Patrol unmanned aerial vehicle flight control management method

Abstract

The invention relates to the technical field of unmanned aerial vehicles, and discloses an inspection unmanned aerial vehicle flight control management method, which comprises the following steps: step 1, an unmanned aerial vehicle obtains environment three-dimensional point cloud data through a laser radar, captures terrain visual information through an RGB camera and a depth camera, obtains real-time position and attitude information of the unmanned aerial vehicle through a GPS and an inertial measurement unit, and obtains the real-time position and attitude information of the unmanned aerial vehicle; and fusing the collected data to construct a spatial grid map, and taking the grid map as input data of path planning. A high-precision space grid map is generated through three-dimensional environment perception and modeling in combination with an unmanned aerial vehicle multi-sensor fusion technology, and a path is updated in real time by using dynamic environment data, and an air route is planned and optimized, so that path dynamic adjustment and energy consumption minimization of the unmanned aerial vehicle in a complex dynamic environment are realized; and the effects of improving the task execution efficiency, reducing the flight energy consumption and improving the flight safety are achieved.

Description

Technical Field

[0001] The present invention relates to the technical field of unmanned aerial vehicles, and in particular to a flight control management method for an inspection unmanned aerial vehicle. Background Art

[0002] With the rapid development of drone technology and the growth of application demand, drones have become an important tool in the field of modern inspection and are widely used in power inspection, oil pipeline monitoring, agricultural and forestry management, urban security and emergency rescue. With the advantages of high maneuverability, wide coverage, flexible operation and low operating cost, drones can efficiently perform inspection tasks in complex environments and solve the problems of high labor costs, low efficiency and limited coverage of traditional inspection methods. However, as the complexity of inspection tasks continues to increase, drones will have problems in flight control management methods, which urgently need further improvement and optimization to adapt to diverse mission requirements.

[0003] The problems with existing UAV flight control management methods are as follows: The existing drone technology mainly relies on preset fixed routes for flight path planning. The path planning method is relatively simple and usually cannot adapt to the dynamic changes in the actual inspection environment. In addition, the path planning process lacks the optimization design of flight energy consumption, which easily leads to path redundancy, seriously affecting the drone's endurance, mission execution efficiency and flight safety.

[0004] When drones are performing inspection tasks, abnormal situations such as low battery, hardware system failure or flight path deviation often occur. However, existing technologies lag behind in battery status monitoring, hardware health assessment and abnormal status response, and fail to achieve real-time monitoring of drone health status and fault prediction. Once the battery power approaches the threshold or the flight control hardware fails, the existing system cannot quickly trigger the emergency response mechanism. In addition, the lack of an automated task takeover mechanism makes the system fault-tolerance low, and the tasks of the faulty drones cannot be efficiently taken over, reducing the task completion rate.

[0005] In inspection tasks, using multiple drones to perform tasks collaboratively is a key means to improve task efficiency. However, existing multi-drone task scheduling methods have problems such as regional division and lack of dynamic adjustment of task allocation. In addition, if a drone fails or runs out of power, the system cannot promptly assign unfinished tasks to neighboring drones. The lack of an intelligent task dynamic takeover mechanism seriously affects the efficiency of multi-drone collaboration and the quality of task completion.

[0006] UAVs need to collect high-definition images, infrared data, and environmental information during inspection missions. However, in existing technologies, data processing mostly relies on offline processing by backhaul ground stations, resulting in delays in data transmission and the inability to achieve real-time anomaly detection and data analysis. At the same time, traditional data processing methods mostly rely on manual screening and analysis, lack intelligent automatic detection methods, have low recognition efficiency, and cannot guarantee the accuracy of detection results.

[0007] Therefore, those skilled in the art provide a flight control management method for an inspection UAV to solve the above-mentioned problems. Summary of the invention

[0008] In view of the deficiencies in the prior art, the present invention provides a flight control management method for an inspection UAV to solve the problems raised in the above background technology.

[0009] To achieve the above objectives, the present invention is implemented through the following technical solutions: a flight control management method for an inspection drone, comprising: Step 1: The drone obtains the 3D point cloud data of the environment through the lidar, captures the terrain visual information through the RGB camera and the depth camera, obtains the real-time position and attitude information of the drone through the GPS and the inertial measurement unit, and then fuses the collected data to construct a spatial grid map, which is used as the input data for path planning; Step 2: Using the grid map obtained by 3D environment perception and modeling as input, determine the starting point, target point and flight constraints of the UAV, generate the initial flight route of the UAV through path planning, monitor the changes of obstacles and dynamic environment during the flight of the UAV in real time, superimpose the detected dynamic environment change information on the grid map, and readjust the route according to the superimposed updated map to obtain the optimized flight path; Step 3: Based on the flight path generated by route planning and dynamic optimization, monitor the status data of the drone in real time during the flight of the route, calculate the battery power as the ratio of the remaining battery capacity to the total battery capacity, use the monitored flight control hardware status information to determine whether there is a hardware anomaly, and use the flight attitude data to determine whether the drone deviates from the route; Step 4: Based on the flight status monitoring, determine whether emergency treatment is needed according to the monitoring results of the drone's battery power status, flight control hardware status and flight attitude information. If the battery power is below the preset threshold, adjust the route to return to the charging station for charging. If the flight control hardware is abnormal, record the abnormal information and stop the mission. If the drone deviates from the route, correct the flight according to the route planning and dynamically optimized path. Step 5: If the UAV performs autonomous emergency response, the task scheduling and collaborative takeover process is triggered. The unfinished task area of ​​the UAV is allocated to the neighboring UAVs through the task area division method, and the task allocation is dynamically adjusted based on the parameters of the UAV. Step 6. During the execution of the task, the drone collects images and environmental data of the inspection area through the onboard sensors. During task scheduling and collaborative takeover, the drone continues to collect data from the unfinished area. The collected data is preliminarily processed by the edge computing module on the drone side. The processed data includes the detection results of the abnormal area and relevant information of the drone's inspection coverage. The processed data is then transmitted back to the ground station for the generation of the inspection report.

[0010] Preferably, the spatial grid map constructed in step 1 is generated by the following method: 3D point cloud data based on LiDAR , rasterize the 3D point cloud data, decompose the environment into voxel units, and generate a 3D grid map G: , in, is the grid cell state, 1 means it is occupied by an obstacle, 0 means it is free; is the index coordinate of the grid cell.

[0011] Preferably, the initial flight route generated in step 2 is dynamically optimized using a real-time updated grid map, and the flight path is updated and planned according to the occupancy status in the grid map; The battery capacity in step 3 is calculated by the following formula: , in, is the percentage of remaining battery capacity, Remaining battery capacity, is the total battery capacity.

[0012] Preferably, the flight control hardware state in step 3 is predicted by the following system state equation: , in, is the current system status, is the system state transition matrix, is the control input vector, is the input matrix, is the system noise, Is the system in time The state vector at time .

[0013] Preferably, the correction of the deviation of the drone from the route in step 4 is provided as a reference by the dynamically updated flight path, and the correction includes adjusting the attitude, flight speed and direction of the drone; The task area division in step 5 is based on the Voronoi segmentation algorithm, and the area division formula is: , in, Assigned to drones mission area, is any point in the mission area, Yes To drones The Euclidean distance of and yes The special points in Indicate point To drones The Euclidean distance.

[0014] Preferably, the task priority of the UAV is dynamically adjusted during the task takeover process, and the task priority is calculated by the following formula: , in, is the task priority, is the estimated completion time of the task, It is the urgency of the task. and is the weight parameter; The data collected in step 6 includes high-definition images, infrared data and environmental parameter information. The collected data is pre-processed in real time through edge computing equipment. The data pre-processing includes compression, anomaly detection and feature extraction.

[0015] Preferably, in step 6, the detection of inspection anomalies uses a deep learning algorithm, and the target detection result is expressed by the following formula: , in, is the target detection result, is the input image, are the parameters of the deep learning network, Represents the forward prediction function of the YOLO algorithm.

[0016] Preferably, the optimized flight route generated by the drone in step 2 aims to minimize the total energy consumption of the path, and the energy consumption calculation formula is: , in, is the total energy consumption of the flight path, is the hovering power of the drone, Forward power of the drone, is the total flight time.

[0017] Preferably, the flight attitude monitoring in step 3 includes the pitch angle, roll angle and yaw angle of the drone, and the attitude calculation formula is: , in, is the attitude matrix of the UAV, is the yaw angle transformation matrix rotating around the z-axis, is the pitch angle transformation matrix rotating around the y-axis, is the roll angle transformation matrix rotating around the x-axis; The UAV determines the degree of flight attitude deviation by calculating the above attitude matrix in real time, and adjusts the flight control parameters according to the deviation to correct the attitude.

[0018] Preferably, when the image data collected by the drone in step 6 is preprocessed by edge computing, the target detection confidence Calculated by the following formula: , in, is the target detection confidence, which indicates the probability that the target detection is a correct result; is the prediction score of target i, indicating the credibility of the model for target i; N is the total number of target categories, and e is the base of the natural logarithm; The drone screens the targets based on the calculated target detection confidence, outputs high-confidence abnormal target information, and transmits the abnormal target information to the ground station for subsequent generation of inspection reports.

[0019] The present invention provides a flight control management method for an inspection drone, which has the following beneficial effects: 1. The present invention generates a high-precision spatial grid map through three-dimensional environmental perception and modeling, combined with drone multi-sensor fusion technology, and uses dynamic environmental data to update the path in real time, perform route planning and optimization, and realize dynamic path adjustment and energy consumption minimization of drones in complex dynamic environments, thereby improving mission execution efficiency, reducing flight energy consumption, and enhancing flight safety.

[0020] 2. The present invention triggers an autonomous emergency response mechanism by real-time monitoring of battery power, hardware health status and flight attitude. At the same time, combined with task area division and dynamic task takeover, it realizes automatic return, task stop recording and takeover of the UAV when the battery is low, the hardware fails or the flight deviates, thereby achieving the effect of uninterrupted missions, improving system fault tolerance and task completion rate.

[0021] 3. The present invention introduces the Voronoi segmentation algorithm to divide the task area, combines task priority calculation with multi-agent collaborative scheduling, and realizes intelligent collaboration and dynamic task allocation of multiple UAVs during task execution, thereby solving task duplication and omissions, improving collaborative work efficiency and intelligent takeover of faulty tasks.

[0022] 4. The present invention combines edge computing technology with deep learning algorithms to realize real-time preprocessing and anomaly detection of data on the drone side, and realizes rapid identification of target data in the inspection area and high-confidence anomaly detection, thereby reducing data processing delays, improving anomaly detection accuracy and the reliability of inspection results. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] Figure 1 It is a flow chart of the present invention. DETAILED DESCRIPTION

[0024] In order to make the technical personnel in the technical field understand the scheme of the present invention, the technical scheme in the embodiment of the present invention will be clearly and completely described below in combination with the drawings in the embodiment of the present invention. Obviously, the described embodiment is a partial embodiment of the present invention, not a complete embodiment. Based on the embodiment of the present invention, other embodiments obtained by ordinary technicians in the field without creative work should fall within the scope of protection of the present invention.

[0025] The present invention is described in detail below in conjunction with the accompanying drawings: Example

[0026] Please refer to the attached Figure 1 The embodiment of the present invention provides a flight control management method for an inspection drone, comprising: Step 1: The drone obtains the 3D point cloud data of the environment through the lidar, captures the terrain visual information through the RGB camera and the depth camera, obtains the real-time position and attitude information of the drone through the GPS and the inertial measurement unit, and then fuses the collected data to construct a spatial grid map, which is used as the input data for path planning; Step 2: Using the grid map obtained by 3D environment perception and modeling as input, determine the starting point, target point and flight constraints of the UAV, generate the initial flight route of the UAV through path planning, monitor the changes of obstacles and dynamic environment during the flight of the UAV in real time, superimpose the detected dynamic environment change information on the grid map, and readjust the route according to the superimposed updated map to obtain the optimized flight path; Step 3: Based on the flight path generated by route planning and dynamic optimization, monitor the status data of the drone in real time during the flight of the route, calculate the battery power as the ratio of the remaining battery capacity to the total battery capacity, use the monitored flight control hardware status information to determine whether there is a hardware anomaly, and use the flight attitude data to determine whether the drone deviates from the route; Step 4: Based on the flight status monitoring, determine whether emergency treatment is needed according to the monitoring results of the drone's battery power status, flight control hardware status and flight attitude information. If the battery power is below the preset threshold, adjust the route to return to the charging station for charging. If the flight control hardware is abnormal, record the abnormal information and stop the mission. If the drone deviates from the route, correct the flight according to the route planning and dynamically optimized path. Step 5: If the UAV performs autonomous emergency response, the task scheduling and collaborative takeover process is triggered. The unfinished task area of ​​the UAV is allocated to the neighboring UAVs through the task area division method, and the task allocation is dynamically adjusted based on the parameters of the UAV. Step 6. During the execution of the task, the drone collects images and environmental data of the inspection area through the onboard sensors. During task scheduling and collaborative takeover, the drone continues to collect data from the unfinished area. The collected data is preliminarily processed by the edge computing module on the drone side. The processed data includes the detection results of the abnormal area and relevant information of the drone's inspection coverage. The processed data is then transmitted back to the ground station for the generation of the inspection report.

[0027] Step 1 Benefits: Achieve high-precision environmental perception and modeling, provide real-time, dynamic three-dimensional environmental information, improve the integrity and accuracy of environmental data through multi-sensor fusion, and provide a reliable input basis for path planning and obstacle detection; Step 2 Benefits: Dynamic planning and real-time optimization of flight routes are achieved to ensure that the drone can adapt to complex dynamic environments, improve the safety and efficiency of flight paths, avoid collisions with obstacles, and effectively reduce flight energy consumption and extend flight time through dynamic path adjustment; Step 3 Benefits: Real-time understanding of the drone's battery and health status, ensuring safe and stable flight of the drone, timely detection of hardware anomalies, reducing the risk of mission failure, and correcting flight path deviations and improving flight accuracy through attitude monitoring; Step 4 Benefits: Provides an autonomous emergency response mechanism to ensure that the drone can return safely or stop the mission in an abnormal state. Through the route correction function, it ensures the stability and accuracy of mission execution, avoids the loss of the drone due to power exhaustion or hardware failure, and improves the system's fault tolerance. Step 5 Benefits: Realize dynamic coordination of multiple UAV tasks, solve the problems of task interruption and area omission, improve the efficiency of task takeover, ensure the integrity and continuity of tasks, reduce manual intervention in the event of UAV failure, and realize automatic task takeover; Benefits of step 6: Realize real-time data collection and intelligent processing, reduce data transmission delays, conduct preliminary analysis through edge computing, quickly identify abnormal areas, improve data analysis efficiency, ensure the integrity and accuracy of inspection data, and provide high-quality results for inspection reports.

[0028] The spatial grid map constructed in step 1 is generated by the following method: 3D point cloud data based on LiDAR , rasterize the 3D point cloud data, decompose the environment into voxel units, and generate a 3D grid map G: , in, is the grid cell state, 1 means it is occupied by an obstacle, 0 means it is free; is the index coordinate of the grid cell.

[0029] The present invention performs rasterization processing on three-dimensional point cloud data based on laser radar, and the generated spatial raster map has the advantages of accurate modeling, efficient calculation, dynamic update and multi-source data fusion, providing high-quality environmental input data for the path planning and dynamic route adjustment of drones. The raster map can accurately reflect the spatial structure of complex environments, significantly improve the computational efficiency and real-time performance of path search, and lay a solid foundation for the intelligent flight control management of drones in dynamic environments.

[0030] The initial flight route generated in step 2 is dynamically optimized using the real-time updated grid map, and the flight path is updated and planned according to the occupancy status in the grid map; The battery charge in step 3 is calculated using the following formula: , in, is the percentage of remaining battery capacity, Remaining battery capacity, is the total battery capacity.

[0031] Real-time updating of grid maps can quickly reflect dynamic changes in the flight environment, such as the appearance of new obstacles or updates to environmental status, allowing drones to avoid risks and adapt to complex dynamic environments; By updating the occupancy status in the grid map in real time, the drone can actively identify obstacle areas on the flight path, replan the path, reduce the risk of obstacle collision, and ensure flight safety; Dynamically adjust the path during flight, continuously optimize the path according to the current environment and mission objectives, shorten the flight distance, reduce unnecessary path redundancy, and improve the execution efficiency of the route; By calculating the ratio of the remaining battery power to the total capacity in real time, the system can accurately obtain the battery health status of the drone, ensuring reliable and accurate power monitoring during flight; Real-time monitoring of battery power helps to detect low battery in time and trigger the return mechanism in advance to prevent the drone from being unable to return safely due to battery exhaustion, thus improving the safety of flight missions; Real-time acquisition of battery status can provide a basis for mission planning. If the battery power is below the preset threshold, the system can dynamically adjust the mission execution path, give priority to return or charging, and ensure that the mission is not interrupted.

[0032] In summary, in the present invention, step 2 realizes dynamic optimization of the flight path by updating the grid map in real time, significantly improving the adaptability, safety and route execution efficiency of the UAV in a complex dynamic environment, while reducing flight energy consumption and improving mission endurance.

[0033] Step 3 accurately calculates the battery power status and monitors the battery health of the drone in real time, providing a basis for dynamic mission adjustment and path optimization, effectively ensuring the flight safety and energy consumption management of the drone.

[0034] The flight control hardware state in step 3 is predicted by the following system state equation: , in, is the current system status, is the system state transition matrix, is the control input vector, is the input matrix, is the system noise, Is the system in time The state vector at time .

[0035] In the present invention, the flight control hardware state prediction method based on the system state equation realizes real-time monitoring of the hardware health status and future trend analysis of the UAV through modeling and prediction of the hardware state, which can improve the accuracy of hardware health monitoring, identify potential faults in advance and trigger emergency response, effectively ensure the flight safety of the UAV and the stability of mission execution. At the same time, through the prediction and repair of abnormal states, the fault tolerance capability of the UAV system is enhanced to ensure that the mission is not interrupted, further improving the reliability and intelligence level of the inspection mission.

[0036] The correction of the drone's deviation from the route in step 4 is a reference for the dynamically updated flight path. The correction includes adjusting the drone's attitude, flight speed and direction; The task area division in step 5 is based on the Voronoi segmentation algorithm, and the area division formula is: , in, Assigned to drones mission area, is any point in the mission area, Yes To drones The Euclidean distance of and yes The special points in Indicate point To drones The Euclidean distance.

[0037] Through dynamic path updates and route corrections, the system can enable the drone to quickly return to the optimal flight path when a path deviation occurs, ensuring the accuracy of route tracking and avoiding incomplete coverage of the mission area due to flight deviations; The route correction is combined with the dynamic adjustment of the drone's attitude, speed and direction to achieve smooth and continuous path repair, reduce the risk of flight jitter or instability caused by rapid adjustment, and improve the safety and stability of the flight process; The Voronoi segmentation algorithm automatically divides the mission area according to the initial position of the UAV and the distribution of the mission area, ensuring that each UAV covers an independent mission area and avoiding area overlap or omission; The rationality of the task area division enables drones to perform tasks efficiently in their respective areas, reduce duplication of work and waste of resources, and improve the overall efficiency of task completion.

[0038] In summary, step 4 achieves accurate correction of the UAV route through dynamically updated flight paths, combines attitude, speed and direction adjustments, improves route tracking accuracy and flight stability, reduces mission execution time and energy waste caused by path deviation, and ensures the autonomous flight capability and safety of the UAV in complex dynamic environments; Step 5 divides the task area through the Voronoi segmentation algorithm to ensure the reasonable allocation and efficient execution of tasks, support dynamic adjustment and takeover of tasks, and improve the efficiency and reliability of multi-UAV collaborative operations.

[0039] During the task takeover process, the task priority of the drone is dynamically adjusted. The task priority is calculated using the following formula: , in, is the task priority, is the estimated completion time of the task, It is the urgency of the task. and is the weight parameter; The data collected in step 6 includes high-definition images, infrared data and environmental parameter information. The collected data is pre-processed in real time through edge computing devices. The data pre-processing includes compression, anomaly detection and feature extraction.

[0040] In the present invention, the task priority of task takeover is dynamically adjusted through weighted calculation of task urgency and completion time, so as to realize intelligent priority sorting of unfinished tasks, improve the decision-making efficiency and resource utilization of task takeover, ensure that high-priority tasks can be completed in time, and optimize the dynamic scheduling capability of multi-UAV collaborative operations.

[0041] The data collection and real-time preprocessing in step 6 uses edge computing to perform real-time compression, anomaly detection and feature extraction of high-definition images, infrared data and environmental parameters, reducing data transmission delays, achieving fast and efficient data analysis and anomaly detection, and improving the real-time and accuracy of inspection tasks.

[0042] In step 6, a deep learning algorithm is used to detect inspection anomalies. The target detection result is expressed by the following formula: , in, is the target detection result, is the input image, are the parameters of the deep learning network, Represents the forward prediction function of the YOLO algorithm.

[0043] The YOLO algorithm has the characteristics of single-stage detection, which combines target detection and bounding box prediction and can complete target detection in a single forward reasoning. Compared with traditional multi-stage methods, YOLO reduces the calculation time of detection and is suitable for real-time anomaly detection when drones are performing tasks.

[0044] By training inspection data through a deep learning network, the YOLO algorithm can learn complex target features and effectively identify abnormal targets of different categories. The detection results contain confidence information, which facilitates the screening of highly reliable abnormal areas and reduces false detections and missed detections.

[0045] The YOLO algorithm can detect various types of abnormal targets and output accurate target categories and location coordinates, which can be used to detect complex multi-target scenes in inspection tasks.

[0046] The optimized flight route generated by the drone in step 2 aims to minimize the total energy consumption of the path. The energy consumption calculation formula is: , in, is the total energy consumption of the flight path, is the hovering power of the drone, Forward power of the drone, is the total flight time.

[0047] The flight attitude monitoring in step 3 includes the pitch angle, roll angle and yaw angle of the drone, and the attitude calculation formula is: , in, is the attitude matrix of the UAV, is the yaw angle transformation matrix rotating around the z-axis, is the pitch angle transformation matrix rotating around the y-axis, is the roll angle transformation matrix rotating around the x-axis; The UAV determines the degree of flight attitude deviation by calculating the above attitude matrix in real time, and adjusts the flight control parameters according to the deviation to correct the attitude.

[0048] Step 2 optimizes the flight route with the goal of minimizing the total energy consumption of the path, effectively reducing energy consumption, increasing flight time, and ensuring that the UAV performs its mission with maximum efficiency in complex environments.

[0049] Step 3 calculates the attitude matrix of the UAV in real time, monitors and corrects the flight attitude deviation, improves flight stability and path tracking accuracy, and reduces the impact of external disturbances on mission execution.

[0050] In step 6, when the image data collected by the drone is preprocessed by edge computing, the target detection confidence Calculated by the following formula: , in, is the target detection confidence, which indicates the probability that the target detection is a correct result; is the prediction score of target i, indicating the credibility of the model for target i; N is the total number of target categories, and e is the base of the natural logarithm; The drone screens the targets based on the calculated target detection confidence, outputs high-confidence abnormal target information, and transmits the abnormal target information to the ground station for subsequent generation of inspection reports.

[0051] In step 6, the present invention performs real-time target detection on the image data through edge computing and calculates the target detection confidence , filtering out high-confidence abnormal target information for output and transmission, which has the following benefits: Improve the reliability and accuracy of target detection results; optimize the data transmission process and reduce the pressure on transmission bandwidth; achieve fast and real-time identification and screening of abnormal targets; support the generation of high-quality inspection reports and provide reliable data support; enhance the autonomous data analysis and processing capabilities of drones and reduce dependence on ground stations.

[0052] Through the screening mechanism of target detection confidence, the present invention realizes efficient and accurate abnormal target recognition and information transmission on the UAV side, significantly improves the data processing efficiency and result reliability of inspection tasks, and provides strong technical support for the automation and intelligence of inspection work.

[0053] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A flight control management method for an inspection drone, characterized in that: include: Step 1: The drone obtains the 3D point cloud data of the environment through the lidar, captures the terrain visual information through the RGB camera and the depth camera, obtains the real-time position and attitude information of the drone through the GPS and the inertial measurement unit, and then fuses the collected data to construct a spatial grid map, which is used as the input data for path planning; Step 2: Using the grid map obtained by 3D environment perception and modeling as input, determine the starting point, target point and flight constraints of the UAV, generate the initial flight route of the UAV through path planning, monitor the changes of obstacles and dynamic environment during the flight of the UAV in real time, superimpose the detected dynamic environment change information on the grid map, and readjust the route according to the superimposed updated map to obtain the optimized flight path; Step 3: Based on the flight path generated by route planning and dynamic optimization, monitor the status data of the drone in real time during the flight of the route, calculate the battery power as the ratio of the remaining battery capacity to the total battery capacity, use the monitored flight control hardware status information to determine whether there is a hardware anomaly, and use the flight attitude data to determine whether the drone deviates from the route; Step 4: Based on the flight status monitoring, determine whether emergency treatment is needed according to the monitoring results of the drone's battery power status, flight control hardware status and flight attitude information. If the battery power is below the preset threshold, adjust the route to return to the charging station for charging. If the flight control hardware is abnormal, record the abnormal information and stop the mission. If the drone deviates from the route, correct the flight according to the route planning and dynamically optimized path. Step 5: If the UAV performs autonomous emergency response, the task scheduling and collaborative takeover process is triggered. The unfinished task area of ​​the UAV is allocated to the neighboring UAVs through the task area division method, and the task allocation is dynamically adjusted based on the parameters of the UAV. Step 6. During the execution of the task, the drone collects images and environmental data of the inspection area through the onboard sensors. During task scheduling and collaborative takeover, the drone continues to collect data from the unfinished area. The collected data is preliminarily processed by the edge computing module on the drone side. The processed data includes the detection results of the abnormal area and relevant information of the drone's inspection coverage. The processed data is then transmitted back to the ground station for the generation of the inspection report.

2. The flight control management method of an inspection drone according to claim 1, characterized in that: The spatial grid map constructed in step 1 is generated by the following method: 3D point cloud data based on LiDAR , rasterize the 3D point cloud data, decompose the environment into voxel units, and generate a 3D grid map G: ， in, is the grid cell state, 1 means it is occupied by an obstacle, 0 means it is free; is the index coordinate of the grid cell.

3. The flight control management method of an inspection drone according to claim 1, characterized in that: The initial flight route generated in step 2 is dynamically optimized using a real-time updated grid map, and the flight path is updated and planned according to the occupancy status in the grid map; The battery capacity in step 3 is calculated by the following formula: ， in, is the percentage of remaining battery capacity, Remaining battery capacity, is the total battery capacity.

4. The flight control management method of an inspection drone according to claim 1, characterized in that: The flight control hardware state in step 3 is predicted by the following system state equation: ， in, is the current system status, is the system state transition matrix, is the control input vector, is the input matrix, is the system noise, Is the system in time The state vector at time .

5. The flight control management method of an inspection drone according to claim 1, characterized in that: The correction of the deviation of the drone from the route in step 4 is to provide a reference for the dynamically updated flight path, and the correction includes adjusting the drone's attitude, flight speed and direction; The task area division in step 5 is based on the Voronoi segmentation algorithm, and the area division formula is: ， in, Assigned to drones mission area, is any point in the mission area, Yes To drones The Euclidean distance of and yes The special points in Indicate point To drones The Euclidean distance.

6. The flight control management method of an inspection drone according to claim 1, characterized in that: During the task takeover process, the task priority of the UAV is dynamically adjusted. The task priority is calculated by the following formula: ， in, is the task priority, is the estimated completion time of the task, It is the urgency of the task. and is the weight parameter; The data collected in step 6 includes high-definition images, infrared data and environmental parameter information. The collected data is pre-processed in real time through edge computing equipment. The data pre-processing includes compression, anomaly detection and feature extraction.

7. The flight control management method of an inspection drone according to claim 1, characterized in that: In step 6, a deep learning algorithm is used to detect inspection anomalies, and the target detection result is expressed by the following formula: ， in, is the target detection result, is the input image, are the parameters of the deep learning network, Represents the forward prediction function of the YOLO algorithm.

8. The flight control management method of an inspection drone according to claim 1, characterized in that: The optimized flight route generated by the UAV in step 2 aims to minimize the total energy consumption of the path, and the energy consumption calculation formula is: ， in, is the total energy consumption of the flight path, is the hovering power of the drone, The forward power of the drone, is the total flight time.

9. The flight control management method of an inspection UAV according to claim 8, characterized in that: The flight attitude monitoring in step 3 includes the pitch angle, roll angle and yaw angle of the UAV, and the attitude calculation formula is: ， in, is the attitude matrix of the UAV, is the yaw angle transformation matrix rotating around the z-axis, is the pitch angle transformation matrix rotating around the y-axis, is the roll angle transformation matrix rotating around the x-axis; The UAV determines the degree of flight attitude deviation by calculating the above attitude matrix in real time, and adjusts the flight control parameters according to the deviation to correct the attitude.

10. The flight control management method of an inspection drone according to claim 1, characterized in that: When the image data collected by the drone in step 6 is preprocessed by edge computing, the target detection confidence Calculated by the following formula: ， in, is the target detection confidence, which indicates the probability that the target detection is a correct result; is the prediction score of target i, indicating the credibility of the model for target i; N is the total number of target categories, and e is the base of the natural logarithm; The drone screens the targets based on the calculated target detection confidence, outputs high-confidence abnormal target information, and transmits the abnormal target information to the ground station for subsequent generation of inspection reports.

Images