A wind power blade narrow cavity negative pressure adsorption centering detection system and a detection method

Abstract

The application discloses a kind of wind power blade narrow cavity negative pressure adsorption centering detection system and detection method, its system includes the robot body for being arranged in blade inner cavity, and terminal controller is connected with robot body bidirectionally by optical fiber communication module communication;Robot body includes bottom plate and the shell of being arranged in the outside of bottom plate, the bottom of bottom plate is equipped with the negative pressure skirt for adhering blade cavity inner wall, and two negative pressure adsorption zones are formed between the bottom surface of bottom plate and negative pressure skirt.This application relies on double-fan negative pressure adsorption, four-wheel differential walking and laser radar positioning, can be matched with optical fiber communication, terminal controller realizes the remote detection of wind power blade inner cavity, effectively replaces artificial and completes the defect detection task in fan blade cavity, reduces operation safety risk;Meanwhile, the center line can be obtained by point cloud, so that the robot can march in the center, optimize imaging quality, avoid inner wall scratch, improve narrow cavity traffic safety, efficient and safe operation, meet the detection needs of intelligent wind power industry.

Description

Technical Field

[0001] This invention relates to the field of adsorption robot technology, specifically to a narrow cavity negative pressure adsorption centering detection system and method for wind turbine blades. Background Technology

[0002] In the wind power industry, wind turbine blades, as the core components of wind power systems, are prone to internal defects such as whitening, wrinkling, cracking, and light transmission after long-term operation due to factors such as airflow impact and environmental erosion. These defects directly affect the safe operation of the unit, thus requiring regular internal cavity inspections. Currently, the industry generally relies on manual entry into the nacelle for regular defect inspection of large wind turbine blades. However, the interior of large wind turbine blades is typically a narrow, enclosed cavity structure. This cavity is not only narrow and dimly lit, but also contains load-bearing structures such as the inner wall of the shell, web, main beam cap, and reinforcing ribs, as well as local protruding obstacles such as sealant joints, cables, sensor components, and rivets. The conditions for manual entry are quite harsh. Therefore, the confined space inside the blade cavity not only limits manual operation and reduces inspection efficiency, but also easily leads to a large number of blind spots during manual inspections, resulting in a very high rate of missed defects. At the same time, manual entry inspection also faces the safety risks of working at height. Therefore, the traditional method of regular manual inspection cannot meet the needs of the wind power industry for intelligent and precise internal cavity inspection.

[0003] To address the technical problems associated with manual inspection, various solutions have been proposed in the industry. For example, patent publication number CN120926041A describes a wind turbine blade inspection method based on a wall-climbing robot and image recognition. This patent utilizes the forward-facing and sweeping cameras of the wall-climbing robot to capture images. Based on camera pose and image stitching algorithms, it achieves panoramic imaging of the blade cavity. Then, image recognition algorithms diagnose fault information within the blade cavity. The fault location is determined by recording the wall-climbing robot's movement. The sweeping camera, equipped with a distance sensor, can roughly estimate the camera's pixel scale, which is used to estimate the size of the fault. While this patent can replace manual inspection inside the blade, it still has the following drawbacks: I. Since the adsorption walking mechanism in the above-mentioned patent is only a conceptual description and has not been engineered to fit the curved surface of the blade cavity and protruding obstacles, and lacks a complete negative pressure seal and obstacle-crossing structure design, it is prone to problems such as unstable adsorption and jamming when facing weld seams and protrusions of the main beam cap.

[0004] Second, when the robot moves through a narrow cavity with varying width, it is easy to deviate to one side, which can cause the shooting distance on one side to be too close and the other side to be too far. This can lead to problems such as image distortion, frequent changes in focus, increased number of repeated shots, and uneven image coverage when the gimbal is shooting, thus reducing the efficiency of defect recognition.

[0005] For example, a device for inspecting the inner cavity of wind turbine blades using an adsorption-type suction cup, published in patent CN121782110A, relies on a vacuum suction cup to achieve adsorption on the inner wall. It is equipped with a multi-circumferentially arranged camera unit, SLAM global navigation, and a charging base station at the blade root, enabling full-coverage image acquisition of the inner cavity and autonomous return-to-base charging, thus improving the level of inspection automation to some extent. However, this patent not only suffers from problems similar to those of the aforementioned patent CN120926041A, but also has unique technical defects: The navigation system in this patent requires real-time acquisition and processing of all 3D point cloud data within the blade's internal cavity. A complete global 3D map must be constructed before path planning and position correction can be completed, resulting in significant computational demands on the embedded hardware. In the context of narrow blade cavities, path calculation delays and lag in motion command response are highly likely. Furthermore, the entire point cloud data is susceptible to interference from noise points, abnormal surface reflections, localized protrusions within the cavity, cable clutter, or obstructions from the robot's own structure, continuously reducing positioning accuracy and causing significant fluctuations in the robot's trajectory. This increases the probability of the robot scraping against internal components such as the web, seams, and reinforcing ribs, significantly raising the risk of equipment and cavity damage. In addition, this patented solution relies solely on a global map for macroscopic path planning, lacking real-time lateral distance monitoring and internal cavity passage safety constraints. In wind turbine blade internal cavity inspection scenarios with strict requirements for imaging uniformity and safe clearances on both sides of the cavity, these deficiencies further amplify safety hazards. Summary of the Invention

[0006] This invention provides a narrow cavity negative pressure adsorption centering detection system and method for wind turbine blades to solve the problems mentioned in the background art.

[0007] To solve the above-mentioned technical problems, the technical solution adopted by the present invention is as follows: A wind turbine blade narrow cavity negative pressure adsorption centering detection system includes a robot body for installation in the inner cavity of the blade, and a terminal controller that is bidirectionally connected to the robot body via an optical fiber communication module. The robot body includes a base plate and an outer shell mounted on the outside of the base plate. The bottom of the base plate is connected to a negative pressure skirt for conforming to the inner wall of the blade cavity. Two negative pressure adsorption zones are formed between the bottom surface of the base plate and the negative pressure skirt. Two centrifugal adsorption fans are fixedly installed on the top of the base plate, with the air inlets of each fan correspondingly positioned within the negative pressure adsorption zones, relying on fan suction to generate adsorption force. Two sets of walking transmission components for moving the robot body along the inner wall of the blade cavity are symmetrically installed on the base plate. A front-facing vision unit for collecting environmental data at the front of the robot is fixedly installed at the center of the front end of the base plate. A fisheye camera and a rear-facing vision unit are mounted at the center of the rear end of the base plate. The fisheye camera monitors the rear environment of the robot, and the rear-facing vision unit has a 90° rotation range to the left and right of the central axis of the robot body for acquiring defect image data of the inner wall of the blade cavity from multiple angles. A lidar is provided on the top of the outer shell, coinciding with the central axis of the robot body, for collecting point cloud data around the robot body. A supplementary lighting unit is also provided on the surface of the outer shell. The terminal controller has a built-in main control chip, which integrates an image processing module for processing image data, a narrow cavity geometry data extraction module for extracting the boundary of point cloud data, and a centering control calculation module for calculating the centering control quantity based on the extracted boundary. The optical fiber communication module is located on the robot body, and the robot body transmits the point cloud data and image data it has collected to the terminal controller through the optical fiber communication module, and receives control commands issued by the terminal controller.

[0008] Preferably, each of the walking transmission components includes a main drive wheel, a driven wheel, and a DC servo motor mounted on the base plate. The main drive wheel and the driven wheel are respectively connected to a synchronous pulley on one side. The two synchronous pulleys are connected by a synchronous belt drive, and the output end of the DC servo motor is connected to the shaft of the main drive wheel.

[0009] Preferably, each of the walking transmission components further includes an idler wheel mounted on the base plate, the teeth of which are used to mesh with the inner teeth of the timing belt.

[0010] Preferably, the negative pressure skirt is made of chemically cross-linked polyethylene foam material, and the surface of the negative pressure skirt is provided with opening areas corresponding to each main drive wheel and driven wheel. The wheel surfaces of the main drive wheel and driven wheel pass through the corresponding opening areas and are in contact with the inner wall of the blade cavity. Both the main drive wheel and driven wheel are hollow rubber wheel structures.

[0011] Preferably, the supplementary lighting unit includes a point light source, a first surface light source, a second surface light source, and a third surface light source disposed on the surface of the housing. Two point light sources and two second surface light sources are provided, and the two point light sources are located at the front end of the housing and on both sides of the front vision unit lens. The two second surface light sources are located on the left and right rear sides of the housing, and the first surface light source is located between the rear vision unit and the lidar. The third surface light source is located at the rear end of the housing and on one side of the rear vision unit lens.

[0012] Preferably, guardrails are provided on the left and right sides and the rear end of the outer shell.

[0013] Preferably, it also includes a battery module located on the base plate, the battery module being used to supply power to the various power-consuming units of the robot body.

[0014] The present invention also provides a centering detection method using the above-mentioned narrow cavity negative pressure adsorption centering detection system for wind turbine blades, comprising the following steps: S1. Place the robot body in the center of the narrow cavity of the wind turbine blade, and use a dual centrifugal adsorption fan to draw air so that the robot body is adsorbed onto the inner wall of the blade cavity. S2. The robot body moves forward along the cavity by controlling the walking transmission component. During the movement, the laser radar collects point cloud data of the inner cavity of the blade in real time, and transmits the collected point cloud data to the terminal controller through the fiber optic communication module. S3. The narrow cavity geometry data extraction module in the terminal controller divides the point cloud data into the left candidate region, the right candidate region, the front safe region and the rear auxiliary region according to the forward direction of the robot body, and extracts the leaf cavity structure lines in the left candidate region and the right candidate region respectively. S4, the centering control calculation module calculates the lateral centering error ey, heading error θC and the safety margin for passage on both sides of the robot body based on the extracted left and right leaf cavity structure lines, and generates motion control commands to send to the robot body to adjust the body in real time to keep it centered along the inner cavity center line. S5. After the robot body moves to the preset image acquisition point, the terminal controller can determine in real time whether the lateral error of the robot body, the heading error, the passage safety margin, the negative pressure adsorption pressure, the body posture status and the working status of the image acquisition unit meet the photo trigger threshold, and issue instructions according to the status of each parameter to perform image acquisition, centering correction or safety protection actions accordingly. S6. Receive the image data collected in S5 through the terminal controller, and perform defect identification, classification and marking on the image data through the image processing module. Simultaneously store the image, location and defect detection data for output of the detection report. S7. After the detection task is completed, control the robot body to move backward and rely on the fisheye camera to collect images of the rear environment in real time to realize obstacle monitoring and collision protection during the backward process, and finally allow the robot body to withdraw from the inner cavity of the blade.

[0015] Preferably, in step S5, the specific control logic of the terminal controller is as follows: If all parameters meet the shooting threshold, a shooting command is issued to drive the front vision unit to acquire a local image of the inner wall of the blade cavity at the front of the body to obtain local image data; and to drive the rear vision unit to rotate and acquire a complete image of the entire inner wall of the blade cavity to obtain complete image data. If any parameter slightly exceeds the limit or does not meet the image capture threshold, the centering adjustment command will be continuously issued until the image capture threshold is met before the image data acquisition of the inner wall of the leaf cavity is performed. If the difference between the left and right sides is too large, the distance between the structures on one side is too close, the boundary confidence is insufficient, the local safety margin is insufficient, or the negative pressure adsorption state is abnormal, the safety strategy will be triggered.

[0016] Preferably, the safety strategy is to control the robot body to decelerate, pause taking pictures, stop moving, retreat to a safe position, or send a manual confirmation request to the terminal controller.

[0017] By adopting the above technical solution, the beneficial effects achieved by the present invention are as follows: This invention utilizes a dual-fan symmetrical negative pressure adsorption structure as its core, ensuring stable attachment of the robot body to the curved wall of the blade cavity. Simultaneously, relying on a four-wheel differential speed walking transmission system, it enables flexible turning and obstacle crossing within confined spaces. Furthermore, with the aid of multi-vision acquisition units, multiple supplementary light sources, and a lidar positioning module mounted on the body, this invention can achieve high-definition video acquisition, defect identification, and precise pose perception within the blade cavity. The low-latency fiber optic communication module and long-distance continuous power supply module effectively solve the challenges of real-time data transmission and uninterrupted operation in complex environments. Employing a matching long-range handheld terminal control system, this invention enables remote control of the robot body, real-time monitoring of operational data, and emergency fault handling, effectively replacing manual labor in defect detection within wind turbine blade cavities. This improves detection efficiency and accuracy, reduces operational safety risks, meets the intelligent and refined detection needs of the wind power industry, and provides reliable technical support for the safe and stable operation of wind power generation equipment.

[0018] This invention enables the negative pressure adsorption inspection robot to maintain relative centering within the narrow space of the blade cavity by calculating the left and right side point cloud lifting lines and local center lines, reducing the equipment's reliance on complex calculations of complete 3D maps or full point clouds. Simultaneously, when the robot rotates to take pictures at a relatively central position, the left and right shooting distances are more balanced, resulting in a more stable image perspective. This helps reduce issues such as near-sidewall image distortion, insufficient detail on the far-sidewall, unbalanced image overlap, and repeated shooting, effectively improving the quality of inspection images. Furthermore, by determining the centering error and safety margin before taking pictures, the problem of frequent focusing, zooming, and invalid shooting can be effectively reduced, thereby shortening the preparation time for shooting at a single inspection point and improving image acquisition efficiency. Moreover, when the robot travels along the blade cavity center line, the safety margin for passage on both sides is more balanced, reducing the likelihood of scraping against the blade cavity inner wall, web, rubber seams, reinforcing ribs, cables, and other internal accessories, thus improving the safety of passage through narrow cavities. Attached Figure Description

[0019] Figure 1 This is a schematic diagram of the overall structure of the robot body of the present invention.

[0020] Figure 2 This is a bottom-view structural diagram of the robot body of the present invention.

[0021] Figure 3 This is a schematic diagram of the internal structure of the robot body of the present invention.

[0022] Figure 4 This is a schematic diagram of the terminal controller structure of the present invention.

[0023] Figure 5 This is a schematic diagram of the main control chip of the present invention.

[0024] Figure 6 This is a schematic diagram showing the robot body of the present invention positioned centrally within the blade cavity for passage.

[0025] Figure 7 This is a schematic diagram of the point cloud ROI extraction for the left and right candidate regions of the present invention.

[0026] Figure 8 This is a diagram comparing the positions of off-center and centered photos.

[0027] Figure 9 This is a schematic diagram of the centering control and photo triggering process of the present invention.

[0028] In the diagram: 100. Robot body; 1. Base plate; 2. Outer shell; 3. Negative pressure skirt; 31. Vent area; 4. Centrifugal adsorption fan; 41. Air inlet; 5. Walking transmission assembly; 51. Main drive wheel; 52. Slave drive wheel; 53. DC servo motor; 54. Synchronous pulley; 55. Synchronous belt; 56. Idler pulley; 6. Front vision unit; 7. Fisheye camera; 8. Rear vision unit; 9. LiDAR; 10. Point light source; 11. First light source; 12. Second light source; 13. Third light source; 14. Guardrail; 15. Battery module; 200. Terminal controller; 21. Main control chip; 22. Image processing module; 23. Narrow cavity geometric data extraction module; 24. Centering control calculation module. Detailed Implementation

[0029] To better understand the above-mentioned objectives, features, and advantages of the present invention, the present invention will be further described below in conjunction with the accompanying drawings and embodiments. It should be noted that, unless otherwise specified, the embodiments and features described in these embodiments can be combined with each other.

[0030] Numerous specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways than those described herein, and therefore the invention is not limited to the specific embodiments disclosed in the following specification. Example 1

[0031] like Figures 1-5 As shown, the present invention provides a narrow cavity negative pressure adsorption centering detection system for wind turbine blades. The detection system includes a robot body 100 for being installed in the inner cavity of the blade, and a terminal controller 200 that is bidirectionally connected to the robot body 100 via an optical fiber communication module.

[0032] The robot body 100 includes a base plate 1 and an outer shell 2 installed on the outside of the base plate 1. The bottom of the base plate 1 is connected to a negative pressure skirt 3 for adhering to the inner wall of the blade cavity. Two negative pressure adsorption zones are formed between the bottom surface of the base plate 1 and the negative pressure skirt 3. Two centrifugal adsorption fans 4 are fixedly installed on the top of the base plate 1. The two centrifugal adsorption fans 4 adopt a symmetrical layout design. The air intake 41 of each centrifugal adsorption fan 4 is respectively set in the negative pressure adsorption zone, and the adsorption force is formed by the fan drawing air. Two sets of walking transmission components 5 are symmetrically installed on the base plate 1 for driving the robot body 100 to move on the inner wall of the blade cavity.

[0033] A front-facing vision unit 6 is fixedly mounted at the center of the front end of the base plate 1. This vision unit is responsible for directional navigation and real-time image acquisition during the robot's movement. It can identify surface defects within a small area of ​​the front-facing field of view, providing core visual support for the robot's autonomous movement and front-facing environmental perception. Furthermore, the lens acquisition end of the front-facing vision unit 6 is fully enclosed with a transparent curved protective cover, which can effectively resist dust and moisture erosion from outdoors and inside the fan cavity, ensuring long-term stable operation of the equipment.

[0034] A fisheye camera 7 and a rear vision unit 8 are mounted in the center of the rear end of the base plate 1. The fisheye camera 7 is used to monitor the rear environment of the robot. Through real-time image acquisition and environmental perception algorithms, it realizes intelligent monitoring of the rear environment during the robot's backward movement, effectively reducing safety hazards such as collisions and jamming during the backward movement, and improving the overall movement safety and operational stability of the robot.

[0035] The rear-mounted vision unit 8 can be equipped with a rotary motor, allowing it to rotate 90° to the left and right around the central axis of the machine body. This enables it to acquire image data of defects on the inner wall of the blade cavity from multiple angles. Specifically, the rear-mounted vision unit 8 is deployed at the top rear of the robot and has a 180° wide-range rotation detection capability. It is the core defect detection vision module of the system. This unit is specifically designed for the detection scenario of the enclosed and low-light interior of the wind turbine blade cavity. Its core function is to complete the full-scene image acquisition, accurate defect identification, and level determination inside the blade cavity, providing multi-dimensional and high-precision visual data support for the detection of defects inside the wind turbine cavity.

[0036] The top of the outer shell 2 is equipped with a LiDAR 9 that coincides with the central axis of the robot body and is used to collect point cloud data around the robot body 100. The LiDAR 9 is used to perform the robot's lateral, longitudinal, and depth three-dimensional positioning tasks inside the blade, enabling precise robot pose perception, obstacle avoidance, and path planning, ensuring operational stability and accuracy. A supplementary lighting unit is provided on the surface of the outer shell 2.

[0037] Combination Figure 5 As shown, the terminal controller 200 has a built-in main control chip 21. The main control chip 21 integrates an image processing module 22 for processing image data, a narrow cavity geometry data extraction module 23 for extracting the boundaries of point cloud data, and a centering control calculation module 24 for calculating centering control values ​​based on the extracted boundaries. Specifically, the terminal controller 200 is a handheld control terminal; as shown... Figure 4 As shown, it integrates two levers, two joysticks, eight function buttons, a power button, an emergency stop button, and a ten-inch LCD screen. It is specially developed for the leaf cavity negative pressure adsorption video inspection robot, which can realize the integration of full-process operation, monitoring and safety management. It can support on-site personnel to accurately control the equipment's actions and acquire operation data in real time.

[0038] Furthermore, the LCD screen interface layout can adopt a "zoned control and centralized display" mode, which is divided into four functional areas: main control area, image display area, historical data query area, and abnormal message display area; comprehensively covering the core needs of remote control system, image processing system, and human-machine interface, realizing the integration of robot full-process control, monitoring, and data management.

[0039] An optical fiber communication module is installed on the robot body 100. The robot body 100 transmits its collected point cloud data and image data to the terminal controller 200 via the optical fiber communication module and receives control commands issued by the terminal controller 200. This invention's communication system uses optical fiber as the core transmission medium, enabling the construction of a highly reliable, low-latency industrial-grade bidirectional communication link to meet the robot's comprehensive requirements for real-time performance, stability, bandwidth, and anti-interference capabilities. This invention integrates a miniaturized optical fiber transceiver interface on the robot body 100, facilitating photoelectric conversion of control commands and high-definition visual data to achieve high-speed transmission with the ground control terminal (i.e., the terminal controller 200).

[0040] Furthermore, the present invention can also be used with a cable winding and unwinding device to follow the robot's movement trajectory and rhythm in real time, thereby achieving automatic winding and unwinding control of the cable harness.

[0041] Combination Figure 3 As shown, each walking transmission assembly 5 includes a main drive wheel 51, a driven wheel 52 and a DC servo motor 53 mounted on the base plate 1. The main drive wheel 51 and the driven wheel 52 are respectively connected to a synchronous wheel 54 on one side. The two synchronous wheels 54 are connected by a synchronous belt 55. The output end of the DC servo motor 53 is connected to the shaft of the main drive wheel 51.

[0042] By arranging two sets of walking transmission components 5, the walking system of the robot body 100 can form a four-wheel differential drive architecture, thereby achieving differentiated speed control of the two drive wheels and ensuring steering flexibility. Specifically, based on the differential drive principle, multi-scenario steering is achieved by adjusting the speed difference between the two motors. That is, when one motor is stopped and the other is running normally, the minimum radius steering is achieved; when the two motors run in opposite directions at the same speed, 360° turning on the spot is achieved, adapting to the steering requirements of the narrow space of the blade cavity and ensuring full coverage of the detection path.

[0043] As a further step, each set of walking transmission components 5 also includes an idler wheel 56 mounted on the base plate 1. The teeth of the idler wheel 56 are used to mesh with the inner teeth of the synchronous belt 55, and the idler wheel 56 is used to tension and limit the synchronous belt 55.

[0044] Combination Figure 2 and Figure 3As shown, the negative pressure skirt 3 is made of chemically cross-linked polyethylene foam (XPE). The surface of the negative pressure skirt 3 is provided with an opening area 31 corresponding to each main drive wheel 51 and driven wheel 52. The wheel surfaces of the main drive wheel 51 and driven wheel 52 pass through the corresponding opening area 31 and are in contact with the inner wall of the blade cavity. Both the main drive wheel 51 and driven wheel 52 are hollow rubber wheel structures.

[0045] Specifically, the negative pressure skirt 3 can be made of XPE, which, based on its excellent elasticity, wear resistance, and environmental resistance, can tightly conform to the curved surface of the inner wall of the blade cavity under complex working conditions, effectively reducing air leakage and ensuring reliable sealing. The overall size of the skirt can be 10mm, and the compression stroke can be designed to be 1-2mm. This design ensures a tight fit between the skirt and the wall surface while avoiding material fatigue and wear caused by excessive compression.

[0046] Furthermore, in this invention, both the main drive wheel 51 and the driven wheel 52 adopt a hollow rubber structure design, which allows the wheel body to effectively buffer the contact impact between the wheel body and the obstacle by relying on the elastic deformation characteristics brought by the hollow structure, so as to achieve smooth crossing of protruding obstacles; and the maximum obstacle crossing height of the whole machine can reach 5mm, which can smoothly pass through common protruding obstacles such as weld seams and rivets on the inner wall of the blade cavity, ensuring the continuity of the robot's movement.

[0047] Combination Figure 1 and Figure 2 As shown, the supplementary lighting unit includes a point light source 10, a first surface light source 11, a second surface light source 12, and a third surface light source 13 disposed on the surface of the housing 2. There are two point light sources 10 and two second surface light sources 12 respectively, and the two point light sources 10 are located at the front end of the housing 2 and on both sides of the lens of the front vision unit 6. The two second surface light sources 12 are located on the left and right rear sides of the housing 2 respectively, and the first surface light source 11 is located between the rear vision unit 8 and the lidar 9. The third surface light source 13 is located at the rear end of the housing 2 and on one side of the lens of the rear vision unit 8.

[0048] Specifically, a set of high-brightness LED point light sources 10 arranged at the front end of the outer shell 2 is used to focus on the core acquisition area directly in front of the front vision unit 6, providing concentrated supplementary lighting for robot navigation and front-end small-scale defect identification, effectively solving the problem of image detail loss in backlight and low-light environments, and ensuring the stability and accuracy of the acquisition by the front vision unit 6. In order to solve the detection environment of the closed low light inside the fan blade cavity, the third light source 13 can be arranged in an arc-shaped surface light source to provide 180° full-coverage surface supplementary lighting for the core acquisition area of ​​the rear vision unit 8, eliminating the light and shadow blind spots in the detection inside the cavity; at the same time, the first light source 11 located at the top of the rear end of the robot and the second light source 12 located on the left and right rear sides of the robot can form a supplementary lighting linkage with the arc-shaped surrounding light source, further expanding the supplementary lighting coverage, improving the uniformity and detail performance of the large-scale image acquisition inside the cavity, realizing blind-angle supplementary lighting in low light, backlight, and complex cavity environments, and ensuring the clarity and detail integrity of image acquisition.

[0049] Combination Figure 1 As shown, guardrails 14 are provided on the left and right sides and the rear end of the outer shell 2. The guardrails 14 are bent metal integral molding structures arranged along the outer contour of the body, which can form physical anti-collision protection for various precision components mounted on the robot body 100, and avoid direct collision with the inner wall protrusion structure when the robot moves or moves backward in the narrow cavity of the blade, effectively reducing the risk of equipment collision damage.

[0050] Combination Figure 3 As shown, the centering detection system also includes a battery module 15 located on the base plate 1. The battery module 15 is used to supply power to each power-consuming unit of the robot body 100. It should be noted that the power supply part of the present invention can also adopt a dual-mode architecture of "main cable power supply + emergency battery (i.e., battery module 15)" to ensure the continuous energy supply of the robot in complex working environments. The battery module 15 mounted on the robot body 100 is used as a redundancy backup. When an unplanned power outage occurs upstream, the system can automatically switch to emergency battery power supply to achieve uninterrupted power transition.

[0051] In summary, the negative pressure adsorption centering detection system proposed in this invention uses a dual-fan symmetrical negative pressure adsorption structure as its core, ensuring that the robot body 100 is stably attached to the curved wall of the blade cavity. Simultaneously, relying on a four-wheel differential speed walking transmission system, the robot body 100 can flexibly turn and overcome protruding obstacles within the confined cavity space. Furthermore, with the help of multi-vision acquisition units, multiple supplementary light sources, and a lidar positioning module mounted on the body, this invention can complete high-definition video acquisition, defect identification, and precise pose perception inside the blade cavity. Through a low-latency fiber optic communication module and a long-distance continuous power supply module, the challenges of real-time data transmission and uninterrupted operation in complex environments can be effectively solved. Using a matching long-distance handheld terminal control system, this invention can achieve remote control of the robot body 100, real-time monitoring of operating data, and emergency fault handling, effectively replacing manual labor in defect detection within the wind turbine blade cavity, improving detection efficiency and accuracy, reducing operational safety risks, and meeting the intelligent and refined detection needs of the wind power industry.

[0052] The development and application of this robot can effectively replace manual labor in the task of detecting defects inside the blade cavity of wind turbines, improve detection efficiency and accuracy, reduce operational safety risks, and provide reliable technical support for the safe and stable operation of wind power generation equipment. Example 2

[0053] Combination Figures 1-9 As shown, the present invention also provides a method for detecting the centering of narrow cavity negative pressure adsorption in wind turbine blades, comprising the following steps: S1. Place the robot body 100 in the center of the narrow cavity inner wall of the wind turbine blade, and use the dual centrifugal adsorption fan 4 to draw air so that the robot body 100 is adsorbed onto the inner wall of the blade cavity. S2. The robot body 100 is controlled to move forward along the cavity by the walking transmission component 5. During the movement, the laser radar 9 collects the point cloud data of the inner cavity of the blade in real time, and transmits the real-time collected point cloud data to the terminal controller 200 through the fiber optic communication module. S3, the narrow cavity geometry data extraction module 23 in the terminal controller 200 divides the point cloud data into the left candidate region, the right candidate region, the front safe region and the rear auxiliary region according to the forward direction of the robot body 100, and extracts the leaf cavity structure line in the left candidate region and the right candidate region respectively. Among them, the blade cavity structure line can be the boundary line of the inner wall of the blade cavity shell, the boundary line of the web, the boundary line near the glue joint, the boundary line of the reinforcing rib, or the local passage boundary line obtained by fitting multiple local structural points. S4, the centering control calculation module 24 calculates the lateral centering error ey, heading error θC and the safety margin for passage on both sides of the robot body 100 based on the extracted left and right side leaf cavity structure lines, and generates motion control commands to send to the robot body 100, adjusting the body in real time to keep it centered along the inner cavity center line. The formula for calculating the lateral centering error ey is: ey=(dR-dL) / 2, where dL is the distance from the robot body 100 to the left blade cavity structure line, and dR is the distance from the robot body 100 to the right blade cavity structure line. The heading error θC is calculated using a fusion function. The calculation formula is: θC=f(θL, θR), where θL is the tilt angle of the left blade cavity structure line and θR is the tilt angle of the right blade cavity structure line. Left-hand traffic safety margin sL=dL; Right-hand traffic safety margin sR=dR; Specifically, this step uses the left and right candidate regions as the main line extraction method instead of performing complex mapping and global planning on the entire point cloud, which can effectively reduce the amount of real-time computation and reduce the impact of a few outliers on the center line judgment. S5. After the robot body 100 moves to the preset image acquisition point, the terminal controller 200 can determine in real time whether the lateral error of the robot body, the heading error, the passage safety margin, the negative pressure adsorption pressure, the body posture status and the working status of the image acquisition unit meet the photo trigger threshold, and issue instructions according to the status of each parameter to perform image acquisition, centering correction or safety protection actions accordingly. The specific control logic of the terminal controller 200 is as follows: If all conditions are met for taking a picture, a shooting command is issued to drive the front vision unit 6 to acquire a local image of the inner wall of the blade cavity at the front of the body to obtain local image data; and to drive the rear vision unit 8 to rotate and acquire a complete image of the entire inner wall of the blade cavity to obtain complete image data. If any item slightly exceeds the limit or does not meet the photo-taking threshold, the centering adjustment command will be continuously issued until the photo-taking threshold is met before the image data acquisition of the inner wall of the leaf cavity is performed. If the difference between the left and right sides is too large, the distance between the structures on one side is too close, the boundary confidence is insufficient, the local safety margin is insufficient, or the negative pressure adsorption state is abnormal, a safety strategy will be triggered. The safety strategy is to control the robot body 100 to decelerate, pause taking pictures, stop moving, retreat to a safe position, or send a manual confirmation request to the terminal controller 200. Specifically, in this step, by linking various parameters such as centering error, heading error, safety margin, and negative pressure adsorption pressure with the image acquisition and shooting trigger conditions, the present invention can form a closed-loop shooting centering control. S6. The terminal controller 200 receives the image data collected in S5, and the image processing module 22 performs defect identification, classification and marking on the image data, and synchronously stores the image, location and defect detection data for output of the detection report. S7. After the detection task is completed, control the robot body 100 to move backward and rely on the fisheye camera 7 to collect images of the rear environment in real time to realize obstacle monitoring and anti-collision protection during the backward process, and finally allow the robot body 100 to withdraw from the inner cavity of the blade.

[0054] The centering detection method proposed in this invention is adapted to an adsorption-type mobile platform, enabling coordinated operation of negative pressure adsorption, walking motor operation, upper-level system data link, and inspection image recording. Furthermore, this invention can be applied to the leading edge cavity, trailing edge cavity, near the web, near the main beam cap, near the seal joint, and other various narrow passageways within the wind turbine blade's internal cavity. It is particularly suitable for inspection scenarios where the robot needs to maintain a centered posture to improve image quality and reduce the risk of cavity abrasion. The centering detection method provided by this invention has the following advantages when inspecting and photographing the blade's internal cavity: First, the centering setting is not only used for navigation, but also to ensure a more balanced shooting distance and angle when rotating the motor or gimbal to take pictures. This effectively reduces problems such as image distortion, frequent focal length changes, increased number of repeated shots, and uneven image coverage, thereby improving image acquisition quality and shooting efficiency.

[0055] Second, by extracting the main structure line from the point clouds of the left and right ROIs, the dependence on the full point cloud can be effectively reduced, the amount of computation can be reduced, and the impact of a few outliers, reflection anomalies and cable points on the centering judgment can also be reduced.

[0056] Third, by calculating the lateral centering error through the distance relationship between the left and right structural lines, and by calculating the center line direction and heading error through the direction relationship between the left and right structural lines, the robot body 100 can travel stably along the local center line of the blade cavity, which promotes a more balanced left and right shooting distance and ensures the image acquisition quality.

[0057] Fourth, by linking the robot's centering state with the photo-taking trigger condition, the photo-taking trigger condition is not met when the robot is biased to one side of the cavity, resulting in an uneven passage gap on both sides. This effectively reduces the problem of repeated shooting and ensures that the robot can only perform image acquisition after calibration and centering.

[0058] Fifth, by defining the central route as the safe route, the safety margin on both sides is more balanced when the robot is moving in the middle position, which effectively reduces the risk of the robot body 100, sensor bracket, camera gimbal, drag cable or lightweight maintenance module scraping against the inner wall of the blade cavity, web plate, rubber seam, reinforcing ribs and other structures.

[0059] In summary, this invention enables the negative pressure adsorption inspection robot to maintain relative centering within the narrow space of the blade cavity by calculating the left and right side point cloud lifting lines and local center lines, reducing the device's reliance on complex calculations of complete 3D maps or full point clouds. Simultaneously, when the robot rotates to take pictures at a relatively central position, the left and right shooting distances are more balanced, resulting in a more stable image perspective. This helps reduce issues such as near-sidewall image distortion, insufficient detail on the far-sidewall, unbalanced image overlap, and repeated reshoots, effectively improving the quality of inspection images. Furthermore, by determining the centering error and safety margin before taking pictures, the problems of frequent focusing, zooming, and invalid shots can be effectively reduced, thereby shortening the preparation time for shooting at a single inspection point and improving image acquisition efficiency. Moreover, when the robot travels along the center line of the blade cavity, the safety margin for passage on both sides is more balanced, reducing the likelihood of scraping against the inner wall of the blade cavity, the web, the sealant joint, the reinforcing ribs, cables, and other internal accessories, thus improving the safety of passage through narrow cavities.

[0060] It is worth noting that the wind turbine blade narrow cavity negative pressure adsorption centering detection method proposed in this invention can also be expanded to include various lightweight cavity maintenance operations such as local marking, light wiping, and light spraying.

[0061] It should be noted that when a component is referred to as being "assembled on," "mounted on," "fixed to," or "set on" another component, it can be directly on the other component or there may be an intermediate component. When a component is considered to be "connected to" another component, it can be directly connected to the other component or there may be an intermediate component present. The terms "vertical," "horizontal," "upper," "lower," "left," "right," and similar expressions used herein are for illustrative purposes only and do not represent the only possible implementation.

[0062] In the description of this specification, the terms "one embodiment," "some embodiments," "specific embodiment," etc., refer to a specific feature, structure, material, or characteristic described in connection with that embodiment or example, which is included in at least one embodiment or example of the present invention. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0063] Although embodiments of the invention have been shown and described, those skilled in the art will understand that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the claims and their equivalents.

Claims

1. A narrow-cavity negative pressure adsorption centering detection system for wind turbine blades, characterized in that, It includes a robot body installed inside the blade cavity, and a terminal controller that is bidirectionally connected to the robot body via an optical fiber communication module. The robot body includes a base plate and an outer shell mounted on the outside of the base plate. The bottom of the base plate is connected to a negative pressure skirt for conforming to the inner wall of the blade cavity. Two negative pressure adsorption zones are formed between the bottom surface of the base plate and the negative pressure skirt. Two centrifugal adsorption fans are mounted on the top of the base plate, with the air inlets of each fan corresponding to the negative pressure adsorption zones, forming adsorption force through fan suction. Two sets of walking transmission components for moving the robot body along the inner wall of the blade cavity are symmetrically mounted on the base plate. A front-facing vision unit for collecting environmental data at the front of the robot is fixed at the center of the front end of the base plate. A fisheye camera and a rear-facing vision unit are mounted at the center of the rear end of the base plate. The fisheye camera monitors the rear environment of the robot, and the rear-facing vision unit has a 90° rotation range to the left and right of the central axis of the robot body for acquiring defect image data of the inner wall of the blade cavity from multiple angles. A lidar is provided on the top of the outer shell, coinciding with the central axis of the robot body, for collecting point cloud data around the robot body. A supplementary lighting unit is also provided on the surface of the outer shell. The terminal controller has a built-in main control chip, which integrates an image processing module for processing image data, a narrow cavity geometric data extraction module for performing boundary extraction on point cloud data, and a centering control calculation module for calculating centering control quantities based on the extracted boundaries. Furthermore, the optical fiber communication module is installed on the robot body. The robot body transmits the point cloud data and image data it has collected to the terminal controller through the optical fiber communication module, and receives control commands issued by the terminal controller.

2. The wind turbine blade narrow cavity negative pressure adsorption centering detection system according to claim 1, characterized in that, Each of the walking transmission components includes a main drive wheel, a driven wheel, and a DC servo motor mounted on the base plate. A synchronous pulley is connected to one side of the main drive wheel and the driven wheel respectively. The two synchronous pulleys are connected by a synchronous belt drive, and the output end of the DC servo motor is connected to the shaft of the main drive wheel.

3. The wind turbine blade narrow cavity negative pressure adsorption centering detection system according to claim 2, characterized in that, Each of the aforementioned walking transmission components also includes an idler wheel mounted on the base plate, the teeth of which are used to mesh with the inner teeth of the timing belt.

4. The wind turbine blade narrow cavity negative pressure adsorption centering detection system according to claim 2, characterized in that, The negative pressure skirt is made of chemically cross-linked polyethylene foam material. The surface of the negative pressure skirt has openings corresponding to each main drive wheel and driven wheel. The wheel surfaces of the main drive wheel and driven wheel pass through the corresponding openings and are in contact with the inner wall of the blade cavity. Both the main drive wheel and driven wheel are hollow rubber wheel structures.

5. The wind turbine blade narrow cavity negative pressure adsorption centering detection system according to claim 1, characterized in that, The supplementary lighting unit includes a point light source, a first surface light source, a second surface light source, and a third surface light source disposed on the surface of the housing. There are two point light sources and two second surface light sources, and both point light sources are located at the front end of the housing and on both sides of the front vision unit lens. The two second surface light sources are located on the left and right rear sides of the housing, and the first surface light source is located between the rear vision unit and the lidar. The third surface light source is located at the rear end of the housing and on one side of the rear vision unit lens.

6. The wind turbine blade narrow cavity negative pressure adsorption centering detection system according to claim 1, characterized in that, Guardrails are provided on the left and right sides and the rear end of the outer shell.

7. The wind turbine blade narrow cavity negative pressure adsorption centering detection system according to claim 1, characterized in that, It also includes a battery module located on the base plate, which is used to supply power to the various power-consuming units of the robot body.

8. A centering detection method using the wind turbine blade narrow cavity negative pressure adsorption centering detection system according to any one of claims 1-7, characterized in that, Includes the following steps: S1. Place the robot body in the center of the narrow cavity of the wind turbine blade, and use a dual centrifugal adsorption fan to draw air so that the robot body is adsorbed onto the inner wall of the blade cavity. S2. The robot body moves forward along the cavity by controlling the walking transmission component. During the movement, the laser radar collects point cloud data of the inner cavity of the blade in real time, and transmits the collected point cloud data to the terminal controller through the fiber optic communication module. S3. The narrow cavity geometry data extraction module in the terminal controller divides the point cloud data into the left candidate region, the right candidate region, the front safe region and the rear auxiliary region according to the forward direction of the robot body, and extracts the leaf cavity structure lines in the left candidate region and the right candidate region respectively. S4, the centering control calculation module calculates the lateral centering error ey, heading error θC and the safety margin for passage on both sides of the robot body based on the extracted left and right leaf cavity structure lines, and generates motion control commands to send to the robot body to adjust the body in real time to keep it centered along the inner cavity center line. S5. After the robot body moves to the preset image acquisition point, the terminal controller can determine in real time whether the lateral error of the robot body, the heading error, the passage safety margin, the negative pressure adsorption pressure, the body posture status and the working status of the image acquisition unit meet the photo trigger threshold, and issue instructions according to the status of each parameter to perform image acquisition, centering correction or safety protection actions accordingly. S6. Receive the image data collected in S5 through the terminal controller, and perform defect identification, classification and marking on the image data through the image processing module. Simultaneously store the image, location and defect detection data for output of the detection report. S7. After the detection task is completed, control the robot body to move backward and rely on the fisheye camera to collect images of the rear environment in real time to realize obstacle monitoring and collision protection during the backward process, and finally allow the robot body to withdraw from the inner cavity of the blade.

9. The detection method according to claim 8, characterized in that, In step S5, the specific control logic of the terminal controller is as follows: If all parameters meet the shooting threshold, a shooting command is issued to drive the front vision unit to acquire a local image of the inner wall of the blade cavity at the front of the body to obtain local image data; and to drive the rear vision unit to rotate and acquire a complete image of the entire inner wall of the blade cavity to obtain complete image data. If any parameter slightly exceeds the limit or does not meet the image capture threshold, the centering adjustment command will be continuously issued until the image capture threshold is met before the image data acquisition of the inner wall of the leaf cavity is performed. If the difference between the left and right sides is too large, the distance between the structures on one side is too close, the boundary confidence is insufficient, the local safety margin is insufficient, or the negative pressure adsorption state is abnormal, the safety strategy will be triggered.

10. The detection method according to claim 9, characterized in that, The safety strategy involves controlling the robot to slow down, pause taking pictures, stop moving, retreat to a safe position, or send a manual confirmation request to the terminal controller.

Images