Airship type unmanned aerial vehicle inspection system and inspection method

Abstract

The invention discloses an airship type unmanned aerial vehicle inspection system and inspection method, and belongs to the field of inspection unmanned aerial vehicles. The system comprises a laser radar, an image acquisition device, an airship capsule filled with helium, a standby safety airbag, a communication module and a ground control center, in the method, the ground control center sets an inspection path and issues the inspection path, and the unmanned aerial vehicle flies according to a preset path. During fault detection, the high-definition camera collects images to recognize line cracks and broken strands, the thermal infrared imager monitors the joint temperature, the gas sensor detects abnormal gas, and multi-source data fusion assists in defect positioning. The routing inspection takes a starting point-terminal point connecting line of a power transmission line as a reference, identifies main obstacles and surrounding nodes to perform global planning, performs key routing inspection on a risk area in combination with historical data, and initializes and guides a flight direction through a Q value to reduce invalid exploration. The four-rotor unmanned aerial vehicle solves the problems of short flight time and poor stability of a four-rotor unmanned aerial vehicle, improves the inspection stability and cruising ability, and improves the inspection efficiency of a power transmission line and the reliability of a power system.

Description

Technical Field

[0001] This invention relates to the field of inspection drones, and more particularly to the field of artificial intelligence technology. Specifically, it relates to an airship-type drone inspection system and method. Background Technology

[0002] The airship-type drone inspection system can fly over power transmission lines along a predetermined path. High-definition cameras capture images of the lines, clearly showing any damage or broken strands. Infrared thermal imagers can detect the temperature of wire joints and insulators, enabling timely detection of overheating faults. Gas detection sensors can monitor for abnormal gases generated by line faults around the transmission lines. By transmitting this data in real time, staff can quickly locate the fault point and take appropriate repair measures, greatly improving the efficiency of power transmission line inspections and the speed of fault handling, thus ensuring the stable operation of the power system.

[0003] Airship-type drones rely on helium gas, lighter than air, within an inflatable bladder to generate static buoyancy. This, combined with a propulsion system, enables controllable flight. They utilize lightweight, high-strength materials such as polyester fiber composite membranes and are equipped with an electrically driven vector propulsion system. Precise flight is achieved through auxiliary gasbag adjustment and control surfaces. Some models employ semi-rigid or rigid structures to enhance stability. Therefore, airship-type drones possess exceptionally long loiter times, large payload capacity, extremely low operating noise, and are suitable for continuous high-altitude monitoring missions, making them ideal for extended-duration aerial operations such as power line inspections.

[0004] Therefore, this invention proposes an airship-type unmanned aerial vehicle (UAV) inspection system and method. It utilizes a camera to collect image data, a lidar to measure the distance to power transmission lines while avoiding bird interference, and an inertial measurement unit (IMU) to monitor the operational status of the airship-type UAV. It also features collision avoidance and anti-sinking safety structures. Through a multi-sensor fusion algorithm, it achieves power transmission line distance monitoring and structural integrity detection. By controlling the airship-type UAV to fly at a near-power transmission line altitude, it performs structural integrity detection, achieving efficient, safe, and accurate inspection operations. Summary of the Invention

[0005] The technical problem to be solved by the present invention is to provide an airship-type unmanned aerial vehicle (UAV) inspection system and inspection method, which solves the problems of short flight time and poor stability of existing quadcopter UAVs. By using the operation mode of an airship, the stability of the inspection UAV is enhanced and the endurance of the UAV inspection is improved.

[0006] To solve the above-mentioned technical problems, the technical solution adopted by the invention is as follows: The present invention provides an airship-type unmanned aerial vehicle (UAV) inspection system and inspection method, including a lidar, an image acquisition device, an airship filled with helium to overcome its own weight and take off, a backup safety airbag device, and a communication module for debugging. According to the location of the power transmission line, the inspection path of the UAV is set at the ground control center, generating inspection path information including a basic path and dynamic adjustment space, and the inspection path information is sent to the UAV through the communication module. The UAV takes off and flies to the power transmission line area according to the preset inspection path.

[0007] This invention proposes a fault detection method for airship-type unmanned aerial vehicles (UAVs). During flight, a high-definition camera in the image acquisition device captures images of the power transmission line. An image recognition processor processes and analyzes the acquired images in real time to identify whether there are cracks, broken strands, or other defects on the surface of the power transmission line. An infrared thermal imager monitors abnormal temperature rises at conductor joints, clamps, and other parts in real time, providing early warnings of overheating faults. Gas detection sensors monitor characteristic gases such as ozone and nitrogen oxides that may be generated in the surrounding environment due to electric arcs or overheating, effectively diagnosing potential insulation faults. Through the fusion and real-time transmission of multi-source sensor data, the system can assist maintenance personnel in quickly locating defects and formulating maintenance strategies, significantly improving inspection efficiency and power system reliability.

[0008] This invention proposes an airship-type UAV inspection method. Based on the starting-end line of the power transmission line inspection, it prioritizes the identification of major obstacles and surrounding nodes on the path for global path planning. Combining historical data detected by image recognition processors and gas detection sensors, it conducts inspections on risk areas with potential line damage, guiding the airship to stay in key areas to improve fault detection accuracy. At the same time, it guides the flight direction by initializing the Q value to reduce ineffective exploration by the airship and achieves efficient global path generation. Attached Figure Description

[0009] The present invention will be further described below with reference to the accompanying drawings and embodiments: Figure 1 This is a structural diagram of the airship-type unmanned aerial vehicle (UAV) inspection system of the present invention; Figure 2 This is a flowchart of the airship-type unmanned aerial vehicle inspection system of the present invention; Figure 3 This is a flowchart of the global path planning method of the present invention; Figure 4 This is a flowchart of the local path planning method of the present invention. Detailed Implementation

[0010] To better understand the purpose, system architecture, and functional implementation of this embodiment, the embodiments and features in the embodiments of this application can be combined with each other without conflict. The exemplary embodiments disclosed in this application will be described below with reference to the accompanying drawings, which include specific technical details disclosed in this embodiment to aid understanding; however, these details should be considered exemplary rather than restrictive. Therefore, those skilled in the art should understand that various improvements and adjustments can be made to the embodiments described herein without departing from the scope and core ideas of the invention. Similarly, for clarity, detailed descriptions of well-known technologies, functions, and structures (such as standard image processing algorithms and common communication protocols) are omitted in the following description.

[0011] Example 1 Figure 1 This is a structural diagram of the airship-type unmanned aerial vehicle (UAV) inspection system of the present invention.

[0012] like Figure 1 As shown, the airship-type unmanned aerial vehicle (UAV) inspection system and inspection method 100 may include S110~S140.

[0013] During operation S110, the ground control center imports the location coordinates of the transmission lines and historical fault data.

[0014] In operation S120, the global path planning module uses a 3D environmental map of the pre-built power transmission line obtained by LiDAR.

[0015] In operation S130, the local path planning module conducts key inspections of potentially unexpected risk areas in the global path.

[0016] During operation of S140, multiple sensors identify and collect line fault information, and the fault information is stored in a database to optimize local inspection paths.

[0017] According to an embodiment of the present invention, in operation S110, starting from the workstation of the ground control center, the operation and maintenance personnel import geographical information such as GPS coordinates, tower numbers and line elevation data of the target transmission line of the system, and at the same time call up historical fault data of the line from the historical inspection database to pre-identify high-risk areas where faults or insulation aging have occurred, which together constitute the initial input for path planning.

[0018] In the operation of S120, the global path planning module constructs a 3D environmental map of the power transmission line through lidar scanning, providing an environmental perception basis for global path generation. First, the lidar is installed in an unobstructed position above the front of the airship, ensuring its scanning field of view is not interfered with by the airship's own structure. From the ground control center, the airship is controlled to perform a pre-scan flight along the power transmission line. The lidar emits laser pulses at a frequency of 10Hz, collecting environmental information such as terrain, towers, trees, and buildings around the power transmission line, generating raw point cloud data.

[0019] The original point cloud data is denoised to remove noise points caused by birds, light reflections, etc. A voxel filtering algorithm is used to downsample the data to reduce the amount of data while retaining key environmental features. Based on the processed point cloud data, a Poisson reconstruction algorithm is used to construct a 3D environmental grid map of the transmission line to balance map accuracy and computational efficiency. The location and outline of obstacles such as conductors, towers, tall trees and buildings of the transmission line are marked on the map.

[0020] Based on the coordinates of the starting tower and ending tower of the transmission line imported in step S110, an initial straight path from the starting point to the ending point is generated in the 3D environment map. The collision relationship between the line connecting the starting point and the ending point (sg line) and all obstacles is calculated, and obstacles that intersect with the sg line are identified as main obstacles. Obstacles that do not intersect are secondary obstacles. Determine the formula for the function. As shown in equation (1) below.

[0021] (1) The main obstacle set is .

[0022] The secondary obstacles are grouped as follows .

[0023] High-voltage power towers, large trees spanning directly above the path, and buildings will be identified as... Plants and distant houses located below the flight path on the side of the route are classified as... .

[0024] Node filtering is performed using SP (Surrounding Points), focusing on key nodes. Obstacle planning detour paths reduces computational complexity. For each identified major obstacle... Find its geometric center point As a reference point, with reference point Centered on the obstacle, a cubic space is constructed. All network nodes within this cubic space are selected; these nodes are the surrounding nodes (SPs) of the obstacle. These nodes represent potential pathpoints where the airship can safely bypass the obstacle. Then, all... The corresponding SP node sets are merged to form the critical node network for path planning, and all of them are directly discarded. and not in any By reducing the search space by more than 50% within the network nodes of the SP range, the speed of path planning can be improved.

[0025] The search algorithm is run on the SP node network: On a sparse network graph consisting of all SP nodes, the A* algorithm is used to search for the optimal path from the starting point S to the ending point G. To further improve search efficiency, a Q-value initialization strategy is adopted. Before the search begins, the direction vector from each SP node to the ending point G is calculated, giving the direction pointing to G a higher initial heuristic value Q. This allows the path search algorithm to be more directional during operation, prioritizing exploration towards the target direction and avoiding a large amount of invalid random searching around the starting point, thus further accelerating the convergence process.

[0026] By combining MO-SP filtering with Q-value initialization, minor obstacles (NMO) and irrelevant grid nodes are ignored, and the path search range is concentrated around the most critical main obstacle. This solves the problem of node explosion in 3D space. Q-value initialization gives the algorithm directionality, enabling it to quickly converge to an approximately optimal global path, which meets the real-time requirements of long-distance airship inspection.

[0027] In step S130, high-risk areas are automatically identified based on historical data and real-time sensing data, and refined local flight strategies are dynamically generated. By inputting the historical fault database of the S110 stage, the system automatically marks all historical fault points on the global path. Real-time data from the high-definition camera, gas sensor, and infrared thermal imager are collected as the airship cruises along the global path. When the airship descends from its cruising altitude and smoothly approaches target components such as insulator strings and clamps, it activates the control surfaces and vector thrusters within a safe distance of 5-10 meters to counteract wind disturbances and maintain stable hovering. This allows the high-definition camera and infrared thermal imager to obtain the highest resolution images, capturing anomalies such as minute cracks, localized overheating, and excessive gas concentrations. Even if there are no historical fault records at a given location, the system immediately marks the current location as a temporary high-risk area and triggers local path planning. When multiple risk areas are adjacent, the system divides them according to risk level, prioritizing real-time alarms > historical faults > model predictions, and automatically plans the inspection sequence. All inspection results are fed back to update the historical fault database. Finally, the local fine-path planning module communicates with the ground station to estimate the time and energy required for fine-tuning the inspection, ensuring the airship's total endurance can support its return to base.

[0028] In step S140, a high-definition visible light camera, employing a high-resolution global shutter, is used to capture detailed images of line components such as insulators, towers, and conductors, identifying physical defects such as damage, corrosion, deformation, and hanging foreign objects. An infrared thermal imager monitors the temperature of conductor joints, clamps, and insulator strings, using an absolute temperature threshold of 80°C and relative temperature difference criteria (temperature change >20K or temperature difference >30% compared to similar components) to diagnose overheating faults. A lidar system is used for preliminary mapping, providing high-precision 3D point clouds in real-time during inspections to measure conductor sag and clearance distances to trees / buildings, mitigating the risk of external damage. Gas sensors monitor the concentration of specific gases, such as ozone (O3), sulfur hexafluoride (SF6), and nitrogen oxides (NOx). x The concentrations of gases such as ozone and SF6 are monitored, with abnormally high ozone concentrations indicating intensified corona discharge and SF6 leakage directly linked to sealing failures in gas-insulated switchgear. By integrating high-precision GNSS and an inertial measurement unit (IMU) into the positioning and attitude unit, precise spatiotemporal stamps and spatial coordinates are added to each frame of sensor data, preparing for data fusion and accurate fault location. Based on the updated fault database, deep learning algorithms are used to analyze the patterns of fault occurrence and changes in high-risk areas, optimizing local inspection paths.

[0029] Figure 2 This is a flowchart of the airship-type unmanned aerial vehicle (UAV) inspection system of the present invention.

[0030] like Figure 2 The flowchart shown illustrates the operation of an airship-type UAV inspection mission 200, including steps such as inspection preparation, cruise flight, data collection and analysis, fault handling, maintenance decision-making, and report archiving. The power operation and maintenance department initiates the inspection mission based on the operation and maintenance plan of the transmission line, clarifying the mission scope, priority, and time requirements. After the mission is approved, the mission parameters are entered into the mission management module of the ground control center. The airship, equipped with sensing modules such as lidar, high-definition cameras, and infrared thermal imagers, as well as communication modules and a power system, is dispatched. The airship's helium pressure and the status of the spare safety airbag are checked to ensure that the power system is fault-free. The path planning algorithm, data processing program, and other software systems are checked and updated to ensure stable system operation.

[0031] The pre-planned global inspection path and path data are imported and stored in the form of latitude and longitude coordinate points and flight commands. Sensor operating parameters are configured, the high-definition camera resolution is set to 4K with a frame rate of 25fps, the infrared thermal imager temperature measurement range is -20℃ to 200℃ with a thermal sensitivity of ≤0.05℃, and the gas detection sensor sampling frequency is 1Hz to ensure that the data acquisition accuracy meets the fault identification requirements. The ground control center establishes a connection with the airship's LoRa communication module, tests the data transmission rate and stability, and starts the remote monitoring interface. Technicians can view the airship status and sensor data monitoring in real time on the interface. Based on GIS (Geographic Information System) coordinates, the optimal energy cruise path connecting each base tower is generated. Around historical fault points and important crossing points, sub-paths for circling or approaching flight are automatically generated, and the airship is instructed to reduce altitude, decelerate, or hover at these points for detailed inspection. The planned path, flight commands, and sensor operating parameters are packaged and sent to the airship through the LoRa communication module.

[0032] After receiving instructions, the airship takes off autonomously. Its flight control system takes off along a preset path. In the initial stage, it climbs to the inspection altitude at a set speed. The system monitors wind speed in real time and performs dynamic compensation through multi-faceted and vector thrust to ensure stable flight and provide a stable platform for the sensors. During the cruise, the ground control center sends path fine-tuning instructions through the LoRa module. The airship's onboard flight control system responds in about 0.5 seconds and adjusts the flight attitude to ensure cruise efficiency.

[0033] The lidar scans the surrounding environment at a frequency of 10Hz, generating 3D point cloud data and updating the environmental map in real time, providing a basis for dynamic path adjustment. High-definition cameras and infrared thermal imagers simultaneously acquire images of the power transmission line. The high-definition camera focuses on the details of the line's appearance, while the infrared thermal imager captures the temperature distribution at the conductor joints and clamps. The data from both are synchronized in time and space through timestamps and location information. Gas monitoring sensors continuously monitor the concentration of gases such as ozone and nitrogen oxides around the line. When the concentration exceeds the threshold, the sampling frequency is automatically increased to 5Hz to ensure the capture of the fault development process. The ship's onboard storage module stores the raw data in real time and performs preliminary preprocessing on the data. The lidar point cloud data is downsampled to retain key features, and the image data is compressed to reduce the bandwidth pressure of data transmission.

[0034] Data from LiDAR, HD cameras, infrared thermal imagers, and gas sensors are fused using a Jetson AGX processor, categorized by timestamp and location information. A multi-sensor fusion algorithm is employed to verify data consistency and eliminate abnormal data caused by sensor malfunctions or environmental interference. Real-time analysis of the fused data is performed using pre-trained deep learning models: a YOLOv8-based conductor defect detection model and a U-Net-based infrared thermal anomaly recognition model. The HD image detection rate for defects such as broken conductor strands and insulator damage is ≥97%, infrared thermal image data identifies street overheating, and gas concentration data identifies faults. Based on the intelligent analysis results and the preset anomaly judgment rules, an anomaly risk level is generated. If the risk is determined to be no, the airship continues to cruise along the original path. If the anomaly or risk is determined, the latitude and longitude coordinates and altitude of the fault point are determined by combining the lidar SLAM data. At the same time, the offset direction and distance of the fault point relative to the power transmission line are marked, and a multi-dimensional data report of the fault point is generated, including key information such as defect images, temperature curves and gas concentrations. The fault report is transmitted back to the ground control center in real time through the LoRa module. At the same time, the backup 5G communication module is activated to ensure that the data is not lost. The fault early warning system of the ground control center automatically pops up an alarm window to notify the operation and maintenance personnel in the form of sound and light prompts and pop-up windows.

[0035] Maintenance personnel at the ground control center review multi-dimensional data reports of fault locations, combining historical fault records and line maintenance specifications to assess the severity and scope of the fault. Specific repair plans are then developed based on the fault type; for example, broken conductor strands require conductor section replacement, overheated joints require clamp repair and application of conductive paste, or insulation faults necessitate insulator replacement. The report generation module at the ground control center integrates data such as the inspection route execution, fault point details, and sensor operating status. The technical head of the maintenance department reviews the accuracy and completeness of the report. After approval, the report is archived in the transmission line maintenance database as a basis for subsequent inspection plans and equipment life assessments.

[0036] Figure 3 This is a flowchart of the global path planning method of the present invention.

[0037] like Figure 3 As shown, the global path planning algorithm 300 for airship-type UAVs includes four steps from mission initiation to airship takeoff and cruise: baseline determination, obstacle recognition, Q-value initialization, and path generation and distribution. This provides a precise global navigation foundation for subsequent inspection tasks. The ground control center triggers the global path planning module based on the inspection task requirements, importing basic data such as the starting and ending coordinates of the power transmission line and its designed route. The path planning algorithm engine is then started, loading the parameter configurations required for algorithm operation, initializing computing resources, and ensuring that the algorithm completes path generation within the specified time.

[0038] Based on the CGCS2000 coordinates of the starting and ending points of the transmission line, a SG line connecting the starting and ending points is constructed in three-dimensional space as a baseline reference line for global path planning. If the transmission line has branches or multiple sections, SG lines for each section need to be constructed separately to ensure that the path planning for each section has a clear baseline. Combined with the rough environmental data from the pre-scanning of LiDAR, it is verified whether the SG line has serious conflicts with the actual terrain. If conflicts exist, the coordinates of the starting and ending points need to be fine-tuned or detour routes need to be planned in advance. Finally, the three-dimensional coordinate point set of the SG line is output for subsequent obstacle recognition and path generation.

[0039] If a pre-built 3D environment map of the transmission line already exists, it is loaded directly. If not, the airship equipped with a lidar performs a rapid scan along the SG line, collecting point cloud data of the surrounding terrain, towers, trees, and buildings. The raw point cloud data is then denoised, downsampled, and reconstructed to generate a 3D environment mesh map with a resolution of 0.5m, clearly marking the location and outline of all obstacles. The spatial intersection of each obstacle with the SG line is calculated. If the shortest distance between an obstacle and the SG line is less than a preset threshold, it is identified as a primary obstacle (MO); otherwise, it is a secondary obstacle (NMO) and can be temporarily excluded from the primary search scope. For each MO, a side length of [missing information] is generated with its geometric center as the origin. ( Network nodes within a cube (grid step size) are used as peripheral nodes (SPs) for subsequent path search. Only these nodes are retained for path calculation, significantly reducing invalid searches.

[0040] In the definition of the target direction function, the current candidate node of the airship is... The starting point is The destination is Calculate the angle between the candidate node and the SG line. As shown in equation (2) below.

[0041] (2) The smaller the value, the closer the candidate node is to the SG line, and closer to the transmission line direction. For non-SP nodes, according to... Assign an initial Q value as shown in equation (3).

[0042] (3) in The initial Q value, Based on the Q value, for non-SP nodes, the initial Q value is set to 0, which forces the airship to prioritize SP nodes close to the SG line in the early stage of path exploration, avoids random exploration, and shortens the path convergence time.

[0043] A path search algorithm incorporating reinforcement learning is employed, using the Q-value as the decision criterion. It searches for the optimal path from the starting point to the destination within the SP node set. The generated path is smoothed, and the B-spline curve algorithm is used to optimize the path curvature, ensuring stable attitude and minimal power consumption during airship flight. Key waypoints are inserted into the path for positioning reference during subsequent local inspections. By simulating the airship's flight along the generated path, the safe distances between the path and all MOs are checked to ensure they meet preset thresholds. If conflicts are found, the algorithm parameters are backtracked and optimized, and the path is regenerated. After successful verification, path data such as waypoint coordinates, flight speed, and dwell time are transmitted via LoRa communication to the airship's onboard flight control system. Simultaneously, the data is visualized on the GIS interface of the ground control center for final confirmation by technical personnel.

[0044] The airship's onboard flight control system receives and analyzes path data, converting waypoint coordinates into local execution commands. Ground technicians send path confirmation commands via remote controller, and the airship transmits its path loading status back. The airship takes off according to the preset path, and during flight, it uses lidar SLAM to compare the deviation between its current position and the path waypoints in real time. When the deviation exceeds 0.5 meters, the path correction mechanism is automatically triggered, adjusting the propeller thrust to return to the preset path. During cruise, the ground control center can monitor the path execution status through real-time position information and sensor data. If significant environmental changes are detected, path fine-tuning commands can be remotely issued. Upon reaching the inspection area, the airship switches to a local fine-tuning inspection and multi-sensor fault detection mode to begin specific inspection and diagnostic tasks.

[0045] Figure 4 This is a flowchart of the local path planning method of the present invention.

[0046] like Figure 4 As shown, the airship-type UAV local path planning method 400 is the core of the intelligent decision-making of the entire inspection system. Based on the global path planning results, local key path planning is triggered when high-risk areas, complex environmental scenarios, and key inspection sections are identified. By calling a reinforcement learning algorithm based on expected information distribution (EID), input parameters such as the coordinates of risk areas and historical fault data in the global path are loaded. By extracting historical fault data, the fault frequency and fault type of the areas in the global path planning are statistically analyzed to generate a fault heat map, marking high-risk, medium-risk, and low-risk areas. Based on the 3D environmental map analysis, the obstacle density and conductor routing complexity within the area are marked as environmental high-risk areas. Areas where key equipment of the transmission line is located are directly marked as equipment high-risk areas, while other areas are low-risk areas. The above-mentioned risk areas are spatially superimposed and sorted according to the priority of high-risk area > medium-risk area > environmental high-risk area > equipment high-risk area > low-risk area to ensure that resources are tilted towards the most critical areas.

[0047] The expected information distribution EID for each risk area is calculated by fusing fault risk value, sensor effectiveness, and information entropy, as shown in equation (4) below.

[0048] (4) in, This represents the fault risk value. For sensor effectiveness, For information entropy, The maximum entropy value is used. The airship dwell time is allocated according to the EID value, as shown in equation (5) below.

[0049] (5) in Set the airship's base dwell time to 5 seconds. For a certain region's EID value, This is the maximum EID value.

[0050] During the airship's stay, data was collected using a multi-sensor collaborative strategy. High-definition cameras employed a "multi-angle + high frame rate" acquisition mode in high-risk areas, capturing images from the front, side, and overhead angles around the tower joint at a frame rate of 30fps to ensure the detection of minute cracks and broken strands. Infrared thermal imagers were switched to continuous recording mode to record temperature change curves in real time, facilitating subsequent analysis of fault development trends. Gas detection sensors increased sampling frequency in areas suspected of insulation faults to accurately capture peak concentrations of gases such as ozone and nitrogen oxides. LiDAR increased point cloud sampling density in complex environments to ensure sufficiently detailed local environmental maps, supporting subsequent obstacle avoidance and defect localization. All collected data was bound to BeiDou positioning information and timestamps and stored in the airship's local altitude cache module. Simultaneously, key frame data, such as fault images or thermal imaging of temperature anomalies, were transmitted back to the ground control center in real time via a LoRa module.

[0051] The airship's onboard processor performs preliminary analysis of the collected raw data. Image data is quickly filtered for defect images using a YOLOv8 network, while temperature data is semantically segmented from thermal imaging images using a U-Net network to mark measurement points exceeding thresholds. Gas data is directly marked by sensors for measurement points exceeding concentration thresholds. After receiving the data, the ground control center conducts further in-depth analysis and verification using historical data and expert experience to determine whether the fault is genuine.

[0052] Based on the type and characteristics of the faults, they are classified into three levels: emergency faults, important faults, and general faults. Emergency faults are those with conductor strand breaks of ≥5mm or joint temperatures exceeding the ambient temperature by 50℃. Important faults are those with minor insulator damage or conductor strand breaks of 2-5mm. General faults are those with minor scratches on the conductor surface or temperatures exceeding the ambient temperature by 30℃ but not reaching 50℃. The system automatically matches predefined maintenance strategies based on the fault level and type. Emergency faults require emergency repair within 2 hours, with timely arrangements including lists of repair personnel, equipment, and materials. Important faults generate maintenance plans for within 72 hours and are handled by regular maintenance teams. General faults generate monthly maintenance plans and are incorporated into the routine maintenance schedule.

[0053] By combining local matching with lidar, the three-dimensional coordinates of the fault point are determined, including its latitude, longitude, and altitude. Simultaneously, the offset direction and distance of the fault point relative to the tower are marked, generating a location feature report for the fault point, including its coordinates, defect image, and peak temperature / gas concentration information. The system automatically extracts key data from this local inspection, including the number of risk areas, fault type and quantity, detailed information for each fault point, and sensor operating status. This data is then compiled into an airship inspection report. The operations and maintenance department reviews the report to confirm data accuracy and the rationality of recommendations. After approval, the system archives the report to the operations and maintenance database.

[0054] Based on the final confirmed operation and maintenance strategy, the system assists in generating detailed execution plans, including automatic triggering of emergency repair resource scheduling for urgent faults, and incorporating maintenance tasks into the operation and maintenance management system for important / general faults, automatically reminding relevant teams to execute according to the plan. Based on historical case studies of fault handling, the system assists in optimizing the maintenance plan for this fault, while updating the handling process and lessons learned from this fault to the operation and maintenance knowledge base for subsequent algorithm optimization and personnel training. Finally, all raw data and analysis results of this partial inspection are archived to the data center, using distributed storage to ensure data security, and backed up to the cloud server.

[0055] The specific embodiments described above do not constitute a limitation on the scope of protection of this disclosure. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can be made according to design requirements and other factors. Although embodiments of this application have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting this application. Those skilled in the art can make changes, modifications, substitutions, and variations to the above embodiments within the scope of this application.

Claims

1. An airship-type unmanned aerial vehicle (UAV) inspection system, characterized in that, It includes an airship airbag, an airship body filled with helium to generate static buoyancy, a lidar, an image acquisition device, an infrared thermal imager, a gas detection sensor, an inertial measurement unit, a backup safety airbag device, and a communication module; The airship's airbag is made of lightweight, high-strength materials such as polyester fiber composite membrane and is equipped with an electrically driven vector propulsion system. Precise flight is achieved through auxiliary airbag adjustment and control surface control. The lidar is used for three-dimensional environmental modeling and distance measurement of the power transmission line, and the image acquisition device includes a high-definition camera for acquiring images of the power transmission line surface; The infrared thermal imager is used to monitor the temperature of wire joints, wire clamps and other parts, and the gas detection sensor is used to monitor the characteristic gases generated in the surrounding environment of the line due to electric arcs or overheating. The inertial measurement unit is used to monitor the airship's operating status, and the backup airbag device is used to provide safety protection in emergency situations; The communication module is used to transmit inspection data.

2. The airship-type unmanned aerial vehicle (UAV) inspection system according to claim 1, characterized in that, The system also includes a ground control center, which imports the location coordinates of the transmission line and historical fault data, generates inspection path information including the basic path and dynamic adjustment space, and sends the inspection path information to the UAV through the communication module. The ground control center also includes a global path planning module and a local path planning module. The global path planning module generates a global inspection path through a 3D environmental map of the transmission line pre-built by lidar. The local path planning module focuses on inspecting risk areas with potential line damage based on historical fault data and real-time perception data.

3. The airship-type unmanned aerial vehicle (UAV) inspection system according to claim 1, characterized in that, The lidar emits laser pulses at a frequency of 10Hz to collect environmental information such as terrain, towers, trees, and buildings around the power transmission line, generating raw point cloud data. After denoising and Poisson reconstruction algorithm, a 3D environmental grid map of the power transmission line is constructed. The location and outline of obstacles such as conductors, towers, tall trees, and buildings of the power transmission line are marked in the 3D environmental grid map.

4. The airship-type unmanned aerial vehicle (UAV) inspection system according to claim 2, characterized in that, The global path planning module uses the connection between the start and end points of the transmission line as a reference, prioritizes the identification of major obstacles and surrounding nodes on the path for global path planning, constructs a cubic space centered on the geometric center point of the major obstacles, selects network nodes within the cubic range as key nodes for path planning, forming a sparse network graph, and uses the A* algorithm to search for the optimal path from the start to the end point on the key node network. The algorithm is given directionality through the Q-value initialization strategy, so that the path search prioritizes exploration in the target direction.

5. The airship-type unmanned aerial vehicle (UAV) inspection system according to claim 2, characterized in that, The local path planning module automatically identifies high-risk areas based on a historical fault database and real-time perception data, and dynamically generates refined local flight strategies. The local path planning module marks historical fault points on the global path. When the airship descends from its cruising altitude, it smoothly approaches target components such as insulator strings and clamps. Within a safe distance of 5-10 meters, it activates the control surfaces and vector thrusters to counteract wind disturbances and maintain stable hovering, allowing the high-definition camera and infrared thermal imager to obtain the highest resolution images.

6. The airship-type unmanned aerial vehicle (UAV) inspection system according to claim 1, characterized in that, The high-definition camera in the image acquisition device is a high-resolution global shutter camera, used to capture detailed images of line components such as insulators, towers and conductors, and to identify physical defects such as damage, corrosion, deformation, and foreign objects hanging. The infrared thermal imager monitors the temperature of conductor joints, clamps, insulator strings, and other parts. It sets an absolute temperature threshold of 80°C and a relative temperature difference criterion of temperature change >20K or temperature difference >30% compared with similar components to diagnose overheating faults.

7. The airship-type unmanned aerial vehicle (UAV) inspection system according to claim 1, characterized in that, The gas detection sensor monitors the concentration of specific gases, such as ozone (O3), sulfur hexafluoride (SF6), and nitrogen oxides (NOx). x The concentrations of gases such as ozone and SF6 are monitored. An abnormally high ozone concentration can indicate an aggravated corona discharge, while an SF6 leak is directly related to a sealing failure of the gas-insulated switchgear.

8. The airship-type unmanned aerial vehicle (UAV) inspection system according to claim 1, characterized in that, The system employs a multi-sensor fusion algorithm to fuse data from lidar, high-definition cameras, infrared thermal imagers, and gas sensors. By running pre-trained deep learning models, a YOLOv8-based conductor defect detection model, and a U-Net-based infrared thermal anomaly recognition model, the system performs real-time analysis of the fused data. The high-definition image recognition rate for defects such as broken conductor strands and damaged insulators is ≥97%, infrared thermal image data is used to identify street overheating, and gas concentration data is used to identify faults.

9. The airship-type unmanned aerial vehicle (UAV) inspection system according to claim 1, characterized in that, The system generates anomaly risk levels based on intelligent analysis results and preset anomaly judgment rules. If the risk is determined to be no, the airship continues to cruise along the original path. If the anomaly or risk is determined to exist, the system uses LiDAR SLAM data to determine the latitude and longitude coordinates and altitude of the fault point, and marks the offset direction and distance of the fault point relative to the power transmission line, generating a multi-dimensional data report of the fault point.

10. A method for inspecting airship-type unmanned aerial vehicles (UAVs), comprising a computer program / instructions, characterized in that, When the computer program / instructions are executed by one or more processors, they implement the steps of the airship-type unmanned aerial vehicle inspection system according to any one of claims 1-9.

Images