Unmanned aerial vehicle inspection method, system and device for high-speed railway tunnel portal

Abstract

This invention discloses a method, system, and apparatus for unmanned aerial vehicle (UAV) inspection of high-speed railway tunnel entrances. The method includes: determining a target event point based on an event inspection route sent by a ground station, and controlling the UAV to fly blindly to the target event point, where the target event point includes the coordinates of the target tunnel entrance; adjusting the UAV's attitude based on the UAV's real-time coordinates and the target tunnel entrance coordinates to obtain a target shooting posture; controlling an intelligent pod to perform visual intelligent recognition of the target tunnel entrance based on the target shooting posture, and determining the center coordinates of the target tunnel entrance; calculating the target distance based on the real-time coordinates and the center coordinates, and adjusting the zoom parameters of the video acquisition module in the intelligent pod to focus on the target tunnel entrance based on the target distance; and controlling the video acquisition module to acquire inspection video of the target tunnel entrance based on a preset acquisition duration to complete the inspection of the target tunnel entrance. This invention automates the inspection of target tunnel entrances and improves inspection accuracy.

Description

Technical Field

[0001] This application relates to the field of unmanned aerial vehicle (UAV) technology, and in particular to a UAV inspection method, system and device for high-speed railway tunnel entrances. Background Technology

[0002] With the rapid development of high-speed railways, safety inspections at high-speed railway tunnel entrances have become a crucial aspect of ensuring train operation safety.

[0003] Currently, existing inspection methods for high-speed railway tunnel entrances typically rely on manually controlled drones to collect video footage from the tunnel entrances. However, manually controlled drone data collection suffers from drawbacks such as inaccurate positioning, lack of automatic zoom adjustment, and difficulty in controlling video dwell time.

[0004] Therefore, existing inspection schemes for high-speed railway tunnel entrances suffer from low automation and insufficient inspection accuracy. Summary of the Invention

[0005] The main purpose of this application is to propose a method, system and device for unmanned aerial vehicle (UAV) inspection of high-speed railway tunnel entrances, aiming to solve the problems of low automation and insufficient inspection accuracy.

[0006] To achieve the above objectives, this application proposes a method for unmanned aerial vehicle (UAV) inspection of high-speed railway tunnel entrances, comprising: The target event point is determined based on the event inspection route sent by the ground station, and the UAV is controlled to fly blindly to the target event point, which includes the coordinates of the target tunnel entrance; The drone's attitude is adjusted based on its real-time coordinates and the target tunnel entrance coordinates to obtain the target shooting posture; Based on the target shooting posture, the intelligent pod is controlled to perform visual intelligent recognition of the target tunnel entrance and determine the center coordinates of the target tunnel entrance. The target distance is calculated based on the real-time coordinates and the center coordinates, and the zoom parameters of the video acquisition module in the smart pod are adjusted according to the target distance to focus on the target tunnel entrance; Based on a preset acquisition duration, the video acquisition module is controlled to acquire inspection videos of the target tunnel entrance in order to complete the inspection of the target tunnel entrance.

[0007] In some embodiments, determining the target event point based on the event inspection route sent by the ground station includes: Receive the event inspection route sent by the ground station. The event inspection route includes the inspection route and multiple event points marked on the inspection route. The event points include the coordinates of the tunnel entrance. Based on the starting point of the inspection route, each event point is classified, and the event points facing the starting point are determined as entry event points, and the event points away from the starting point are determined as exit event points. The target event point is determined from the plurality of entry event points and the plurality of exit event points.

[0008] In some embodiments, the inspection route further includes an endpoint; determining the target event point from the plurality of entry event points and the plurality of exit event points includes: An entrance sorting list is obtained by sorting the entrance event points in ascending order based on the distance between each entrance event point and the starting point, and an exit sorting list is obtained by sorting the exit event points in ascending order based on the distance between each exit event point and the ending point. The exit sorting list is added to the end of the inlet sorting list and the sequence number is updated to obtain the target sorting list; The index of the first element in the target sorted list is taken as the current index. The entry event point or exit event point corresponding to the current sequence number is determined as the target event point; When the inspection of the target tunnel entrance corresponding to the target event point is completed, determine whether the current sequence number is the final sequence number of the target sorting list; If not, increment the current sequence number by 1, and based on the incremented current sequence number, re-execute the step of determining the entry event point or exit event point corresponding to the current sequence number as the target event point; If so, then complete the inspection of all tunnel entrances in the event inspection route.

[0009] In some embodiments, controlling the drone to fly blindly to the target event point includes: The drone is controlled to fly blindly based on the inspection route and the coordinates of the target tunnel entrance, and the distance between the real-time coordinates and the coordinates of the target tunnel entrance is calculated to obtain the ready-to-collect distance; Determine whether the prepared sampling distance is less than the preset sampling distance; If the prepared collection distance is not less than the preset collection distance, the step of controlling the UAV to fly blindly according to the inspection route and the coordinates of the target tunnel entrance will continue to be executed. If the prepared collection distance is less than the preset collection distance, then the UAV is determined to fly blindly to the target event point.

[0010] In some embodiments, adjusting the attitude of the UAV based on its real-time coordinates and the coordinates of the target tunnel entrance to obtain the target shooting posture includes: Transform the real-time coordinates and the target tunnel entrance coordinates to the WGS84 geocentric coordinate system; In the WGS84 geocentric coordinate system, the relative position of the real-time coordinates and the target tunnel entrance coordinates is calculated to obtain the target relative position; The attitude of the drone and the angle of the smart pod are adjusted according to the relative position of the target to obtain the target shooting posture.

[0011] In some embodiments, controlling the intelligent pod to perform visual intelligent recognition of the target tunnel entrance based on the target shooting posture control, and determining the center coordinates of the target tunnel entrance includes: The drone is controlled to maintain the target shooting posture, and the image acquisition module of the intelligent pod is controlled to continuously acquire multiple on-site image frames; A preset tunnel entrance visual recognition model is invoked to identify multiple on-site image frames to obtain multiple tunnel entrance candidate boxes, each of which includes a confidence level. Based on the confidence level of each of the candidate tunnel entrances, a target tunnel entrance is determined from the candidate tunnel entrances. The geometric center pixel coordinates of the target box at the tunnel entrance in the corresponding on-site image frame are calculated to obtain the center coordinates.

[0012] In some embodiments, the step of calculating the target distance based on the real-time coordinates and the center coordinates, and adjusting the zoom parameters of the video acquisition module in the smart pod according to the target distance to focus on the target tunnel entrance includes: The target distance is obtained by calculating the distance between the real-time coordinates and the center coordinates; The target zoom parameters are calculated based on the target distance and the preset zoom algorithm. The video acquisition module is controlled to perform zoom adjustment based on the target zoom parameters in order to focus on the target tunnel entrance.

[0013] In some embodiments, controlling the video acquisition module to acquire inspection video of the target tunnel entrance based on a preset acquisition duration, in order to complete the inspection of the target tunnel entrance, includes: Based on the preset acquisition duration, the video acquisition module is controlled to continuously record the inspection video at the target tunnel entrance; The recording of the inspection video ends when the preset collection time expires; The inspection video is bound to the target event point and stored to complete the inspection of the target tunnel entrance.

[0014] This application further proposes a drone inspection system for high-speed railway tunnel entrances, the drone inspection system for high-speed railway tunnel entrances includes a ground station and a drone, the drone includes a controller and a smart pod; the drone inspection system for high-speed railway tunnel entrances is capable of performing the drone inspection method for high-speed railway tunnel entrances described above.

[0015] This application further proposes a drone inspection device for high-speed railway tunnel entrances, comprising: At least one processor; and, A memory communicatively connected to the at least one processor; wherein, The memory stores instructions that are executed by the at least one processor, which enable the at least one processor to perform the aforementioned unmanned aerial vehicle (UAV) inspection method for high-speed railway tunnel entrances.

[0016] This technical solution determines the target event point based on the event inspection route sent by the ground station and controls the UAV to fly blindly to the target event point, which includes the coordinates of the target tunnel entrance. The UAV's attitude is adjusted based on its real-time coordinates and the target tunnel entrance coordinates to obtain the target shooting posture. Based on the target shooting posture, the intelligent pod is controlled to perform visual intelligent recognition of the target tunnel entrance to determine its center coordinates. The target distance is calculated based on the real-time coordinates and the center coordinates, and the zoom parameters of the video acquisition module in the intelligent pod are adjusted according to the target distance to focus on the target tunnel entrance. Based on a preset acquisition duration, the video acquisition module is controlled to acquire inspection videos of the target tunnel entrance to complete the inspection. By relying on the event inspection route to achieve an integrated operation mode of autonomous blind flight, attitude adjustment, visual intelligent recognition and ranging linkage adjustment of zoom parameters, and timed acquisition of inspection videos, the automated operation of target tunnel entrance inspection is realized, effectively improving the inspection accuracy of the target tunnel entrance. Attached Figure Description

[0017] Figure 1 This is a flowchart illustrating an embodiment of the unmanned aerial vehicle (UAV) inspection method for high-speed railway tunnel entrances according to this application. Figure 2 This is a flowchart illustrating another embodiment of the drone inspection method for high-speed railway tunnel entrances according to this application. Figure 3 This is a flowchart illustrating another embodiment of the drone inspection method for high-speed railway tunnel entrances according to this application. Figure 4 This is a flowchart illustrating another embodiment of the drone inspection method for high-speed railway tunnel entrances according to this application. Figure 5 This is a flowchart illustrating another embodiment of the drone inspection method for high-speed railway tunnel entrances according to this application. Figure 6 This is a flowchart illustrating another embodiment of the drone inspection method for high-speed railway tunnel entrances according to this application. Figure 7 This is a flowchart illustrating another embodiment of the drone inspection method for high-speed railway tunnel entrances according to this application. Figure 8 This is a flowchart illustrating another embodiment of the drone inspection method for high-speed railway tunnel entrances according to this application. Figure 9 This is a schematic diagram of a structural embodiment of the unmanned aerial vehicle (UAV) inspection system for high-speed railway tunnel entrances according to this application. Figure 10 This is a schematic diagram of an embodiment of the unmanned aerial vehicle (UAV) inspection device for high-speed railway tunnel entrances according to this application.

[0018] The realization of the objective, functional features and advantages of the present invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0019] The solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0020] It should be noted that all directional indications (such as up, down, left, right, front, back, etc.) in the embodiments of the present invention are only used to explain the relative positional relationship and movement of each component in a certain specific posture (as shown in the figure). If the specific posture changes, the directional indication will also change accordingly.

[0021] It should also be noted that when a component is described as "fixed to" or "set on" another component, it can be directly on the other component or there may be an intervening component present. When a component is described as "connected to" another component, it can be directly connected to the other component or there may be an intervening component present.

[0022] Furthermore, the use of terms such as "first" and "second" in this invention is for descriptive purposes only and should not be construed as indicating or implying their relative importance or implicitly specifying the number of technical features indicated. Therefore, a feature defined with "first" or "second" may explicitly or implicitly include at least one of that feature. Additionally, the technical solutions of the various embodiments can be combined with each other, but only on the basis of being achievable by those skilled in the art. When the combination of technical solutions is contradictory or impossible to implement, such a combination of technical solutions should be considered non-existent and not within the scope of protection claimed by this invention.

[0023] This application proposes a method for unmanned aerial vehicle (UAV) inspection of high-speed railway tunnel entrances, referring to... Figure 1 , Figure 1 This is a flowchart illustrating an embodiment of the drone inspection method for high-speed railway tunnel entrances according to this application. In some embodiments, the drone inspection method for high-speed railway tunnel entrances includes: Step S110: Determine the target event point based on the event inspection route sent by the ground station, and control the UAV to fly blindly to the target event point, which includes the coordinates of the target tunnel entrance; Step S120: Adjust the drone's attitude according to the drone's real-time coordinates and the target tunnel entrance coordinates to obtain the target shooting posture; Step S130: Based on the target shooting posture control, control the intelligent pod to perform visual intelligent recognition of the target tunnel entrance and determine the center coordinates of the target tunnel entrance; Step S140: Calculate the target distance based on the real-time coordinates and the center coordinates, and adjust the zoom parameters of the video acquisition module in the smart pod according to the target distance to focus on the target tunnel entrance; Step S150: Based on the preset acquisition duration, control the video acquisition module to acquire inspection video of the target tunnel entrance in order to complete the inspection of the target tunnel entrance.

[0024] In this embodiment, the UAV inspection method for high-speed railway tunnel entrances can be configured as an operating program or functional software, or encapsulated as airborne firmware, control driver, or independent functional module; then it is deployed to the UAV's controller so that the controller can run the UAV inspection method for high-speed railway tunnel entrances. Alternatively, it can be deployed to other types of intelligent control devices (e.g., ground maintenance terminals, industrial control hosts, remote dispatch and management platforms) so that these devices can run the UAV inspection method for high-speed railway tunnel entrances.

[0025] Understandably, a high-speed railway tunnel entrance drone inspection system comprises a ground station and drones. The drones consist of a controller and a smart pod. The ground station primarily generates and edits inspection routes, delineates event points, and receives real-time inspection videos, enabling remote scheduling, status monitoring, and data storage management. The drones, acting as aerial inspection vehicles, autonomously fly and arrive at designated points according to the assigned inspection routes, carrying out the overall inspection operation. The controller, the core control unit of the drone, is responsible for parsing inspection commands, controlling flight attitude, planning flight paths, and coordinating the operation of various functional components. The smart pod is equipped with cameras, image acquisition modules, video acquisition modules, and other functional components, enabling intelligent target identification at the tunnel entrance, distance calculation, zoom and focus adjustment, and standardized video acquisition for on-site perception and data collection.

[0026] When operating a drone inspection method for high-speed railway tunnel entrances, the target event point can be determined first based on the event inspection route sent by the ground station, and the drone can then be controlled to fly blindly to the target event point. The target event point includes the coordinates of the target tunnel entrance. For example, based on the drone's flight range, the user can pre-collect a railway KML (Keyhole Markup Language) file along a section of high-speed railway (e.g., the drone's flight range can be more than twice the distance of that section of high-speed railway). The railway KML file can include the trajectory coordinate data of that section of high-speed railway and the geographical coordinates of each tunnel entrance. Then, the user can import the railway KML file into the ground station. At this point, the ground station can parse the railway KML file, generate an inspection route based on the trajectory coordinate data of that section of high-speed railway, and mark event points on the inspection route according to the geographical coordinates of each tunnel entrance. Then, based on the inspection route and the multiple event points marked on the inspection route, an event inspection route is generated. Finally, the ground station can send the event inspection route to the drone. At this point, the controller in the drone can receive the event inspection route and then determine the target event point based on the event inspection route sent by the ground station.

[0027] For example, after the ground station sends the event inspection route to the drone, the user can also issue inspection commands to the drone through the ground station. These commands could include a first inspection command to inspect a specific tunnel entrance (e.g., tunnel entrance A), a second inspection command to inspect several tunnel entrances (e.g., tunnel entrance A, tunnel entrance B, and tunnel entrance C), and a third inspection command to inspect all tunnel entrances. If the first inspection command is received, the drone can respond by identifying the event point corresponding to that tunnel entrance (e.g., tunnel entrance A) as the target event point. If the second inspection command is received, the drone can respond by sequentially identifying the event points corresponding to the several tunnel entrances (e.g., tunnel entrance A, tunnel entrance B, and tunnel entrance C) as target event points (e.g., first identifying the event point corresponding to tunnel entrance A as the target event point, then identifying the event point corresponding to tunnel entrance B as the target event point after completing the inspection of tunnel entrance A, and finally identifying the event point corresponding to tunnel entrance C as the target event point after completing the inspection of tunnel entrance B). Similarly, upon receiving a third inspection command, the system can respond by sequentially identifying all event points corresponding to tunnel entrances as target event points. After identifying the target event points, the system can control the drone to fly blindly to them. For example, based on the event inspection route and the coordinates of the target tunnel entrance, the drone can be controlled to fly blindly to the vicinity of the target tunnel entrance.

[0028] After the drone is blindly flown to the target event point, its attitude can be adjusted based on the drone's real-time coordinates and the target tunnel entrance coordinates to obtain the target shooting posture. For example, the drone can also include a positioning module that can collect the drone's real-time coordinates. First, the relative position between the real-time coordinates and the target tunnel entrance coordinates is calculated, and then the drone's attitude is adjusted based on the relative position to obtain the target shooting posture.

[0029] After obtaining the target's shooting posture, the intelligent pod can be controlled to perform visual intelligent recognition of the target tunnel entrance based on this posture, determining the center coordinates of the tunnel entrance. For example, the drone can be controlled to maintain the target shooting posture to ensure it is in the optimal shooting position. The intelligent pod can include at least one camera. The camera on the intelligent pod scans the target tunnel entrance and performs visual intelligent recognition to identify the center of the tunnel entrance, then determines its center coordinates. The drone is a vertical takeoff and landing (VTOL) fixed-wing UAV. While maintaining the target shooting posture, the drone is not stationary but continues to fly forward, performing recognition as it flies. VTOL fixed-wing UAVs have long endurance and large payload capabilities, enabling them to perform long-range inspection missions.

[0030] After determining the center coordinates, the target distance can be calculated based on the real-time coordinates and the center coordinates. The zoom parameters of the video acquisition module in the smart pod are then adjusted according to this target distance to focus on the tunnel entrance. For example, the target distance between the drone and the tunnel entrance is calculated by combining the real-time coordinates and the center coordinates. The zoom parameters of the video acquisition module in the smart pod are then adjusted based on this target distance to adjust the camera's focus so that it is focused on the tunnel entrance. Since the drone is not stationary, the target distance is constantly changing, therefore, the adjustment of the zoom parameters of the video acquisition module in the smart pod is a continuous process.

[0031] After focusing on the target tunnel entrance, the video acquisition module can be controlled to collect inspection video of the target tunnel entrance based on a preset acquisition duration to complete the inspection. For example, the preset acquisition duration can be customized by the user, such as 3 seconds. The video acquisition module calls the camera to capture video of the target tunnel entrance for 3 seconds, thus obtaining the inspection video and completing the inspection of the target tunnel entrance. After acquiring the inspection video, the controller can also associate the inspection video with the corresponding target event point and then send the associated inspection video back to the ground station. In this way, the user can view the inspection video through the ground station.

[0032] The technical solution of this application is based on a preset acquisition duration, which controls the video acquisition module to acquire inspection videos of the target tunnel entrance in order to complete the inspection of the target tunnel entrance; by relying on the event inspection route to realize an integrated operation mode of autonomous blind flight, attitude adjustment, visual intelligent recognition and distance measurement linkage adjustment of zoom parameters, and timed acquisition of inspection videos, the inspection of the target tunnel entrance is automated, effectively improving the inspection accuracy of the target tunnel entrance.

[0033] Reference Figure 2 , Figure 2 This is a flowchart illustrating another embodiment of the UAV inspection method for high-speed railway tunnel entrances according to this application. In some embodiments, the aforementioned determination of the target event point based on the event inspection route sent by the ground station includes: Step S160: Receive the event inspection route sent by the ground station. The event inspection route includes the inspection route and multiple event points marked on the inspection route. The event points include the coordinates of the tunnel entrance. Step S161: Based on the starting point of the inspection route, classify each event point, determine the event points facing the starting point as entry event points, and determine the event points away from the starting point as exit event points. Step S162: Determine the target event point from multiple entry event points and multiple exit event points.

[0034] In this embodiment, as Figure 2 As shown, when determining the target event point based on the event inspection route sent by the ground station in step S110, the event inspection route sent by the ground station can be received first. The event inspection route includes the inspection route itself and multiple event points marked on the inspection route. The event points include the coordinates of the tunnel entrance. For example, the UAV controller receives the event inspection route sent by the ground station via a wireless communication link. The event inspection route is constructed based on a railway KML file, containing a complete flight trajectory (inspection route), and multiple event points are marked sequentially along the inspection route. The coordinates of the corresponding tunnel entrance are recorded for each event point.

[0035] After obtaining the inspection route, event points can be classified based on the starting point of the route. Event points facing the starting point are designated as entry event points, and those moving away from the starting point are designated as exit event points. For example, the controller uses the starting position of the entire inspection route as a reference point to compare the orientation of each event point relative to the starting point. If the tunnel entrance corresponding to the event point faces the starting direction of the route, it is designated as an entry event point; if the tunnel entrance corresponding to the event point moves away from the starting direction of the route, it is designated as an exit event point.

[0036] The target event point is determined from multiple entry and exit event points. For example, after the ground station sends the event inspection route to the drone, the user can also issue inspection commands to the drone through the ground station. For instance, if the inspection command is for a single entry or exit event point, the controller can determine the target event point from that entry or exit event point. If the command is for all entry event points, the controller can determine the target event point from each entry event point sequentially. If the command is for all exit event points, the controller can determine the target event point from each exit event point sequentially. If the command is for both entry and exit event points, the controller can first determine the target event point from each entry event point, then determine the target event point from the corresponding exit event point, and so on; or, the controller can first determine the target event point from each entry event point sequentially, and then determine the target event point from each exit event point sequentially. When inspecting multiple event points, it is necessary to complete the inspection of the previous target event point before determining the next target event point.

[0037] Reference Figure 3 , Figure 3 This is a flowchart illustrating another embodiment of the UAV inspection method for high-speed railway tunnel entrances according to this application. In some embodiments, the inspection route further includes an endpoint; the aforementioned determination of the target event point from multiple entrance event points and multiple exit event points includes: Step S170: Sort the entry points in ascending order based on the distance between each entry event point and the starting point to obtain an entry sort list, and sort the exit points in ascending order based on the distance between each exit event point and the ending point to obtain an exit sort list. Step S171: Add the exit sorting list to the end of the entry sorting list and update the sequence number to obtain the target sorting list; Step S172: Take the first index of the target sorting list as the current index; Step S173: Determine the entry event point or exit event point corresponding to the current sequence number as the target event point; Step S174: When the inspection is completed at the target tunnel entrance corresponding to the target event point, determine whether the current sequence number is the final sequence number of the target sorting list; Step S175: If not, increment the current sequence number by 1, and re-execute the step of determining the entry event point or exit event point corresponding to the current sequence number as the target event point based on the incremented current sequence number. Step S176: If yes, then complete the inspection of all tunnel entrances in the event inspection route.

[0038] In this embodiment, as Figure 3As shown, during step S162, each entry event point and each exit event point can be sorted. The inspection route also includes the endpoint. After the ground station sends the event inspection route to the drone, the user can also issue inspection commands to the drone through the ground station. If the inspection command is for inspecting all entry event points and all exit event points, the controller can respond to the command, sorting each entry event point in ascending order based on its distance from the starting point to obtain an entry sorting list, and then sorting each exit event point in ascending order based on its distance from the endpoint to obtain an exit sorting list. For example: the controller calculates the linear distance of each entry event point relative to the starting point, sorts them in ascending order from closest to furthest distance to generate an entry sorting list; then it calculates the linear distance of each exit event point relative to the endpoint, sorting them in ascending order from closest to furthest distance to obtain an exit sorting list.

[0039] The exit sorting list is added to the end of the entrance sorting list and its sequence number is updated to obtain the target sorting list. For example, the controller concatenates the sorted exit sorting list to the end of the entrance sorting list and re-numbers them sequentially to form a complete target sorting list, determining the unified inspection order of all tunnel entrances. For instance, if there are 5 entrance event points, the sequence numbers of the entrance sorting list would be 1, 2, 3, 4, 5. If there are 5 exit event points, the sequence numbers of the exit sorting list would be 1, 2, 3, 4, 5. The target sorting list would then have the entrance event points numbered 1, 2, 3, 4, 5, and the exit event points numbered 6, 7, 8, 9, 10.

[0040] Use the first index of the target sort list as the current index. For example, the controller selects the first index of the target sort list and sets it as the initial current index.

[0041] The entry or exit event point corresponding to the current sequence number is determined as the target event point. For example, the controller matches the corresponding entry or exit event point based on the current sequence number and determines the entry or exit event point as the target event point to be inspected in this inspection.

[0042] When the inspection of the target tunnel entrance corresponding to the target event point is completed, it is determined whether the current sequence number is the final sequence number in the target sorting list. For example, after the inspection of the target tunnel entrance corresponding to the target event point is completed, the controller can compare the current sequence number with the final sequence number at the end of the target sorting list to determine whether all sequence numbers have been traversed.

[0043] If not, increment the current sequence number by 1, and re-execute the step of determining the entry or exit event point corresponding to the current sequence number as the target event point based on the incremented current sequence number. For example, if the current sequence number is not the final sequence number of the target sorting list, it means that all sequence numbers have not been traversed. In this case, the controller can increment the value of the current sequence number by 1, and re-execute step S173 based on the new current sequence number obtained after incrementing.

[0044] If so, the inspection of all tunnel entrances in the event inspection route is completed. For example, if the current sequence number is the final sequence number in the target sorting list, it means that all sequence numbers have been traversed. At this time, the controller determines that the inspection of all tunnel entrances in the event inspection route has been completed.

[0045] Reference Figure 4 , Figure 4 This is a flowchart illustrating another embodiment of the unmanned aerial vehicle (UAV) inspection method for high-speed railway tunnel entrances according to this application. In some embodiments, the aforementioned control of the UAV to blindly fly to the target event point includes: Step S180: Control the UAV to fly blindly according to the inspection route and the coordinates of the target tunnel entrance, and calculate the distance between the real-time coordinates and the coordinates of the target tunnel entrance to obtain the ready-to-collect distance. Step S181: Determine whether the prepared sampling distance is less than the preset sampling distance; Step S182: If the prepared collection distance is not less than the preset collection distance, continue to execute the step of controlling the drone to fly blindly according to the inspection route and the coordinates of the target tunnel entrance. Step S183: If the distance to be collected is less than the preset distance, then the UAV will fly blindly to the target event point.

[0046] In this embodiment, as Figure 4 As shown, when controlling the UAV to blindly fly to the target event point in step S110, the UAV can be controlled to fly blindly based on the inspection route and the coordinates of the target tunnel entrance. The distance between the real-time coordinates and the target tunnel entrance coordinates is calculated to obtain the prepared data acquisition distance. For example, the controller relies on the inspection route and the target tunnel entrance coordinates to autonomously control the UAV to fly along the planned path without manual remote control. During flight, the UAV's positioning module can collect its own real-time coordinates, the controller can obtain these real-time coordinates, and then calculate the straight-line distance between the real-time coordinates and the target tunnel entrance coordinates to obtain the prepared data acquisition distance.

[0047] After obtaining the planned sampling distance, it can be determined whether the planned sampling distance is less than the preset sampling distance. The preset sampling distance can be customized by the user; for example, the preset sampling distance can be set to 200 meters.

[0048] If the planned data collection distance is not less than the preset collection distance, the controller will continue to execute the steps of controlling the drone to fly blindly based on the inspection route and the coordinates of the target tunnel entrance. For example, if the planned data collection distance is not less than the preset collection distance, it can be determined that the drone has not yet reached the compliant inspection range of the target tunnel entrance. At this time, the controller can continue to execute the steps of controlling the drone to fly blindly based on the inspection route and the coordinates of the target tunnel entrance to control the drone to continue flying blindly to the target event point.

[0049] If the planned data collection distance is less than the preset data collection distance, the drone will be determined to have blindly flown to the target event point. For example, if the planned data collection distance is less than the preset data collection distance, it can be determined that the drone has entered the compliance inspection range of the target tunnel entrance. At this time, the controller can determine that the drone has blindly flown to the target event point.

[0050] Reference Figure 5 , Figure 5 This is a flowchart illustrating another embodiment of the UAV inspection method for high-speed railway tunnel entrances according to this application. In some embodiments, the aforementioned adjustment of the UAV's attitude based on the UAV's real-time coordinates and the target tunnel entrance coordinates to obtain the target shooting posture includes: Step S190: Convert the real-time coordinates and the target tunnel entrance coordinates to the WGS84 geocentric coordinate system; Step S191: In the WGS84 geocentric coordinate system, calculate the relative position between the real-time coordinates and the target tunnel entrance coordinates to obtain the target relative position; Step S192: Adjust the attitude of the drone and the angle of the smart pod according to the relative position of the target to obtain the target shooting posture.

[0051] In this embodiment, as Figure 5 As shown, during step S120, the real-time coordinates and the target tunnel entrance coordinates can be converted to the WGS84 geocentric coordinate system (World Geodetic System-1984 Coordinate System). The controller can uniformly convert the real-time coordinates and the target tunnel entrance coordinates to the WGS84 geocentric coordinate system to complete the coordinate mapping and eliminate the position calculation deviation caused by different coordinate systems.

[0052] In the WGS84 geocentric coordinate system, the relative position of the target is obtained by calculating the relative position between the real-time coordinates and the target tunnel entrance coordinates. For example, the real-time coordinates and the target tunnel entrance coordinates can be spatial coordinates, which include the X-axis, Y-axis, and Z-axis. The relative position of each axis can be calculated separately to obtain the target's relative position.

[0053] Assume the real-time coordinates are (X1, Y1, Z1) and the target tunnel entrance coordinates are (X2, Y2, Z2); then ΔX = X2 - X1, ΔY = Y2 - Y1, ΔZ = Z2 - Z1; and the target's relative position is (ΔX, ΔY, ΔZ).

[0054] The drone's attitude and the pod's angle are adjusted based on the target's relative position to obtain the target's shooting posture. For example, when the target's relative position is (ΔX, ΔY, ΔZ), the controller can call a preset flight control algorithm to adjust the drone's attitude and the pod's angle according to (ΔX, ΔY, ΔZ), placing the pod at the optimal shooting angle to obtain the target's shooting posture. The flight control algorithm can be a PID (Proportional-Integral-Derivative) closed-loop attitude control algorithm.

[0055] Reference Figure 6 , Figure 6 This is a flowchart illustrating another embodiment of the UAV inspection method for high-speed railway tunnel entrances according to this application. In some embodiments, the aforementioned step of controlling the intelligent pod to perform visual intelligent recognition of the target tunnel entrance based on the target shooting posture and determining the center coordinates of the target tunnel entrance includes: Step S200: Control the drone to maintain the target shooting posture, and control the image acquisition module of the smart pod to continuously acquire multiple on-site image frames; Step S201: Call the preset tunnel entrance visual recognition model to recognize multiple on-site image frames to obtain multiple tunnel entrance candidate boxes, and the tunnel entrance candidate boxes include confidence scores. Step S202: Determine the target box of the tunnel entrance from multiple candidate boxes based on the confidence level of each candidate box. Step S203: Calculate the geometric center pixel coordinates of the target box at the tunnel entrance in the corresponding on-site image frame to obtain the center coordinates.

[0056] In this embodiment, as Figure 6 As shown, during step S130, the drone can be controlled to maintain the target shooting posture. The drone is controlled to maintain the target shooting posture, and the image acquisition module of the intelligent pod is controlled to continuously acquire multiple on-site image frames. For example, the controller keeps the drone flying stably forward while maintaining the target shooting posture, avoiding attitude swaying that could interfere with the imaging effect. Simultaneously, the image acquisition module inside the intelligent pod is controlled to continuously capture on-site images at a fixed frame rate, continuously outputting multiple on-site image frames.

[0057] A preset tunnel entrance visual recognition model is invoked to identify multiple on-site image frames, resulting in multiple tunnel entrance candidate boxes. Each candidate box includes a confidence level. For example, the controller invokes the preset tunnel entrance visual recognition model to perform feature detection and target matching on each on-site image frame in sequence, identifying and outputting multiple tunnel entrance candidate boxes surrounding the tunnel entrance area. Each candidate box is accompanied by a confidence level representing the reliability of the recognition.

[0058] Based on the confidence level of each candidate tunnel entrance bounding box, the target tunnel entrance bounding box is determined from multiple candidate boxes. For example, the controller uses the confidence level as the filtering criterion, compares the confidence levels of all candidate tunnel entrance bounding boxes, eliminates interference boxes with low confidence, and selects the box with the highest confidence level as the target tunnel entrance bounding box. When multiple candidate tunnel entrance bounding boxes have equal or similar confidence levels, the candidate box closest to the center of the on-site image frame can be preferentially selected as the target tunnel entrance bounding box to reduce false detections and interference from edge targets.

[0059] Calculate the geometric center pixel coordinates of the tunnel entrance target frame in the corresponding field image frame to obtain the center coordinates. For example, the controller can first determine the location of the tunnel entrance target frame in the corresponding field image frame, and then use the pixels of the field image frame as the measurement reference to calculate the position of the geometric center point of the tunnel entrance target frame outline, and convert the corresponding pixel coordinates to obtain the center coordinates of the target tunnel entrance.

[0060] Assuming the target bounding box at the tunnel entrance is I in the corresponding on-site image frame, the formula for calculating the center coordinates is: ; , ; Where Ct represents the set of contour points of the target box at the tunnel entrance, N represents the total number of contour points, and (x,y) represents the pixel coordinates in the on-site image frame I(x,y). Let θ represent the gradient magnitude of the image frame I at pixel coordinates (x, y), and let θ represent the edge detection threshold; then the center coordinates are (xc, yc).

[0061] Specifically, the gradient components in the horizontal and vertical directions are calculated based on the grayscale difference between adjacent pixels, and then the gradient magnitude is calculated using the gradient components in the horizontal and vertical directions.

[0062] Reference Figure 7 , Figure 7 This is a flowchart illustrating another embodiment of the UAV inspection method for high-speed railway tunnel entrances according to this application. In some embodiments, the aforementioned calculation of the target distance based on real-time coordinates and center coordinates, and adjustment of the zoom parameters of the video acquisition module in the intelligent pod to focus on the target tunnel entrance based on the target distance, includes: Step S210: Calculate the distance between the real-time coordinates and the center coordinates to obtain the target distance; Step S211: Calculate the target zoom parameters based on the target distance and the preset zoom algorithm; Step S212: Control the video acquisition module to complete the zoom adjustment according to the target zoom parameters in order to focus on the target tunnel entrance.

[0063] In this embodiment, as Figure 7 As shown, during step S140, the target distance is first calculated by measuring the distance between the real-time coordinates and the center coordinates. The controller can first convert the center coordinates to the WGS84 geocentric coordinate system, and then use the three-dimensional Euclidean distance formula to calculate the distance between the real-time coordinates and the center coordinates, thereby obtaining the target distance.

[0064] The target zoom parameters are calculated based on the target distance and a preset zoom algorithm. For example, the controller substitutes the target distance into the preset zoom algorithm to calculate the target zoom parameters.

[0065] The controller controls the video acquisition module to adjust the zoom level based on the target zoom parameters to focus on the target tunnel entrance. For example, the controller controls the video acquisition module to drive the camera to perform optical zoom based on the target zoom parameters to focus on the target tunnel entrance.

[0066] For example, if the real-time coordinates are (X1, Y1, Z1) and the center coordinates are (xc, yc), after converting the center coordinates to the WGS84 geocentric coordinate system, we get (Xc, Yc, Zc).

[0067] The target distance is obtained by calculating the distance between the real-time coordinates and the center coordinates: ; Where D represents the target distance.

[0068] Based on the target distance and the preset zoom algorithm, the target zoom parameters are calculated as follows: ; Where Z represents the target zoom parameter, Zmin represents the minimum zoom parameter, and k represents the scaling factor.

[0069] The minimum zoom parameter is the parameter measured when the camera leaves the factory, and the scaling factor is the distance-zoom mapping factor measured when the camera leaves the factory.

[0070] Reference Figure 8 , Figure 8 This is a flowchart illustrating another embodiment of the UAV inspection method for high-speed railway tunnel entrances according to this application. In some embodiments, the aforementioned step of controlling the video acquisition module to acquire inspection videos of the target tunnel entrance based on a preset acquisition duration to complete the inspection of the target tunnel entrance includes: Step S220: Based on the preset acquisition duration, control the video acquisition module to continuously record the inspection video of the target tunnel entrance; Step S221: When the preset collection time ends, stop recording the inspection video; Step S222: Bind and store the inspection video with the target event point to complete the inspection of the target tunnel entrance.

[0071] In this embodiment, as Figure 8 As shown, during step S150, the video acquisition module can be controlled to continuously record the inspection video of the target tunnel entrance based on a preset acquisition duration. After the camera focuses on the target tunnel entrance, the controller can control the video acquisition module to continuously record the inspection video of the target tunnel entrance. The controller continuously acquires real-time images of the tunnel entrance according to a preset acquisition duration (e.g., 3 seconds) as the recording cycle, and the video frame sequence is synchronously written to the buffer to ensure that the recording process is continuous and uninterrupted.

[0072] The recording of the inspection video ends when the preset acquisition time is reached. For example, the controller can use a built-in timer to count the recording time in real time. When the recording time reaches the preset acquisition time, it will automatically control the video acquisition module to stop recording and terminate the video recording process, at which point a complete inspection video can be obtained.

[0073] The inspection video is bound to and stored with target event points to complete the inspection of the target tunnel entrance. For example, after obtaining the complete inspection video, the controller can bind and store the video with the target event points to complete the inspection of the target tunnel entrance. Furthermore, the controller can also send the inspection video bound to the target event points to the ground station, allowing users to view the inspection video through the ground station.

[0074] The technical solution of this application is based on a preset acquisition duration, which controls the video acquisition module to acquire inspection videos of the target tunnel entrance in order to complete the inspection of the target tunnel entrance; by relying on the event inspection route to realize an integrated operation mode of autonomous blind flight, attitude adjustment, visual intelligent recognition and distance measurement linkage adjustment of zoom parameters, and timed acquisition of inspection videos, the inspection of the target tunnel entrance is automated, effectively improving the inspection accuracy of the target tunnel entrance.

[0075] This application further proposes a drone inspection system for high-speed railway tunnel entrances, referring to... Figure 9 , Figure 9 This is a schematic diagram of the structure of an embodiment of the unmanned aerial vehicle (UAV) inspection system for high-speed railway tunnel entrances according to this application. In some embodiments, the UAV inspection system for high-speed railway tunnel entrances includes a ground station and a UAV, wherein the UAV includes a controller and a smart pod; the UAV inspection system for high-speed railway tunnel entrances is capable of performing the UAV inspection method for high-speed railway tunnel entrances described above.

[0076] In this embodiment, as Figure 9 As shown, the UAV inspection system at the entrance of a high-speed railway tunnel includes a ground station and UAVs. The UAVs consist of a controller and a smart pod. The ground station is primarily used to generate and edit inspection routes, delineate event points, and receive real-time inspection videos, enabling remote scheduling, status monitoring, and data storage management. The UAVs, serving as the aerial inspection platform, autonomously fly and arrive at designated points according to the issued inspection routes, carrying out the overall inspection operation. The controller, the core control unit of the UAV, is responsible for parsing inspection commands, controlling flight attitude, planning flight paths, and coordinating the collaborative operation of various functional components. The smart pod is equipped with cameras, image acquisition modules, video acquisition modules, and other functional components, enabling on-site perception and data collection tasks such as intelligent target identification at the tunnel entrance, distance calculation, zoom and focus adjustment, and standardized video acquisition. The UAVs are vertical take-off and landing (VTOL) fixed-wing UAVs. VTOL fixed-wing UAVs possess long endurance and high payload capabilities, enabling long-range inspection missions.

[0077] This application further proposes a drone inspection device for high-speed railway tunnel entrances, referring to... Figure 10 , Figure 10 This is a schematic diagram of a structure of an embodiment of the drone inspection device for high-speed railway tunnel entrances according to this application. In some embodiments, the drone inspection device for high-speed railway tunnel entrances includes: At least one processor; and, A memory that is communicatively connected to at least one processor; wherein, The memory stores instructions that are executed by the at least one processor, which enable the at least one processor to perform the unmanned aerial vehicle (UAV) inspection method for high-speed railway tunnel entrances as described above.

[0078] In this embodiment, refer to Figure 10 The drone inspection device for high-speed railway tunnel entrances in this application embodiment can be a processor capable of running a drone inspection method for high-speed railway tunnel entrances; there is at least one processor. For example... Figure 10As shown, the UAV inspection device at the entrance of the high-speed railway tunnel may include: a processor 1001 (e.g., CPU), a network interface 1004, a user interface 1003, a memory 1005, and a communication bus 1002. The communication bus 1002 is used to establish communication between these components. The user interface 1003 may include a display screen and an input unit, such as a keyboard; optionally, the user interface 1003 may also include a standard wired interface or a wireless interface. The network interface 1004 may optionally include a standard wired interface or a wireless interface (such as a Wi-Fi interface). The memory 1005 may be high-speed RAM or non-volatile memory, such as a disk drive. Optionally, the memory 1005 may also be a storage device independent of the aforementioned processor 1001.

[0079] Those skilled in the art will understand that Figure 10 The structure of the drone inspection device at the entrance of a high-speed railway tunnel shown in the figure does not constitute a limitation on the drone inspection device at the entrance of a high-speed railway tunnel. It may include more or fewer components than shown, or combine certain components, or have different component arrangements.

[0080] like Figure 10 As shown, the memory 1005, which serves as a computer storage medium, may include an operating system, a network communication module, a user interface module, and computer programs.

[0081] exist Figure 10 In the UAV inspection device for the high-speed railway tunnel entrance shown, the network interface 1004 is mainly used to connect to the backend server and communicate with the backend server; the user interface 1003 is mainly used to connect to the client (user end) and communicate with the client; and the processor 1001 can be used to call the computer program stored in the memory 1005. When the computer program is called and executed by the processor 1001, it implements the steps of the above-mentioned UAV inspection method for the high-speed railway tunnel entrance.

[0082] The above description is only a part or preferred embodiment of the present invention. Neither the text nor the drawings should limit the scope of protection of the present invention. All equivalent structural transformations made using the content of the present invention specification and drawings under the overall concept of the present invention, or direct / indirect applications in other related technical fields, are included within the scope of protection of the present invention.

Claims

1. A method for unmanned aerial vehicle (UAV) inspection of high-speed railway tunnel entrances, characterized in that, include: The target event point is determined based on the event inspection route sent by the ground station, and the UAV is controlled to fly blindly to the target event point, which includes the coordinates of the target tunnel entrance; The drone's attitude is adjusted based on its real-time coordinates and the target tunnel entrance coordinates to obtain the target shooting posture; Based on the target shooting posture, the intelligent pod is controlled to perform visual intelligent recognition of the target tunnel entrance and determine the center coordinates of the target tunnel entrance. The target distance is calculated based on the real-time coordinates and the center coordinates, and the zoom parameters of the video acquisition module in the smart pod are adjusted according to the target distance to focus on the target tunnel entrance; Based on a preset acquisition duration, the video acquisition module is controlled to acquire inspection videos of the target tunnel entrance in order to complete the inspection of the target tunnel entrance. The process of determining the target event point based on the event inspection route sent by the ground station includes: Receive the event inspection route sent by the ground station. The event inspection route includes the inspection route and multiple event points marked on the inspection route. The event points include the coordinates of the tunnel entrance. Based on the starting point of the inspection route, each event point is classified, and the event points facing the starting point are determined as entry event points, and the event points away from the starting point are determined as exit event points. The target event point is determined from the plurality of entry event points and the plurality of exit event points.

2. The method for unmanned aerial vehicle (UAV) inspection of high-speed railway tunnel entrances according to claim 1, characterized in that, The inspection route also includes a destination; determining the target event point from the plurality of entry event points and the plurality of exit event points includes: An entrance sorting list is obtained by sorting the entrance event points in ascending order based on the distance between each entrance event point and the starting point, and an exit sorting list is obtained by sorting the exit event points in ascending order based on the distance between each exit event point and the ending point. The exit sorting list is added to the end of the inlet sorting list and the sequence number is updated to obtain the target sorting list; The index of the first element in the target sorted list is taken as the current index. The entry event point or exit event point corresponding to the current sequence number is determined as the target event point; When the inspection of the target tunnel entrance corresponding to the target event point is completed, determine whether the current sequence number is the final sequence number of the target sorting list; If not, increment the current sequence number by 1, and based on the incremented current sequence number, re-execute the step of determining the entry event point or exit event point corresponding to the current sequence number as the target event point; If so, then complete the inspection of all tunnel entrances in the event inspection route.

3. The method for unmanned aerial vehicle (UAV) inspection of high-speed railway tunnel entrances according to claim 1, characterized in that, The process of controlling the drone to fly blindly to the target event point includes: The drone is controlled to fly blindly based on the inspection route and the coordinates of the target tunnel entrance, and the distance between the real-time coordinates and the coordinates of the target tunnel entrance is calculated to obtain the ready-to-collect distance; Determine whether the prepared sampling distance is less than the preset sampling distance; If the prepared collection distance is not less than the preset collection distance, the step of controlling the UAV to fly blindly according to the inspection route and the coordinates of the target tunnel entrance will continue to be executed. If the prepared collection distance is less than the preset collection distance, then the UAV is determined to fly blindly to the target event point.

4. The method for unmanned aerial vehicle (UAV) inspection of high-speed railway tunnel entrances according to claim 3, characterized in that, The step of adjusting the drone's attitude based on the drone's real-time coordinates and the target tunnel entrance coordinates to obtain the target shooting posture includes: Transform the real-time coordinates and the target tunnel entrance coordinates to the WGS84 geocentric coordinate system; In the WGS84 geocentric coordinate system, the relative position of the real-time coordinates and the target tunnel entrance coordinates is calculated to obtain the target relative position; The attitude of the drone and the angle of the smart pod are adjusted according to the relative position of the target to obtain the target shooting posture.

5. The method for unmanned aerial vehicle (UAV) inspection of high-speed railway tunnel entrances according to claim 4, characterized in that, The step of controlling the intelligent pod to perform visual intelligent recognition of the target tunnel entrance based on the target shooting posture, and determining the center coordinates of the target tunnel entrance, includes: The drone is controlled to maintain the target shooting posture, and the image acquisition module of the intelligent pod is controlled to continuously acquire multiple on-site image frames; A preset tunnel entrance visual recognition model is invoked to identify multiple on-site image frames to obtain multiple tunnel entrance candidate boxes, each of which includes a confidence level. Based on the confidence level of each of the candidate tunnel entrances, a target tunnel entrance is determined from the candidate tunnel entrances. The geometric center pixel coordinates of the target box at the tunnel entrance in the corresponding on-site image frame are calculated to obtain the center coordinates.

6. The method for unmanned aerial vehicle (UAV) inspection of high-speed railway tunnel entrances according to claim 5, characterized in that, The step of calculating the target distance based on the real-time coordinates and the center coordinates, and adjusting the zoom parameters of the video acquisition module in the smart pod according to the target distance to focus on the target tunnel entrance includes: The target distance is obtained by calculating the distance between the real-time coordinates and the center coordinates; The target zoom parameters are calculated based on the target distance and the preset zoom algorithm. The video acquisition module is controlled to perform zoom adjustment based on the target zoom parameters in order to focus on the target tunnel entrance.

7. The method for unmanned aerial vehicle (UAV) inspection of high-speed railway tunnel entrances according to claim 6, characterized in that, The step of controlling the video acquisition module to acquire inspection video of the target tunnel entrance based on a preset acquisition duration, in order to complete the inspection of the target tunnel entrance, includes: Based on the preset acquisition duration, the video acquisition module is controlled to continuously record the inspection video at the target tunnel entrance; The recording of the inspection video ends when the preset collection time expires; The inspection video is bound to the target event point and stored to complete the inspection of the target tunnel entrance.

8. A drone inspection system for high-speed railway tunnel entrances, characterized in that, The unmanned aerial vehicle (UAV) inspection system at the entrance of a high-speed railway tunnel includes a ground station and a UAV, wherein the UAV includes a controller and a smart pod; the UAV inspection system at the entrance of a high-speed railway tunnel is capable of performing the UAV inspection method at the entrance of a high-speed railway tunnel as described in any one of claims 1 to 7.

9. A drone inspection device for high-speed railway tunnel entrances, characterized in that, include: At least one processor; as well as, A memory communicatively connected to the at least one processor; wherein, The memory stores instructions that are executed by the at least one processor to enable the at least one processor to perform the unmanned aerial vehicle (UAV) inspection method for high-speed railway tunnel entrances as described in any one of claims 1 to 7.

Images