calculate its pixel coordinates in the image. Then, combine the camera intrinsic and extrinsic parameters to calculate the preliminary pose data. Next, identify the concentric circular markers through Hough circle transformation and correct the preliminary data. Finally, introduce adaptive parameter adjustment of Kalman filter - when the pose deviation is greater than 1 mm, increase the proportional coefficient to speed up convergence; when the deviation is less than 0.5 mm, reduce the proportional coefficient to avoid oscillation. The final calculated lateral deviation, height deviation, and angle deviation have a precision of ±0.05 mm and ±0.005°, respectively. Fifth, start dynamic fine-tuning and docking preparation. Input the pose deviation data into the PID controller, which uses differentiated control strategies for different degrees of freedom: "fast approximation + slow convergence" mode for lateral deviation, "uniform descent + real-time correction" mode for height deviation, and angle deviation adjustment prior to position deviation to ensure that the UAV attitude meets the standard first. During the fine-tuning process, real-time rotor speed feedback data is collected. If the speed of a certain rotor approaches the maximum value, automatically adjust the speed of other rotors for compensation to avoid overloading of a single rotor. When the pose deviation is less than the docking tolerance for 10 consecutive frames, the UAV sends a "docking ready" signal to the charging nest and waits for confirmation from the charging nest.

[0032] The high-definition camera continuously captures an image stream containing an optical guide pattern and transmits it to an on-board image processing unit. The image processing unit first pre-processes the image, then uses a special algorithm to identify the two-dimensional code and calculate its position in the image, and simultaneously identifies the surrounding concentric circle markers through Hough circle transformation, and calculates the pose deviation data between the UAV charging interface and the charging nest charging interface by combining the binocular vision principle, including the horizontal two-direction lateral deviation, the vertical height deviation, and the angular deviation around the three axes, a total of six degrees of freedom deviation data, to ensure that the deviation detection accuracy meets the docking requirements.

[0033] The pose deviation data is input to a proportional-integral-derivative controller, which uses an incremental algorithm, where each coefficient is determined by offline tuning, and parameters are set for different degrees of freedom of the deviation. The controller generates fine-tuning control instructions based on the deviation data, which are transmitted to the electronic speed controller system of the UAV through a special bus, controlling the rotation speed of each rotor and the deflection angle of the rudder, allowing the UAV to make fine movements on the hovering reference, gradually reducing the pose deviation. When the pose deviation detected by the high-definition camera in multiple consecutive frames is less than the system's set docking tolerance, the controller outputs a descent command, controlling the UAV to slowly descend along the vertical direction until the charging male head at the bottom of the UAV is tightly inserted into the charging female socket on the charging nest, completing the physical connection. At this time, the pressure sensor at the charging interface feeds back the pressure signal, confirming that the connection is in place.

[0034] The charging process is started after the physical connection is completed. The core innovation of this stage is "safe handshake + dynamic charging curve + fault prediction". The specific steps are as follows: first, a multi-round safe handshake protocol is executed. After detecting the confirmation signal of the charging interface pressure sensor, the charging nest master controller first sends a "connection verification" signal through the communication loop. The unmanned aerial vehicle power management unit responds and feeds back the interface contact resistance detection data (calculated through a small current test). If the contact resistance is greater than the threshold, the charging nest sends a "redocking retry" signal. If the contact resistance is qualified, the charging nest master controller sends a formal handshake signal, including the maximum output power of the charging nest, the safety protection threshold and other information. Second, the precise interaction of battery information is realized. The unmanned aerial vehicle power management unit responds to the handshake signal within 100 ms. In addition to the type, voltage and remaining capacity of the battery information, the key data such as the number of battery cycles, historical fault records and the current maximum temperature of the battery are added. If the battery cycle number exceeds 1000 times or there is a history of overcharging, the maximum allowed charging current is automatically reduced by 20%. Third, a dynamic charging curve is constructed. The charging nest master controller calls the basic charging curve from the internal memory according to the received battery information, and dynamically corrects the curve through real-time battery temperature. When the battery temperature is lower than 0℃, the constant current charging phase time is prolonged and the initial charging current is reduced. When the temperature is higher than 45℃, the constant voltage charging mode is started immediately, and "charging rate self-learning" is introduced. If the battery is normal in the past 3 times of charging, the constant current stage current can be increased by 10% this time. If there is a record of rapid temperature rise once, the current is reduced by 15%. Fourth, real-time monitoring and early warning of the charging process is started. The charging nest monitors the charging current, voltage and battery temperature at a frequency of 10 Hz. The unmanned aerial vehicle power management unit synchronously monitors the voltage balance of the battery cells. Both sides exchange monitoring data through CAN bus every 500 ms to establish a "double early warning mechanism". When the charging current fluctuates more than ±10%, the charging nest immediately reduces the current and sends a warning. When the voltage rise rate of a certain battery cell exceeds the threshold, the unmanned aerial vehicle immediately feeds back, and the charging nest suspends charging for troubleshooting. Fifth, dynamic adjustment of the charging process is performed. If the actual charging parameters deviate from the curve by more than ±5%, the charging nest adjusts the PWM duty cycle of the DC-DC power supply module through the PID algorithm, and adjusts the speed of the cooling fan in real time according to the change of the battery temperature to ensure that the internal temperature of the charging nest does not exceed 50℃. When thunderstorm weather is detected (through the meteorological sensor carried by the charging nest), the charging current is automatically reduced to 50% of the standard value, and the insulation monitoring is strengthened.

[0035] The charging nest master controller calls a matching charging curve from the internal memory according to the received battery information, the curve being a multi-stage charging curve based on battery characteristics, including a constant current charging stage, a constant voltage charging stage and a trickle charging stage. The master controller sets the initial charging voltage and charging current according to the charging curve, controls the power module inside the charging nest to start output, and uses the constant current charging mode to charge the drone battery. During the charging process, the current sensor and voltage sensor inside the charging nest monitor the actual charging parameters in the charging loop in real time at a fixed frequency, compare the monitoring data with the corresponding expected values on the charging curve, and if there is a deviation, adjust the output of the power module to make the actual charging parameters always track the charging curve, ensuring the stability of the charging process.

[0036] In the charging process, the core innovation of battery equalization control is "adaptive equalization strategy + efficiency optimization". The specific steps are as follows: first, start multi-dimensional equalization state monitoring. The unmanned aerial vehicle power management unit collects the voltage data of each battery through a special battery monitoring chip at a frequency of 5Hz, and collects the temperature data of each battery through a thermistor, constructs a "voltage-temperature-time" three-dimensional monitoring matrix, and avoids the limitation of judging the equalization state only by voltage. If the voltage of a certain battery is high but the temperature is low, it may be a measurement error rather than true equalization; second, execute dynamic adjustment of equalization threshold. The two-level safety threshold preset by the system introduces temperature correction: at 25°C standard environment, the first safety threshold is 200mV, and the second safety threshold is 50mV; when the temperature is lower than 0°C, the first threshold is expanded to 250mV to avoid misjudgment caused by battery voltage fluctuation at low temperature; when the temperature is higher than 45°C, the first threshold is reduced to 150mV to start equalization in advance to prevent thermal runaway; third, realize intelligent switching of equalization mode. After the charging nest main controller receives the equalization request, it first selects the equalization mode according to the voltage difference: if the difference is less than 300mV, use "small current continuous equalization" (transfer current 0.1A) to avoid large current impact; if the difference is greater than 300mV, use "pulse equalization" (0.5A current for 1 second, pause for 2 seconds) to speed up the equalization speed, and adjust the mode through the battery temperature data - the battery whose temperature exceeds 40°C is prohibited from discharging large current; fourth, execute closed-loop control of equalization process. The equalization module collects the voltage of each battery every 1 second, adjusts the transfer current size through difference feedback, and if the voltage of a certain battery decreases too fast, the current of the corresponding conversion circuit is immediately reduced. When the voltage difference decreases to the second safety threshold, instead of stopping equalization immediately, it enters the "maintenance stage" to continue small current equalization for 30 seconds to ensure stable equalization effect, and then closes the equalization module and restores the main charging loop. When restoring, use "gradient boost" method to avoid impact on the battery caused by sudden change of charging current; fifth, record equalization data for optimization. The starting time, difference change, and transferred power of each equalization process are stored in the charging nest local database and uploaded to the system server regularly to update the equalization algorithm parameters. For batteries that equalize frequently, the system automatically marks them as "to be maintained" to remind the maintenance personnel to check.

[0037] The charging nest master controller controls the relay of the main charging circuit to be disconnected immediately after receiving the balancing request, suspends the main charging process, and starts the independent balancing module. The balancing module adopts an energy transfer type balancing scheme and contains multiple conversion circuits, which can transfer the charge of the battery cell with higher voltage to the battery cell with lower voltage. The balancing module determines the battery cell that needs to be adjusted by collecting the voltage data of each battery cell, controls the corresponding conversion circuit to work, and controls the transfer current in the safe range. In the balancing process, the voltage of each battery cell is monitored in real time, and when the voltage difference is lower than the second safety threshold, the balancing module is closed, the relay of the main charging circuit is re-closed, and the main charging process is resumed, so that the voltages of the battery cells in the battery pack are kept consistent and the cycle life of the battery is prolonged.

[0038] The termination of the charging process and the subsequent operation core innovation lies in "multi-condition confirmation + task seamless connection", the specific steps are as follows: first, start the multi-dimensional judgment of the charging completion, the charging nest main controller judges the battery state through three ways at the same time: current judgment (the charging current stabilizes at a small current value for 30 seconds), voltage judgment (the battery voltage reaches the rated voltage and the fluctuation is less than 0.1V within 10 seconds), time judgment (the charging time reaches 120% of the theoretical full charging time of the battery), any two of the three conditions can be judged as full, avoiding the misjudgment of a single condition - such as the battery aging leading to the early decline of the current but not full; second, execute the safe termination process, the charging nest main controller first sends a "pre-termination" signal to the unmanned aerial vehicle power management unit, after the unmanned aerial vehicle feedbacks the current state of the battery, the main controller controls the power module to gradually reduce the output voltage to 0V, and then cuts off the charging circuit, avoiding the generation of electric arc caused by voltage surge, and sends a charging report containing the charged power, charging time, and balancing times to the unmanned aerial vehicle; third, realize the cooperative control of unlocking and taking off, after receiving the termination signal, the unmanned aerial vehicle power management unit controls the on-board relay to disconnect the charging circuit, and feedbacks the "circuit disconnection confirmation" to the charging nest, after receiving it, the charging nest main controller controls the electromagnetic lock device to be de-energized and demagnetized, and detects whether the unmanned aerial vehicle landing gear is still in the locked area through the infrared sensor, and sends a "take-off permission" signal after confirming the unlocking; fourth, execute the task breakpoint recovery preparation, after detecting that the mechanical lock is released, the unmanned aerial vehicle controls the rotor to generate lift at low speed, slowly takes off from the charging nest platform, flies to the 5-meter standby area above and hovers, and sends a "charging completion" signal to the system server through the 5G module, the server immediately pushes the breakpoint information of the unfinished inspection task of the unmanned aerial vehicle (such as having inspected the 12th tower, the next step is to inspect the 13th tower), including the key inspection area of the section, historical fault point and other data; fifth, start the pre-flight self-check, the unmanned aerial vehicle completes the rapid self-check of the battery state, sensor working state and communication link quality in the standby area, if all indicators are normal, it automatically plans the shortest path from the standby area to the task breakpoint, and waits for the server to send the take-off instruction; if there is an abnormality (such as a sensor failure), it immediately sends a "fault alarm" to the server, and waits for further instructions, the whole process does not need manual intervention, realizing the seamless connection of the inspection task.

[0039] After receiving the charging termination signal, the power management unit of the UAV controls the on-board relay to disconnect the charging circuit, avoiding overcharging of the battery. The main controller of the charging nest then controls the internal electromagnetic locking device to be powered off and demagnetized, releasing the mechanical locking of the UAV landing gear. After detecting the release of the mechanical locking through the sensor on the landing gear, the UAV generates lift by controlling the rotors through the flight control module, autonomously takes off from the charging nest platform, flies to the standby area above the charging nest and hovers, and sends a charging completion signal to the system server through the communication module, waiting for the next task instruction. Thus, the entire autonomous charging process is completed.

[0040] Embodiment 2 As shown in Figure 2 The autonomous charging system for power transmission line UAVs disclosed in this embodiment is used to implement the autonomous charging method described in Embodiment 1. The system is deployed in the corridor of high-voltage power transmission lines, covering a certain length of line area, and a plurality of charging nest sites are provided to form a grid charging network, which can provide autonomous charging services for multiple UAVs at the same time. The system as a whole adopts a three-level architecture of "UAV subsystem - charging nest subsystem - system server", and each subsystem realizes data interaction through a dedicated communication network to ensure that the data transmission rate and delay meet the requirements, and has data encryption and fault tolerance mechanisms to ensure the safety and reliability of system operation.

[0041] The unmanned aerial vehicle subsystem is the core execution unit for realizing autonomous inspection and autonomous charging. Each unmanned aerial vehicle subsystem corresponds to a unique device identifier, facilitating unified management by the system server. The structural design and method of the subsystem are deeply matched, and the specific module details are as follows: The state monitoring module serves as the "perception core" and is provided with a new time series data processing unit and an error correction unit. The power supply monitoring unit is provided with a special chip supporting 16-channel synchronous sampling. The battery monitoring chip can simultaneously collect data of 12 battery cells. The positioning and navigation unit is integrated with a multi-mode satellite navigation module supporting a signal strength weighting algorithm. The visual odometry module is provided with a transmission line tower feature library. The feature point matching is realized at a high speed through a hardware acceleration unit. The fusion algorithm unit is provided with an FPGA chip for realizing parallel operation of extended Kalman filtering, ensuring that the positioning data update frequency reaches 100 Hz. The high-performance microcontroller of the charging decision module is provided with a neural network acceleration unit for quickly reasoning the air route wind resistance model. An environment correction submodule and a historical decision database are newly added. The environment correction submodule can receive temperature and humidity sensor data to dynamically adjust the energy consumption coefficient. The historical decision database stores the results and feedback data of the past 100 charging decisions, and the threshold calculation parameters are optimized through a simple iterative algorithm. The flight control and navigation module is provided with a dual-core architecture. The main core is responsible for attitude control, and the slave core is responsible for path planning. The path planning chip is provided with a special obstacle priority processing unit, which can classify obstacles in real time and output an optimized path. The laser radar and visible light camera of the environment perception module are triggered synchronously through hardware. The data fusion unit adopts a heterogeneous computing architecture, ensuring that the delay of obstacle identification and map generation is less than 50 ms. The precise docking control module is provided with a multi-modal guidance switching unit and an adaptive PID control unit. The multi-modal guidance switching unit can automatically select visible light or infrared guidance mode according to light sensor data. The adaptive PID control unit realizes dynamic adjustment of filtering parameters through a software algorithm. The 5G module of the onboard communication module supports network slicing function, and allocates a high-priority transmission channel for charging-related data. The LoRa module supports adaptive communication rate, and automatically adjusts the transmission frequency according to the distance, ensuring stable communication during docking.

[0042] The charging decision module adopts a high-performance microcontroller as the core. The microcontroller has sufficient storage capacity and can store the air route wind resistance model and various threshold parameters. The microcontroller receives state data output by the state monitoring module in real time through a special bus, calls the air route wind resistance model to predict the return wind resistance, calculates the estimated compensation power and adaptive threshold, generates a charging request signal by continuously comparing the remaining power and the threshold, and has a fault diagnosis function. When an abnormal power is detected, the charging request signal can be forcibly generated to ensure the safety of the unmanned aerial vehicle.

[0043] The flight control and navigation module is the "control center" of the UAV, which is composed of a flight control mainboard, a path planning chip and an actuator. The flight control mainboard integrates various sensors, supports basic flight functions such as attitude control and hovering control, and ensures control accuracy. The path planning chip has strong parallel computing capability, can run initial path planning algorithm and local path re-planning algorithm, combines with the obstacle data input by the environment perception module, generates and updates flight trajectory in real time, and sends control instructions to the actuator to control the flight state of the UAV. The environment perception module, as the "eyes" of the module, includes a solid-state laser radar, a visible light camera and a data fusion unit. The environment data collected by the laser radar and the camera are processed by the data fusion unit to generate a comprehensive obstacle map, which provides a basis for path planning.

[0044] The precision docking control module is responsible for realizing the precise docking of the UAV and the charging nest, which is composed of an optical recognition unit, a pose solving unit and a flight fine-tuning unit. The optical recognition unit includes a high-definition camera on the belly and an image preprocessing chip. The camera has automatic focusing and exposure adjustment functions, can clearly capture the optical guide pattern of the charging nest under different lighting conditions, and the image preprocessing chip optimizes the image processing to improve the subsequent recognition accuracy. The pose solving unit uses a special chip to identify the guide pattern through a special algorithm, and calculates the six-degree-of-freedom pose deviation data combined with the calibration parameters to ensure the calculation speed and accuracy. The flight fine-tuning unit converts the pose deviation data into control instructions based on the PID control algorithm, and transmits them to the flight control mainboard through the high-speed bus to realize the fine attitude adjustment of the UAV and ensure the docking accuracy.

[0045] The onboard communication module adopts a "5G+LoRa" dual-mode communication scheme. The 5G module supports independent networking mode, has the characteristics of high speed and low delay, and is used to transmit state data, charging requests, high-definition images and other large-capacity data. The LoRa module has long-distance communication capability and is used to realize short-distance data interaction with the charging nest in areas where 5G signal coverage is poor, ensuring the continuity of communication. The module supports high-strength data encryption algorithm to prevent data from being stolen or tampered with.

[0046] The charging nest system is the "energy supply station" of the unmanned aerial vehicle. The structural design provides hardware support for the creative steps of the method. Each charging nest is fixedly installed on a special support of the power transmission line tower. The support is equipped with a horizontal adjustment device at the bottom to ensure that the horizontal error of the docking platform is less than 0.1°. The nest body structure adopts a double-layer design. The outer layer is a wind and rainproof shell, and the inner layer is a heat insulation layer. A temperature control system is installed inside to ensure that the core module works stably at a temperature of 20-40℃. The specific module details are as follows: The charging female socket of the charging interface module is provided with a contact resistance detection circuit. The connection state of the interface is detected in real time through a small current of 10mA. The conical guide column of the guide mechanism is made of elastic material and can absorb slight impact during docking. The electromagnetic suction disc of the locking device is provided with a Hall sensor for detecting the adsorption state. The nest end guiding module is provided with a guiding mode switching unit and a weather adapting unit. The guiding mode switching unit can select the guiding mode according to the data of the light sensor and the infrared detector. The weather adapting unit receives the data of the temperature and humidity sensor and the rainfall sensor carried by the charging nest. In bad weather, the LED brightness and the infrared output power are automatically enhanced. The power input unit of the nest end power management module supports photovoltaic and mains dual power supply and has an automatic switching function. The DC-DC conversion unit adopts a digital control architecture and can quickly adjust the output parameters through software. The charging control unit is provided with a dynamic charging curve library and a safety warning unit. The dynamic charging curve library stores the charging parameters of 10 kinds of common unmanned aerial vehicle batteries. The safety warning unit can monitor 12 kinds of fault types such as overcurrent, overvoltage and overtemperature of the charging loop. The safety protection module is provided with a lightning protection counter and a fault positioning unit. The lightning protection counter records the number of lightning strikes for reference. The fault positioning unit can quickly locate the fault module and send alarm information. An emergency power-off device is also provided to remotely cut off the power supply in an emergency. An environmental monitoring unit is also installed inside the nest body, which includes temperature and humidity, light, rainfall and wind speed sensors. The data is uploaded to the system server in real time to provide environmental data support for the charging decision and docking control of the unmanned aerial vehicle.

[0047] The charging interface module is the key part of energy transmission, which includes the charging female socket, the guide mechanism and the locking device. The charging female socket adopts a waterproof aviation plug, which is provided with sufficient conductive terminals inside for charging loop and communication loop. The terminals are specially treated to reduce the contact resistance and ensure stable large current transmission. The guide mechanism is composed of multiple conical guide columns, which cooperate with the guide holes of the unmanned aerial vehicle charging male head to realize coarse positioning during docking. The locking device adopts an electromagnetic suction disc structure, which generates sufficient adsorption force after being powered on to firmly fix the unmanned aerial vehicle landing gear and prevent displacement of the unmanned aerial vehicle during charging due to external force.

[0048] The nest end guiding module adopts an optical guiding mode and is composed of a high-brightness LED array, a driving circuit and a protective glass. The LED array is fixedly installed in a center region of a nest body docking platform in a specific geometric pattern to form an optical guiding mark for being recognized by the unmanned aerial vehicle. The LED is of a high-brightness patch type, and the driving circuit adopts a constant current driving mode to ensure the stability of the brightness of the LED. The protective glass is made of high-strength material and has an anti-reflection treatment on the surface, which not only protects the LED array from being damaged, but also does not affect the optical guiding effect. When the unmanned aerial vehicle approaches the charging nest, the nest end guiding module is automatically started, and the LED array continuously emits light to provide a clear guiding mark for the unmanned aerial vehicle.

[0049] The nest end communication module corresponds to the on-board communication module of the unmanned aerial vehicle subsystem and also adopts a "5G+LoRa" dual-mode scheme. The 5G module is used for long-distance data interaction with the system server to transmit information such as the charging nest state and charging data. The LoRa module is used for short-distance communication with the approaching unmanned aerial vehicle to transmit real-time data such as guiding signals and handshake signals. The module has an automatic switching function. When it is detected that the unmanned aerial vehicle enters a specific range, it is automatically switched to the appropriate communication mode to ensure the real-time nature of the communication during the docking process.

[0050] The nest end power management module is the "energy management center" of the charging nest and is composed of a power input unit, a DC-DC conversion unit, a charging control unit and an equalization module. The power input unit supports multiple input modes and has overvoltage and undervoltage protection functions. The DC-DC conversion unit adopts a high-efficiency topology structure and has a high conversion efficiency, which can convert the input voltage into an adjustable DC voltage that meets the demand of the battery of the unmanned aerial vehicle, and the output current range covers the charging demand of the batteries of unmanned aerial vehicles with different capacities. The charging control unit adopts a high-performance microcontroller, which is responsible for communication with the power management unit of the unmanned aerial vehicle, executes the charging handshake protocol, calls the charging curve, controls the charging process, and monitors the charging parameters in real time through the current sensor and the voltage sensor to realize closed-loop control of the charging process. The equalization module adopts an energy transfer type structure and can actively equalize the unbalanced battery cells of the unmanned aerial vehicle to ensure charging safety.

[0051] The safety protection module is a guarantee for the safe operation of the charging nest and includes functions such as overcurrent protection, overvoltage protection, short circuit protection, overtemperature protection and lightning protection. The overcurrent protection is realized by a fuse and an electronic circuit breaker, which immediately trips when the charging current exceeds the safety value. The overvoltage protection is triggered when the charging voltage exceeds the threshold value to cut off the power output. The short circuit protection uses a fast-response element to quickly cut off the circuit when a short circuit fault is detected. The overtemperature protection uses a temperature sensor to monitor the internal temperature of the nest body. When the temperature exceeds the warning value, the cooling device is started. If the temperature continues to rise, the charging is stopped. The lightning protection uses a special lightning arrester that can withstand standard waveform lightning surges to protect the internal electronic equipment from being damaged.

[0052] The system server as the "command center" of the whole system, its software function design fully supports the creative implementation of the method, the industrial server adopts dual machine hot standby architecture, ensures that the system has no single point of failure, the storage device adopts SSD and HDD hybrid storage scheme, SSD is used to store real-time data and core algorithm, HDD is used to store historical data and video image, the core function modules of the system management software are as follows: the charging nest management module adds charging nest state evaluation unit and resource scheduling optimization unit, the state evaluation unit gives a health score to the charging nest through multi-dimensional data (charging success rate, fault times, equipment aging degree), the resource scheduling optimization unit uses genetic algorithm to allocate the optimal charging nest for the unmanned aerial vehicle, and considers factors such as distance, waiting time, charging nest health status; the unmanned aerial vehicle management module adds task breakpoint management unit and performance analysis unit, the task breakpoint management unit stores the inspection task progress and breakpoint information of the unmanned aerial vehicle, supports seamless task recovery, and the performance analysis unit analyzes the battery attenuation and sensor working state through the state data uploaded by the unmanned aerial vehicle, and provides basis for operation and maintenance; the task scheduling module adopts distributed scheduling architecture, supports task allocation of multiple unmanned aerial vehicles working at the same time, adds emergency scheduling unit, when a certain unmanned aerial vehicle fails, automatically schedules adjacent unmanned aerial vehicles to take over its task; the data statistics module adds algorithm optimization unit, regularly updates the built-in wind resistance model, charging curve, balancing algorithm and other parameters of the system by analyzing historical charging data, docking data and fault data, realizes the continuous optimization of system performance; the remote monitoring module supports 3D visual display, can display the information of unmanned aerial vehicle position, charging nest state, power transmission line terrain in real time, the fault diagnosis module adopts the combination of expert system and machine learning, can quickly locate the fault reason and give the solution, the operation and maintenance personnel can realize the remote management and control of the whole system through the web page or mobile APP.

[0053] The system server connects each charging nest subsystem and unmanned aerial vehicle subsystem through a dedicated communication network, receives the data uploaded by each device in real time, including the position, state and resource information of the charging nest, and the position, power and task progress information of the unmanned aerial vehicle, arranges these information into charging nest network data, stores it in the database and updates it at fixed frequency. When the server receives the charging request signal sent by the unmanned aerial vehicle, based on the charging nest network data, the optimal charging nest is selected by using the optimization algorithm, and its position information and state information are sent to the unmanned aerial vehicle, providing data support for the path planning of the unmanned aerial vehicle.

[0054] At the same time, the system server has a task scheduling function, which can dynamically allocate inspection tasks according to the charging completion and task progress of each unmanned aerial vehicle, to ensure the continuity of the entire transmission line inspection work. The data statistics function can automatically record the operation data of each unmanned aerial vehicle and charging nest, and generate various reports to provide decision basis for operation and maintenance personnel. In addition, the server also has remote monitoring and fault diagnosis functions, and operation and maintenance personnel can view the operation state of each device in real time through the web page. When a device fault is detected, the system automatically sends an alarm message to the mobile terminal of the operation and maintenance personnel, facilitating timely maintenance.

[0055] In the actual operation process of the whole system, each subsystem realizes the full-process automation and performance improvement of the unmanned aerial vehicle self-charging of the transmission line through the cooperative work of the creative technology. In the inspection process of the unmanned aerial vehicle, the precise perception of power and position is realized through the state monitoring of time sequence fusion, the charging decision based on multi-factor coupling ensures the accurate judgment of charging opportunity, the path planning based on priority avoidance reduces energy consumption while ensuring safety, the precise docking success rate is improved to 99.9% through multi-modal guidance, the battery life is extended by more than 20% through the charging process of dynamic curve and adaptive balancing strategy, and the seamless task connection after charging improves the inspection efficiency by 30%. The charging nest subsystem provides stable and reliable energy supply for the unmanned aerial vehicle through double power supply, dynamic guidance and multiple safety protection, and its environmental monitoring capability provides key support for the autonomous decision of the unmanned aerial vehicle. The system server realizes the efficient operation and continuous evolution of the entire charging network through intelligent scheduling, performance analysis and algorithm optimization. Compared with the traditional technology, the creative design of the system effectively solves the core pain points of low precision, poor adaptability, insufficient safety and low efficiency, greatly reduces the cost of manual intervention, improves the efficiency and safety of transmission line inspection, and provides reliable technical support for intelligent operation and maintenance of high-voltage transmission lines.

[0056] From the technical common sense, the present application can be realized by other embodiments without departing from the spirit or essential characteristics thereof. Therefore, the above disclosed embodiments are merely illustrative in all aspects and are not the only ones. All changes within the scope of the present application or within the scope equivalent to the present application are included in the present application.

Claims

1. A power transmission line drone autonomous charging method, characterized in that, The method comprises the following steps: Real-time monitoring of the power level and location information of the unmanned aerial vehicle to obtain state data; Based on the state data, when the power level is lower than a preset threshold, a charging request signal is generated; In response to the charging request signal, a navigation path is calculated according to the location information and the location of the charging nest to obtain path data; The unmanned aerial vehicle is controlled to fly to the charging nest according to the path data, and the flight trajectory is adjusted in real time during the flight process by using environmental perception data to obtain an updated trajectory; When the unmanned aerial vehicle reaches the charging nest based on the updated trajectory, a docking operation is performed to physically connect the unmanned aerial vehicle with the charging interface of the charging nest; After the connection is completed, a charging process is started to charge the battery of the unmanned aerial vehicle.

2. The method of claim 1, wherein, The step of real-time monitoring of the power level and location information of the unmanned aerial vehicle to obtain state data specifically comprises: The power management unit on board the unmanned aerial vehicle collects voltage data and current data of the battery at a preset sampling frequency; The voltage data and current data are fused and calculated by using a pre-established battery model to obtain an accurate real-time remaining power estimate; Synchronously, the navigation system of the unmanned aerial vehicle receives global satellite navigation system signals, reads angular velocity and acceleration data output by the inertial measurement unit, and obtains displacement data calculated by the visual odometry; A Kalman filtering algorithm is used to fuse the global satellite navigation system signals, inertial measurement unit data and visual odometry data to eliminate errors of a single sensor and obtain high-precision fused positioning data; The real-time remaining power estimate and the high-precision fused positioning data are combined to form the state data.

3. The method of claim 2, wherein, The step of generating a charging request signal when the power level is lower than a preset threshold based on the state data specifically comprises: A flight route wind resistance model pre-stored in the memory of the unmanned aerial vehicle is called, and the flight route wind resistance model stores historical wind speed and direction data under different flight routes; The high-precision fused positioning data are input into the flight route wind resistance model to predict the wind resistance that the unmanned aerial vehicle may encounter on the way back to the charging nest from the current location, and prediction wind resistance data are obtained; Based on the prediction wind resistance data, an estimated compensation power required to overcome the wind resistance is calculated, the preset threshold is added to the estimated compensation power to obtain an adaptive threshold that is dynamically adjusted; The real-time remaining power estimate and the adaptive threshold are continuously compared, and when the real-time remaining power estimate is lower than the adaptive threshold, the charging request signal is triggered to be generated.

4. The method of claim 1, wherein, The step of controlling the unmanned aerial vehicle to fly to the charging nest according to the path data and adjusting the flight trajectory in real time during the flight process by using environmental perception data to obtain an updated trajectory specifically comprises: When the unmanned aerial vehicle flies along the path data, a laser radar sensor carried thereon is started to scan the front airspace to obtain original point cloud data; At the same time, a visible light camera is started to collect image sequences of the front airspace; the original point cloud data are clustered and classified to identify the obstacle profile in the point cloud data to obtain preliminary information of the obstacle; applying a deep learning object detection algorithm to the image sequence to identify birds, temporary building dynamic obstacles in the image sequence, to obtain a visual recognition result; fusing the obstacle preliminary information and the visual recognition result to confirm the position, speed and motion trend of the dynamic obstacle, and generating a comprehensive obstacle map; superimposing the comprehensive obstacle map and a three-dimensional electronic map called from an on-board database to identify potential collision risks in front of the path data; based on the position and motion trend of the potential collision risks, using a local path re-planning algorithm to calculate a new flight path that can safely avoid all dynamic and static obstacles as the updated trajectory.

5. The method of claim 1, wherein, When the UAV reaches the charging nest based on the updated trajectory, the docking operation is performed to physically connect the UAV and the charging interface of the charging nest, specifically including: After the UAV flies directly above the charging nest and enters a hovering state, the on-board pitch axis is adjusted to align the high-definition camera installed on the belly with a specific optical guide pattern preset on the platform of the charging nest; a continuous image stream containing the specific optical guide pattern is captured through the high-definition camera; a computer vision algorithm is applied to each frame of image to calculate the six-degree-of-freedom pose deviation data of the lateral deviation in two horizontal directions, the height deviation in the vertical direction, and the angle deviation around three axes between the charging interface of the UAV and the charging interface of the charging nest; the six-degree-of-freedom pose deviation data is input into a proportional-integral-derivative controller to generate fine-tuning control instructions for correcting the position and attitude of the UAV; the fine-tuning control instructions are executed to control the rotor speed and rudder surface of the UAV to make the UAV move finely on the hovering reference, gradually reducing the six-degree-of-freedom pose deviation; when the pose deviation detected by the high-definition camera is less than the system set docking tolerance, the UAV is controlled to slowly descend in the vertical direction until the charging male head at the bottom of the UAV is tightly inserted into the charging female seat on the charging nest, and the physical connection is completed.

6. The method of claim 1, wherein, After the connection is completed, the charging process is started to charge the battery of the UAV, specifically including: after detecting that the physical connection is completed, the main controller of the charging nest sends a handshake signal to the power management unit of the UAV to start a charging handshake protocol; the power management unit of the UAV responds to the handshake signal to feed back the type information, current voltage information and maximum allowed charging current information of the battery; the main controller of the charging nest calls a matching charging curve from the internal memory according to the type information, current voltage information and maximum allowed charging current information of the battery, and sets the initial charging voltage and charging current according to the charging curve; the constant current charging mode is used to start charging the battery of the UAV, and during the charging process, the actual charging current and actual charging voltage in the charging loop are monitored in real time through the current sensor and voltage sensor inside the charging nest; The actual charging current and actual charging voltage are compared with the expected values ​​corresponding to the charging curve. If there is a deviation, the output of the charging nest power module is adjusted to ensure that the actual charging current and actual charging voltage always track the charging curve.

7. The method of claim 6, wherein, The initiation of the charging process, after the step of charging the drone battery, specifically includes: During the charging process, the drone's power management unit continuously monitors the voltage of multiple cells inside the battery to obtain a set of cell voltage data. Calculate the difference between the maximum and minimum voltage values ​​in the battery cell voltage data set to obtain the voltage range; When the voltage difference exceeds the first safety threshold, the UAV power management unit determines that there is an imbalance in the battery and sends an active balancing request to the charging nest main controller through the charging communication link. In response to the active balancing request, the charging nest main controller suspends the main charging circuit and starts an independent balancing module. This balancing module applies a discharge load to the higher voltage cells or transfers charge from the higher voltage cells to the lower voltage cells until the voltage difference is lower than the second safety threshold. Then, the balancing module is shut down and the charging of the main charging circuit is resumed.

8. The method of claim 1, wherein, After the step of initiating the charging process to charge the drone battery after the connection is completed, the method further includes: The charging nest main controller starts a timer when charging begins to record the charging duration. Meanwhile, the charging nest's current sensor continuously monitors the charging current. When it detects that the charging current is continuously decreasing and stabilizes at a preset small current value, the charging nest's main controller determines that the battery is fully charged. The charging nest main controller sends a charging termination signal to the drone power management unit and controls the charging nest power module to stop outputting power. Upon receiving a charging termination signal, the drone's power management unit controls the onboard relay to disconnect the charging circuit. The main controller of the charging station de-energizes and demagnetizes the electromagnetic locking device inside, releasing the mechanical lock on the drone's landing gear. After detecting that the mechanical lock has been released, the drone controls its rotor to generate lift, autonomously takes off from the charging station platform, and hovers in the waiting area above the charging station, waiting for the next mission command.

9. A power line unmanned aerial vehicle autonomous charging system for implementing the power line unmanned aerial vehicle autonomous charging method of any one of claims 1 to 8, characterized in that, include: Unmanned aerial vehicle (UAV) subsystem, charging station subsystem, and system server; The unmanned aerial vehicle (UAV) subsystem includes: The status monitoring module is used to monitor the drone's battery level and location information in real time to obtain status data; The charging decision module is used to generate a charging request signal based on the status data when the power level is lower than a preset threshold. The flight control and navigation module is used to respond to the charging request signal, calculate the navigation path based on the location information and the location of the charging station, obtain path data, control the UAV to fly according to the path data, and adjust the flight trajectory in real time using environmental perception data during the flight to obtain an updated trajectory. The precision docking control module is used to perform a docking operation when the UAV arrives at the charging station based on the updated trajectory, so that the UAV and the charging station's charging interface are physically connected. An airborne communication module is used for data interaction with the charging subsystem and the system server; The charging nest system includes: a nest structure; A charging interface module is arranged on the nest structure, and is configured to charge the battery of the UAV after the physical connection is established; A nest end guiding module is configured to generate a guiding signal when the UAV approaches, and to assist the UAV subsystem to complete the precise docking; A nest end communication module is configured to interact with the UAV subsystem and the system server; A nest end power management module is configured to manage the electric energy and control the charging process; The system server is connected with the onboard communication module of the UAV subsystem and the nest end communication module of the charging nest subsystem through a communication network, and is configured to maintain charging nest network data, which at least includes the location, state and resource information of all charging nests in the system, and to provide data support required by the path planning of the flight control and navigation module of the UAV subsystem.

10. The power line drone autonomous charging system of claim 9, wherein, The nest end guiding module of the charging nest subsystem is an optical guiding module, and specifically includes: A high-brightness LED array is fixedly installed on the docking platform of the nest structure in a specific geometric pattern, and constitutes an optical guiding mark for the UAV to recognize; The precise docking control module of the UAV subsystem includes: A high-speed visual recognition unit is configured to capture an image containing the optical guiding mark when the UAV flies to the charging nest; A pose solving unit is connected to the high-speed visual recognition unit, and is configured to process the captured image and solve the pose deviation data of the UAV charging interface and the charging nest charging interface in six degrees of freedom, including the lateral deviation in two horizontal directions, the height deviation in the vertical direction, and the angle deviation around three axes; A flight fine-tuning unit is connected to the pose solving unit, and is configured to convert the pose deviation data in six degrees of freedom into fine-tuning instructions for controlling the thrust of each rotor and the rudder surface of the UAV, and to drive the UAV to move accurately to eliminate the pose deviation until the docking condition is met.

Images